From 781e07249c6e5698ccd6887b90ba3fa1f60edb2b Mon Sep 17 00:00:00 2001 From: "Jiang, Beibei" Date: Tue, 6 Sep 2022 10:45:57 +0000 Subject: [PATCH] ModelArts 2021430 (GA) UMN 25072022 provided by R&D (third review) Reviewed-by: gtema Co-authored-by: Jiang, Beibei Co-committed-by: Jiang, Beibei --- docs/modelarts/umn/ALL_META.TXT.json | 538 ++++---- docs/modelarts/umn/CLASS.TXT.json | 842 +++++++------ .../umn/en-us_image_0000001156920825.png | Bin 0 -> 28863 bytes .../umn/en-us_image_0000001157080841.png | Bin 0 -> 8163 bytes .../umn/en-us_image_0000001157080843.png | Bin 0 -> 10762 bytes .../umn/en-us_image_0000001157080915.png | Bin 0 -> 16516 bytes .../umn/en-us_image_0000001278234781.png | Bin 0 -> 49064 bytes .../umn/en-us_image_0000001281686748.png | Bin 0 -> 25214 bytes .../umn/en-us_image_0000001290603082.png | Bin 0 -> 38137 bytes .../umn/en-us_image_0000001340184197.png | Bin 0 -> 144308 bytes .../umn/en-us_image_0000001340265309.png | Bin 0 -> 154806 bytes docs/modelarts/umn/modelarts_01_0001.html | 1 + docs/modelarts/umn/modelarts_01_0003.html | 1 - docs/modelarts/umn/modelarts_01_0006.html | 8 +- docs/modelarts/umn/modelarts_01_0012.html | 3 +- docs/modelarts/umn/modelarts_01_0013.html | 4 +- docs/modelarts/umn/modelarts_01_0015.html | 5 +- docs/modelarts/umn/modelarts_05_0003.html | 10 +- docs/modelarts/umn/modelarts_05_0012.html | 3 +- docs/modelarts/umn/modelarts_05_0013.html | 2 +- docs/modelarts/umn/modelarts_05_0014.html | 4 +- docs/modelarts/umn/modelarts_05_0022.html | 30 + docs/modelarts/umn/modelarts_05_0024.html | 56 + docs/modelarts/umn/modelarts_05_0032.html | 2 +- docs/modelarts/umn/modelarts_05_0063.html | 8 +- docs/modelarts/umn/modelarts_05_0067.html | 4 +- docs/modelarts/umn/modelarts_05_0081.html | 2 +- docs/modelarts/umn/modelarts_05_0090.html | 18 +- docs/modelarts/umn/modelarts_05_0128.html | 2 +- docs/modelarts/umn/modelarts_08_0002.html | 1 + docs/modelarts/umn/modelarts_08_0003.html | 4 +- docs/modelarts/umn/modelarts_08_0005.html | 2 +- docs/modelarts/umn/modelarts_08_0007.html | 6 +- docs/modelarts/umn/modelarts_21_0003.html | 4 +- docs/modelarts/umn/modelarts_21_0004.html | 2 +- docs/modelarts/umn/modelarts_21_0006.html | 4 +- docs/modelarts/umn/modelarts_21_0009.html | 4 +- docs/modelarts/umn/modelarts_21_0010.html | 2 +- docs/modelarts/umn/modelarts_21_0012.html | 4 +- docs/modelarts/umn/modelarts_21_0016.html | 4 +- docs/modelarts/umn/modelarts_21_0019.html | 55 +- docs/modelarts/umn/modelarts_21_0038.html | 538 +++----- docs/modelarts/umn/modelarts_21_0057.html | 2 +- docs/modelarts/umn/modelarts_21_0072.html | 2 +- docs/modelarts/umn/modelarts_21_0077.html | 2 +- docs/modelarts/umn/modelarts_21_0079.html | 2 +- docs/modelarts/umn/modelarts_21_0080.html | 2 +- docs/modelarts/umn/modelarts_21_0083.html | 2 +- docs/modelarts/umn/modelarts_21_0084.html | 2 +- docs/modelarts/umn/modelarts_21_0085.html | 2 +- docs/modelarts/umn/modelarts_23_0002.html | 121 +- docs/modelarts/umn/modelarts_23_0004.html | 132 +- docs/modelarts/umn/modelarts_23_0006.html | 17 - docs/modelarts/umn/modelarts_23_0008.html | 230 +--- docs/modelarts/umn/modelarts_23_0009.html | 483 +++----- docs/modelarts/umn/modelarts_23_0010.html | 4 - docs/modelarts/umn/modelarts_23_0011.html | 2 +- docs/modelarts/umn/modelarts_23_0012.html | 4 +- docs/modelarts/umn/modelarts_23_0013.html | 2 +- docs/modelarts/umn/modelarts_23_0014.html | 2 +- docs/modelarts/umn/modelarts_23_0018.html | 4 +- docs/modelarts/umn/modelarts_23_0033.html | 2 +- docs/modelarts/umn/modelarts_23_0034.html | 2 +- docs/modelarts/umn/modelarts_23_0039.html | 93 +- docs/modelarts/umn/modelarts_23_0044.html | 23 +- docs/modelarts/umn/modelarts_23_0051.html | 2 - docs/modelarts/umn/modelarts_23_0052.html | 18 +- docs/modelarts/umn/modelarts_23_0060.html | 15 +- docs/modelarts/umn/modelarts_23_0061.html | 12 +- docs/modelarts/umn/modelarts_23_0063.html | 8 +- docs/modelarts/umn/modelarts_23_0066.html | 102 +- docs/modelarts/umn/modelarts_23_0076.html | 5 - docs/modelarts/umn/modelarts_23_0079.html | 4 +- docs/modelarts/umn/modelarts_23_0080.html | 6 +- docs/modelarts/umn/modelarts_23_0085.html | 2 +- docs/modelarts/umn/modelarts_23_0087.html | 2 +- docs/modelarts/umn/modelarts_23_0091.html | 62 +- docs/modelarts/umn/modelarts_23_0092.html | 994 +++++---------- docs/modelarts/umn/modelarts_23_0093.html | 361 ++---- docs/modelarts/umn/modelarts_23_0100.html | 2 +- docs/modelarts/umn/modelarts_23_0102.html | 4 +- docs/modelarts/umn/modelarts_23_0157.html | 89 +- docs/modelarts/umn/modelarts_23_0161.html | 12 +- docs/modelarts/umn/modelarts_23_0162.html | 12 +- docs/modelarts/umn/modelarts_23_0163.html | 10 +- docs/modelarts/umn/modelarts_23_0164.html | 10 +- docs/modelarts/umn/modelarts_23_0165.html | 10 +- docs/modelarts/umn/modelarts_23_0166.html | 10 +- docs/modelarts/umn/modelarts_23_0167.html | 10 +- docs/modelarts/umn/modelarts_23_0168.html | 10 +- docs/modelarts/umn/modelarts_23_0169.html | 10 +- docs/modelarts/umn/modelarts_23_0170.html | 10 +- docs/modelarts/umn/modelarts_23_0173.html | 958 +++++---------- docs/modelarts/umn/modelarts_23_0175.html | 483 +++----- docs/modelarts/umn/modelarts_23_0176.html | 1085 ++++++----------- docs/modelarts/umn/modelarts_23_0177.html | 91 +- docs/modelarts/umn/modelarts_23_0178.html | 186 +-- docs/modelarts/umn/modelarts_23_0179.html | 135 +- docs/modelarts/umn/modelarts_23_0181.html | 2 +- docs/modelarts/umn/modelarts_23_0182.html | 4 +- docs/modelarts/umn/modelarts_23_0183.html | 1 + docs/modelarts/umn/modelarts_23_0187.html | 17 - docs/modelarts/umn/modelarts_23_0207.html | 89 +- docs/modelarts/umn/modelarts_23_0210.html | 3 +- docs/modelarts/umn/modelarts_23_0211.html | 2 +- docs/modelarts/umn/modelarts_23_0217.html | 12 +- docs/modelarts/umn/modelarts_23_0219.html | 6 +- docs/modelarts/umn/modelarts_23_0238.html | 28 +- docs/modelarts/umn/modelarts_23_0239.html | 4 +- docs/modelarts/umn/modelarts_23_0332.html | 21 +- docs/modelarts/umn/modelarts_23_0333.html | 10 +- docs/modelarts/umn/modelarts_23_0335.html | 93 +- docs/modelarts/umn/modelarts_23_0336.html | 2 +- .../umn/public_sys-resources/commonrtl.css | 2 +- 114 files changed, 3102 insertions(+), 5234 deletions(-) create mode 100644 docs/modelarts/umn/en-us_image_0000001156920825.png create mode 100644 docs/modelarts/umn/en-us_image_0000001157080841.png create mode 100644 docs/modelarts/umn/en-us_image_0000001157080843.png create mode 100644 docs/modelarts/umn/en-us_image_0000001157080915.png create mode 100644 docs/modelarts/umn/en-us_image_0000001278234781.png create mode 100644 docs/modelarts/umn/en-us_image_0000001281686748.png create mode 100644 docs/modelarts/umn/en-us_image_0000001290603082.png create mode 100644 docs/modelarts/umn/en-us_image_0000001340184197.png create mode 100644 docs/modelarts/umn/en-us_image_0000001340265309.png create mode 100644 docs/modelarts/umn/modelarts_05_0022.html create mode 100644 docs/modelarts/umn/modelarts_05_0024.html diff --git a/docs/modelarts/umn/ALL_META.TXT.json b/docs/modelarts/umn/ALL_META.TXT.json index 982e94efe..9adc8b160 100644 --- a/docs/modelarts/umn/ALL_META.TXT.json +++ b/docs/modelarts/umn/ALL_META.TXT.json @@ -123,7 +123,7 @@ "uri":"modelarts_01_0006.html", "product_code":"modelarts", "code":"13", - "des":"ModelArts uses Object Storage Service (OBS) to store data and model backups and snapshots. OBS provides secure, reliable, low-cost storage. For more details, see Object S", + "des":"ModelArts uses Identity and Access Management (IAM) for authentication and authorization. For more information about IAM, see Identity and Access Management User Guide.Mo", "doc_type":"usermanual", "kw":"Related Services,Service Overview,User Guide", "title":"Related Services", @@ -549,20 +549,10 @@ "title":"Object Detection", "githuburl":"" }, - { - "uri":"modelarts_23_0345.html", - "product_code":"modelarts", - "code":"56", - "des":"Training a model uses a large number of labeled images. Therefore, label images before the model training. You can label images on the ModelArts management console. Alter", - "doc_type":"usermanual", - "kw":"Image Segmentation,Labeling Data,User Guide", - "title":"Image Segmentation", - "githuburl":"" - }, { "uri":"modelarts_23_0013.html", "product_code":"modelarts", - "code":"57", + "code":"56", "des":"Model training requires a large amount of labeled data. Therefore, before the model training, add labels to the files that are not labeled. In addition, you can modify, d", "doc_type":"usermanual", "kw":"Text Classification,Labeling Data,User Guide", @@ -572,7 +562,7 @@ { "uri":"modelarts_23_0014.html", "product_code":"modelarts", - "code":"58", + "code":"57", "des":"Named entity recognition assigns labels to named entities in text, such as time and locations. Before labeling, you need to understand the following:A label name can cont", "doc_type":"usermanual", "kw":"Named Entity Recognition,Labeling Data,User Guide", @@ -582,7 +572,7 @@ { "uri":"modelarts_23_0211.html", "product_code":"modelarts", - "code":"59", + "code":"58", "des":"Triplet labeling is suitable for scenarios where structured information, such as subjects, predicates, and objects, needs to be labeled in statements. With this function,", "doc_type":"usermanual", "kw":"Text Triplet,Labeling Data,User Guide", @@ -592,7 +582,7 @@ { "uri":"modelarts_23_0015.html", "product_code":"modelarts", - "code":"60", + "code":"59", "des":"Model training requires a large amount of labeled data. Therefore, before the model training, label the unlabeled audio files. ModelArts enables you to label audio files ", "doc_type":"usermanual", "kw":"Sound Classification,Labeling Data,User Guide", @@ -602,7 +592,7 @@ { "uri":"modelarts_23_0016.html", "product_code":"modelarts", - "code":"61", + "code":"60", "des":"Model training requires a large amount of labeled data. Therefore, before the model training, label the unlabeled audio files. ModelArts enables you to label audio files ", "doc_type":"usermanual", "kw":"Speech Labeling,Labeling Data,User Guide", @@ -612,27 +602,17 @@ { "uri":"modelarts_23_0017.html", "product_code":"modelarts", - "code":"62", + "code":"61", "des":"Model training requires a large amount of labeled data. Therefore, before the model training, label the unlabeled audio files. ModelArts enables you to label audio files.", "doc_type":"usermanual", "kw":"Speech Paragraph Labeling,Labeling Data,User Guide", "title":"Speech Paragraph Labeling", "githuburl":"" }, - { - "uri":"modelarts_23_0282.html", - "product_code":"modelarts", - "code":"63", - "des":"Model training requires a large amount of labeled video data. Therefore, before the model training, label the unlabeled video files. ModelArts enables you to label video ", - "doc_type":"usermanual", - "kw":"Video Labeling,Labeling Data,User Guide", - "title":"Video Labeling", - "githuburl":"" - }, { "uri":"modelarts_23_0005.html", "product_code":"modelarts", - "code":"64", + "code":"62", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Importing Data", @@ -642,7 +622,7 @@ { "uri":"modelarts_23_0006.html", "product_code":"modelarts", - "code":"65", + "code":"63", "des":"After a dataset is created, you can directly synchronize data from the dataset. Alternatively, you can import more data by importing the dataset. Data can be imported fro", "doc_type":"usermanual", "kw":"Import Operation,Importing Data,User Guide", @@ -652,7 +632,7 @@ { "uri":"modelarts_23_0008.html", "product_code":"modelarts", - "code":"66", + "code":"64", "des":"When a dataset is imported, the data storage directory and file name must comply with the ModelArts specifications if the data to be used is stored in OBS.Only the follow", "doc_type":"usermanual", "kw":"Specifications for Importing Data from an OBS Directory,Importing Data,User Guide", @@ -662,7 +642,7 @@ { "uri":"modelarts_23_0009.html", "product_code":"modelarts", - "code":"67", + "code":"65", "des":"The manifest file defines the mapping between labeling objects and content. The Manifest file import mode means that the manifest file is used for dataset import. The man", "doc_type":"usermanual", "kw":"Specifications for Importing the Manifest File,Importing Data,User Guide", @@ -672,7 +652,7 @@ { "uri":"modelarts_23_0214.html", "product_code":"modelarts", - "code":"68", + "code":"66", "des":"A dataset includes labeled and unlabeled data. You can select images or filter data based on the filter criteria and export to a new dataset or the specified OBS director", "doc_type":"usermanual", "kw":"Exporting Data,Data Management,User Guide", @@ -682,7 +662,7 @@ { "uri":"modelarts_23_0020.html", "product_code":"modelarts", - "code":"69", + "code":"67", "des":"For a created dataset, you can modify its basic information to match service changes.You have created a dataset.Log in to the ModelArts management console. In the left na", "doc_type":"usermanual", "kw":"Modifying a Dataset,Data Management,User Guide", @@ -692,7 +672,7 @@ { "uri":"modelarts_23_0018.html", "product_code":"modelarts", - "code":"70", + "code":"68", "des":"ModelArts distinguishes data of the same source according to versions labeled at different time, which facilitates the selection of dataset versions during subsequent mod", "doc_type":"usermanual", "kw":"Publishing a Dataset,Data Management,User Guide", @@ -702,7 +682,7 @@ { "uri":"modelarts_23_0021.html", "product_code":"modelarts", - "code":"71", + "code":"69", "des":"If a dataset is no longer in use, you can delete it to release resources.After a dataset is deleted, if you need to delete the data in the dataset input and output paths ", "doc_type":"usermanual", "kw":"Deleting a Dataset,Data Management,User Guide", @@ -712,7 +692,7 @@ { "uri":"modelarts_23_0019.html", "product_code":"modelarts", - "code":"72", + "code":"70", "des":"After labeling data, you can publish the dataset to multiple versions for management. For the published versions, you can view the dataset version updates, set the curren", "doc_type":"usermanual", "kw":"Managing Dataset Versions,Data Management,User Guide", @@ -720,9 +700,59 @@ "githuburl":"" }, { - "uri":"modelarts_23_0032.html", + "uri":"modelarts_23_0180.html", + "product_code":"modelarts", + "code":"71", + "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", + "doc_type":"usermanual", + "kw":"Team Labeling", + "title":"Team Labeling", + "githuburl":"" + }, + { + "uri":"modelarts_23_0181.html", + "product_code":"modelarts", + "code":"72", + "des":"Generally, a small data labeling task can be completed by an individual. However, team work is required to label a large dataset. ModelArts provides the team labeling fun", + "doc_type":"usermanual", + "kw":"Introduction to Team Labeling,Team Labeling,User Guide", + "title":"Introduction to Team Labeling", + "githuburl":"" + }, + { + "uri":"modelarts_23_0182.html", "product_code":"modelarts", "code":"73", + "des":"Team labeling is managed in a unit of teams. To enable team labeling for a dataset, a team must be specified. Multiple members can be added to a team.An account can have ", + "doc_type":"usermanual", + "kw":"Team Management,Team Labeling,User Guide", + "title":"Team Management", + "githuburl":"" + }, + { + "uri":"modelarts_23_0183.html", + "product_code":"modelarts", + "code":"74", + "des":"There is no member in a new team. You need to add members who will participate in a team labeling task.A maximum of 100 members can be added to a team. If there are more ", + "doc_type":"usermanual", + "kw":"Member Management,Team Labeling,User Guide", + "title":"Member Management", + "githuburl":"" + }, + { + "uri":"modelarts_23_0210.html", + "product_code":"modelarts", + "code":"75", + "des":"For datasets with team labeling enabled, you can create team labeling tasks and assign the labeling tasks to different teams so that team members can complete the labelin", + "doc_type":"usermanual", + "kw":"Managing Team Labeling Tasks,Team Labeling,User Guide", + "title":"Managing Team Labeling Tasks", + "githuburl":"" + }, + { + "uri":"modelarts_23_0032.html", + "product_code":"modelarts", + "code":"76", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"DevEnviron (Notebook)", @@ -732,7 +762,7 @@ { "uri":"modelarts_23_0033.html", "product_code":"modelarts", - "code":"74", + "code":"77", "des":"ModelArts integrates the open-source Jupyter Notebook to provide you with online interactive development and debugging environments. You can use the Notebook on the Model", "doc_type":"usermanual", "kw":"Introduction to Notebook,DevEnviron (Notebook),User Guide", @@ -742,7 +772,7 @@ { "uri":"modelarts_23_0111.html", "product_code":"modelarts", - "code":"75", + "code":"78", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Managing Notebook Instances", @@ -752,7 +782,7 @@ { "uri":"modelarts_23_0034.html", "product_code":"modelarts", - "code":"76", + "code":"79", "des":"Before developing a model, create a notebook instance, open it, and perform encoding.You will be charged as long as your notebook instance is in the Running status. We re", "doc_type":"usermanual", "kw":"Creating a Notebook Instance,Managing Notebook Instances,User Guide", @@ -762,7 +792,7 @@ { "uri":"modelarts_23_0325.html", "product_code":"modelarts", - "code":"77", + "code":"80", "des":"You can open a created notebook instance (that is, an instance in the Running state) and start coding in the development environment.Go to the Jupyter Notebook page.In th", "doc_type":"usermanual", "kw":"Opening a Notebook Instance,Managing Notebook Instances,User Guide", @@ -772,7 +802,7 @@ { "uri":"modelarts_23_0041.html", "product_code":"modelarts", - "code":"78", + "code":"81", "des":"You can stop unwanted notebook instances to prevent unnecessary fees. You can also start a notebook instance that is in the Stopped state to use it again.Log in to the Mo", "doc_type":"usermanual", "kw":"Starting or Stopping a Notebook Instance,Managing Notebook Instances,User Guide", @@ -782,7 +812,7 @@ { "uri":"modelarts_23_0042.html", "product_code":"modelarts", - "code":"79", + "code":"82", "des":"You can delete notebook instances that are no longer used to release resources.Log in to the ModelArts management console. In the left navigation pane, choose DevEnviron ", "doc_type":"usermanual", "kw":"Deleting a Notebook Instance,Managing Notebook Instances,User Guide", @@ -792,7 +822,7 @@ { "uri":"modelarts_23_0035.html", "product_code":"modelarts", - "code":"80", + "code":"83", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Using Jupyter Notebook", @@ -802,7 +832,7 @@ { "uri":"modelarts_23_0326.html", "product_code":"modelarts", - "code":"81", + "code":"84", "des":"Jupyter Notebook is a web-based application for interactive computing. It can be applied to full-process computing: development, documentation, running code, and presenti", "doc_type":"usermanual", "kw":"Introduction to Jupyter Notebook,Using Jupyter Notebook,User Guide", @@ -812,7 +842,7 @@ { "uri":"modelarts_23_0120.html", "product_code":"modelarts", - "code":"82", + "code":"85", "des":"This section describes common operations on Jupyter Notebook.In the notebook instance list, locate the row where the target notebook instance resides and click Open in th", "doc_type":"usermanual", "kw":"Common Operations on Jupyter Notebook,Using Jupyter Notebook,User Guide", @@ -822,7 +852,7 @@ { "uri":"modelarts_23_0327.html", "product_code":"modelarts", - "code":"83", + "code":"86", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Configuring the Jupyter Notebook Environment", @@ -832,7 +862,7 @@ { "uri":"modelarts_23_0117.html", "product_code":"modelarts", - "code":"84", + "code":"87", "des":"For developers who are used to coding, the terminal function is very convenient and practical. This section describes how to enable the terminal function in a notebook in", "doc_type":"usermanual", "kw":"Using the Notebook Terminal Function,Configuring the Jupyter Notebook Environment,User Guide", @@ -842,7 +872,7 @@ { "uri":"modelarts_23_0280.html", "product_code":"modelarts", - "code":"85", + "code":"88", "des":"For a GPU-based notebook instance, you can switch different versions of CUDA on the Terminal page of Jupyter.CPU-based notebook instances do not use CUDA. Therefore, the ", "doc_type":"usermanual", "kw":"Switching the CUDA Version on the Terminal Page of a GPU-based Notebook Instance,Configuring the Jup", @@ -852,7 +882,7 @@ { "uri":"modelarts_23_0040.html", "product_code":"modelarts", - "code":"86", + "code":"89", "des":"Multiple environments have been installed in ModelArts notebook instances, including TensorFlow. You can use pip install to install external libraries from a Jupyter note", "doc_type":"usermanual", "kw":"Installing External Libraries and Kernels in Notebook Instances,Configuring the Jupyter Notebook Env", @@ -862,7 +892,7 @@ { "uri":"modelarts_23_0039.html", "product_code":"modelarts", - "code":"87", + "code":"90", "des":"In notebook instances, you can use ModelArts SDKs to manage OBS, training jobs, models, and real-time services.For details about how to use ModelArts SDKs, see ModelArts ", "doc_type":"usermanual", "kw":"Using ModelArts SDKs,Using Jupyter Notebook,User Guide", @@ -872,7 +902,7 @@ { "uri":"modelarts_23_0038.html", "product_code":"modelarts", - "code":"88", + "code":"91", "des":"If you specify Storage Path during notebook instance creation, your compiled code will be automatically stored in your specified OBS bucket. If code invocation among diff", "doc_type":"usermanual", "kw":"Synchronizing Files with OBS,Using Jupyter Notebook,User Guide", @@ -882,7 +912,7 @@ { "uri":"modelarts_23_0037.html", "product_code":"modelarts", - "code":"89", + "code":"92", "des":"After code compiling is finished, you can save the entered code as a .py file which can be used for starting training jobs.Create and open a notebook instance or open an ", "doc_type":"usermanual", "kw":"Using the Convert to Python File Function,Using Jupyter Notebook,User Guide", @@ -892,7 +922,7 @@ { "uri":"modelarts_23_0330.html", "product_code":"modelarts", - "code":"90", + "code":"93", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Using JupyterLab", @@ -902,7 +932,7 @@ { "uri":"modelarts_23_0209.html", "product_code":"modelarts", - "code":"91", + "code":"94", "des":"JupyterLab is an interactive development environment. It is a next-generation product of Jupyter Notebook. JupyterLab enables you to compile notebooks, operate terminals,", "doc_type":"usermanual", "kw":"Introduction to JupyterLab and Common Operations,Using JupyterLab,User Guide", @@ -912,7 +942,7 @@ { "uri":"modelarts_23_0331.html", "product_code":"modelarts", - "code":"92", + "code":"95", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Uploading and Downloading Data", @@ -922,7 +952,7 @@ { "uri":"modelarts_23_0332.html", "product_code":"modelarts", - "code":"93", + "code":"96", "des":"On the JupyterLab page, click Upload Files to upload a file. For details, see Uploading a File in Introduction to JupyterLab and Common Operations. If a message is displa", "doc_type":"usermanual", "kw":"Uploading Data to JupyterLab,Uploading and Downloading Data,User Guide", @@ -932,7 +962,7 @@ { "uri":"modelarts_23_0333.html", "product_code":"modelarts", - "code":"94", + "code":"97", "des":"Only files within 100 MB in JupyterLab can be downloaded to a local PC. You can perform operations in different scenarios based on the storage location selected when crea", "doc_type":"usermanual", "kw":"Downloading a File from JupyterLab,Uploading and Downloading Data,User Guide", @@ -942,7 +972,7 @@ { "uri":"modelarts_23_0335.html", "product_code":"modelarts", - "code":"95", + "code":"98", "des":"In notebook instances, you can use ModelArts SDKs to manage OBS, training jobs, models, and real-time services.For details about how to use ModelArts SDKs, see ModelArts ", "doc_type":"usermanual", "kw":"Using ModelArts SDKs,Using JupyterLab,User Guide", @@ -952,7 +982,7 @@ { "uri":"modelarts_23_0336.html", "product_code":"modelarts", - "code":"96", + "code":"99", "des":"If you specify Storage Path during notebook instance creation, your compiled code will be automatically stored in your specified OBS bucket. If code invocation among diff", "doc_type":"usermanual", "kw":"Synchronizing Files with OBS,Using JupyterLab,User Guide", @@ -962,7 +992,7 @@ { "uri":"modelarts_23_0043.html", "product_code":"modelarts", - "code":"97", + "code":"100", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Training Management", @@ -972,7 +1002,7 @@ { "uri":"modelarts_23_0044.html", "product_code":"modelarts", - "code":"98", + "code":"101", "des":"ModelArts provides model training for you to view the training effect, based on which you can adjust your model parameters. You can select resource pools (CPU or GPU) wit", "doc_type":"usermanual", "kw":"Introduction to Model Training,Training Management,User Guide", @@ -982,7 +1012,7 @@ { "uri":"modelarts_23_0156.html", "product_code":"modelarts", - "code":"99", + "code":"102", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Built-in Algorithms", @@ -992,7 +1022,7 @@ { "uri":"modelarts_23_0045.html", "product_code":"modelarts", - "code":"100", + "code":"103", "des":"Based on the frequently-used AI engines in the industry, ModelArts provides built-in algorithms to meet a wide range of your requirements. You can directly select the alg", "doc_type":"usermanual", "kw":"Introduction to Built-in Algorithms,Built-in Algorithms,User Guide", @@ -1002,7 +1032,7 @@ { "uri":"modelarts_23_0157.html", "product_code":"modelarts", - "code":"101", + "code":"104", "des":"The built-in algorithms provided by ModelArts can be used for image classification, object detection, and image semantic segmentation. The requirements for the datasets v", "doc_type":"usermanual", "kw":"Requirements on Datasets,Built-in Algorithms,User Guide", @@ -1012,7 +1042,7 @@ { "uri":"modelarts_23_0158.html", "product_code":"modelarts", - "code":"102", + "code":"105", "des":"This section describes the built-in algorithms supported by ModelArts and the running parameters supported by each algorithm. You can set running parameters for a trainin", "doc_type":"usermanual", "kw":"Algorithms and Their Running Parameters,Built-in Algorithms,User Guide", @@ -1022,7 +1052,7 @@ { "uri":"modelarts_23_0235.html", "product_code":"modelarts", - "code":"103", + "code":"106", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Creating a Training Job", @@ -1032,7 +1062,7 @@ { "uri":"modelarts_23_0046.html", "product_code":"modelarts", - "code":"104", + "code":"107", "des":"ModelArts supports multiple types of training jobs during the entire AI development process. Select a creation mode based on the algorithm source.Built-inIf you do not kn", "doc_type":"usermanual", "kw":"Introduction to Training Jobs,Creating a Training Job,User Guide", @@ -1042,7 +1072,7 @@ { "uri":"modelarts_23_0237.html", "product_code":"modelarts", - "code":"105", + "code":"108", "des":"If you do not have the algorithm development capability, you can use the built-in algorithms of ModelArts. After simple parameter adjustment, you can create a training jo", "doc_type":"usermanual", "kw":"Using Built-in Algorithms to Train Models,Creating a Training Job,User Guide", @@ -1052,7 +1082,7 @@ { "uri":"modelarts_23_0238.html", "product_code":"modelarts", - "code":"106", + "code":"109", "des":"If you use frequently-used frameworks, such as TensorFlow and MXNet, to develop algorithms locally, you can select Frequently-used to create training jobs and build model", "doc_type":"usermanual", "kw":"Using Frequently-used Frameworks to Train Models,Creating a Training Job,User Guide", @@ -1062,7 +1092,7 @@ { "uri":"modelarts_23_0239.html", "product_code":"modelarts", - "code":"107", + "code":"110", "des":"If the framework used for algorithm development is not a frequently-used framework, you can build an algorithm into a custom image and use the custom image to create a tr", "doc_type":"usermanual", "kw":"Using Custom Images to Train Models,Creating a Training Job,User Guide", @@ -1072,7 +1102,7 @@ { "uri":"modelarts_23_0159.html", "product_code":"modelarts", - "code":"108", + "code":"111", "des":"In the training job list, click Stop in the Operation column for a training job in the Running state to stop a running training job.If you have selected Save Training Par", "doc_type":"usermanual", "kw":"Stopping or Deleting a Job,Training Management,User Guide", @@ -1082,7 +1112,7 @@ { "uri":"modelarts_23_0047.html", "product_code":"modelarts", - "code":"109", + "code":"112", "des":"During model building, you may need to frequently tune the data, training parameters, or the model based on the training results to obtain a satisfactory model. ModelArts", "doc_type":"usermanual", "kw":"Managing Training Job Versions,Training Management,User Guide", @@ -1092,7 +1122,7 @@ { "uri":"modelarts_23_0048.html", "product_code":"modelarts", - "code":"110", + "code":"113", "des":"After a training job finishes, you can manage the training job versions and check whether the training result of the job is satisfactory by viewing the job details.In the", "doc_type":"usermanual", "kw":"Viewing Job Details,Training Management,User Guide", @@ -1102,7 +1132,7 @@ { "uri":"modelarts_23_0049.html", "product_code":"modelarts", - "code":"111", + "code":"114", "des":"You can store the parameter settings in ModelArts during job creation so that you can use the stored settings to create follow-up training jobs, which makes job creation ", "doc_type":"usermanual", "kw":"Managing Job Parameters,Training Management,User Guide", @@ -1112,7 +1142,7 @@ { "uri":"modelarts_23_0050.html", "product_code":"modelarts", - "code":"112", + "code":"115", "des":"You can create visualization jobs of TensorBoard and MindInsight types on ModelArts.TensorBoard supports training jobs based on the TensorFlow engine, and MindInsight sup", "doc_type":"usermanual", "kw":"Managing Visualization Jobs,Training Management,User Guide", @@ -1122,7 +1152,7 @@ { "uri":"modelarts_23_0051.html", "product_code":"modelarts", - "code":"113", + "code":"116", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Model Management", @@ -1132,7 +1162,7 @@ { "uri":"modelarts_23_0052.html", "product_code":"modelarts", - "code":"114", + "code":"117", "des":"AI model development and optimization require frequent iterations and debugging. Changes in datasets, training code, or parameters may affect the quality of models. If th", "doc_type":"usermanual", "kw":"Introduction to Model Management,Model Management,User Guide", @@ -1142,7 +1172,7 @@ { "uri":"modelarts_23_0204.html", "product_code":"modelarts", - "code":"115", + "code":"118", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Importing a Model", @@ -1152,7 +1182,7 @@ { "uri":"modelarts_23_0054.html", "product_code":"modelarts", - "code":"116", + "code":"119", "des":"You can create a training job on ModelArts and perform training to obtain a satisfactory model. Then import the model to Model Management for unified management. In addit", "doc_type":"usermanual", "kw":"Importing a Meta Model from a Training Job,Importing a Model,User Guide", @@ -1162,7 +1192,7 @@ { "uri":"modelarts_23_0205.html", "product_code":"modelarts", - "code":"117", + "code":"120", "des":"Because the configurations of models with the same functions are similar, ModelArts integrates the configurations of such models into a common template. By using this tem", "doc_type":"usermanual", "kw":"Importing a Meta Model from a Template,Importing a Model,User Guide", @@ -1172,7 +1202,7 @@ { "uri":"modelarts_23_0206.html", "product_code":"modelarts", - "code":"118", + "code":"121", "des":"For AI engines that are not supported by ModelArts, you can import the model you compile to ModelArts from custom images.For details about the specifications and descript", "doc_type":"usermanual", "kw":"Importing a Meta Model from a Container Image,Importing a Model,User Guide", @@ -1182,7 +1212,7 @@ { "uri":"modelarts_23_0207.html", "product_code":"modelarts", - "code":"119", + "code":"122", "des":"In scenarios where frequently-used frameworks are used for model development and training, you can import the model to ModelArts for unified management.The model has been", "doc_type":"usermanual", "kw":"Importing a Meta Model from OBS,Importing a Model,User Guide", @@ -1192,67 +1222,17 @@ { "uri":"modelarts_23_0055.html", "product_code":"modelarts", - "code":"120", + "code":"123", "des":"To facilitate source tracing and repeated model tuning, ModelArts provides the model version management function. You can manage models based on versions.You have importe", "doc_type":"usermanual", "kw":"Managing Model Versions,Model Management,User Guide", "title":"Managing Model Versions", "githuburl":"" }, - { - "uri":"modelarts_23_0106.html", - "product_code":"modelarts", - "code":"121", - "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", - "doc_type":"usermanual", - "kw":"Model Compression and Conversion", - "title":"Model Compression and Conversion", - "githuburl":"" - }, - { - "uri":"modelarts_23_0107.html", - "product_code":"modelarts", - "code":"122", - "des":"To obtain higher computing power, you can deploy the models created on ModelArts or a local PC on the Ascend chip. In this case, you need to compress or convert the model", - "doc_type":"usermanual", - "kw":"Compressing and Converting Models,Model Compression and Conversion,User Guide", - "title":"Compressing and Converting Models", - "githuburl":"" - }, - { - "uri":"modelarts_23_0108.html", - "product_code":"modelarts", - "code":"123", - "des":"During model conversion, the model input directory must comply with certain specifications. This section describes how to upload your model package to OBS.The requirement", - "doc_type":"usermanual", - "kw":"Model Input Path Specifications,Model Compression and Conversion,User Guide", - "title":"Model Input Path Specifications", - "githuburl":"" - }, - { - "uri":"modelarts_23_0109.html", - "product_code":"modelarts", - "code":"124", - "des":"The following describes the output path of the model run on the Ascend chip after conversion:For TensorFlow-based models, the output path must comply with the following s", - "doc_type":"usermanual", - "kw":"Model Output Path Description,Model Compression and Conversion,User Guide", - "title":"Model Output Path Description", - "githuburl":"" - }, - { - "uri":"modelarts_23_0110.html", - "product_code":"modelarts", - "code":"125", - "des":"ModelArts provides the following conversion templates based on different AI frameworks:TF-FrozenGraph-To-Ascend-C32Convert the model trained by the TensorFlow framework a", - "doc_type":"usermanual", - "kw":"Conversion Templates,Model Compression and Conversion,User Guide", - "title":"Conversion Templates", - "githuburl":"" - }, { "uri":"modelarts_23_0057.html", "product_code":"modelarts", - "code":"126", + "code":"124", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Model Deployment", @@ -1262,7 +1242,7 @@ { "uri":"modelarts_23_0058.html", "product_code":"modelarts", - "code":"127", + "code":"125", "des":"After a training job is complete and a model is generated, you can deploy the model on the Service Deployment page. You can also deploy the model imported from OBS. Model", "doc_type":"usermanual", "kw":"Introduction to Model Deployment,Model Deployment,User Guide", @@ -1272,7 +1252,7 @@ { "uri":"modelarts_23_0059.html", "product_code":"modelarts", - "code":"128", + "code":"126", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Real-Time Services", @@ -1282,7 +1262,7 @@ { "uri":"modelarts_23_0060.html", "product_code":"modelarts", - "code":"129", + "code":"127", "des":"After a model is prepared, you can deploy the model as a real-time service and predict and call the service.A maximum of one real-time service can be deployed.Data has be", "doc_type":"usermanual", "kw":"Deploying a Model as a Real-Time Service,Real-Time Services,User Guide", @@ -1292,7 +1272,7 @@ { "uri":"modelarts_23_0061.html", "product_code":"modelarts", - "code":"130", + "code":"128", "des":"After a model is deployed as a real-time service, you can access the service page to view its details.Log in to the ModelArts management console and choose Service Deploy", "doc_type":"usermanual", "kw":"Viewing Service Details,Real-Time Services,User Guide", @@ -1302,7 +1282,7 @@ { "uri":"modelarts_23_0062.html", "product_code":"modelarts", - "code":"131", + "code":"129", "des":"After a model is deployed as a real-time service, you can debug code or add files for testing on the Prediction tab page. Based on the input request (JSON text or file) d", "doc_type":"usermanual", "kw":"Testing a Service,Real-Time Services,User Guide", @@ -1312,7 +1292,7 @@ { "uri":"modelarts_23_0063.html", "product_code":"modelarts", - "code":"132", + "code":"130", "des":"If a real-time service is in the Running state, the real-time service has been deployed successfully. This service provides a standard RESTful API for users to call. Befo", "doc_type":"usermanual", "kw":"Accessing a Real-Time Service (Token-based Authentication),Real-Time Services,User Guide", @@ -1322,7 +1302,7 @@ { "uri":"modelarts_23_0065.html", "product_code":"modelarts", - "code":"133", + "code":"131", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Batch Services", @@ -1332,7 +1312,7 @@ { "uri":"modelarts_23_0066.html", "product_code":"modelarts", - "code":"134", + "code":"132", "des":"After a model is prepared, you can deploy it as a batch service. The Service Deployment > Batch Services page lists all batch services. You can enter a service name in th", "doc_type":"usermanual", "kw":"Deploying a Model as a Batch Service,Batch Services,User Guide", @@ -1342,7 +1322,7 @@ { "uri":"modelarts_23_0067.html", "product_code":"modelarts", - "code":"135", + "code":"133", "des":"When deploying a batch service, you can select the location of the output data directory. You can view the running result of the batch service that is in the Running comp", "doc_type":"usermanual", "kw":"Viewing the Batch Service Prediction Result,Batch Services,User Guide", @@ -1352,7 +1332,7 @@ { "uri":"modelarts_23_0071.html", "product_code":"modelarts", - "code":"136", + "code":"134", "des":"For a deployed service, you can modify its basic information to match service changes. You can modify the basic information about a service in either of the following way", "doc_type":"usermanual", "kw":"Modifying a Service,Model Deployment,User Guide", @@ -1362,7 +1342,7 @@ { "uri":"modelarts_23_0072.html", "product_code":"modelarts", - "code":"137", + "code":"135", "des":"You can start services in the Successful, Abnormal, or Stopped status. Services in the Deploying status cannot be started. A service is billed when it is started and in t", "doc_type":"usermanual", "kw":"Starting or Stopping a Service,Model Deployment,User Guide", @@ -1372,7 +1352,7 @@ { "uri":"modelarts_23_0073.html", "product_code":"modelarts", - "code":"138", + "code":"136", "des":"If a service is no longer in use, you can delete it to release resources.Log in to the ModelArts management console and choose Service Deployment from the left navigation", "doc_type":"usermanual", "kw":"Deleting a Service,Model Deployment,User Guide", @@ -1382,7 +1362,7 @@ { "uri":"modelarts_23_0076.html", "product_code":"modelarts", - "code":"139", + "code":"137", "des":"When using ModelArts to implement AI Development Lifecycle, you can use two different resource pools to train and deploy models.Public Resource Pool: provides public larg", "doc_type":"usermanual", "kw":"Resource Pools,User Guide", @@ -1392,7 +1372,7 @@ { "uri":"modelarts_23_0083.html", "product_code":"modelarts", - "code":"140", + "code":"138", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Custom Images", @@ -1402,7 +1382,7 @@ { "uri":"modelarts_23_0084.html", "product_code":"modelarts", - "code":"141", + "code":"139", "des":"ModelArts provides multiple frequently-used built-in engines. However, when users have special requirements for the deep learning engine and development library, the buil", "doc_type":"usermanual", "kw":"Introduction to Custom Images,Custom Images,User Guide", @@ -1412,7 +1392,7 @@ { "uri":"modelarts_23_0085.html", "product_code":"modelarts", - "code":"142", + "code":"140", "des":"ModelArts allows you to use custom images to create training jobs and import models. Before creating and uploading a custom image, understand the following information:So", "doc_type":"usermanual", "kw":"Creating and Uploading a Custom Image,Custom Images,User Guide", @@ -1422,7 +1402,7 @@ { "uri":"modelarts_23_0216.html", "product_code":"modelarts", - "code":"143", + "code":"141", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"For Training Models", @@ -1432,7 +1412,7 @@ { "uri":"modelarts_23_0217.html", "product_code":"modelarts", - "code":"144", + "code":"142", "des":"When creating an image using locally developed models and training scripts, ensure that they meet the specifications defined by ModelArts.Custom images cannot contain mal", "doc_type":"usermanual", "kw":"Specifications for Custom Images Used for Training Jobs,For Training Models,User Guide", @@ -1442,7 +1422,7 @@ { "uri":"modelarts_23_0087.html", "product_code":"modelarts", - "code":"145", + "code":"143", "des":"After creating and uploading a custom image to SWR, you can use the image to create a training job on the ModelArts management console to complete model training.You have", "doc_type":"usermanual", "kw":"Creating a Training Job Using a Custom Image (GPU),For Training Models,User Guide", @@ -1452,7 +1432,7 @@ { "uri":"modelarts_23_0218.html", "product_code":"modelarts", - "code":"146", + "code":"144", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"For Importing Models", @@ -1462,7 +1442,7 @@ { "uri":"modelarts_23_0219.html", "product_code":"modelarts", - "code":"147", + "code":"145", "des":"When creating an image using locally developed models, ensure that they meet the specifications defined by ModelArts.Custom images cannot contain malicious code.The size ", "doc_type":"usermanual", "kw":"Specifications for Custom Images Used for Importing Models,For Importing Models,User Guide", @@ -1472,7 +1452,7 @@ { "uri":"modelarts_23_0086.html", "product_code":"modelarts", - "code":"148", + "code":"146", "des":"After creating and uploading a custom image to SWR, you can use the image to import a model and deploy the model as a service on the ModelArts management console.You have", "doc_type":"usermanual", "kw":"Importing a Model Using a Custom Image,For Importing Models,User Guide", @@ -1482,7 +1462,7 @@ { "uri":"modelarts_23_0090.html", "product_code":"modelarts", - "code":"149", + "code":"147", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Model Package Specifications", @@ -1492,7 +1472,7 @@ { "uri":"modelarts_23_0091.html", "product_code":"modelarts", - "code":"150", + "code":"148", "des":"When you import models in Model Management, if the meta model is imported from OBS or a container image, the model package must meet the following specifications:The mode", "doc_type":"usermanual", "kw":"Model Package Specifications,Model Package Specifications,User Guide", @@ -1502,7 +1482,7 @@ { "uri":"modelarts_23_0092.html", "product_code":"modelarts", - "code":"151", + "code":"149", "des":"A model developer needs to compile a configuration file when publishing a model. The model configuration file describes the model usage, computing framework, precision, i", "doc_type":"usermanual", "kw":"Specifications for Compiling the Model Configuration File,Model Package Specifications,User Guide", @@ -1512,7 +1492,7 @@ { "uri":"modelarts_23_0093.html", "product_code":"modelarts", - "code":"152", + "code":"150", "des":"This section describes how to compile model inference code in ModelArts. The following also provides an example of inference code for the TensorFlow engine and an example", "doc_type":"usermanual", "kw":"Specifications for Compiling Model Inference Code,Model Package Specifications,User Guide", @@ -1522,7 +1502,7 @@ { "uri":"modelarts_23_0097.html", "product_code":"modelarts", - "code":"153", + "code":"151", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Model Templates", @@ -1532,7 +1512,7 @@ { "uri":"modelarts_23_0098.html", "product_code":"modelarts", - "code":"154", + "code":"152", "des":"Because the configurations of models with the same functions are similar, ModelArts integrates the configurations of such models into a common template. By using this tem", "doc_type":"usermanual", "kw":"Introduction to Model Templates,Model Templates,User Guide", @@ -1542,7 +1522,7 @@ { "uri":"modelarts_23_0118.html", "product_code":"modelarts", - "code":"155", + "code":"153", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Template Description", @@ -1552,7 +1532,7 @@ { "uri":"modelarts_23_0161.html", "product_code":"modelarts", - "code":"156", + "code":"154", "des":"AI engine: TensorFlow 1.8; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or appl", "doc_type":"usermanual", "kw":"TensorFlow-py27 General Template,Template Description,User Guide", @@ -1562,7 +1542,7 @@ { "uri":"modelarts_23_0162.html", "product_code":"modelarts", - "code":"157", + "code":"155", "des":"AI engine: TensorFlow 1.8; Environment: Python 3.6; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or appl", "doc_type":"usermanual", "kw":"TensorFlow-py36 General Template,Template Description,User Guide", @@ -1572,7 +1552,7 @@ { "uri":"modelarts_23_0163.html", "product_code":"modelarts", - "code":"158", + "code":"156", "des":"AI engine: MXNet 1.2.1; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or applica", "doc_type":"usermanual", "kw":"MXNet-py27 General Template,Template Description,User Guide", @@ -1582,7 +1562,7 @@ { "uri":"modelarts_23_0164.html", "product_code":"modelarts", - "code":"159", + "code":"157", "des":"AI engine: MXNet 1.2.1; Environment: Python 3.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or applica", "doc_type":"usermanual", "kw":"MXNet-py37 General Template,Template Description,User Guide", @@ -1592,7 +1572,7 @@ { "uri":"modelarts_23_0165.html", "product_code":"modelarts", - "code":"160", + "code":"158", "des":"AI engine: PyTorch 1.0; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or applica", "doc_type":"usermanual", "kw":"PyTorch-py27 General Template,Template Description,User Guide", @@ -1602,7 +1582,7 @@ { "uri":"modelarts_23_0166.html", "product_code":"modelarts", - "code":"161", + "code":"159", "des":"AI engine: PyTorch 1.0; Environment: Python 3.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or applica", "doc_type":"usermanual", "kw":"PyTorch-py37 General Template,Template Description,User Guide", @@ -1612,7 +1592,7 @@ { "uri":"modelarts_23_0167.html", "product_code":"modelarts", - "code":"162", + "code":"160", "des":"AI engine: CPU-based Caffe 1.0; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or", "doc_type":"usermanual", "kw":"Caffe-CPU-py27 General Template,Template Description,User Guide", @@ -1622,7 +1602,7 @@ { "uri":"modelarts_23_0168.html", "product_code":"modelarts", - "code":"163", + "code":"161", "des":"AI engine: GPU-based Caffe 1.0; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or", "doc_type":"usermanual", "kw":"Caffe-GPU-py27 General Template,Template Description,User Guide", @@ -1632,7 +1612,7 @@ { "uri":"modelarts_23_0169.html", "product_code":"modelarts", - "code":"164", + "code":"162", "des":"AI engine: CPU-based Caffe 1.0; Environment: Python 3.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or", "doc_type":"usermanual", "kw":"Caffe-CPU-py37 General Template,Template Description,User Guide", @@ -1642,7 +1622,7 @@ { "uri":"modelarts_23_0170.html", "product_code":"modelarts", - "code":"165", + "code":"163", "des":"AI engine: GPU-based Caffe 1.0; Environment: Python 3.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or", "doc_type":"usermanual", "kw":"Caffe-GPU-py37 General Template,Template Description,User Guide", @@ -1652,7 +1632,7 @@ { "uri":"modelarts_23_0099.html", "product_code":"modelarts", - "code":"166", + "code":"164", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Input and Output Modes", @@ -1662,7 +1642,7 @@ { "uri":"modelarts_23_0100.html", "product_code":"modelarts", - "code":"167", + "code":"165", "des":"This is a built-in input and output mode for object detection. The models using this mode are identified as object detection models. The prediction request path is /, the", "doc_type":"usermanual", "kw":"Built-in Object Detection Mode,Input and Output Modes,User Guide", @@ -1672,7 +1652,7 @@ { "uri":"modelarts_23_0101.html", "product_code":"modelarts", - "code":"168", + "code":"166", "des":"The built-in image processing input and output mode can be applied to models such as image classification, object detection, and image semantic segmentation. The predicti", "doc_type":"usermanual", "kw":"Built-in Image Processing Mode,Input and Output Modes,User Guide", @@ -1682,7 +1662,7 @@ { "uri":"modelarts_23_0102.html", "product_code":"modelarts", - "code":"169", + "code":"167", "des":"This is a built-in input and output mode for predictive analytics. The models using this mode are identified as predictive analytics models. The prediction request path i", "doc_type":"usermanual", "kw":"Built-in Predictive Analytics Mode,Input and Output Modes,User Guide", @@ -1692,7 +1672,7 @@ { "uri":"modelarts_23_0103.html", "product_code":"modelarts", - "code":"170", + "code":"168", "des":"The undefined mode does not define the input and output mode. The input and output mode is determined by the model. Select this mode only when the existing input and outp", "doc_type":"usermanual", "kw":"Undefined Mode,Input and Output Modes,User Guide", @@ -1702,7 +1682,7 @@ { "uri":"modelarts_23_0172.html", "product_code":"modelarts", - "code":"171", + "code":"169", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Examples of Custom Scripts", @@ -1712,7 +1692,7 @@ { "uri":"modelarts_23_0173.html", "product_code":"modelarts", - "code":"172", + "code":"170", "des":"TensorFlow has two types of APIs: Keras and tf. Keras and tf use different code for training and saving models, but the same code for inference.", "doc_type":"usermanual", "kw":"TensorFlow,Examples of Custom Scripts,User Guide", @@ -1722,7 +1702,7 @@ { "uri":"modelarts_23_0175.html", "product_code":"modelarts", - "code":"173", + "code":"171", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"PyTorch,Examples of Custom Scripts,User Guide", @@ -1732,7 +1712,7 @@ { "uri":"modelarts_23_0176.html", "product_code":"modelarts", - "code":"174", + "code":"172", "des":"lenet_train_test.prototxt filelenet_solver.prototxt fileTrain the model.The caffemodel file is generated after model training. Rewrite the lenet_train_test.prototxt file ", "doc_type":"usermanual", "kw":"Caffe,Examples of Custom Scripts,User Guide", @@ -1742,8 +1722,8 @@ { "uri":"modelarts_23_0177.html", "product_code":"modelarts", - "code":"175", - "des":"After the model is saved, it must be uploaded to the OBS directory before being published. The config.json and customize_service.py files must be contained during publish", + "code":"173", + "des":"Before training, download the iris.csv dataset, decompress it, and upload it to the /home/ma-user/work/ directory of the notebook instance. Download the iris.csv dataset ", "doc_type":"usermanual", "kw":"XGBoost,Examples of Custom Scripts,User Guide", "title":"XGBoost", @@ -1752,7 +1732,7 @@ { "uri":"modelarts_23_0178.html", "product_code":"modelarts", - "code":"176", + "code":"174", "des":"After the model is saved, it must be uploaded to the OBS directory before being published. The config.json configuration and customize_service.py must be contained during", "doc_type":"usermanual", "kw":"PySpark,Examples of Custom Scripts,User Guide", @@ -1762,8 +1742,8 @@ { "uri":"modelarts_23_0179.html", "product_code":"modelarts", - "code":"177", - "des":"After the model is saved, it must be uploaded to the OBS directory before being published. The config.json and customize_service.py files must be contained during publish", + "code":"175", + "des":"Before training, download the iris.csv dataset, decompress it, and upload it to the /home/ma-user/work/ directory of the notebook instance. Download the iris.csv dataset ", "doc_type":"usermanual", "kw":"Scikit Learn,Examples of Custom Scripts,User Guide", "title":"Scikit Learn", @@ -1772,7 +1752,7 @@ { "uri":"modelarts_23_0077.html", "product_code":"modelarts", - "code":"178", + "code":"176", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Permissions Management", @@ -1782,7 +1762,7 @@ { "uri":"modelarts_23_0078.html", "product_code":"modelarts", - "code":"179", + "code":"177", "des":"A fine-grained policy is a set of permissions defining which operations on which cloud services can be performed. Each policy can define multiple permissions. After a pol", "doc_type":"usermanual", "kw":"Basic Concepts,Permissions Management,User Guide", @@ -1792,7 +1772,7 @@ { "uri":"modelarts_23_0079.html", "product_code":"modelarts", - "code":"180", + "code":"178", "des":"A fine-grained policy consists of the policy version (the Version field) and statement (the Statement field).Version: Distinguishes between role-based access control (RBA", "doc_type":"usermanual", "kw":"Creating a User and Granting Permissions,Permissions Management,User Guide", @@ -1802,7 +1782,7 @@ { "uri":"modelarts_23_0080.html", "product_code":"modelarts", - "code":"181", + "code":"179", "des":"If default policies cannot meet the requirements on fine-grained access control, you can create custom policies and assign the policies to the user group.You can create c", "doc_type":"usermanual", "kw":"Creating a Custom Policy,Permissions Management,User Guide", @@ -1812,7 +1792,7 @@ { "uri":"modelarts_23_0186.html", "product_code":"modelarts", - "code":"182", + "code":"180", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Monitoring", @@ -1822,7 +1802,7 @@ { "uri":"modelarts_23_0187.html", "product_code":"modelarts", - "code":"183", + "code":"181", "des":"The cloud service platform provides Cloud Eye to help you better understand the status of your ModelArts real-time services and models. You can use Cloud Eye to automatic", "doc_type":"usermanual", "kw":"ModelArts Metrics,Monitoring,User Guide", @@ -1832,7 +1812,7 @@ { "uri":"modelarts_23_0188.html", "product_code":"modelarts", - "code":"184", + "code":"182", "des":"Setting alarm rules allows you to customize the monitored objects and notification policies so that you can know the status of ModelArts real-time services and models in ", "doc_type":"usermanual", "kw":"Setting Alarm Rules,Monitoring,User Guide", @@ -1842,7 +1822,7 @@ { "uri":"modelarts_23_0189.html", "product_code":"modelarts", - "code":"185", + "code":"183", "des":"Cloud Eye on the cloud service platform monitors the status of ModelArts real-time services and model loads. You can obtain the monitoring metrics of each ModelArts real-", "doc_type":"usermanual", "kw":"Viewing Monitoring Metrics,Monitoring,User Guide", @@ -1852,7 +1832,7 @@ { "uri":"modelarts_23_0249.html", "product_code":"modelarts", - "code":"186", + "code":"184", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Audit Logs", @@ -1862,7 +1842,7 @@ { "uri":"modelarts_23_0250.html", "product_code":"modelarts", - "code":"187", + "code":"185", "des":"CTS is available on the public cloud platform. With CTS, you can record operations associated with ModelArts for later query, audit, and backtrack operations.CTS has been", "doc_type":"usermanual", "kw":"Key Operations Recorded by CTS,Audit Logs,User Guide", @@ -1872,7 +1852,7 @@ { "uri":"modelarts_23_0251.html", "product_code":"modelarts", - "code":"188", + "code":"186", "des":"After CTS is enabled, CTS starts recording operations related to ModelArts. The CTS management console stores the last seven days of operation records. This section descr", "doc_type":"usermanual", "kw":"Viewing Audit Logs,Audit Logs,User Guide", @@ -1882,7 +1862,7 @@ { "uri":"modelarts_05_0000.html", "product_code":"modelarts", - "code":"189", + "code":"187", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"FAQs", @@ -1892,7 +1872,7 @@ { "uri":"modelarts_05_0014.html", "product_code":"modelarts", - "code":"190", + "code":"188", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"General Issues", @@ -1902,7 +1882,7 @@ { "uri":"modelarts_05_0001.html", "product_code":"modelarts", - "code":"191", + "code":"189", "des":"ModelArts is a one-stop development platform for AI developers. With data preprocessing, semi-automated data labeling, distributed training, automated model building, and", "doc_type":"usermanual", "kw":"What Is ModelArts?,General Issues,User Guide", @@ -1912,17 +1892,17 @@ { "uri":"modelarts_05_0003.html", "product_code":"modelarts", - "code":"192", - "des":"ModelArts uses Object Storage Service (OBS) to store data and model backups and snapshots. OBS provides secure, reliable, low-cost storage. For more details, see Object S", + "code":"190", + "des":"ModelArts uses Identity and Access Management (IAM) for authentication and authorization. For more information about IAM, see Identity and Access Management User Guide.Mo", "doc_type":"usermanual", - "kw":"What Are the Relationships Between ModelArts and Other Services,General Issues,User Guide", - "title":"What Are the Relationships Between ModelArts and Other Services", + "kw":"What Are The Relationships Between ModelArts And Other Services,General Issues,User Guide", + "title":"What Are The Relationships Between ModelArts And Other Services", "githuburl":"" }, { "uri":"modelarts_05_0004.html", "product_code":"modelarts", - "code":"193", + "code":"191", "des":"Log in to the console, enter the My Credentials page, and choose Access Keys > Create Access Key.In the Create Access Key dialog box that is displayed, use the login pass", "doc_type":"usermanual", "kw":"How Do I Obtain Access Keys?,General Issues,User Guide", @@ -1932,7 +1912,7 @@ { "uri":"modelarts_05_0013.html", "product_code":"modelarts", - "code":"194", + "code":"192", "des":"Before using ModelArts to develop AI models, data needs to be uploaded to an OBS bucket. You can log in to the OBS console to create an OBS bucket, create a folder, and u", "doc_type":"usermanual", "kw":"How Do I Upload Data to OBS?,General Issues,User Guide", @@ -1942,7 +1922,7 @@ { "uri":"modelarts_05_0128.html", "product_code":"modelarts", - "code":"195", + "code":"193", "des":"Supported AI frameworks and versions of ModelArts vary slightly based on the development environment, training jobs, and model inference (model management and deployment)", "doc_type":"usermanual", "kw":"Which AI Frameworks Does ModelArts Support?,General Issues,User Guide", @@ -1952,28 +1932,18 @@ { "uri":"modelarts_21_0055.html", "product_code":"modelarts", - "code":"196", + "code":"194", "des":"For common users, ModelArts provides the predictive analytics function of ExeML to train models based on structured data.For advanced users, ModelArts provides the notebo", "doc_type":"usermanual", "kw":"How Do I Use ModelArts to Train Models Based on Structured Data?,General Issues,User Guide", "title":"How Do I Use ModelArts to Train Models Based on Structured Data?", "githuburl":"" }, - { - "uri":"modelarts_21_0056.html", - "product_code":"modelarts", - "code":"197", - "des":"If an OBS directory needs to be specified for using ModelArts functions, such as creating training jobs and datasets, ensure that the OBS bucket and ModelArts are in the ", - "doc_type":"usermanual", - "kw":"Why Cannot I Find the OBS Bucket on ModelArts After Uploading Data to OBS?,General Issues,User Guide", - "title":"Why Cannot I Find the OBS Bucket on ModelArts After Uploading Data to OBS?", - "githuburl":"" - }, { "uri":"modelarts_21_0057.html", "product_code":"modelarts", - "code":"198", - "des":"No. The current ModelArts version does not support multiple projects. Customers can only use it in the default eu-de project.", + "code":"195", + "des":"The current version supports multiple projects.", "doc_type":"usermanual", "kw":"Does ModelArts Support Multiple Projects?,General Issues,User Guide", "title":"Does ModelArts Support Multiple Projects?", @@ -1982,7 +1952,7 @@ { "uri":"modelarts_21_0058.html", "product_code":"modelarts", - "code":"199", + "code":"196", "des":"To view all files stored in OBS when using notebook instances or training jobs, use either of the following methods:OBS consoleLog in to OBS console using the current acc", "doc_type":"usermanual", "kw":"How Do I View All Files in an OBS Directory on ModelArts?,General Issues,User Guide", @@ -1992,7 +1962,7 @@ { "uri":"modelarts_21_0059.html", "product_code":"modelarts", - "code":"200", + "code":"197", "des":"No. The current ModelArts version does not support encrypted files stored in OBS.", "doc_type":"usermanual", "kw":"Does ModelArts Support Encrypted Files Stored in OBS?,General Issues,User Guide", @@ -2002,7 +1972,7 @@ { "uri":"modelarts_05_0015.html", "product_code":"modelarts", - "code":"201", + "code":"198", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"ExeML", @@ -2012,7 +1982,7 @@ { "uri":"modelarts_05_0002.html", "product_code":"modelarts", - "code":"202", + "code":"199", "des":"ExeML is the process of automating model design, parameter tuning, and model training, compression, and deployment with the labeled data. The process is free of coding an", "doc_type":"usermanual", "kw":"What Is ExeML?,ExeML,User Guide", @@ -2022,7 +1992,7 @@ { "uri":"modelarts_05_0018.html", "product_code":"modelarts", - "code":"203", + "code":"200", "des":"Image classification is an image processing method that separates different classes of targets according to the features reflected in the images. With quantitative analys", "doc_type":"usermanual", "kw":"What Are Image Classification and Object Detection?,ExeML,User Guide", @@ -2032,7 +2002,7 @@ { "uri":"modelarts_05_0005.html", "product_code":"modelarts", - "code":"204", + "code":"201", "des":"The Train button turns to be available when the training images for an image classification project are classified into at least two categories, and each category contain", "doc_type":"usermanual", "kw":"What Should I Do When the Train Button Is Unavailable After I Create an Image Classification Project", @@ -2042,7 +2012,7 @@ { "uri":"modelarts_05_0006.html", "product_code":"modelarts", - "code":"205", + "code":"202", "des":"Yes. You can add multiple labels to an image.", "doc_type":"usermanual", "kw":"Can I Add Multiple Labels to an Image for an Object Detection Project?,ExeML,User Guide", @@ -2052,7 +2022,7 @@ { "uri":"modelarts_05_0008.html", "product_code":"modelarts", - "code":"206", + "code":"203", "des":"Models created in ExeML are deployed as real-time services. You can add images or compile code to test the services, as well as call the APIs using the URLs.After model d", "doc_type":"usermanual", "kw":"What Type of Service Is Deployed in ExeML?,ExeML,User Guide", @@ -2062,7 +2032,7 @@ { "uri":"modelarts_05_0010.html", "product_code":"modelarts", - "code":"207", + "code":"204", "des":"Images in JPG, JPEG, PNG, or BMP format are supported.", "doc_type":"usermanual", "kw":"What Formats of Images Are Supported by Object Detection or Image Classification Projects?,ExeML,Use", @@ -2072,7 +2042,7 @@ { "uri":"modelarts_21_0062.html", "product_code":"modelarts", - "code":"208", + "code":"205", "des":"Data files cannot be stored in the root directory of an OBS bucket.The name of files in a dataset consists of letters, digits, hyphens (-), and underscores (_), and the f", "doc_type":"usermanual", "kw":"What Are the Requirements for Training Data When You Create a Predictive Analytics Project in ExeML?", @@ -2082,7 +2052,7 @@ { "uri":"modelarts_21_0061.html", "product_code":"modelarts", - "code":"209", + "code":"206", "des":"The model cannot be downloaded. However, you can view the model or deploy the model as a real-time service on the model management page.", "doc_type":"usermanual", "kw":"Can I Download a Model After It Is Automatically Trained?,ExeML,User Guide", @@ -2092,7 +2062,7 @@ { "uri":"modelarts_21_0060.html", "product_code":"modelarts", - "code":"210", + "code":"207", "des":"Each round of training generates a training version in an ExeML project. If a training result is unsatisfactory (for example, unsatisfactory about the training precision)", "doc_type":"usermanual", "kw":"How Do I Perform Incremental Training in an ExeML Project?,ExeML,User Guide", @@ -2102,7 +2072,7 @@ { "uri":"modelarts_05_0101.html", "product_code":"modelarts", - "code":"211", + "code":"208", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Data Management", @@ -2112,7 +2082,7 @@ { "uri":"modelarts_21_0063.html", "product_code":"modelarts", - "code":"212", + "code":"209", "des":"For the data management function, there are limits on the image size when you upload images to the datasets whose labeling type is object detection or image classificatio", "doc_type":"usermanual", "kw":"Are There Size Limits for Images to be Uploaded?,Data Management,User Guide", @@ -2122,7 +2092,7 @@ { "uri":"modelarts_05_0103.html", "product_code":"modelarts", - "code":"213", + "code":"210", "des":"Failed to use the manifest file of the published dataset to import data again.Data has been changed in the OBS directory of the published dataset, for example, images hav", "doc_type":"usermanual", "kw":"Why Does Data Fail to Be Imported Using the Manifest File?,Data Management,User Guide", @@ -2132,7 +2102,7 @@ { "uri":"modelarts_05_0067.html", "product_code":"modelarts", - "code":"214", + "code":"211", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Notebook", @@ -2142,7 +2112,7 @@ { "uri":"modelarts_05_0071.html", "product_code":"modelarts", - "code":"215", + "code":"212", "des":"Log in to the ModelArts management console, and choose DevEnviron > Notebooks.In the notebook list, click Open in the Operation column of the target notebook instance to ", "doc_type":"usermanual", "kw":"How Do I Enable the Terminal Function in DevEnviron of ModelArts?,Notebook,User Guide", @@ -2150,10 +2120,10 @@ "githuburl":"" }, { - "uri":"modelarts_21_0064.html", + "uri":"modelarts_05_0022.html", "product_code":"modelarts", - "code":"216", - "des":"Log in to the ModelArts management console, and choose DevEnviron > Notebooks.In the notebook list, click Open in the Operation column of the target notebook instance to ", + "code":"213", + "des":"Multiple environments have been integrated into ModelArts Notebook. These environments contain Jupyter Notebook and Python packages, including TensorFlow, MXNet, Caffe, P", "doc_type":"usermanual", "kw":"How Do I Install External Libraries in a Notebook Instance?,Notebook,User Guide", "title":"How Do I Install External Libraries in a Notebook Instance?", @@ -2162,7 +2132,7 @@ { "uri":"modelarts_21_0065.html", "product_code":"modelarts", - "code":"217", + "code":"214", "des":"Notebook instances in DevEnviron support the Keras engine. The Keras engine is not supported in job training and model deployment (inference).Keras is an advanced neural ", "doc_type":"usermanual", "kw":"Is the Keras Engine Supported?,Notebook,User Guide", @@ -2172,7 +2142,7 @@ { "uri":"modelarts_21_0066.html", "product_code":"modelarts", - "code":"218", + "code":"215", "des":"After the training code is debugged in a notebook instance, if you need to use the training code for training jobs on ModelArts, convert the ipynb file into a Python file", "doc_type":"usermanual", "kw":"How Do I Use Training Code in Training Jobs After Debugging the Code in a Notebook Instance?,Noteboo", @@ -2182,17 +2152,27 @@ { "uri":"modelarts_21_0067.html", "product_code":"modelarts", - "code":"219", + "code":"216", "des":"In the notebook instance, error message \"No Space left...\" is displayed after the pip install command is run.You are advised to run the pip install --no-cache ** command", "doc_type":"usermanual", "kw":"What Should I Do When the System Displays an Error Message Indicating that No Space Left After I Run", "title":"What Should I Do When the System Displays an Error Message Indicating that No Space Left After I Run the pip install Command?", "githuburl":"" }, + { + "uri":"modelarts_05_0024.html", + "product_code":"modelarts", + "code":"217", + "des":"In a notebook instance, you can call the ModelArts MoXing API or SDK to exchange data with OBS for uploading a file to OBS or downloading a file from OBS to the notebook ", + "doc_type":"faq", + "kw":"How Do I Upload a File from a Notebook Instance to OBS or Download a File from OBS to a Notebook Ins", + "title":"How Do I Upload a File from a Notebook Instance to OBS or Download a File from OBS to a Notebook Instance?", + "githuburl":"" + }, { "uri":"modelarts_21_0068.html", "product_code":"modelarts", - "code":"220", + "code":"218", "des":"Small files (files smaller than 100 MB)Open a notebook instance and click Upload in the upper right corner to upload a local file to the notebook instance.Upload a small ", "doc_type":"usermanual", "kw":"How Do I Upload Local Files to a Notebook Instance?,Notebook,User Guide", @@ -2202,7 +2182,7 @@ { "uri":"modelarts_05_0045.html", "product_code":"modelarts", - "code":"221", + "code":"219", "des":"If you use OBS to store the notebook instance, after you click upload, the data is directly uploaded to the target OBS path, that is, the OBS path specified when the note", "doc_type":"usermanual", "kw":"Where Will the Data Be Uploaded to?,Notebook,User Guide", @@ -2212,7 +2192,7 @@ { "uri":"modelarts_21_0069.html", "product_code":"modelarts", - "code":"222", + "code":"220", "des":"The following uses the TensorFlow-1.8 engine as an example. The operations on other engines are similar. You only need to replace the engine name and version number in th", "doc_type":"usermanual", "kw":"Should I Access the Python Environment Same as the Notebook Kernel of the Current Instance in the Te", @@ -2222,7 +2202,7 @@ { "uri":"modelarts_21_0070.html", "product_code":"modelarts", - "code":"223", + "code":"221", "des":"If a notebook instance fails to execute code, you can locate and rectify the fault based on the following scenarios:If the execution of a cell is suspended or lasts for a", "doc_type":"usermanual", "kw":"What Do I Do If a Notebook Instance Fails to Execute Code?,Notebook,User Guide", @@ -2232,7 +2212,7 @@ { "uri":"modelarts_21_0071.html", "product_code":"modelarts", - "code":"224", + "code":"222", "des":"Currently, Terminal in ModelArts DevEnviron does not support apt-get. You can use a custom imagecustom image to support it.", "doc_type":"usermanual", "kw":"Does ModelArts DevEnviron Support apt-get?,Notebook,User Guide", @@ -2242,7 +2222,7 @@ { "uri":"modelarts_05_0080.html", "product_code":"modelarts", - "code":"225", + "code":"223", "des":"/cache is a temporary directory and will not be saved. After an instance using OBS storage is stopped, data in the ~work directory will be deleted. After a notebook insta", "doc_type":"usermanual", "kw":"Do Files in /cache Still Exist After a Notebook Instance is Stopped or Restarted? How Do I Avoid a R", @@ -2252,7 +2232,7 @@ { "uri":"modelarts_05_0081.html", "product_code":"modelarts", - "code":"226", + "code":"224", "des":"Log in to the ModelArts management console, and choose DevEnviron > Notebooks.In the Operation column of the target notebook instance in the notebook list, click Open to ", "doc_type":"usermanual", "kw":"Where Is Data Stored After the Sync OBS Function Is Used?,Notebook,User Guide", @@ -2262,7 +2242,7 @@ { "uri":"modelarts_21_0072.html", "product_code":"modelarts", - "code":"227", + "code":"225", "des":"If you select GPU when creating a notebook instance, perform the following operations to view GPU usage:Log in to the ModelArts management console, and choose DevEnviron ", "doc_type":"usermanual", "kw":"How Do I View GPU Usage on the Notebook?,Notebook,User Guide", @@ -2272,7 +2252,7 @@ { "uri":"modelarts_21_0073.html", "product_code":"modelarts", - "code":"228", + "code":"226", "des":"When creating a notebook instance, select the target Python development environment. Python2 and Python3 are supported, corresponding to Python 2.7 and Python 3.6, respec", "doc_type":"usermanual", "kw":"What Python Development Environments Does Notebook Support?,Notebook,User Guide", @@ -2282,7 +2262,7 @@ { "uri":"modelarts_21_0074.html", "product_code":"modelarts", - "code":"229", + "code":"227", "des":"The python2 environment of ModelArts supports Caffe, but the python3 environment does not support it.", "doc_type":"usermanual", "kw":"Does ModelArts Support the Caffe Engine?,Notebook,User Guide", @@ -2292,7 +2272,7 @@ { "uri":"modelarts_21_0075.html", "product_code":"modelarts", - "code":"230", + "code":"228", "des":"For security purposes, notebook instances do not support sudo privilege escalation.", "doc_type":"usermanual", "kw":"Is sudo Privilege Escalation Supported?,Notebook,User Guide", @@ -2302,7 +2282,7 @@ { "uri":"modelarts_05_0030.html", "product_code":"modelarts", - "code":"231", + "code":"229", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Training Jobs", @@ -2312,7 +2292,7 @@ { "uri":"modelarts_05_0031.html", "product_code":"modelarts", - "code":"232", + "code":"230", "des":"The code directory for creating a training job has limits on the size and number of files.Delete the files except the code from the code directory or save the files in ot", "doc_type":"usermanual", "kw":"What Can I Do If the Message \"Object directory size/quantity exceeds the limit\" Is Displayed When I ", @@ -2322,7 +2302,7 @@ { "uri":"modelarts_05_0032.html", "product_code":"modelarts", - "code":"233", + "code":"231", "des":"When you use ModelArts, your data is stored in the OBS bucket. The data has a corresponding OBS path, for example, bucket_name/dir/image.jpg. ModelArts training jobs run ", "doc_type":"usermanual", "kw":"What Can I Do If \"No such file or directory\" Is Displayed In the Training Job Log?,Training Jobs,Use", @@ -2332,7 +2312,7 @@ { "uri":"modelarts_05_0063.html", "product_code":"modelarts", - "code":"234", + "code":"232", "des":"When a model references a dependency package, select a frequently-used framework to create training jobs. In addition, place the required file or installation package in ", "doc_type":"usermanual", "kw":"How Do I Create a Training Job When a Dependency Package Is Referenced in a Model?,Training Jobs,Use", @@ -2342,7 +2322,7 @@ { "uri":"modelarts_21_0077.html", "product_code":"modelarts", - "code":"235", + "code":"233", "des":"Pay attention to the following when setting training parameters:When setting running parameters for creating a training job, you only need to set the corresponding parame", "doc_type":"usermanual", "kw":"What Should I Know When Setting Training Parameters?,Training Jobs,User Guide", @@ -2352,7 +2332,7 @@ { "uri":"modelarts_21_0078.html", "product_code":"modelarts", - "code":"236", + "code":"234", "des":"In the left navigation pane of the ModelArts management console, choose Training Management > Training Jobs to go to the Training Jobs page. In the training job list, cli", "doc_type":"usermanual", "kw":"How Do I Check Resource Usage of a Training Job?,Training Jobs,User Guide", @@ -2362,8 +2342,8 @@ { "uri":"modelarts_05_0090.html", "product_code":"modelarts", - "code":"237", - "des":"When creating a training job, you can select CPU, GPU, or Ascend resources based on the size of the training job.ModelArts mounts the disk to the /cache directory. You ca", + "code":"235", + "des":"When creating a training job, you can select CPU, GPUresources based on the size of the training job.ModelArts mounts the disk to the /cache directory. You can use this d", "doc_type":"usermanual", "kw":"What Are Sizes of the /cache Directories for Different Resource Specifications in the Training Envir", "title":"What Are Sizes of the /cache Directories for Different Resource Specifications in the Training Environment?", @@ -2372,7 +2352,7 @@ { "uri":"modelarts_21_0079.html", "product_code":"modelarts", - "code":"238", + "code":"236", "des":"In the script of the training job boot file, run the following commands to obtain the sizes of the to-be-copied and copied folders. Then determine whether folder copy is ", "doc_type":"usermanual", "kw":"How Do I Check Whether Folder Copy Is Complete During Job Training?,Training Jobs,User Guide", @@ -2382,7 +2362,7 @@ { "uri":"modelarts_21_0080.html", "product_code":"modelarts", - "code":"239", + "code":"237", "des":"Training job parameters can be automatically generated in the background or manually entered by users. Perform the following operations to obtain training job parameters:", "doc_type":"usermanual", "kw":"How Do I Obtain Training Job Parameters from the Boot File of the Training Job?,Training Jobs,User G", @@ -2392,7 +2372,7 @@ { "uri":"modelarts_21_0081.html", "product_code":"modelarts", - "code":"240", + "code":"238", "des":"ModelArts does not support access to the background of a training job.", "doc_type":"usermanual", "kw":"How Do I Access the Background of a Training Job?,Training Jobs,User Guide", @@ -2402,7 +2382,7 @@ { "uri":"modelarts_21_0082.html", "product_code":"modelarts", - "code":"241", + "code":"239", "des":"Storage directories of ModelArts training jobs do not affect each other. Environments are isolated from each other, and data of other jobs cannot be viewed.", "doc_type":"usermanual", "kw":"Is There Any Conflict When Models of Two Training Jobs Are Saved in the Same Directory of a Containe", @@ -2412,7 +2392,7 @@ { "uri":"modelarts_21_0083.html", "product_code":"modelarts", - "code":"242", + "code":"240", "des":"In a training job, only three valid digits are retained in a training output log. When the value of loss is too small, the value is displayed as 0.000. Log content is as ", "doc_type":"usermanual", "kw":"Only Three Valid Digits Are Retained in a Training Output Log. Can the Value of loss Be Changed?,Tra", @@ -2422,7 +2402,7 @@ { "uri":"modelarts_21_0084.html", "product_code":"modelarts", - "code":"243", + "code":"241", "des":"If you cannot access the corresponding folder by using os.system('cd xxx') in the boot script of the training job, you are advised to use the following method:", "doc_type":"usermanual", "kw":"Why Can't I Use os.system ('cd xxx') to Access the Corresponding Folder During Job Training?,Trainin", @@ -2432,7 +2412,7 @@ { "uri":"modelarts_21_0085.html", "product_code":"modelarts", - "code":"244", + "code":"242", "des":"ModelArts enables you to invoke a shell script, and you can use Python to invoke .sh. The procedure is as follows:Upload the .sh script to an OBS bucket. For example, upl", "doc_type":"usermanual", "kw":"How Do I Invoke a Shell Script in a Training Job to Execute the .sh File?,Training Jobs,User Guide", @@ -2442,7 +2422,7 @@ { "uri":"modelarts_05_0016.html", "product_code":"modelarts", - "code":"245", + "code":"243", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Model Management", @@ -2452,7 +2432,7 @@ { "uri":"modelarts_21_0086.html", "product_code":"modelarts", - "code":"246", + "code":"244", "des":"ModelArts does not support the import of models in .h5 format. You can convert the models in .h5 format of Keras to the TensorFlow format and then import the models to Mo", "doc_type":"usermanual", "kw":"How Do I Import the .h5 Model of Keras to ModelArts?,Model Management,User Guide", @@ -2462,7 +2442,7 @@ { "uri":"modelarts_05_0124.html", "product_code":"modelarts", - "code":"247", + "code":"245", "des":"ModelArts allows you to upload local models to OBS or import models stored in OBS directly into ModelArts.For details about how to import a model from OBS, see Importing ", "doc_type":"usermanual", "kw":"How Do I Import a Model Downloaded from OBS to ModelArts?,Model Management,User Guide", @@ -2472,7 +2452,7 @@ { "uri":"modelarts_05_0017.html", "product_code":"modelarts", - "code":"248", + "code":"246", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Service Deployment", @@ -2482,8 +2462,8 @@ { "uri":"modelarts_05_0012.html", "product_code":"modelarts", - "code":"249", - "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", + "code":"247", + "des":"Currently, models can only be deployed as real-time services and batch services.", "doc_type":"usermanual", "kw":"What Types of Services Can Models Be Deployed as on ModelArts?,Service Deployment,User Guide", "title":"What Types of Services Can Models Be Deployed as on ModelArts?", @@ -2492,7 +2472,7 @@ { "uri":"modelarts_05_0100.html", "product_code":"modelarts", - "code":"250", + "code":"248", "des":"Before importing a model, you need to place the corresponding inference code and configuration file in the model folder. When encoding with Python, you are advised to use", "doc_type":"usermanual", "kw":"What Should I Do If a Conflict Occurs When Deploying a Model As a Real-Time Service?,Service Deploym", @@ -2502,7 +2482,7 @@ { "uri":"modelarts_04_0099.html", "product_code":"modelarts", - "code":"251", + "code":"249", "des":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "doc_type":"usermanual", "kw":"Change History,User Guide", diff --git a/docs/modelarts/umn/CLASS.TXT.json b/docs/modelarts/umn/CLASS.TXT.json index fb1da5400..61f078bef 100644 --- a/docs/modelarts/umn/CLASS.TXT.json +++ b/docs/modelarts/umn/CLASS.TXT.json @@ -108,7 +108,7 @@ "code":"12" }, { - "desc":"ModelArts uses Object Storage Service (OBS) to store data and model backups and snapshots. OBS provides secure, reliable, low-cost storage. For more details, see Object S", + "desc":"ModelArts uses Identity and Access Management (IAM) for authentication and authorization. For more information about IAM, see Identity and Access Management User Guide.Mo", "product_code":"modelarts", "title":"Related Services", "uri":"modelarts_01_0006.html", @@ -494,15 +494,6 @@ "p_code":"53", "code":"55" }, - { - "desc":"Training a model uses a large number of labeled images. Therefore, label images before the model training. You can label images on the ModelArts management console. Alter", - "product_code":"modelarts", - "title":"Image Segmentation", - "uri":"modelarts_23_0345.html", - "doc_type":"usermanual", - "p_code":"53", - "code":"56" - }, { "desc":"Model training requires a large amount of labeled data. Therefore, before the model training, add labels to the files that are not labeled. In addition, you can modify, d", "product_code":"modelarts", @@ -510,7 +501,7 @@ "uri":"modelarts_23_0013.html", "doc_type":"usermanual", "p_code":"53", - "code":"57" + "code":"56" }, { "desc":"Named entity recognition assigns labels to named entities in text, such as time and locations. Before labeling, you need to understand the following:A label name can cont", @@ -519,7 +510,7 @@ "uri":"modelarts_23_0014.html", "doc_type":"usermanual", "p_code":"53", - "code":"58" + "code":"57" }, { "desc":"Triplet labeling is suitable for scenarios where structured information, such as subjects, predicates, and objects, needs to be labeled in statements. With this function,", @@ -528,7 +519,7 @@ "uri":"modelarts_23_0211.html", "doc_type":"usermanual", "p_code":"53", - "code":"59" + "code":"58" }, { "desc":"Model training requires a large amount of labeled data. Therefore, before the model training, label the unlabeled audio files. ModelArts enables you to label audio files ", @@ -537,7 +528,7 @@ "uri":"modelarts_23_0015.html", "doc_type":"usermanual", "p_code":"53", - "code":"60" + "code":"59" }, { "desc":"Model training requires a large amount of labeled data. Therefore, before the model training, label the unlabeled audio files. ModelArts enables you to label audio files ", @@ -546,7 +537,7 @@ "uri":"modelarts_23_0016.html", "doc_type":"usermanual", "p_code":"53", - "code":"61" + "code":"60" }, { "desc":"Model training requires a large amount of labeled data. Therefore, before the model training, label the unlabeled audio files. ModelArts enables you to label audio files.", @@ -555,16 +546,7 @@ "uri":"modelarts_23_0017.html", "doc_type":"usermanual", "p_code":"53", - "code":"62" - }, - { - "desc":"Model training requires a large amount of labeled video data. Therefore, before the model training, label the unlabeled video files. ModelArts enables you to label video ", - "product_code":"modelarts", - "title":"Video Labeling", - "uri":"modelarts_23_0282.html", - "doc_type":"usermanual", - "p_code":"53", - "code":"63" + "code":"61" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -573,7 +555,7 @@ "uri":"modelarts_23_0005.html", "doc_type":"usermanual", "p_code":"50", - "code":"64" + "code":"62" }, { "desc":"After a dataset is created, you can directly synchronize data from the dataset. Alternatively, you can import more data by importing the dataset. Data can be imported fro", @@ -581,8 +563,8 @@ "title":"Import Operation", "uri":"modelarts_23_0006.html", "doc_type":"usermanual", - "p_code":"64", - "code":"65" + "p_code":"62", + "code":"63" }, { "desc":"When a dataset is imported, the data storage directory and file name must comply with the ModelArts specifications if the data to be used is stored in OBS.Only the follow", @@ -590,8 +572,8 @@ "title":"Specifications for Importing Data from an OBS Directory", "uri":"modelarts_23_0008.html", "doc_type":"usermanual", - "p_code":"64", - "code":"66" + "p_code":"62", + "code":"64" }, { "desc":"The manifest file defines the mapping between labeling objects and content. The Manifest file import mode means that the manifest file is used for dataset import. The man", @@ -599,8 +581,8 @@ "title":"Specifications for Importing the Manifest File", "uri":"modelarts_23_0009.html", "doc_type":"usermanual", - "p_code":"64", - "code":"67" + "p_code":"62", + "code":"65" }, { "desc":"A dataset includes labeled and unlabeled data. You can select images or filter data based on the filter criteria and export to a new dataset or the specified OBS director", @@ -609,7 +591,7 @@ "uri":"modelarts_23_0214.html", "doc_type":"usermanual", "p_code":"50", - "code":"68" + "code":"66" }, { "desc":"For a created dataset, you can modify its basic information to match service changes.You have created a dataset.Log in to the ModelArts management console. In the left na", @@ -618,7 +600,7 @@ "uri":"modelarts_23_0020.html", "doc_type":"usermanual", "p_code":"50", - "code":"69" + "code":"67" }, { "desc":"ModelArts distinguishes data of the same source according to versions labeled at different time, which facilitates the selection of dataset versions during subsequent mod", @@ -627,7 +609,7 @@ "uri":"modelarts_23_0018.html", "doc_type":"usermanual", "p_code":"50", - "code":"70" + "code":"68" }, { "desc":"If a dataset is no longer in use, you can delete it to release resources.After a dataset is deleted, if you need to delete the data in the dataset input and output paths ", @@ -636,7 +618,7 @@ "uri":"modelarts_23_0021.html", "doc_type":"usermanual", "p_code":"50", - "code":"71" + "code":"69" }, { "desc":"After labeling data, you can publish the dataset to multiple versions for management. For the published versions, you can view the dataset version updates, set the curren", @@ -645,8 +627,53 @@ "uri":"modelarts_23_0019.html", "doc_type":"usermanual", "p_code":"50", + "code":"70" + }, + { + "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", + "product_code":"modelarts", + "title":"Team Labeling", + "uri":"modelarts_23_0180.html", + "doc_type":"usermanual", + "p_code":"50", + "code":"71" + }, + { + "desc":"Generally, a small data labeling task can be completed by an individual. However, team work is required to label a large dataset. ModelArts provides the team labeling fun", + "product_code":"modelarts", + "title":"Introduction to Team Labeling", + "uri":"modelarts_23_0181.html", + "doc_type":"usermanual", + "p_code":"71", "code":"72" }, + { + "desc":"Team labeling is managed in a unit of teams. To enable team labeling for a dataset, a team must be specified. Multiple members can be added to a team.An account can have ", + "product_code":"modelarts", + "title":"Team Management", + "uri":"modelarts_23_0182.html", + "doc_type":"usermanual", + "p_code":"71", + "code":"73" + }, + { + "desc":"There is no member in a new team. You need to add members who will participate in a team labeling task.A maximum of 100 members can be added to a team. If there are more ", + "product_code":"modelarts", + "title":"Member Management", + "uri":"modelarts_23_0183.html", + "doc_type":"usermanual", + "p_code":"71", + "code":"74" + }, + { + "desc":"For datasets with team labeling enabled, you can create team labeling tasks and assign the labeling tasks to different teams so that team members can complete the labelin", + "product_code":"modelarts", + "title":"Managing Team Labeling Tasks", + "uri":"modelarts_23_0210.html", + "doc_type":"usermanual", + "p_code":"71", + "code":"75" + }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "product_code":"modelarts", @@ -654,7 +681,7 @@ "uri":"modelarts_23_0032.html", "doc_type":"usermanual", "p_code":"", - "code":"73" + "code":"76" }, { "desc":"ModelArts integrates the open-source Jupyter Notebook to provide you with online interactive development and debugging environments. You can use the Notebook on the Model", @@ -662,8 +689,8 @@ "title":"Introduction to Notebook", "uri":"modelarts_23_0033.html", "doc_type":"usermanual", - "p_code":"73", - "code":"74" + "p_code":"76", + "code":"77" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -671,8 +698,8 @@ "title":"Managing Notebook Instances", "uri":"modelarts_23_0111.html", "doc_type":"usermanual", - "p_code":"73", - "code":"75" + "p_code":"76", + "code":"78" }, { "desc":"Before developing a model, create a notebook instance, open it, and perform encoding.You will be charged as long as your notebook instance is in the Running status. We re", @@ -680,8 +707,8 @@ "title":"Creating a Notebook Instance", "uri":"modelarts_23_0034.html", "doc_type":"usermanual", - "p_code":"75", - "code":"76" + "p_code":"78", + "code":"79" }, { "desc":"You can open a created notebook instance (that is, an instance in the Running state) and start coding in the development environment.Go to the Jupyter Notebook page.In th", @@ -689,8 +716,8 @@ "title":"Opening a Notebook Instance", "uri":"modelarts_23_0325.html", "doc_type":"usermanual", - "p_code":"75", - "code":"77" + "p_code":"78", + "code":"80" }, { "desc":"You can stop unwanted notebook instances to prevent unnecessary fees. You can also start a notebook instance that is in the Stopped state to use it again.Log in to the Mo", @@ -698,8 +725,8 @@ "title":"Starting or Stopping a Notebook Instance", "uri":"modelarts_23_0041.html", "doc_type":"usermanual", - "p_code":"75", - "code":"78" + "p_code":"78", + "code":"81" }, { "desc":"You can delete notebook instances that are no longer used to release resources.Log in to the ModelArts management console. In the left navigation pane, choose DevEnviron ", @@ -707,8 +734,8 @@ "title":"Deleting a Notebook Instance", "uri":"modelarts_23_0042.html", "doc_type":"usermanual", - "p_code":"75", - "code":"79" + "p_code":"78", + "code":"82" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -716,8 +743,8 @@ "title":"Using Jupyter Notebook", "uri":"modelarts_23_0035.html", "doc_type":"usermanual", - "p_code":"73", - "code":"80" + "p_code":"76", + "code":"83" }, { "desc":"Jupyter Notebook is a web-based application for interactive computing. It can be applied to full-process computing: development, documentation, running code, and presenti", @@ -725,8 +752,8 @@ "title":"Introduction to Jupyter Notebook", "uri":"modelarts_23_0326.html", "doc_type":"usermanual", - "p_code":"80", - "code":"81" + "p_code":"83", + "code":"84" }, { "desc":"This section describes common operations on Jupyter Notebook.In the notebook instance list, locate the row where the target notebook instance resides and click Open in th", @@ -734,8 +761,8 @@ "title":"Common Operations on Jupyter Notebook", "uri":"modelarts_23_0120.html", "doc_type":"usermanual", - "p_code":"80", - "code":"82" + "p_code":"83", + "code":"85" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -743,8 +770,8 @@ "title":"Configuring the Jupyter Notebook Environment", "uri":"modelarts_23_0327.html", "doc_type":"usermanual", - "p_code":"80", - "code":"83" + "p_code":"83", + "code":"86" }, { "desc":"For developers who are used to coding, the terminal function is very convenient and practical. This section describes how to enable the terminal function in a notebook in", @@ -752,8 +779,8 @@ "title":"Using the Notebook Terminal Function", "uri":"modelarts_23_0117.html", "doc_type":"usermanual", - "p_code":"83", - "code":"84" + "p_code":"86", + "code":"87" }, { "desc":"For a GPU-based notebook instance, you can switch different versions of CUDA on the Terminal page of Jupyter.CPU-based notebook instances do not use CUDA. Therefore, the ", @@ -761,8 +788,8 @@ "title":"Switching the CUDA Version on the Terminal Page of a GPU-based Notebook Instance", "uri":"modelarts_23_0280.html", "doc_type":"usermanual", - "p_code":"83", - "code":"85" + "p_code":"86", + "code":"88" }, { "desc":"Multiple environments have been installed in ModelArts notebook instances, including TensorFlow. You can use pip install to install external libraries from a Jupyter note", @@ -770,8 +797,8 @@ "title":"Installing External Libraries and Kernels in Notebook Instances", "uri":"modelarts_23_0040.html", "doc_type":"usermanual", - "p_code":"83", - "code":"86" + "p_code":"86", + "code":"89" }, { "desc":"In notebook instances, you can use ModelArts SDKs to manage OBS, training jobs, models, and real-time services.For details about how to use ModelArts SDKs, see ModelArts ", @@ -779,8 +806,8 @@ "title":"Using ModelArts SDKs", "uri":"modelarts_23_0039.html", "doc_type":"usermanual", - "p_code":"80", - "code":"87" + "p_code":"83", + "code":"90" }, { "desc":"If you specify Storage Path during notebook instance creation, your compiled code will be automatically stored in your specified OBS bucket. If code invocation among diff", @@ -788,8 +815,8 @@ "title":"Synchronizing Files with OBS", "uri":"modelarts_23_0038.html", "doc_type":"usermanual", - "p_code":"80", - "code":"88" + "p_code":"83", + "code":"91" }, { "desc":"After code compiling is finished, you can save the entered code as a .py file which can be used for starting training jobs.Create and open a notebook instance or open an ", @@ -797,8 +824,8 @@ "title":"Using the Convert to Python File Function", "uri":"modelarts_23_0037.html", "doc_type":"usermanual", - "p_code":"80", - "code":"89" + "p_code":"83", + "code":"92" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -806,8 +833,8 @@ "title":"Using JupyterLab", "uri":"modelarts_23_0330.html", "doc_type":"usermanual", - "p_code":"73", - "code":"90" + "p_code":"76", + "code":"93" }, { "desc":"JupyterLab is an interactive development environment. It is a next-generation product of Jupyter Notebook. JupyterLab enables you to compile notebooks, operate terminals,", @@ -815,8 +842,8 @@ "title":"Introduction to JupyterLab and Common Operations", "uri":"modelarts_23_0209.html", "doc_type":"usermanual", - "p_code":"90", - "code":"91" + "p_code":"93", + "code":"94" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -824,8 +851,8 @@ "title":"Uploading and Downloading Data", "uri":"modelarts_23_0331.html", "doc_type":"usermanual", - "p_code":"90", - "code":"92" + "p_code":"93", + "code":"95" }, { "desc":"On the JupyterLab page, click Upload Files to upload a file. For details, see Uploading a File in Introduction to JupyterLab and Common Operations. If a message is displa", @@ -833,8 +860,8 @@ "title":"Uploading Data to JupyterLab", "uri":"modelarts_23_0332.html", "doc_type":"usermanual", - "p_code":"92", - "code":"93" + "p_code":"95", + "code":"96" }, { "desc":"Only files within 100 MB in JupyterLab can be downloaded to a local PC. You can perform operations in different scenarios based on the storage location selected when crea", @@ -842,8 +869,8 @@ "title":"Downloading a File from JupyterLab", "uri":"modelarts_23_0333.html", "doc_type":"usermanual", - "p_code":"92", - "code":"94" + "p_code":"95", + "code":"97" }, { "desc":"In notebook instances, you can use ModelArts SDKs to manage OBS, training jobs, models, and real-time services.For details about how to use ModelArts SDKs, see ModelArts ", @@ -851,8 +878,8 @@ "title":"Using ModelArts SDKs", "uri":"modelarts_23_0335.html", "doc_type":"usermanual", - "p_code":"90", - "code":"95" + "p_code":"93", + "code":"98" }, { "desc":"If you specify Storage Path during notebook instance creation, your compiled code will be automatically stored in your specified OBS bucket. If code invocation among diff", @@ -860,8 +887,8 @@ "title":"Synchronizing Files with OBS", "uri":"modelarts_23_0336.html", "doc_type":"usermanual", - "p_code":"90", - "code":"96" + "p_code":"93", + "code":"99" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -870,7 +897,7 @@ "uri":"modelarts_23_0043.html", "doc_type":"usermanual", "p_code":"", - "code":"97" + "code":"100" }, { "desc":"ModelArts provides model training for you to view the training effect, based on which you can adjust your model parameters. You can select resource pools (CPU or GPU) wit", @@ -878,8 +905,8 @@ "title":"Introduction to Model Training", "uri":"modelarts_23_0044.html", "doc_type":"usermanual", - "p_code":"97", - "code":"98" + "p_code":"100", + "code":"101" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -887,8 +914,8 @@ "title":"Built-in Algorithms", "uri":"modelarts_23_0156.html", "doc_type":"usermanual", - "p_code":"97", - "code":"99" + "p_code":"100", + "code":"102" }, { "desc":"Based on the frequently-used AI engines in the industry, ModelArts provides built-in algorithms to meet a wide range of your requirements. You can directly select the alg", @@ -896,8 +923,8 @@ "title":"Introduction to Built-in Algorithms", "uri":"modelarts_23_0045.html", "doc_type":"usermanual", - "p_code":"99", - "code":"100" + "p_code":"102", + "code":"103" }, { "desc":"The built-in algorithms provided by ModelArts can be used for image classification, object detection, and image semantic segmentation. The requirements for the datasets v", @@ -905,8 +932,8 @@ "title":"Requirements on Datasets", "uri":"modelarts_23_0157.html", "doc_type":"usermanual", - "p_code":"99", - "code":"101" + "p_code":"102", + "code":"104" }, { "desc":"This section describes the built-in algorithms supported by ModelArts and the running parameters supported by each algorithm. You can set running parameters for a trainin", @@ -914,8 +941,8 @@ "title":"Algorithms and Their Running Parameters", "uri":"modelarts_23_0158.html", "doc_type":"usermanual", - "p_code":"99", - "code":"102" + "p_code":"102", + "code":"105" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -923,8 +950,8 @@ "title":"Creating a Training Job", "uri":"modelarts_23_0235.html", "doc_type":"usermanual", - "p_code":"97", - "code":"103" + "p_code":"100", + "code":"106" }, { "desc":"ModelArts supports multiple types of training jobs during the entire AI development process. Select a creation mode based on the algorithm source.Built-inIf you do not kn", @@ -932,8 +959,8 @@ "title":"Introduction to Training Jobs", "uri":"modelarts_23_0046.html", "doc_type":"usermanual", - "p_code":"103", - "code":"104" + "p_code":"106", + "code":"107" }, { "desc":"If you do not have the algorithm development capability, you can use the built-in algorithms of ModelArts. After simple parameter adjustment, you can create a training jo", @@ -941,8 +968,8 @@ "title":"Using Built-in Algorithms to Train Models", "uri":"modelarts_23_0237.html", "doc_type":"usermanual", - "p_code":"103", - "code":"105" + "p_code":"106", + "code":"108" }, { "desc":"If you use frequently-used frameworks, such as TensorFlow and MXNet, to develop algorithms locally, you can select Frequently-used to create training jobs and build model", @@ -950,8 +977,8 @@ "title":"Using Frequently-used Frameworks to Train Models", "uri":"modelarts_23_0238.html", "doc_type":"usermanual", - "p_code":"103", - "code":"106" + "p_code":"106", + "code":"109" }, { "desc":"If the framework used for algorithm development is not a frequently-used framework, you can build an algorithm into a custom image and use the custom image to create a tr", @@ -959,8 +986,8 @@ "title":"Using Custom Images to Train Models", "uri":"modelarts_23_0239.html", "doc_type":"usermanual", - "p_code":"103", - "code":"107" + "p_code":"106", + "code":"110" }, { "desc":"In the training job list, click Stop in the Operation column for a training job in the Running state to stop a running training job.If you have selected Save Training Par", @@ -968,8 +995,8 @@ "title":"Stopping or Deleting a Job", "uri":"modelarts_23_0159.html", "doc_type":"usermanual", - "p_code":"97", - "code":"108" + "p_code":"100", + "code":"111" }, { "desc":"During model building, you may need to frequently tune the data, training parameters, or the model based on the training results to obtain a satisfactory model. ModelArts", @@ -977,8 +1004,8 @@ "title":"Managing Training Job Versions", "uri":"modelarts_23_0047.html", "doc_type":"usermanual", - "p_code":"97", - "code":"109" + "p_code":"100", + "code":"112" }, { "desc":"After a training job finishes, you can manage the training job versions and check whether the training result of the job is satisfactory by viewing the job details.In the", @@ -986,8 +1013,8 @@ "title":"Viewing Job Details", "uri":"modelarts_23_0048.html", "doc_type":"usermanual", - "p_code":"97", - "code":"110" + "p_code":"100", + "code":"113" }, { "desc":"You can store the parameter settings in ModelArts during job creation so that you can use the stored settings to create follow-up training jobs, which makes job creation ", @@ -995,8 +1022,8 @@ "title":"Managing Job Parameters", "uri":"modelarts_23_0049.html", "doc_type":"usermanual", - "p_code":"97", - "code":"111" + "p_code":"100", + "code":"114" }, { "desc":"You can create visualization jobs of TensorBoard and MindInsight types on ModelArts.TensorBoard supports training jobs based on the TensorFlow engine, and MindInsight sup", @@ -1004,8 +1031,8 @@ "title":"Managing Visualization Jobs", "uri":"modelarts_23_0050.html", "doc_type":"usermanual", - "p_code":"97", - "code":"112" + "p_code":"100", + "code":"115" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1014,7 +1041,7 @@ "uri":"modelarts_23_0051.html", "doc_type":"usermanual", "p_code":"", - "code":"113" + "code":"116" }, { "desc":"AI model development and optimization require frequent iterations and debugging. Changes in datasets, training code, or parameters may affect the quality of models. If th", @@ -1022,8 +1049,8 @@ "title":"Introduction to Model Management", "uri":"modelarts_23_0052.html", "doc_type":"usermanual", - "p_code":"113", - "code":"114" + "p_code":"116", + "code":"117" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1031,8 +1058,8 @@ "title":"Importing a Model", "uri":"modelarts_23_0204.html", "doc_type":"usermanual", - "p_code":"113", - "code":"115" + "p_code":"116", + "code":"118" }, { "desc":"You can create a training job on ModelArts and perform training to obtain a satisfactory model. Then import the model to Model Management for unified management. In addit", @@ -1040,8 +1067,8 @@ "title":"Importing a Meta Model from a Training Job", "uri":"modelarts_23_0054.html", "doc_type":"usermanual", - "p_code":"115", - "code":"116" + "p_code":"118", + "code":"119" }, { "desc":"Because the configurations of models with the same functions are similar, ModelArts integrates the configurations of such models into a common template. By using this tem", @@ -1049,8 +1076,8 @@ "title":"Importing a Meta Model from a Template", "uri":"modelarts_23_0205.html", "doc_type":"usermanual", - "p_code":"115", - "code":"117" + "p_code":"118", + "code":"120" }, { "desc":"For AI engines that are not supported by ModelArts, you can import the model you compile to ModelArts from custom images.For details about the specifications and descript", @@ -1058,8 +1085,8 @@ "title":"Importing a Meta Model from a Container Image", "uri":"modelarts_23_0206.html", "doc_type":"usermanual", - "p_code":"115", - "code":"118" + "p_code":"118", + "code":"121" }, { "desc":"In scenarios where frequently-used frameworks are used for model development and training, you can import the model to ModelArts for unified management.The model has been", @@ -1067,8 +1094,8 @@ "title":"Importing a Meta Model from OBS", "uri":"modelarts_23_0207.html", "doc_type":"usermanual", - "p_code":"115", - "code":"119" + "p_code":"118", + "code":"122" }, { "desc":"To facilitate source tracing and repeated model tuning, ModelArts provides the model version management function. You can manage models based on versions.You have importe", @@ -1076,54 +1103,9 @@ "title":"Managing Model Versions", "uri":"modelarts_23_0055.html", "doc_type":"usermanual", - "p_code":"113", - "code":"120" - }, - { - "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", - "product_code":"modelarts", - "title":"Model Compression and Conversion", - "uri":"modelarts_23_0106.html", - "doc_type":"usermanual", - "p_code":"113", - "code":"121" - }, - { - "desc":"To obtain higher computing power, you can deploy the models created on ModelArts or a local PC on the Ascend chip. In this case, you need to compress or convert the model", - "product_code":"modelarts", - "title":"Compressing and Converting Models", - "uri":"modelarts_23_0107.html", - "doc_type":"usermanual", - "p_code":"121", - "code":"122" - }, - { - "desc":"During model conversion, the model input directory must comply with certain specifications. This section describes how to upload your model package to OBS.The requirement", - "product_code":"modelarts", - "title":"Model Input Path Specifications", - "uri":"modelarts_23_0108.html", - "doc_type":"usermanual", - "p_code":"121", + "p_code":"116", "code":"123" }, - { - "desc":"The following describes the output path of the model run on the Ascend chip after conversion:For TensorFlow-based models, the output path must comply with the following s", - "product_code":"modelarts", - "title":"Model Output Path Description", - "uri":"modelarts_23_0109.html", - "doc_type":"usermanual", - "p_code":"121", - "code":"124" - }, - { - "desc":"ModelArts provides the following conversion templates based on different AI frameworks:TF-FrozenGraph-To-Ascend-C32Convert the model trained by the TensorFlow framework a", - "product_code":"modelarts", - "title":"Conversion Templates", - "uri":"modelarts_23_0110.html", - "doc_type":"usermanual", - "p_code":"121", - "code":"125" - }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", "product_code":"modelarts", @@ -1131,7 +1113,7 @@ "uri":"modelarts_23_0057.html", "doc_type":"usermanual", "p_code":"", - "code":"126" + "code":"124" }, { "desc":"After a training job is complete and a model is generated, you can deploy the model on the Service Deployment page. You can also deploy the model imported from OBS. Model", @@ -1139,8 +1121,8 @@ "title":"Introduction to Model Deployment", "uri":"modelarts_23_0058.html", "doc_type":"usermanual", - "p_code":"126", - "code":"127" + "p_code":"124", + "code":"125" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1148,8 +1130,8 @@ "title":"Real-Time Services", "uri":"modelarts_23_0059.html", "doc_type":"usermanual", - "p_code":"126", - "code":"128" + "p_code":"124", + "code":"126" }, { "desc":"After a model is prepared, you can deploy the model as a real-time service and predict and call the service.A maximum of one real-time service can be deployed.Data has be", @@ -1157,8 +1139,8 @@ "title":"Deploying a Model as a Real-Time Service", "uri":"modelarts_23_0060.html", "doc_type":"usermanual", - "p_code":"128", - "code":"129" + "p_code":"126", + "code":"127" }, { "desc":"After a model is deployed as a real-time service, you can access the service page to view its details.Log in to the ModelArts management console and choose Service Deploy", @@ -1166,8 +1148,8 @@ "title":"Viewing Service Details", "uri":"modelarts_23_0061.html", "doc_type":"usermanual", - "p_code":"128", - "code":"130" + "p_code":"126", + "code":"128" }, { "desc":"After a model is deployed as a real-time service, you can debug code or add files for testing on the Prediction tab page. Based on the input request (JSON text or file) d", @@ -1175,8 +1157,8 @@ "title":"Testing a Service", "uri":"modelarts_23_0062.html", "doc_type":"usermanual", - "p_code":"128", - "code":"131" + "p_code":"126", + "code":"129" }, { "desc":"If a real-time service is in the Running state, the real-time service has been deployed successfully. This service provides a standard RESTful API for users to call. Befo", @@ -1184,8 +1166,8 @@ "title":"Accessing a Real-Time Service (Token-based Authentication)", "uri":"modelarts_23_0063.html", "doc_type":"usermanual", - "p_code":"128", - "code":"132" + "p_code":"126", + "code":"130" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1193,8 +1175,8 @@ "title":"Batch Services", "uri":"modelarts_23_0065.html", "doc_type":"usermanual", - "p_code":"126", - "code":"133" + "p_code":"124", + "code":"131" }, { "desc":"After a model is prepared, you can deploy it as a batch service. The Service Deployment > Batch Services page lists all batch services. You can enter a service name in th", @@ -1202,8 +1184,8 @@ "title":"Deploying a Model as a Batch Service", "uri":"modelarts_23_0066.html", "doc_type":"usermanual", - "p_code":"133", - "code":"134" + "p_code":"131", + "code":"132" }, { "desc":"When deploying a batch service, you can select the location of the output data directory. You can view the running result of the batch service that is in the Running comp", @@ -1211,8 +1193,8 @@ "title":"Viewing the Batch Service Prediction Result", "uri":"modelarts_23_0067.html", "doc_type":"usermanual", - "p_code":"133", - "code":"135" + "p_code":"131", + "code":"133" }, { "desc":"For a deployed service, you can modify its basic information to match service changes. You can modify the basic information about a service in either of the following way", @@ -1220,8 +1202,8 @@ "title":"Modifying a Service", "uri":"modelarts_23_0071.html", "doc_type":"usermanual", - "p_code":"126", - "code":"136" + "p_code":"124", + "code":"134" }, { "desc":"You can start services in the Successful, Abnormal, or Stopped status. Services in the Deploying status cannot be started. A service is billed when it is started and in t", @@ -1229,8 +1211,8 @@ "title":"Starting or Stopping a Service", "uri":"modelarts_23_0072.html", "doc_type":"usermanual", - "p_code":"126", - "code":"137" + "p_code":"124", + "code":"135" }, { "desc":"If a service is no longer in use, you can delete it to release resources.Log in to the ModelArts management console and choose Service Deployment from the left navigation", @@ -1238,8 +1220,8 @@ "title":"Deleting a Service", "uri":"modelarts_23_0073.html", "doc_type":"usermanual", - "p_code":"126", - "code":"138" + "p_code":"124", + "code":"136" }, { "desc":"When using ModelArts to implement AI Development Lifecycle, you can use two different resource pools to train and deploy models.Public Resource Pool: provides public larg", @@ -1248,7 +1230,7 @@ "uri":"modelarts_23_0076.html", "doc_type":"usermanual", "p_code":"", - "code":"139" + "code":"137" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1257,7 +1239,7 @@ "uri":"modelarts_23_0083.html", "doc_type":"usermanual", "p_code":"", - "code":"140" + "code":"138" }, { "desc":"ModelArts provides multiple frequently-used built-in engines. However, when users have special requirements for the deep learning engine and development library, the buil", @@ -1265,8 +1247,8 @@ "title":"Introduction to Custom Images", "uri":"modelarts_23_0084.html", "doc_type":"usermanual", - "p_code":"140", - "code":"141" + "p_code":"138", + "code":"139" }, { "desc":"ModelArts allows you to use custom images to create training jobs and import models. Before creating and uploading a custom image, understand the following information:So", @@ -1274,8 +1256,8 @@ "title":"Creating and Uploading a Custom Image", "uri":"modelarts_23_0085.html", "doc_type":"usermanual", - "p_code":"140", - "code":"142" + "p_code":"138", + "code":"140" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1283,8 +1265,8 @@ "title":"For Training Models", "uri":"modelarts_23_0216.html", "doc_type":"usermanual", - "p_code":"140", - "code":"143" + "p_code":"138", + "code":"141" }, { "desc":"When creating an image using locally developed models and training scripts, ensure that they meet the specifications defined by ModelArts.Custom images cannot contain mal", @@ -1292,8 +1274,8 @@ "title":"Specifications for Custom Images Used for Training Jobs", "uri":"modelarts_23_0217.html", "doc_type":"usermanual", - "p_code":"143", - "code":"144" + "p_code":"141", + "code":"142" }, { "desc":"After creating and uploading a custom image to SWR, you can use the image to create a training job on the ModelArts management console to complete model training.You have", @@ -1301,8 +1283,8 @@ "title":"Creating a Training Job Using a Custom Image (GPU)", "uri":"modelarts_23_0087.html", "doc_type":"usermanual", - "p_code":"143", - "code":"145" + "p_code":"141", + "code":"143" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1310,8 +1292,8 @@ "title":"For Importing Models", "uri":"modelarts_23_0218.html", "doc_type":"usermanual", - "p_code":"140", - "code":"146" + "p_code":"138", + "code":"144" }, { "desc":"When creating an image using locally developed models, ensure that they meet the specifications defined by ModelArts.Custom images cannot contain malicious code.The size ", @@ -1319,8 +1301,8 @@ "title":"Specifications for Custom Images Used for Importing Models", "uri":"modelarts_23_0219.html", "doc_type":"usermanual", - "p_code":"146", - "code":"147" + "p_code":"144", + "code":"145" }, { "desc":"After creating and uploading a custom image to SWR, you can use the image to import a model and deploy the model as a service on the ModelArts management console.You have", @@ -1328,8 +1310,8 @@ "title":"Importing a Model Using a Custom Image", "uri":"modelarts_23_0086.html", "doc_type":"usermanual", - "p_code":"146", - "code":"148" + "p_code":"144", + "code":"146" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1338,7 +1320,7 @@ "uri":"modelarts_23_0090.html", "doc_type":"usermanual", "p_code":"", - "code":"149" + "code":"147" }, { "desc":"When you import models in Model Management, if the meta model is imported from OBS or a container image, the model package must meet the following specifications:The mode", @@ -1346,8 +1328,8 @@ "title":"Model Package Specifications", "uri":"modelarts_23_0091.html", "doc_type":"usermanual", - "p_code":"149", - "code":"150" + "p_code":"147", + "code":"148" }, { "desc":"A model developer needs to compile a configuration file when publishing a model. The model configuration file describes the model usage, computing framework, precision, i", @@ -1355,8 +1337,8 @@ "title":"Specifications for Compiling the Model Configuration File", "uri":"modelarts_23_0092.html", "doc_type":"usermanual", - "p_code":"149", - "code":"151" + "p_code":"147", + "code":"149" }, { "desc":"This section describes how to compile model inference code in ModelArts. The following also provides an example of inference code for the TensorFlow engine and an example", @@ -1364,8 +1346,8 @@ "title":"Specifications for Compiling Model Inference Code", "uri":"modelarts_23_0093.html", "doc_type":"usermanual", - "p_code":"149", - "code":"152" + "p_code":"147", + "code":"150" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1374,7 +1356,7 @@ "uri":"modelarts_23_0097.html", "doc_type":"usermanual", "p_code":"", - "code":"153" + "code":"151" }, { "desc":"Because the configurations of models with the same functions are similar, ModelArts integrates the configurations of such models into a common template. By using this tem", @@ -1382,8 +1364,8 @@ "title":"Introduction to Model Templates", "uri":"modelarts_23_0098.html", "doc_type":"usermanual", - "p_code":"153", - "code":"154" + "p_code":"151", + "code":"152" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1391,8 +1373,8 @@ "title":"Template Description", "uri":"modelarts_23_0118.html", "doc_type":"usermanual", - "p_code":"153", - "code":"155" + "p_code":"151", + "code":"153" }, { "desc":"AI engine: TensorFlow 1.8; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or appl", @@ -1400,8 +1382,8 @@ "title":"TensorFlow-py27 General Template", "uri":"modelarts_23_0161.html", "doc_type":"usermanual", - "p_code":"155", - "code":"156" + "p_code":"153", + "code":"154" }, { "desc":"AI engine: TensorFlow 1.8; Environment: Python 3.6; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or appl", @@ -1409,8 +1391,8 @@ "title":"TensorFlow-py36 General Template", "uri":"modelarts_23_0162.html", "doc_type":"usermanual", - "p_code":"155", - "code":"157" + "p_code":"153", + "code":"155" }, { "desc":"AI engine: MXNet 1.2.1; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or applica", @@ -1418,8 +1400,8 @@ "title":"MXNet-py27 General Template", "uri":"modelarts_23_0163.html", "doc_type":"usermanual", - "p_code":"155", - "code":"158" + "p_code":"153", + "code":"156" }, { "desc":"AI engine: MXNet 1.2.1; Environment: Python 3.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or applica", @@ -1427,8 +1409,8 @@ "title":"MXNet-py37 General Template", "uri":"modelarts_23_0164.html", "doc_type":"usermanual", - "p_code":"155", - "code":"159" + "p_code":"153", + "code":"157" }, { "desc":"AI engine: PyTorch 1.0; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or applica", @@ -1436,8 +1418,8 @@ "title":"PyTorch-py27 General Template", "uri":"modelarts_23_0165.html", "doc_type":"usermanual", - "p_code":"155", - "code":"160" + "p_code":"153", + "code":"158" }, { "desc":"AI engine: PyTorch 1.0; Environment: Python 3.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or applica", @@ -1445,8 +1427,8 @@ "title":"PyTorch-py37 General Template", "uri":"modelarts_23_0166.html", "doc_type":"usermanual", - "p_code":"155", - "code":"161" + "p_code":"153", + "code":"159" }, { "desc":"AI engine: CPU-based Caffe 1.0; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or", @@ -1454,8 +1436,8 @@ "title":"Caffe-CPU-py27 General Template", "uri":"modelarts_23_0167.html", "doc_type":"usermanual", - "p_code":"155", - "code":"162" + "p_code":"153", + "code":"160" }, { "desc":"AI engine: GPU-based Caffe 1.0; Environment: Python 2.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or", @@ -1463,8 +1445,8 @@ "title":"Caffe-GPU-py27 General Template", "uri":"modelarts_23_0168.html", "doc_type":"usermanual", - "p_code":"155", - "code":"163" + "p_code":"153", + "code":"161" }, { "desc":"AI engine: CPU-based Caffe 1.0; Environment: Python 3.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or", @@ -1472,8 +1454,8 @@ "title":"Caffe-CPU-py37 General Template", "uri":"modelarts_23_0169.html", "doc_type":"usermanual", - "p_code":"155", - "code":"164" + "p_code":"153", + "code":"162" }, { "desc":"AI engine: GPU-based Caffe 1.0; Environment: Python 3.7; Input and output mode: undefined mode. Select an appropriate input and output mode based on the model function or", @@ -1481,8 +1463,8 @@ "title":"Caffe-GPU-py37 General Template", "uri":"modelarts_23_0170.html", "doc_type":"usermanual", - "p_code":"155", - "code":"165" + "p_code":"153", + "code":"163" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1490,8 +1472,8 @@ "title":"Input and Output Modes", "uri":"modelarts_23_0099.html", "doc_type":"usermanual", - "p_code":"153", - "code":"166" + "p_code":"151", + "code":"164" }, { "desc":"This is a built-in input and output mode for object detection. The models using this mode are identified as object detection models. The prediction request path is /, the", @@ -1499,8 +1481,8 @@ "title":"Built-in Object Detection Mode", "uri":"modelarts_23_0100.html", "doc_type":"usermanual", - "p_code":"166", - "code":"167" + "p_code":"164", + "code":"165" }, { "desc":"The built-in image processing input and output mode can be applied to models such as image classification, object detection, and image semantic segmentation. The predicti", @@ -1508,8 +1490,8 @@ "title":"Built-in Image Processing Mode", "uri":"modelarts_23_0101.html", "doc_type":"usermanual", - "p_code":"166", - "code":"168" + "p_code":"164", + "code":"166" }, { "desc":"This is a built-in input and output mode for predictive analytics. The models using this mode are identified as predictive analytics models. The prediction request path i", @@ -1517,8 +1499,8 @@ "title":"Built-in Predictive Analytics Mode", "uri":"modelarts_23_0102.html", "doc_type":"usermanual", - "p_code":"166", - "code":"169" + "p_code":"164", + "code":"167" }, { "desc":"The undefined mode does not define the input and output mode. The input and output mode is determined by the model. Select this mode only when the existing input and outp", @@ -1526,8 +1508,8 @@ "title":"Undefined Mode", "uri":"modelarts_23_0103.html", "doc_type":"usermanual", - "p_code":"166", - "code":"170" + "p_code":"164", + "code":"168" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1536,7 +1518,7 @@ "uri":"modelarts_23_0172.html", "doc_type":"usermanual", "p_code":"", - "code":"171" + "code":"169" }, { "desc":"TensorFlow has two types of APIs: Keras and tf. Keras and tf use different code for training and saving models, but the same code for inference.", @@ -1544,8 +1526,8 @@ "title":"TensorFlow", "uri":"modelarts_23_0173.html", "doc_type":"usermanual", - "p_code":"171", - "code":"172" + "p_code":"169", + "code":"170" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1553,8 +1535,8 @@ "title":"PyTorch", "uri":"modelarts_23_0175.html", "doc_type":"usermanual", - "p_code":"171", - "code":"173" + "p_code":"169", + "code":"171" }, { "desc":"lenet_train_test.prototxt filelenet_solver.prototxt fileTrain the model.The caffemodel file is generated after model training. Rewrite the lenet_train_test.prototxt file ", @@ -1562,17 +1544,17 @@ "title":"Caffe", "uri":"modelarts_23_0176.html", "doc_type":"usermanual", - "p_code":"171", - "code":"174" + "p_code":"169", + "code":"172" }, { - "desc":"After the model is saved, it must be uploaded to the OBS directory before being published. The config.json and customize_service.py files must be contained during publish", + "desc":"Before training, download the iris.csv dataset, decompress it, and upload it to the /home/ma-user/work/ directory of the notebook instance. Download the iris.csv dataset ", "product_code":"modelarts", "title":"XGBoost", "uri":"modelarts_23_0177.html", "doc_type":"usermanual", - "p_code":"171", - "code":"175" + "p_code":"169", + "code":"173" }, { "desc":"After the model is saved, it must be uploaded to the OBS directory before being published. The config.json configuration and customize_service.py must be contained during", @@ -1580,17 +1562,17 @@ "title":"PySpark", "uri":"modelarts_23_0178.html", "doc_type":"usermanual", - "p_code":"171", - "code":"176" + "p_code":"169", + "code":"174" }, { - "desc":"After the model is saved, it must be uploaded to the OBS directory before being published. The config.json and customize_service.py files must be contained during publish", + "desc":"Before training, download the iris.csv dataset, decompress it, and upload it to the /home/ma-user/work/ directory of the notebook instance. Download the iris.csv dataset ", "product_code":"modelarts", "title":"Scikit Learn", "uri":"modelarts_23_0179.html", "doc_type":"usermanual", - "p_code":"171", - "code":"177" + "p_code":"169", + "code":"175" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1599,7 +1581,7 @@ "uri":"modelarts_23_0077.html", "doc_type":"usermanual", "p_code":"", - "code":"178" + "code":"176" }, { "desc":"A fine-grained policy is a set of permissions defining which operations on which cloud services can be performed. Each policy can define multiple permissions. After a pol", @@ -1607,8 +1589,8 @@ "title":"Basic Concepts", "uri":"modelarts_23_0078.html", "doc_type":"usermanual", - "p_code":"178", - "code":"179" + "p_code":"176", + "code":"177" }, { "desc":"A fine-grained policy consists of the policy version (the Version field) and statement (the Statement field).Version: Distinguishes between role-based access control (RBA", @@ -1616,8 +1598,8 @@ "title":"Creating a User and Granting Permissions", "uri":"modelarts_23_0079.html", "doc_type":"usermanual", - "p_code":"178", - "code":"180" + "p_code":"176", + "code":"178" }, { "desc":"If default policies cannot meet the requirements on fine-grained access control, you can create custom policies and assign the policies to the user group.You can create c", @@ -1625,8 +1607,8 @@ "title":"Creating a Custom Policy", "uri":"modelarts_23_0080.html", "doc_type":"usermanual", - "p_code":"178", - "code":"181" + "p_code":"176", + "code":"179" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1635,7 +1617,7 @@ "uri":"modelarts_23_0186.html", "doc_type":"usermanual", "p_code":"", - "code":"182" + "code":"180" }, { "desc":"The cloud service platform provides Cloud Eye to help you better understand the status of your ModelArts real-time services and models. You can use Cloud Eye to automatic", @@ -1643,8 +1625,8 @@ "title":"ModelArts Metrics", "uri":"modelarts_23_0187.html", "doc_type":"usermanual", - "p_code":"182", - "code":"183" + "p_code":"180", + "code":"181" }, { "desc":"Setting alarm rules allows you to customize the monitored objects and notification policies so that you can know the status of ModelArts real-time services and models in ", @@ -1652,8 +1634,8 @@ "title":"Setting Alarm Rules", "uri":"modelarts_23_0188.html", "doc_type":"usermanual", - "p_code":"182", - "code":"184" + "p_code":"180", + "code":"182" }, { "desc":"Cloud Eye on the cloud service platform monitors the status of ModelArts real-time services and model loads. You can obtain the monitoring metrics of each ModelArts real-", @@ -1661,8 +1643,8 @@ "title":"Viewing Monitoring Metrics", "uri":"modelarts_23_0189.html", "doc_type":"usermanual", - "p_code":"182", - "code":"185" + "p_code":"180", + "code":"183" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1671,7 +1653,7 @@ "uri":"modelarts_23_0249.html", "doc_type":"usermanual", "p_code":"", - "code":"186" + "code":"184" }, { "desc":"CTS is available on the public cloud platform. With CTS, you can record operations associated with ModelArts for later query, audit, and backtrack operations.CTS has been", @@ -1679,8 +1661,8 @@ "title":"Key Operations Recorded by CTS", "uri":"modelarts_23_0250.html", "doc_type":"usermanual", - "p_code":"186", - "code":"187" + "p_code":"184", + "code":"185" }, { "desc":"After CTS is enabled, CTS starts recording operations related to ModelArts. The CTS management console stores the last seven days of operation records. This section descr", @@ -1688,8 +1670,8 @@ "title":"Viewing Audit Logs", "uri":"modelarts_23_0251.html", "doc_type":"usermanual", - "p_code":"186", - "code":"188" + "p_code":"184", + "code":"186" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1698,7 +1680,7 @@ "uri":"modelarts_05_0000.html", "doc_type":"usermanual", "p_code":"", - "code":"189" + "code":"187" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1706,8 +1688,8 @@ "title":"General Issues", "uri":"modelarts_05_0014.html", "doc_type":"usermanual", - "p_code":"189", - "code":"190" + "p_code":"187", + "code":"188" }, { "desc":"ModelArts is a one-stop development platform for AI developers. With data preprocessing, semi-automated data labeling, distributed training, automated model building, and", @@ -1715,17 +1697,17 @@ "title":"What Is ModelArts?", "uri":"modelarts_05_0001.html", "doc_type":"usermanual", - "p_code":"190", - "code":"191" + "p_code":"188", + "code":"189" }, { - "desc":"ModelArts uses Object Storage Service (OBS) to store data and model backups and snapshots. OBS provides secure, reliable, low-cost storage. For more details, see Object S", + "desc":"ModelArts uses Identity and Access Management (IAM) for authentication and authorization. For more information about IAM, see Identity and Access Management User Guide.Mo", "product_code":"modelarts", - "title":"What Are the Relationships Between ModelArts and Other Services", + "title":"What Are The Relationships Between ModelArts And Other Services", "uri":"modelarts_05_0003.html", "doc_type":"usermanual", - "p_code":"190", - "code":"192" + "p_code":"188", + "code":"190" }, { "desc":"Log in to the console, enter the My Credentials page, and choose Access Keys > Create Access Key.In the Create Access Key dialog box that is displayed, use the login pass", @@ -1733,8 +1715,8 @@ "title":"How Do I Obtain Access Keys?", "uri":"modelarts_05_0004.html", "doc_type":"usermanual", - "p_code":"190", - "code":"193" + "p_code":"188", + "code":"191" }, { "desc":"Before using ModelArts to develop AI models, data needs to be uploaded to an OBS bucket. You can log in to the OBS console to create an OBS bucket, create a folder, and u", @@ -1742,8 +1724,8 @@ "title":"How Do I Upload Data to OBS?", "uri":"modelarts_05_0013.html", "doc_type":"usermanual", - "p_code":"190", - "code":"194" + "p_code":"188", + "code":"192" }, { "desc":"Supported AI frameworks and versions of ModelArts vary slightly based on the development environment, training jobs, and model inference (model management and deployment)", @@ -1751,8 +1733,8 @@ "title":"Which AI Frameworks Does ModelArts Support?", "uri":"modelarts_05_0128.html", "doc_type":"usermanual", - "p_code":"190", - "code":"195" + "p_code":"188", + "code":"193" }, { "desc":"For common users, ModelArts provides the predictive analytics function of ExeML to train models based on structured data.For advanced users, ModelArts provides the notebo", @@ -1760,26 +1742,17 @@ "title":"How Do I Use ModelArts to Train Models Based on Structured Data?", "uri":"modelarts_21_0055.html", "doc_type":"usermanual", - "p_code":"190", - "code":"196" + "p_code":"188", + "code":"194" }, { - "desc":"If an OBS directory needs to be specified for using ModelArts functions, such as creating training jobs and datasets, ensure that the OBS bucket and ModelArts are in the ", - "product_code":"modelarts", - "title":"Why Cannot I Find the OBS Bucket on ModelArts After Uploading Data to OBS?", - "uri":"modelarts_21_0056.html", - "doc_type":"usermanual", - "p_code":"190", - "code":"197" - }, - { - "desc":"No. The current ModelArts version does not support multiple projects. Customers can only use it in the default eu-de project.", + "desc":"The current version supports multiple projects.", "product_code":"modelarts", "title":"Does ModelArts Support Multiple Projects?", "uri":"modelarts_21_0057.html", "doc_type":"usermanual", - "p_code":"190", - "code":"198" + "p_code":"188", + "code":"195" }, { "desc":"To view all files stored in OBS when using notebook instances or training jobs, use either of the following methods:OBS consoleLog in to OBS console using the current acc", @@ -1787,8 +1760,8 @@ "title":"How Do I View All Files in an OBS Directory on ModelArts?", "uri":"modelarts_21_0058.html", "doc_type":"usermanual", - "p_code":"190", - "code":"199" + "p_code":"188", + "code":"196" }, { "desc":"No. The current ModelArts version does not support encrypted files stored in OBS.", @@ -1796,8 +1769,8 @@ "title":"Does ModelArts Support Encrypted Files Stored in OBS?", "uri":"modelarts_21_0059.html", "doc_type":"usermanual", - "p_code":"190", - "code":"200" + "p_code":"188", + "code":"197" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1805,8 +1778,8 @@ "title":"ExeML", "uri":"modelarts_05_0015.html", "doc_type":"usermanual", - "p_code":"189", - "code":"201" + "p_code":"187", + "code":"198" }, { "desc":"ExeML is the process of automating model design, parameter tuning, and model training, compression, and deployment with the labeled data. The process is free of coding an", @@ -1814,8 +1787,8 @@ "title":"What Is ExeML?", "uri":"modelarts_05_0002.html", "doc_type":"usermanual", - "p_code":"201", - "code":"202" + "p_code":"198", + "code":"199" }, { "desc":"Image classification is an image processing method that separates different classes of targets according to the features reflected in the images. With quantitative analys", @@ -1823,8 +1796,8 @@ "title":"What Are Image Classification and Object Detection?", "uri":"modelarts_05_0018.html", "doc_type":"usermanual", - "p_code":"201", - "code":"203" + "p_code":"198", + "code":"200" }, { "desc":"The Train button turns to be available when the training images for an image classification project are classified into at least two categories, and each category contain", @@ -1832,8 +1805,8 @@ "title":"What Should I Do When the Train Button Is Unavailable After I Create an Image Classification Project and Label the Images?", "uri":"modelarts_05_0005.html", "doc_type":"usermanual", - "p_code":"201", - "code":"204" + "p_code":"198", + "code":"201" }, { "desc":"Yes. You can add multiple labels to an image.", @@ -1841,8 +1814,8 @@ "title":"Can I Add Multiple Labels to an Image for an Object Detection Project?", "uri":"modelarts_05_0006.html", "doc_type":"usermanual", - "p_code":"201", - "code":"205" + "p_code":"198", + "code":"202" }, { "desc":"Models created in ExeML are deployed as real-time services. You can add images or compile code to test the services, as well as call the APIs using the URLs.After model d", @@ -1850,8 +1823,8 @@ "title":"What Type of Service Is Deployed in ExeML?", "uri":"modelarts_05_0008.html", "doc_type":"usermanual", - "p_code":"201", - "code":"206" + "p_code":"198", + "code":"203" }, { "desc":"Images in JPG, JPEG, PNG, or BMP format are supported.", @@ -1859,8 +1832,8 @@ "title":"What Formats of Images Are Supported by Object Detection or Image Classification Projects?", "uri":"modelarts_05_0010.html", "doc_type":"usermanual", - "p_code":"201", - "code":"207" + "p_code":"198", + "code":"204" }, { "desc":"Data files cannot be stored in the root directory of an OBS bucket.The name of files in a dataset consists of letters, digits, hyphens (-), and underscores (_), and the f", @@ -1868,8 +1841,8 @@ "title":"What Are the Requirements for Training Data When You Create a Predictive Analytics Project in ExeML?", "uri":"modelarts_21_0062.html", "doc_type":"usermanual", - "p_code":"201", - "code":"208" + "p_code":"198", + "code":"205" }, { "desc":"The model cannot be downloaded. However, you can view the model or deploy the model as a real-time service on the model management page.", @@ -1877,8 +1850,8 @@ "title":"Can I Download a Model After It Is Automatically Trained?", "uri":"modelarts_21_0061.html", "doc_type":"usermanual", - "p_code":"201", - "code":"209" + "p_code":"198", + "code":"206" }, { "desc":"Each round of training generates a training version in an ExeML project. If a training result is unsatisfactory (for example, unsatisfactory about the training precision)", @@ -1886,8 +1859,8 @@ "title":"How Do I Perform Incremental Training in an ExeML Project?", "uri":"modelarts_21_0060.html", "doc_type":"usermanual", - "p_code":"201", - "code":"210" + "p_code":"198", + "code":"207" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1895,8 +1868,8 @@ "title":"Data Management", "uri":"modelarts_05_0101.html", "doc_type":"usermanual", - "p_code":"189", - "code":"211" + "p_code":"187", + "code":"208" }, { "desc":"For the data management function, there are limits on the image size when you upload images to the datasets whose labeling type is object detection or image classificatio", @@ -1904,8 +1877,8 @@ "title":"Are There Size Limits for Images to be Uploaded?", "uri":"modelarts_21_0063.html", "doc_type":"usermanual", - "p_code":"211", - "code":"212" + "p_code":"208", + "code":"209" }, { "desc":"Failed to use the manifest file of the published dataset to import data again.Data has been changed in the OBS directory of the published dataset, for example, images hav", @@ -1913,8 +1886,8 @@ "title":"Why Does Data Fail to Be Imported Using the Manifest File?", "uri":"modelarts_05_0103.html", "doc_type":"usermanual", - "p_code":"211", - "code":"213" + "p_code":"208", + "code":"210" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -1922,8 +1895,8 @@ "title":"Notebook", "uri":"modelarts_05_0067.html", "doc_type":"usermanual", - "p_code":"189", - "code":"214" + "p_code":"187", + "code":"211" }, { "desc":"Log in to the ModelArts management console, and choose DevEnviron > Notebooks.In the notebook list, click Open in the Operation column of the target notebook instance to ", @@ -1931,17 +1904,17 @@ "title":"How Do I Enable the Terminal Function in DevEnviron of ModelArts?", "uri":"modelarts_05_0071.html", "doc_type":"usermanual", - "p_code":"214", - "code":"215" + "p_code":"211", + "code":"212" }, { - "desc":"Log in to the ModelArts management console, and choose DevEnviron > Notebooks.In the notebook list, click Open in the Operation column of the target notebook instance to ", + "desc":"Multiple environments have been integrated into ModelArts Notebook. These environments contain Jupyter Notebook and Python packages, including TensorFlow, MXNet, Caffe, P", "product_code":"modelarts", "title":"How Do I Install External Libraries in a Notebook Instance?", - "uri":"modelarts_21_0064.html", + "uri":"modelarts_05_0022.html", "doc_type":"usermanual", - "p_code":"214", - "code":"216" + "p_code":"211", + "code":"213" }, { "desc":"Notebook instances in DevEnviron support the Keras engine. The Keras engine is not supported in job training and model deployment (inference).Keras is an advanced neural ", @@ -1949,8 +1922,8 @@ "title":"Is the Keras Engine Supported?", "uri":"modelarts_21_0065.html", "doc_type":"usermanual", - "p_code":"214", - "code":"217" + "p_code":"211", + "code":"214" }, { "desc":"After the training code is debugged in a notebook instance, if you need to use the training code for training jobs on ModelArts, convert the ipynb file into a Python file", @@ -1958,8 +1931,8 @@ "title":"How Do I Use Training Code in Training Jobs After Debugging the Code in a Notebook Instance?", "uri":"modelarts_21_0066.html", "doc_type":"usermanual", - "p_code":"214", - "code":"218" + "p_code":"211", + "code":"215" }, { "desc":"In the notebook instance, error message \"No Space left...\" is displayed after the pip install command is run.You are advised to run the pip install --no-cache ** command", @@ -1967,8 +1940,17 @@ "title":"What Should I Do When the System Displays an Error Message Indicating that No Space Left After I Run the pip install Command?", "uri":"modelarts_21_0067.html", "doc_type":"usermanual", - "p_code":"214", - "code":"219" + "p_code":"211", + "code":"216" + }, + { + "desc":"In a notebook instance, you can call the ModelArts MoXing API or SDK to exchange data with OBS for uploading a file to OBS or downloading a file from OBS to the notebook ", + "product_code":"modelarts", + "title":"How Do I Upload a File from a Notebook Instance to OBS or Download a File from OBS to a Notebook Instance?", + "uri":"modelarts_05_0024.html", + "doc_type":"faq", + "p_code":"211", + "code":"217" }, { "desc":"Small files (files smaller than 100 MB)Open a notebook instance and click Upload in the upper right corner to upload a local file to the notebook instance.Upload a small ", @@ -1976,8 +1958,8 @@ "title":"How Do I Upload Local Files to a Notebook Instance?", "uri":"modelarts_21_0068.html", "doc_type":"usermanual", - "p_code":"214", - "code":"220" + "p_code":"211", + "code":"218" }, { "desc":"If you use OBS to store the notebook instance, after you click upload, the data is directly uploaded to the target OBS path, that is, the OBS path specified when the note", @@ -1985,8 +1967,8 @@ "title":"Where Will the Data Be Uploaded to?", "uri":"modelarts_05_0045.html", "doc_type":"usermanual", - "p_code":"214", - "code":"221" + "p_code":"211", + "code":"219" }, { "desc":"The following uses the TensorFlow-1.8 engine as an example. The operations on other engines are similar. You only need to replace the engine name and version number in th", @@ -1994,8 +1976,8 @@ "title":"Should I Access the Python Environment Same as the Notebook Kernel of the Current Instance in the Terminal?", "uri":"modelarts_21_0069.html", "doc_type":"usermanual", - "p_code":"214", - "code":"222" + "p_code":"211", + "code":"220" }, { "desc":"If a notebook instance fails to execute code, you can locate and rectify the fault based on the following scenarios:If the execution of a cell is suspended or lasts for a", @@ -2003,8 +1985,8 @@ "title":"What Do I Do If a Notebook Instance Fails to Execute Code?", "uri":"modelarts_21_0070.html", "doc_type":"usermanual", - "p_code":"214", - "code":"223" + "p_code":"211", + "code":"221" }, { "desc":"Currently, Terminal in ModelArts DevEnviron does not support apt-get. You can use a custom imagecustom image to support it.", @@ -2012,8 +1994,8 @@ "title":"Does ModelArts DevEnviron Support apt-get?", "uri":"modelarts_21_0071.html", "doc_type":"usermanual", - "p_code":"214", - "code":"224" + "p_code":"211", + "code":"222" }, { "desc":"/cache is a temporary directory and will not be saved. After an instance using OBS storage is stopped, data in the ~work directory will be deleted. After a notebook insta", @@ -2021,8 +2003,8 @@ "title":"Do Files in /cache Still Exist After a Notebook Instance is Stopped or Restarted? How Do I Avoid a Restart?", "uri":"modelarts_05_0080.html", "doc_type":"usermanual", - "p_code":"214", - "code":"225" + "p_code":"211", + "code":"223" }, { "desc":"Log in to the ModelArts management console, and choose DevEnviron > Notebooks.In the Operation column of the target notebook instance in the notebook list, click Open to ", @@ -2030,8 +2012,8 @@ "title":"Where Is Data Stored After the Sync OBS Function Is Used?", "uri":"modelarts_05_0081.html", "doc_type":"usermanual", - "p_code":"214", - "code":"226" + "p_code":"211", + "code":"224" }, { "desc":"If you select GPU when creating a notebook instance, perform the following operations to view GPU usage:Log in to the ModelArts management console, and choose DevEnviron ", @@ -2039,8 +2021,8 @@ "title":"How Do I View GPU Usage on the Notebook?", "uri":"modelarts_21_0072.html", "doc_type":"usermanual", - "p_code":"214", - "code":"227" + "p_code":"211", + "code":"225" }, { "desc":"When creating a notebook instance, select the target Python development environment. Python2 and Python3 are supported, corresponding to Python 2.7 and Python 3.6, respec", @@ -2048,8 +2030,8 @@ "title":"What Python Development Environments Does Notebook Support?", "uri":"modelarts_21_0073.html", "doc_type":"usermanual", - "p_code":"214", - "code":"228" + "p_code":"211", + "code":"226" }, { "desc":"The python2 environment of ModelArts supports Caffe, but the python3 environment does not support it.", @@ -2057,8 +2039,8 @@ "title":"Does ModelArts Support the Caffe Engine?", "uri":"modelarts_21_0074.html", "doc_type":"usermanual", - "p_code":"214", - "code":"229" + "p_code":"211", + "code":"227" }, { "desc":"For security purposes, notebook instances do not support sudo privilege escalation.", @@ -2066,8 +2048,8 @@ "title":"Is sudo Privilege Escalation Supported?", "uri":"modelarts_21_0075.html", "doc_type":"usermanual", - "p_code":"214", - "code":"230" + "p_code":"211", + "code":"228" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -2075,8 +2057,8 @@ "title":"Training Jobs", "uri":"modelarts_05_0030.html", "doc_type":"usermanual", - "p_code":"189", - "code":"231" + "p_code":"187", + "code":"229" }, { "desc":"The code directory for creating a training job has limits on the size and number of files.Delete the files except the code from the code directory or save the files in ot", @@ -2084,8 +2066,8 @@ "title":"What Can I Do If the Message \"Object directory size/quantity exceeds the limit\" Is Displayed When I Create a Training Job?", "uri":"modelarts_05_0031.html", "doc_type":"usermanual", - "p_code":"231", - "code":"232" + "p_code":"229", + "code":"230" }, { "desc":"When you use ModelArts, your data is stored in the OBS bucket. The data has a corresponding OBS path, for example, bucket_name/dir/image.jpg. ModelArts training jobs run ", @@ -2093,8 +2075,8 @@ "title":"What Can I Do If \"No such file or directory\" Is Displayed In the Training Job Log?", "uri":"modelarts_05_0032.html", "doc_type":"usermanual", - "p_code":"231", - "code":"233" + "p_code":"229", + "code":"231" }, { "desc":"When a model references a dependency package, select a frequently-used framework to create training jobs. In addition, place the required file or installation package in ", @@ -2102,8 +2084,8 @@ "title":"How Do I Create a Training Job When a Dependency Package Is Referenced in a Model?", "uri":"modelarts_05_0063.html", "doc_type":"usermanual", - "p_code":"231", - "code":"234" + "p_code":"229", + "code":"232" }, { "desc":"Pay attention to the following when setting training parameters:When setting running parameters for creating a training job, you only need to set the corresponding parame", @@ -2111,8 +2093,8 @@ "title":"What Should I Know When Setting Training Parameters?", "uri":"modelarts_21_0077.html", "doc_type":"usermanual", - "p_code":"231", - "code":"235" + "p_code":"229", + "code":"233" }, { "desc":"In the left navigation pane of the ModelArts management console, choose Training Management > Training Jobs to go to the Training Jobs page. In the training job list, cli", @@ -2120,17 +2102,17 @@ "title":"How Do I Check Resource Usage of a Training Job?", "uri":"modelarts_21_0078.html", "doc_type":"usermanual", - "p_code":"231", - "code":"236" + "p_code":"229", + "code":"234" }, { - "desc":"When creating a training job, you can select CPU, GPU, or Ascend resources based on the size of the training job.ModelArts mounts the disk to the /cache directory. You ca", + "desc":"When creating a training job, you can select CPU, GPUresources based on the size of the training job.ModelArts mounts the disk to the /cache directory. You can use this d", "product_code":"modelarts", "title":"What Are Sizes of the /cache Directories for Different Resource Specifications in the Training Environment?", "uri":"modelarts_05_0090.html", "doc_type":"usermanual", - "p_code":"231", - "code":"237" + "p_code":"229", + "code":"235" }, { "desc":"In the script of the training job boot file, run the following commands to obtain the sizes of the to-be-copied and copied folders. Then determine whether folder copy is ", @@ -2138,8 +2120,8 @@ "title":"How Do I Check Whether Folder Copy Is Complete During Job Training?", "uri":"modelarts_21_0079.html", "doc_type":"usermanual", - "p_code":"231", - "code":"238" + "p_code":"229", + "code":"236" }, { "desc":"Training job parameters can be automatically generated in the background or manually entered by users. Perform the following operations to obtain training job parameters:", @@ -2147,8 +2129,8 @@ "title":"How Do I Obtain Training Job Parameters from the Boot File of the Training Job?", "uri":"modelarts_21_0080.html", "doc_type":"usermanual", - "p_code":"231", - "code":"239" + "p_code":"229", + "code":"237" }, { "desc":"ModelArts does not support access to the background of a training job.", @@ -2156,8 +2138,8 @@ "title":"How Do I Access the Background of a Training Job?", "uri":"modelarts_21_0081.html", "doc_type":"usermanual", - "p_code":"231", - "code":"240" + "p_code":"229", + "code":"238" }, { "desc":"Storage directories of ModelArts training jobs do not affect each other. Environments are isolated from each other, and data of other jobs cannot be viewed.", @@ -2165,8 +2147,8 @@ "title":"Is There Any Conflict When Models of Two Training Jobs Are Saved in the Same Directory of a Container?", "uri":"modelarts_21_0082.html", "doc_type":"usermanual", - "p_code":"231", - "code":"241" + "p_code":"229", + "code":"239" }, { "desc":"In a training job, only three valid digits are retained in a training output log. When the value of loss is too small, the value is displayed as 0.000. Log content is as ", @@ -2174,8 +2156,8 @@ "title":"Only Three Valid Digits Are Retained in a Training Output Log. Can the Value of loss Be Changed?", "uri":"modelarts_21_0083.html", "doc_type":"usermanual", - "p_code":"231", - "code":"242" + "p_code":"229", + "code":"240" }, { "desc":"If you cannot access the corresponding folder by using os.system('cd xxx') in the boot script of the training job, you are advised to use the following method:", @@ -2183,8 +2165,8 @@ "title":"Why Can't I Use os.system ('cd xxx') to Access the Corresponding Folder During Job Training?", "uri":"modelarts_21_0084.html", "doc_type":"usermanual", - "p_code":"231", - "code":"243" + "p_code":"229", + "code":"241" }, { "desc":"ModelArts enables you to invoke a shell script, and you can use Python to invoke .sh. The procedure is as follows:Upload the .sh script to an OBS bucket. For example, upl", @@ -2192,8 +2174,8 @@ "title":"How Do I Invoke a Shell Script in a Training Job to Execute the .sh File?", "uri":"modelarts_21_0085.html", "doc_type":"usermanual", - "p_code":"231", - "code":"244" + "p_code":"229", + "code":"242" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -2201,8 +2183,8 @@ "title":"Model Management", "uri":"modelarts_05_0016.html", "doc_type":"usermanual", - "p_code":"189", - "code":"245" + "p_code":"187", + "code":"243" }, { "desc":"ModelArts does not support the import of models in .h5 format. You can convert the models in .h5 format of Keras to the TensorFlow format and then import the models to Mo", @@ -2210,8 +2192,8 @@ "title":"How Do I Import the .h5 Model of Keras to ModelArts?", "uri":"modelarts_21_0086.html", "doc_type":"usermanual", - "p_code":"245", - "code":"246" + "p_code":"243", + "code":"244" }, { "desc":"ModelArts allows you to upload local models to OBS or import models stored in OBS directly into ModelArts.For details about how to import a model from OBS, see Importing ", @@ -2219,8 +2201,8 @@ "title":"How Do I Import a Model Downloaded from OBS to ModelArts?", "uri":"modelarts_05_0124.html", "doc_type":"usermanual", - "p_code":"245", - "code":"247" + "p_code":"243", + "code":"245" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -2228,17 +2210,17 @@ "title":"Service Deployment", "uri":"modelarts_05_0017.html", "doc_type":"usermanual", - "p_code":"189", - "code":"248" + "p_code":"187", + "code":"246" }, { - "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", + "desc":"Currently, models can only be deployed as real-time services and batch services.", "product_code":"modelarts", "title":"What Types of Services Can Models Be Deployed as on ModelArts?", "uri":"modelarts_05_0012.html", "doc_type":"usermanual", - "p_code":"248", - "code":"249" + "p_code":"246", + "code":"247" }, { "desc":"Before importing a model, you need to place the corresponding inference code and configuration file in the model folder. When encoding with Python, you are advised to use", @@ -2246,8 +2228,8 @@ "title":"What Should I Do If a Conflict Occurs When Deploying a Model As a Real-Time Service?", "uri":"modelarts_05_0100.html", "doc_type":"usermanual", - "p_code":"248", - "code":"250" + "p_code":"246", + "code":"248" }, { "desc":"HUAWEI CLOUD Help Center presents technical documents to help you quickly get started with HUAWEI CLOUD services. The technical documents include Service Overview, Price Details, Purchase Guide, User Guide, API Reference, Best Practices, FAQs, and Videos.", @@ -2256,6 +2238,6 @@ "uri":"modelarts_04_0099.html", "doc_type":"usermanual", "p_code":"", - "code":"251" + "code":"249" } ] \ No newline at end of file diff --git a/docs/modelarts/umn/en-us_image_0000001156920825.png b/docs/modelarts/umn/en-us_image_0000001156920825.png new file mode 100644 index 0000000000000000000000000000000000000000..51988655853db3d43682815930898f2efd61414f GIT binary patch literal 28863 zcmeFZbx>T<^DhV_cz^%_f&_vGx8R!KHrU`6+}&M*6Wn2N8QdKP2ofBEyEC}Evy<=d zy|=G+Yir*hTf6`4R88HvXYRezr%#`AKHdH430DM2VxSSB!NI{{NK1(+!@(ig!@lD-{il+!PUf|KN%px&Qa~9G@an!aGd|W;a5$` z+~MHtfzo2%fbM$7kXHpf&5qOOCytkbmO()sCp%mxEKiQ0gr5e0y;w6Tp8S?~`#RAy zSiQ@$Cm!imfGa*pP@20${_Xm_1!?r+!JgU&Ya8>1+y=ykfOFY0sjVi7CxkHD5097L zLS?h1kWyDwC2fPcre|MjviWNppQ?t{h7RDf{#%%2aBvrBC_Mj;{sM0hF#mn{JU$TN zPu%7hbRdIn@q3L^`PK-+5w`us#a3555+7grI_sOv!3yuU3(^L=(v507CEY_v3)a%M z;DL=5_~vDzyh!xo8(9M_U&-xQ;_91!kJh22gy(j2e1liptfkl&Tg622thwKudcx|8e(Oq}2;{7-tvh&u!T(Jg1I0|9nI7>nVKT9^EU$R+)! z%I{jO9gz}6*SXvq4rDe@)NKv!73w@Dnq3%8T*X;Z`wb&vjQtdh(Np=;Ex9^{M!(Zt ziL#cY5MU2o#|URykDt3$x%RUci?4m>X}$yOin^U>lwr`ZTQa(?jwDL1;lHF`Pu-^@ zYT=aX!P?EaWnzr}?wA--g*A!dx%N*BAnGFjjGWa>PMOb4r5fTUew5!$5pDKSi7G*P z8(Ea~&@E5NdL|u!3AA;0jAfk^OZgET1=Ma?c6dx-FfVDP>F_7spWL*U#!wv;;D6Db z@gDU$-V(CZTN7y@vn}jplC7@(VZGKGU^(|y`I0(%SXAf~H*I&PSw$I=Y?f>+dBjsx)h{N5D`Exe@s65@0W#b zvPxrkd&Thtbaw47L1XTu&uEPr2)3Syi^2}Q`h7R(Ut`S`4K2PwZ4W-p?V@1uS*)x@ zv@w`#cT(3c>oq{9|FV_#=3vJ*GHLk)g^*=9i5{hBmptaWrUV=87UBKgO|rH3f)uFN z(XWGQD}0;}f%ys1c>F+}?(Q88K~DW_G#7<(!$U;s0g5f2K?>_K+>=6Lxz&i`rT3fM zk#|#`afrL(w=`yVobzf*Dz8T)3jd~^?@2F?J>@4Yxx~{S@Lb|Xg8apDjjxRXIJ1_0 z(}sZ*rK=tRhyh1Q_d2HMKk@CSaHpS>P%ofZV4@_QP1_`+7ITs9_vTpAP*wz zPAf7lLF?m!gJtF5Q@p)nkN_t|Xs85jEyu=W7&R&K77U7zrM1et&p z-PNGkOy?&Ou1LX`qL9XV-Tb&gI--3$a1@`dsjnvelVMVQWlN-I6;Dr*ZMqdatwGyw zgSfM>Ye$~4t60}Ge6z2laV{P_9X&LI|9df z#GuzfexLeNT*=-xr>k-+)zn$C+>=~XqP7`v;_LTKC{w^HQkYKQRuj_@K@IM?hI5=m z2|kP;CWDh3934&2F*F2!SU9a4W6rJ(UeitF513=>Se^%7k*>rhh<~(o4HEwyI!pDG ziU6<|SKTAw(VwUCIf$5kDK%YXarnDbSv~nHbE?~IgC5*Jf@crHM78sNMi}ea{!A_8 z>`6*czAO(>xi|{P(GjyzpXWt?I4(Ajrv;KWRhPQVB zS80no9k;FyTVKEN-UOa~bWR;AfjwiG#+Nc)>&U~B`v{+v4^db3PDJH}wXYN>i#vC8 zz4-PjBuz1&&a-B{Oi|Tctmz@QW5023*yV`RO8zN1tmhpyqkD11IhYM8$10r25kpvXwhP3GDyu29*0MDXCeGv1D+hon2^9~OFR4bS3 z7RKA<`9l0sW%LgbB_}zV5-MN4YuS=${Io@9%I1`9Y@)URAf5-rWn=d&nRfm;tKS`I zi2QF!U?RkF<{^UG+;d~m=nqQN-Ue7*19LqS&`Qn{07O}G*R#b)f~ClnoqdEd^+Kge4ETh@U-Z{&Q)8J*Pxxo zE!#V^8G1E!k#O_!*k`-X>jlUFuv6~qv!|L~i8~mh^WgIHI!qFOOy!lSLA<1QcF#Kze0CeqWomf67gNI;uW(7Q--?d>oTB|}QBuEHRNvlN1Ox5X0M zaFRKvBLuNuK~~$#wx!A^1v{H1Rx@Y$3cx-*`$>*M{g5{EYv0z5n=JW>YEA0l{m=Vp zasEcXr#tjmNWMT1pR-%h2wUY>H}8t;8{g_^kW7iCUVq;(DVeG3p2w@(RP6*H{byGm zAlg8PM*OMSzJ99Y6z5^CWv}Yv8Aa-tuV=F=+*lje$l3H4|JjzqM51euTY)?EISKd} zif_5PAf+l}&V$dqUe3ajdhh0Ijjh_BX8`Jp*{@;HlhqbVYH=KUTp)R6RNBFG8d`3x zn4WIW;Ck6W!Fc+p(sf>0^8)1ipeDts=3w5?G&31`6jIdshH z;J@^LgiXJ$97)ivH_sqf;JLDmM~Ou{&ENCnG==6%$KniKqgcOcn~cX|{@ZBD#G}Ii zW*21C#-mvm`)c_u?xtH0-EBU(FRtCt_~-$KvN4Lq>KTPLKcjV`&>VptF1&cpS3{~cjK3iJPj z3;qA21-Jo@{|p`sI==h=FZqX;|0lhOj7_yQgmK;MMY)0E!~|8zNM^Lj1+2jbSkZsO zR^Qf%0unj4k3knsrLq?(LW{NT4A%!t?*TzDa65M02b|-|ZTf`Nn4)4+=0@i3MrdJ` zvzYd_ubf@x*lKpN1KBfSe?^CfKk4KepME1Em%pTGEhMX={)d5-(OGk*B?0~zV=ux- z-jAYPf;>2X2NV?ZDr1vvJBylnJn{t~^%-bQH~Ud;37QOMcA`I4O}{@pC*b|zw^-R! ztv`)k9aE}J@86pHQscpt#7MW@niq>EGdKAs+Up^QRpwDB_L-r$+?yAVu`D-)%gH;! zmc+UM?QqdXkfWiEd`>D&tG$OZEFMg35fD~j0TFGk|A}9*U%>W4cl|ezIti#&6NpAY zpfY%XPn7Bm$nbhF-|Bu3PkUYI<;Sy6>)gbYUInb6}!eq+lqWL>s$I|Mg9#< zLgDB6rS(kOs>s+$8Lc@!Dd_|n#=Uf_$oo|e zic##(w(g`uKlR@<#9W>-v{~#UKU&P|WRMhMFOZ4vasHeh`CfjXyjZrkq9aXB(qW^1 zg}Ecafj+qN?OGd^=g_zF+yaAXXz@8a%^!V4Vcdfwwx7=|gnXLrVD-oJe&Ud`dI@HO ziYPS!g0JJ3a_a)u%WY_tzcD`}M-gqYo4Oz#vr*xAaEYNh8F$F-mvaLQ0%m5p!Xwv; z!l#BM$gZnOgFp>q$p-B&I-kFhopH73+xAOHW2_zrn&0rWS&C^f(>o-O53Qyvd&*cmg}OL1Z+0uouMwL6--h;QfLYCdee{%C59LZulx-SubgPyB(fx( z%m{hGk2b<)()#n_ppnjeE|h78bAuD@cU@_f8))9#5HO{=`P)FtG?+W@*E?@|1#fW@ zwevdh;*cBc%n%3a@GRrXB%4zTX2| zkC)|LcpT!vjf~}9?NDWl8NfEDwofO#33+=z6QI?9Te~%md2qx10gJyRH*rc)IIhad zshiuiq>l~^Z!&LMZpJ-x;X&tA8g+y*qb&#a6G{Z(08EnkahAx>0(hWiVIjdzH+tq`OFD*&oy_AT{)ks_Y!Y z{zjNA#E~RSzDM%v^La__Bk+Fvu|lj{vCranIDux0dsuP4QIPscf8p|K!Aum+pS~#1 z4_z!GR2ZMm1!rg1r!^m>Nb)MiM27A>DP4$!>$4qul2SKlJ#gqT)z@wBnsY5)c~o$NQc(skT?(W}r*qRdCMRKPB48=k&%#MDD z)PrNb>3s#gic4n*7*Z-JWImA>HNFHq+40+Zrt^5OW+%)RNkc(CgS(}(JM9N}Q*pA<#YZdPvzgf0<@9tL zV4cBQ>}Xe5)HQXDl~^VU|0!K*SE%F)*!>aiIYES%?{CM`JG+EjVj^b z-K3r_fGrPGfI;MYVmm&hZ9?^%$ZjHAEF7AZVu_vnUIyK+h5dsNY5GrFMG{2C;+~bs zI=gE1fou8_*Pl~;EKa0TuVX{VD9sDvTo~vq{;HKkd`Y@hvPkx26tGp(csk!(lgHaB zQim=b;3YZC8s?#rQe%S;AkG>?3wdCr&>qhCZk7Q&)~VXF*J>ScXEqaup3=n zNyV>vAWCBLcVZdv-d?Iu$%lvX6E}zFnr`xuavZ$C6WTFw}{sVc3GwNsL**y}h69PcnF44f) zmC_ig79E;Ciznkes-<_yde>>;_MG%HgT|L)Gu1VUc(<7K0o|M4=I4b>?v89&o3a!_ zhyv%T^Sgn%Vga6}49k^jM?{m#s1Zrjm^m!J)>Z?wi>21z&vS+MWu9m}Cp#fZCwPap z-bTd5f>McL!t}Z(=WuJrlN=P01of%1P$OV^A=%|0CMw1(_b~Hq_PvsFG>BlUD~Ql8 zi~R}6lS3m+ptL|+a^iz!>-}jGW>#sQliK+qlBFu=Uv$dHou+=%e1?2fjlO(ZF6D)X`RMC7oDsyo{c^P%0Cm&i)7l+kJmb!U;c zm`m8h0I~jk8#pDEA+g{#F2!;fYWlV7`w%CioPKBLh0!>2Iz3_ZQ4T`i4kFJR`DAPbFCo>S_4I{79kDtVUwB_fwkxh^}e ziV1y>`+iw+;u4v*?Gj&y!5{|xI-_jrVt@(63ejtTl>j)=jL$q?~6(yT?J4wv@(^yA*e)^L^PHqLx{p*h{@>yN4y4PfU7Gh;R8F+JVs7mA#a} zPtSGG9k=dWhHD}5M@e4c0zDMQe*BYXYwUUj_xZ{p&~cNt9>wHp#~L0xx777pq)=^a zrPZEOH`*w{ex2^q+nh(*_xR-OPh~_qgqw#f4Dke$xLIF*s>%)D>0mSC?yMNpIK#&u&Bv|+P-_EMPhaKkDzW3#R@$?P zdi@{@n+@jAzR{sgBkn^j1!Z#5_J0ztY`@BVTpU6@*4={V$VfRgqS; zcK2G@lTE`MV(#22;UyDvw+ZW#9gH}*%3HTwXdJ00+N@Ofg2d>yPnJIipf?Q3`Ib60 z-aj03+EJ|oZ^13_{f{8&z#@_Kh={Ekfcl9tVUl&@&dR*eJ$`97CxqDh4ecErEe{9i zIA@5QKH4%YzqXI?X9_K3$I~s0Md05UAK-1aJE8}-B*z6(GZ6HIC?KMi>0+;#H||;w zlH?4ngE?kIxLTvS=L=VW2HhVtB?>aMgg30Z8FV!fY)6eYpXP4^+%Q4AJB`CJi3{~I z(!w^2cz#m6XDCFR>V|l#6<>R#SKF}^IKa2%-OI~Q)5|(mG{llY;+we)8#S{Hvtq_S zb;_px`i--`2*@X)rr~~=^dG9O5+qaP6kFaq+<;P{6S}=bBTE*vy7VdX&){Wv$F0hS zSK&!w$|^!cVp<$x@4u)iI9Te1zu2ve2RsV!^psC3^bUxZr()~4g*$UqDorkm5URh)Q547#ddIWV{$IU6}uY)GmM)!_dZj)Ee z_is@ix5i3f(Op}C>HVH}ryGtNH?AS9Q=Av;Ul?}f@A0TG3c>yOF2BGaj(9iaQ1J&c z)m_OX&9o2-mm20>%1=?iP2Raprnfqd#W6^M5-RUr9-Wh&0W|64m6&YpY5qnaRm$3- zFpjk*Ejyl?WHfvN6$2>JqFW}qjfUx->JLO&p1-wxqFH-0mS0@ecMz-F!V{Q)rccqu zC{wXPr3^*lvhynn(SM^br`9o*osv$n>eW5Y?rR@JcQ6Nrn&J=EXT)m}2BX|;&|PSi z735w1zgDWQsGubFFg5v*D$U^1FII z>GZQUt_dHs^5WD1dQTm)df3xnp`zKPf@;YfY7aoMSK%rz-D9X#k|X>R5xgrr?^FRX z&c?ci`BRj|(a|oQC_0T9LffqtHF8$#0_B!DMNW@=!^8LT0`SN?ScECJc%_;>9Pngt zthfeIv~UaGB>>%OW2T56XHDv@KRwWt zSk+h1Kg{4bjdqQt4GE?N0kYKR0HWdFfUj*x|YwOgluR!6gLzB4{9w$Vud zmbA#(U1b#Lf@LRCzjdwT_l9yF)Nz_z@WyMM?mABIDsR^@93*fKXWHEjYNrd@-I5jT zpzS)zyNYrwHwMu6u3X>&Ypw$?)y$p=QFl40R~6QIvg?ZraIP2fht^M{1&<0sCJk?K zr4JXK^SP!^bcPKW%ddPfduWQ4xG)rdxX-*ulqzmS-IWiyli6r9Wp{5}PPeTnXICpx zPlk~!cckO=16D`hCLQNBg;}a7M!6V+2gMkwL_cG|JOcU9O6ED*9 zXepy-bj+e6B&rMd?J=!QReAB$ef7nlXEqk&=r41mru4LG8phg5GdWj{lLZl{p@Dr2 z=3D#I=ow~m!wJO2zpCP}&cIJlXRWQfjb%T+=k9J2XO@RxlX)M)bnt%AN(y~6Nfr)W zrz+FfExPFCA|1iww$KJEi{UF*7GUPh0j}P|%el8NjAXDAN9eV_j3vSGrmY!)gUlpp zoK=&o4xls=1!j;*!Vj4M*WlJR%9{9>dbbnO$!l^+|29Y#BSDKOTEUO%YWJ)hmk% zCgj;07vTIMb$}$EceLN`!{$2H`qT%oq{qS5Rp4F4W|zU#chQbrjUPfFW(TvHno3oD zCktdI7xBPO(10tostq>_YjVt>lg3E@G#I%g98r9)|BNYcC<4c(b$K|ndsK2qbl3$> z2}-5d#r9I2`gLJG_tqiJO;}A(MGN0a>!fAJSC01!Mm*>bb&=elt5oXLiFi}=kGzxN z0f7e878X^3qnNJY?Be!hM&>BUR ze*0Rtl)PhNM4{k;atf`?G1;zRAu1A^s3c^Elv5qQgmz~-1+D-*=5gQ_i#1a_-`wr8 zw=XfLkL=fDyH@26BXMfg}|?G zn}4uZ_bqmQ?3RV$sEa(D73B--6WYcS5KjC^ILZqNR{L2l9hq&}e3&UQgO3+zr5d&KOV`+EHPZ6*)O<7PAFSYH!h1SJ zF&g@5=Lx8n^e+yHzY;_}rO7d!p3Q2BXd zDKYhOi|;*^&L$rlDl-{cP-UhIJ#7RqM(;QmoAf$bS$Y^K-WK36Id;t3JsENIsAPq7 zS;kJDaT4v%AqDAf*`Wx|6$Q^Z^uN<>X&gRGF?5Wr(oNzs(~LW;>e?*8%F zBGs}ta8i!^zH~VWiTTGJ5atH>Z4Cr*4v;@iO;OEsc@AZ+J5YX!o^_4&Q}!*dzYHUY z%&FZ(&DX~48((+K{6n0a@U`>V!my7&zux%iB|<;a`Wr4bth=-)cf+Z~Tz))(7lZE| z^*y%+-nMM)G<5zdniLCG!FtnH6tV>m_ro{*gT0XrM*X_c+7!4_1B!>9HDi$_qaLzE zRP0lna8Pd0SRW$saZzC=(xfm)r;eVpz>r*F2%Ss!uXbRkK_jfk^q5(+�FA2^LtO zU}8P6B>fngAaB*t6fnmB_+M|Wy;HeIBhOeLrdVgf=sE@VZDIwqV4lya_w|3dJ#T3L zxjnX^6)3BVHiRxvMpz*xqWQi^(0;b@Vtk2#(jj|=wE4{OcvpT{I(_ZsYt3e+k`_Cu zADNRU(@n&AG3(>0_|o&hBRji-XzG5)uaQIGf?KoXqZ%KiVB;3XUm0PELja(HFG1FIq^yPCq zg_w-YaSNA9;#D}-H&n(1O!hcWd5`jtg`KENimx!^m*6&xF%cqp<;v@cNZ7WzMlLZp z|4dabI|^dik#NFbf=c`yoP&kUfrRYH08i^ko|oh$$5Ey+KP!OB#j+ZG&8|%_*A2VC zK;RT3ja2022bW5-*Yfs-e_n%*uJ)NXtshAbqK;V04Yn#^x;h72M)bJ8QfG4Te4H$~ z{%g%vrnDA&$pziL8UU{&cd(chyb{^?sB~*$ob|wJVY!y_{N(>*^wMc8e8{TGU)FK8 z#py|`en>~KtZ8B7p15UFvgZ#L9NYl=glKWXTycpjFWwM+RjGsw$U2w3$uKi7K=&Hh zT#y0!a0U@2@dRMv2DG%;u{p;O(>R%#GlzY4iDmi4l2xi;!Bs)Dt3r{TZc1_FY#R$nl zAJbJXF5=&<*J{02tqAYwxJRun-~sUZ&^wiS4G%Oxb{gMpch+-<2a^U|MjBh5Co19dgW{7op>tJtwW-{bPLoyvWQm zRJ7v<)I7YKxaV3OzgO<1_AZQpK|6r(}{rVygJ1yZkh;jFcJzYcoDkH_2Dw%n{dI-ItA7Cy2n zZ1?T5FbMM!M#1qbVX16%N_zmDPB*(?f)7s@PDQb?4DCGaqK14k)o}C_(%9^CAEY)j zlBF?ieWm^~(SFLURfi|lJ;viE#Y|md% z^69@F6Yc!LKa9~1K8MzuZ=T8eYAIzAN-z1lFChS+Y1jfJH0NFoC|EDgUI5Xl0X=AK zYy9vLU3kvwixj#oG2U$$`8jMnWH^K!ZBzXdysi{L1m8$30iP1sPa{mR(K@{Pr-)5y zA3m@3gD~k+Zy@8wp5ymk>$CP+#a5ct<-0aXvqx^rkKfEUA0+yIzGWLp9cacX7M#CZvIYWRbm1(cvY@E> zo1&o7*nY$oqd$=P>j+*azNq7BQKXKNgQ&eMU=uuh8%RV|0=MwoEXsf*G!;oOEfR#0 z5b`O~@(W@~&~a8iWEpM=5{8r2MQefx4cHabK zxUn6f=xU9tZ+82kd6RZMwIKg%YrJjaUG>yNwgg(o4&~H*WLt77>r!Za^&pBgE~ZK> zA$$rxDRbHBz9YTi*w-8{UaT=k0Xg5uk7)E9GG|&2F}*`|#a}ORT^LJ2#PF1Nr&uit zJ<<1ZTl8x_&=Egfphn*jiu%PsIKWG0NqmK2uXg& zN2-qnBVRyi*sr9Zkh=%}JG0K9vAex! z!R9&Y#8HHwYjK$EpJq)p0rOPl^-5`I$yS9qrc?|ZM5$ol_d}x=8KkxOKYexWC=Bk8 zsJnVsij;=9DzmSe@QQM zi##TSg{ttkohXTUd9QiDdoBo-$Ay`@hy!~cW@fQA`ko(+Q6^Zjbia)n%2o2)aelT1 ziTB<`6T)-_BpMw025quCnItv>hX-fcS+cjRyrFek`(>WuytS6X`$Q9!-h8CDvo_2N z3aI{}KW2+G8W8#4j7x6E7qOxXxH*NUp8msdgW$`Fvru+JZ?sqE-B8-v=-$1PqKyAo%F=sXE(uB|2HK1=d& za474?uvi^6%gW=)0ljUpNnVvlZQ@3(48HTpGPlpMrJx%p$`N{F@TSWeDLc1~1XdkQRQ<}s>ZiYy}M_Mj0|x_SEI(X7#P z{?&qNPG*rSE&K_7z`mgTJH%@9x9#u{v{)d63EpfdW;=~vI8iWG`th(rJ$(MQ{{uLf zk&k^7aKdn#+ZX1Zsj#J^SJaEW5qXCHQB2cVBOQ7&_NX5@V zr_dT1qp)KiLOq4Gii=HW(Kn|Ht;K5&3ut-#nIufoRF6nQv*Mg-DSXv_{vT-I;RWBH zdP`iAEB0ciMnhi^w=*>8DoP|ZUM&i@^)ou-;91hTwpEqJYYj>ZO{z_gdya=C!S72J znlq)1a9re6dr5@OB{7EWOza7%V!yPeBx>P zz3w`tIXWaf38A{WN&P&}?kQR>v<%pN7>%5|j(IU(xP3efx!Fkzn!ZR0IHdx|6UE{$ z6(;pIsTkm}YPuY@%6mSSeWdQ77=y-3v31W4l&U!>sQC;%R=k^>FRBoNb`5TRXajR{ zk&LCi1@)?i3{RJ8FiXC;S%rUTT2XYWDA4sM$e76|_L z@<93zBl*9A%KsW-&Hlf*ljN#g6)93fkOTU#eVLgi# z{Jahmza5%P82A>+`b1s~#-uhTM&XHN`$mDlnmGk&?k5|NU)5?fHUp2tHdwm$MlB!(}30O<_13(sxi>9}j3BI<^2qnI5b#?~v>KEy(ihu%y+Q8$<5Uha;Yu%T&HgK4wy6T+oy9 zxu{naLcJaqFY5F4Own!xZBvq$iK#Nz^i_5-fuPvH1VgnK+fY#3;z&FDy(*>);fn_<0|HIHb1rtlW2dw4yJ`mB<>3$0y zwk*zNnxQPJMdA+fc8UO{w(@Iag=y;Ri63;nK^)HJXEShR^!B&eFJc@t$C>)l? zLhAK~VcA?@O|B9+!8CqLt!ztiM(WtjeXR#z`2~8%9LH0h<6?6G5tr<8(G&iDNS~te zdlMF|<|qa!zJpOMO{kL+S-{;g$I+%RhCcp!w6=-2w~`)uvzhbJ>oeZ-$szLelR{_i z@tS781nJ)zCYP(o&PUGYXBI#CSB*l*UK+X=hrm(Ev{PtWIvW@9g9ez|^6 zr+&FAv>AJ3ews3`HD}ycZJ0j4j(7IQ_sAk;_6a(l@SZo-vnNXGF#2V0&2P$@XT4alJsYtHA14$JxhQ zi#qUbk}db~E7n)p99XcL!v&fU|2Byw8RE>1q1L(70LsaKk=eFtImI){ z2$Xe~f*w{Nxv8aCmJ?ipQu2&qjXI8EDLU%PkPJTgsTE*p# z!Wd;ySEM}JUEvx`3(LDo96=K5%N-;7EUX^N6n9ol)Zv1lfpB%abU7>Y+4^^*+#I3xW z*7}wYv=9B`qs`+rS*4)o@i72w7Vg$4YDhFwR1 ze;wlZz;91{8%CXY-RbJ;I6mFno_?I`da3c(_(LLq#{Sj!IOP&+LG1a_fpFqL@@G1fxOl-b-7mJC{Iv&L}ai2 zHQ<2Kkx@yzmm+mf;u4MCs?qb{LZ@p5hIlMbmY(hQ{9ZEg4P!dM$I#BAV*-`kh#aT4 zpx5~~mJg@Ut2Vnkm^Y#lC@*_sYCpdkyh-8_< z>p_UZBQ1N{QgT)pJeBQBh(kqDYbII`>DDGCU}VgDU0@c?BG_+2KP=w2Ja|gb7ko4?5Am-zKZKC?impAh*G~Z8V>s!^X8^itc=f2uICbp77|2%QC{o4>GV5L-0 z`b=IE=B-R1csH#O6<#~lm}0$GULR{&L;EKfnOUX_C2^a5z{W-F#YeNAD015djTJr2mox#0w0f!aIz z>V~+o5trS(_Q{0!Mazz>r65T&~) zkC*H(JAH6wWqoV+b?)z^JOPN;bn;D0$s`9dvlIO8NLiUiOGN2WmMnKBm0)DaBEE*{ zOV+!!^n3dVdwAN*4I=zGuo^_Bb&uD6uu?(`xUk9ial@O)ymEX={GC~2;?(Q>DNIs) zl_p~cQS9d)SHW%GuOdrU5zD0Y9De^6?rfFIE#LN2-QM?9>%LFveH&}%r~D78Ey?m2 z`7gJv9-oBp{;95fKG~u5*)yH_`1bfUefYfB$#Qe8(LaH-sYm!Jkw_-#-g?%)>|OuK zt7YO#6DjHQ@gS9a`l^6-R}JS4WGFlr_|ITqOZV#Y4jvy8wriv@Gr=ng|0U03^-lb>f|Yg5L*h`%4+0Sq;IE(03SeVmYO$cxl+nZe>X zSekT0dGbunu}Lj1eL%&b^M{{TtP&_ZG#<|=oIqwbwfXd!a9#eo;zBZ918~rECj5g@R5=w>q5ibD#ORVd7Dz|9KrQ#B{}m|X_v*&D;|p2Q1ihzT z*DwY>9_jCqp+)W!_^Hdh-mA|ZsO;Zs^K|xr4MRQ~#9^qM`c8cxb&^AE%bp%mu7b^5 zg?+CMxe#Tn*Ryn6H*JXdBkkTJTiR1mny2_1v#Z!Lg|5NCP`m7rS4Dal0M1AwV1rwy zloH!6)m_Zg3g$bc>1nwX@TP}hWc(I8ifuh*vuH*Mca?wZ(i?bWb{g2Fzg9dzL=>M0 z^NwpLudEL5$`?5JZaRJOSV0*+{{|YworcBD9*2vLz4`}M^0NeULw=#+V7C}vOs-fG z9y~#-_Js+aLtb{x*1QM=MglVzFBn|g5@ASAVG7R*u2-tWCK5eN7TZQS!4 zy8xC;Bsl*T#Q%FX%>Vdffd6a4%>UXY#v{FTyI;RP`6Gp%{Qki@U!FQmr`!Yww+W;4 zuuIS{f{k#$^4^mm{rqj(R{)4>MTC3*pB`M)lK|m+U^gtKPBX#zyZ_UEfd9|mKb#C; zj$AK{&|@qI+fT`|?QKTvjC_e>Y!qy~4w`C2ybO;9Nz#sywG}^3_7Itr917SX-)ACv z9W72tdEuDw1BW|7`KHFJW7S7NQjcuvSX`p|Ge5u9TIbuDf3q!2!<(yjp141vLvQ`XV?s;4 z1Cxw#`xj!8iiRx_%qKn%@P9*?FpMyUEqckYu57?6sJeO#)bN=@y2m!mk0G`45{1X> z30xlQ4P3rTjc$_0nsXNN8VY2l1T090bE7o74a2f#VOVcT22#Jkn{hX0A@n4^X}VH$=YVJ$v5BNdC}pK#%JNbPcOlme;ceD z{Tv$KFpYKsJqs>oQ*K-3iffIAk`)G8L`&ffnTtW)j#K}BOW_Z~33%Hn9dHnp=58NU zN?=>42eE*aYDvu?OJ3s-aDQG`?q(d3`*>vY*le_!JZ_Us_|3RiDfWJxW3DJI6JH76 znJME~A9O4td&!cdR?>bAg)rlm9*J~8rjX!~MZDQzmH7J4=n(U^swPu7gp*7vt*-qL zE|FsKaE^Xcq6hDCMrvP#pxAq|AB|nCom;7EnN4g}OEDx~A8jdT{W|Z*TQ&`dND`1HClQsLBqxy|IcE?=kti7fo1BrHgCq$OB}>jpP?Xd}X>x2bc9YXo z>-o<4?mTm6?%bKV(?5843aWS2u3h_Gd%f#jyG?F~jjf`v;ELr9rv|;9`LRgFXZ7*; zCpJRjvP?Kz#b-!~%{ddbR|Db2ADjgbRYfdg4z~+_jw-Si>;GBd{m>@-nYs1U>e;w6 ztA0zs#6eY+5{LG)-PaBX9v>3s^H=SJ%V$8z_5_Yk0nkT{Z)Yb0?yIA)A_Qgv%AF!- zbpyWIN-g5@UcV_nojpFGe0AmpjEf1nixc^AJgNJA0p+;7!t|iqF;igs&He0+UI)p1 z^6W7fQq4yRKw5DoG;Jrz;BtUUu+LZJyH^@KvY$o8I_>jg+G9$!wzt{(F)it$j)Kli zvz1IX8`xgs^oAn7@Jy@GfL5y~Oe^ux3)_~2i7M0CZ(L4=setK49*?k(t)ve7+RmDk zW@tsdxVp0o-6(M;K84|yBx)+ppV;t1Avq2g#ii6g4mNE zwtTq!FAFe^Zwh3;`w&(m$^2M`egC@gj-%Aw;2*^NcRzo&USwYSZc9XaK9V};Q0Kkw ziIqX@|5L5+o^l&^^)%vT`gO}AWJ9Js;_B(1yGmgxH6(M-K5cll7vfkK7U(xX>-ozS zPUiSabR*+opdTiK!~r&ZfZnQPUE(rd(DX3FeXU0}2U^`CZt>)};7N%sl0cepr*N}Z zRR0kgs+U^L)3(k_-v_J0%@*Bfr}t1DuQ(UJ{G_t#EWjJI|KR6WGh4|npHV5iBpZ}0 zz|nFMfDcv;#RhX${D8Mh^;z(udop4Aw#^r1jpLBJXt_unejI#RQ=YBSfgK$8 zLpA%MQuBnHXvgTqB&2ShKeL?fIw{QMTL25!ZO6Sfbj-;XhB$sliEqMr>%U9OXP1-SWP8?oh$Lg-yIM(DU>Q%NP=p?qH*ToHR zv<9H)Ak&aS++sB=Rk>(kqrBO0TM*pyju>bOasCX3*gT*I?^^y2la{;KG#=~yuP_AZ zlSzan%(@DCB1rU-{j{F2>wm&d%W${~y(KiDc0F87tt)8v^1U~Q5o^}ZNME@rXjiQN zk2*hC7Xg^_8XrvPck%}}APp7APJ(c$v5fHp{Q|9jKX`yPe(5aL@iw~ji;0iXBr6Qe zU?-}@$8}q^k?DNaj?FmQL-8FcbGZ)+w*iOg!Z&EQhQUSICng_P@a0|jO3QPMlwSik z>$N?EB468!T_ET!*1WPmE>-*puh)OyfxIbtU3XPxr=8|Da45Z=!YXPU!j7>Xj*gSx z{uU>307&Ct^A|Y4$D+Z%k&gGz>tk2|5Rfs&U(p$Wu7RCvU4NZUar$Ku#-mIa+WEZ3 zQnOHg77Pl*L5JUWmP6w7Wg(rJNm%C;|DMcmyXQ48{&sZ~#a8r^Z%Y6SxUdr^BD)CzJr+ab<9-+~pj#6tGWvrdD732q zs0w<6_Xn5+DndZqwwv<3!{;cp~}&uRebF!pQ0{`0)0nDIcVcz(7HNCAbCO^$*T_iJ z_c*`iD*{>}AA@xmlZY>Tu$$C%qepPe<(yKbbQ$!WP_qqq68-ueD!zwYsN9{lxo}*( z+W4Q)k|Nteo>rwrV#mD1B94O1z~eDU|1;iFbJ>nJ!>4I?(qn{4@9$Ae-*Kd!9KV4K z&w@9}+l8v_x2;3zK@I@q0$dze7QSq}y5@i$#`~JgDR6NjuTxs~c=xpW@=|km36egS zWn&vYBj6`O%my$ogb+<=O|MCO`|Ltwo;^+CsMq#FV^~mmy`y@F=e%$VOd}8SFZ$7w5Bj7=29sTZ ztU|Un!q=SuKs)Agp>i~Awm*Z}MsoTrs^mSG_m;)cPiq~X;fwL558(M@g%aRewFy~ga2Q_JeoI7K%Tmm@sG z4g%9@73PnkMdX_tT&XG;?h8VA={yjcifr{ETw>g4R>duYCeOIw{}YbU+<+!b2-!qxyptb&wN?3i2X8Mj2Q8;c`4=IiyIK5hq_x#)Z!yY z-u^mI3sduX0u9m8^#$fm9~+cKQcho)xykDB(ii1+uih?h1x#AIF}H(vB=5K!yre@_ zQ7}@f6~~O6z0c(tK0C%PmyyhM9;UICt}p~;UvTjI874@)iAiU2{vO=pB#d!k<5Wk) zX@jjPY-lOK0F0g^%>{zaE+EX;Z^TsPRC&~^+3%p@3k8fxPo zgQH`+gWk6=xL{ZgWxwdm4^G?kuaN-?hF|(%9CtS+BnqWanWQX-wk7WxJ4rJ8SbUK% z^{IKTb&fq9EeL13D&2oboUq$2*%JQQrB{;pF;UUEJ~i{8@p0I2^Sauo|I7K|j_&=w zdr7L~_M@k63eDPkW>!$~OIyW!iFVaO!kT`3#bzNeXCMm9K!~0*AGG;-dS}PCpU0Y@ z+VdwbeGAsxMO7=no~*9{xWpV$6I@58A)OM*%yV}EG6`$T!3FfCs^i%&uZ$#O8~q|E zo_gU%!D-!=c82BwXv?6 zU8zIXJ5dQHgWM%3SJeo47tdv{9%4sc^8qI{*c(s_<6n1kim7r4FU&@*m1ZFaGa4QO zMen_5oO>;GfHy_)v?bSz)a~3G#4hgUm@B+U#szyH?*KzCES6GAtc*=35$Snh(lrWE zu>iFT?+N2_kDR}T2LpCITT!{j|9=FC8G`435?p$SWMGD*y7 zA;&`M3PR2>C}%9=0UNlyUiuxr)|~=}%tGm>2CosquTUZiTw|C$_~2bLNxWQ6D)sem zT>;2M4}jG~j+keq--=;3JXs22TCh_7F{?yiD&n$G;IO%Gv$*ly-v!!k2zVGdZ)@{@ zJvW7YCVn$yo;ptKXu(M{;=-@Tq1-WTZDBG(l4TNk^Q&&#u2uhjF+ey?m7RM()4ec( z<>g2O<@bKvJ< zWxP+E|5>2x`0blIF|Liys+r|{P9Cl6S<}N1N-Ua)?DSdo8{;;^39g$?&v>2vSvRg=d8gP*Mx2*imEdE@$CQPeU;F23Kp5+5*`FpZ zOR{f739an=dVdC{+qWhpSdYKRpMpU>FI7DkI+}3-EZJv3KrP~Rt3%BHU$HQKzUa93 zE1R4$4i!ik{I@We^{XNK=NR{%-kA6ufl~s%Q_TMh!S-p4w?a(z0^SX>1A6aG({DX#Dppy0Lpcku%|=sD-(S@2^oG-~+}j}7{c{htFZ)7gB1V@?~?O}X`; zR26i4*?@*y?d$rZ%rR;{9&yFlnXO_+Cq@o^eVJe#zyZ^lFpB0=A$IzC2EpPWaf!*O zd}acadWAhrr&_SF^SJd*YiqU~YJ^*k(^u9Gz&8`k{Y*_Ky?U-b1wsA7_aqq?I^dxH zp4P~vw6*>4F`aXG>hm0ru;NR@4{x!It7GaCbw|Hsu~<^CJW9iBh>i^YCa;E5+qMke zy*T&|dvyh{MJ@o?{GS1jK*}it)mJ|Pp#w^a9V!}oYh>Eq`&lla6|oxTg>tWPIa=>- z*Q+oNmqJcw*sQ)e^E4V?8XkMT%ryVet7To0oaRnxm~?vUEGbcKvX8_wu&$Ly>beV+aNf$(5xv>d?t4eMs_*?67ENPk45F*RiW-z3xyVHdEp?#&HHR zcJ^e`;@b~5C#|kAS#oARhs1Z_+5W-yEb z<3Xz{cI^%yVX;5lOaL41zsxT2r6I@nG64WzVZ8c>DzR$03b48TwdY%z_oGP9& zE=r6+Lz)~UBibb7h>Wcoo^sZBY~^Ao2wJ?q%qqwy(JEV~->?-D`^D6A9XCg)n88P+ zLZj!jX(m}~_`>!*jDePEL7&3enCP>qn~dPOS23yf%(jJ zVS`e{A*f*gLL~DAk5{f`>X*RJ>0QK*^Uo~(-GCe-B3c<~I(Qz=DY0s5T2ndpmc=^b zc^vNJ+keIGKS+eSJO$NsJ%qY&??dQVG1G(a8l#eAwKFMn(iJ_6`I0WK;4YjZ%e z*1n1dyjx+6V}Oz6;RJ$FXR;~EFXDZ|;d{Dg;%augZypF;Fm@ph`KY!Lxjl;lie9IB z<285IDuVVp^bnZY2i@Wcet1OnnW~(lY~-?o;D6Y^~i{^-ZUWS280#0WT5#yG69a~!oL8z3sZw4 zb%x$rsv6(8Co0f@M=lmYVZT!t{sQ8n@Bg0sd#ImQs}Om>*dP$t5vIi|&=6%c4H*w( z6NwMI%QSV=i&qk-yKExEZ~4YOKfa^irNlCpX~3M-UUwfMo0ryUxTb=XVP<-%R&d-j zg`FgEiri^1)bCR^^@QHd(P`7?U$np@W2tU~5CDlX<`D(0T5p0yKw?!Q!?fZ$(6K zub+tC{p;*P8F>#)m$odJyQv76b8P;>CA8FkuGC`ha z1#VgqqLzg4)c4X;KlyTD-_tP3H`~*e`|OE)Tr}~7#L1)6=r(g?iu_=RORrdw=&S|T z_+~d?G17Tu%I&f*q!6zlc%2w)-nmJbE#diiGBo4V8AO3z#}8%PYFJDFJbw+9flT>m zKrbMSImf0N;jaU%nE+EFy%!Jp>HCd4YKv`+wYsfGz|mO;Z_26_wREMwO+)$rhE7re z2xwqXtZR{#uSI^XXj*>*uc6bW`Ytlx&_B)4j%uxfuZKlCcJPYxY}dNU53fBd&*PGM zj_bTFH|w(?QYQGkQFN>nF6I_zUUrOHP9NMaH12jQ@r^vbaekfOu4Gp_UQ9da$RpmW zShzytNoGwohw@>JAk$|k$JG`?*Ak2RO znE!Mfj;1@0r~4mK5OBJtJXXWEhG^zwn!8vl-yQ?h^B>D|p>fQAlpaCpP zJEz-FA#&cPC1xgy{@&;}aQS&N&u+l2#f$=0u`_FYG6A+v#MbU4%$1g+M2e3%O9&*~ z##?f>#mDIL!NF}<_MDz~YbEa7sF$(jCG(33dw-R=As(&QdCh3pn7vUhftrBdhi1=; zU28hDV{2_s6bxmw#5tBPq9`W&-d{z*^scksDGz39HL#Plza0n4t8d+GI0b>D-==dH zn2H*K?N2!&yru|R$T|_y2SNiot%LB>BO)+1TCr*RGl?bz0B zzX7Rs$%bM|VSTl6obw{UtnqUZ;Jvrt-E9}8ol|ZkHQrpVp_xfZa^pvT4S8|Mv!Mn2;_N{ z1yxPq?mYlp^eWSFcYGJCscfv2dbwoxYgxT$X|B|4gzUnLKDrG_qIXPL?lG#Hx?p;%x_O;L6S@4qV2~dmH@!3Sl zG)^8=4PDJnn|`&=Ivbg$z1Oi@9c%0o-90+!<6G!{%W&#hTfcqfjOP&2m$2LO_0q28 zJhFDadg0Es!A04M(FjU^`k`53ldxFP)~R9RTy#1~Udj~FQTqayjA#!*>CCQfA*4ej zQWar#U4aw`-Sc+a^Nv1#g-F9wz$2pBkJ0zYBxn#W$%+(HwE?V0^2@$a zb!0}n1UrSdI5t9#E7hlf6xelTrcH!9%+GqCL#Bg+WB&0Ez55xoOLD~$?$AV%H>;n^ z=7sOj+>k400W@|6WKo;`N`@TcNf1^S>>|?Zl4stk7tG&zhg5vf=6y@hr>U$oRpb-R zt06AXfS;0_xgnefF5$B3!aC{Pe6!S=t%i4Fe|zDn4h*9r zJprx*AN`U=8GK0E0-7!#eHir!7m?)5ZxZ;**Eidw%fb9d7TB}nL?tJq;E>Ld8W}5M zjw>{MEFp+ub%pO|zw)mQk zl&l*|IFJ^O96TK?_^y>GrrX{tNjRW`i{EFubZ@3!#_!Ql!_VVX%Pzozn^srqLgzGVt=u@BU%^w#>qF-@R7=z#H;!Fm!ZDMZ5;Tl(*6m1v%aA!;3T`s^L?yQ2b<@g)Zsm!Q zJs!W?Q)`G)$LXw=Ouj4?`}^{0`7=N46jVv(WUf*e4TLPGr|YMOS+;1eQ)NuCORc;c zQfYZ!!dJ%OX-{W1ef3Vbz`?);piB{5nZ4}@Y{t;)UB)MKUl)Kxk&olew8xiurQV5% zdzHY^nZ-4}$x8ldf`l(*U@GEU5y@{W*k*dmp36|IP|ZXEJ^<2XwdFIs+E%08^w-*k z3KZn8>w=>Jw=90;CRVb+SF(e(Bn3A$R>_4L9t7b?{_-M9%vaE}>rxGh0 zM%DD+x|L_VjGKNt4w4fVE7~1!Ftl9&Q2_vL-+1gtC$$jPGS})Moq+IV+jF}R-AgC@ z*+B-$Bww%0fC{uhbqmK{kRLU$e#fRa5?9X(F%8k6{%vCbUzDG&4?pVl&;cwy6=F=r z*Ne^Mq_rtl$jGy7yRhIwp-zi8Hph@|Er`V!-_2B%Hpi5!)E@c8dO_@diudLYqZU}z z#I9@7bzr#gzHvuzp{f2^{xV)(cym3D>L{wrb0M*L&Lh}Nu!&cOwOv*|ZBg2&OL%E7;F4KMZSvO3fZW73&%3@msnj-JJG!`U|A!HQN~ zNxcRTZW2{W$faMf=4~Gyrv6jM%bLWK6A zsW(5~#wZ*V&lU6n4Dg+h2>>4fWRHOE&;nZoLb!qDL38UDm;Bdrn1&}6~R}W6+`f75{K)B`rVQdGU}Hp zZd$VV7=>$@92N(8-H*^_2onA;)nAO}2eMP?Yn6NMp%e4O{H_~~K15ALIggY(Wb;ks z&+?h}N`BkbCZx{}9?BLAd(hN*(+3UtBTc9o7pwqlzdKsB9IVp6e!@1YZ^!8r(m&c_ zl==d31S;z?u>H~k?|0i+5N6<;+Ba=nKn^S4VTU%V%0w&MyN^f_xBuY&h%!%`7jP~k zNk|sr?e|;}tDxq|4WPd<5qh(snWl<`ujkw8p(pW25(7wH^;Ab@%@_MLWB^tMNd=97 z!G_uBH&QvmnoWK_qWeF~DnIYL9OFosFLl#}G3dBpQl@QLh9ol++|qXhREGl@baj3; zLrJ|@^an?bf*sqTZhU}yR_-r8?T)XGp8a0HE-YaK%{)ORx|E!SPYlEj#^ryzgyH2I zofp5(p_CZk&`HFONYOH2iHO^zWf?JxyT3qD`_1_sKI1o=L7mU5G!a?-+{z;zt|WgSC)&Ya^*Qe(iv={j`R+ZaRof<72K2 zi_?kNJK&*QDmBOW@Cb`Mgui-Esm~O)Wy#dFEAHm-4f6Dc+p$`peK3qXBQvFNaB2$oT+-zU zRXMyD@-dhorFiHF$uE*|WGueO86lux`ELDHjixqy2^sD3c47U#AbysMflP@`QT3C= zp%h1jCfu}}4V6MNw}d!FMwxo%1vK!F6$Ln$l$8@1n+BH) zupRpmp%!2n0%8yZ*jLVOJL;kr)6T3TT*Py3gQBkZz+e2}44NIQ6993FKHcv&%Yr@8 zti}^%D2$hDiK>x3PeerCgAJ}1Wtpz1{*R7Lc%v1L$yd(lp}SmDO53m$YIzT;y>7EiI|x-vqIH+ajx6=@x$E%upZ#PM% zCj)gJP`so&dj%Lwb7F9`eVVprQ@}UH6fHA5JObrSX`*QLL!75Y$H0CcoTjSSN;XzT zYKo|+XEr&TKpj(lMz5I;-r;=%@*`xMvs*;j3-1R6wgA@-I(hr&o>RcP-_l4?y{{n8 z6C{TXAlCB|!+msB@=9I`mNrAY5n`vk)ttVFs^Wlk2)=q$S|n zdz{S6_bZ4mW3>SdwLmS?$%DT=YsC)nHZNWNemdLM$Mq*uMC)z&n6!FjNMHw_3ybrd zl;~fzzGhX)jysRMMcLO6u9%Ll5?s$poLr9B^|0Xk>!n?u3+7JUN446zodL?9KBq#a zT3<7QIpyz!%NteZ^R6aOJ;>aI0)(Ou9WW8bE`N{mOE6Pwc2|BK)F!^TeZDA$*T7X# z!l<#XALQq`PfIK6n%t&DP1P!u`Hri2tIqd&+o=4e&?aEwDZSPaRm(pd5aZoy!Gi_w z7<4W{=phOIqC^6(Z}K*n4vs)!kczO5w&R z3^Jvry(IX99rHrsL-y5qi_XBX?tdu62!xxl_}W#m;G>z78Bqqn-ZFl0 z4B!WiB`08bc&>uWf6YVe7CJck@#)Zp+mL}GjeQ>GQ`J#&@m64OA|TGg&>2gMPPd?) z9%lYKwz@fVLixw^=OJy>Rpsf-1Av#xM$Rb7C7lrE=BL@|r!W8MjWqVh8>w0d99^*S zrBkj%UsIX68d_@u&>5{ZWSiA@S)U)9zp}M^QrH>jRpv1S^=@BYx6!)H U`Hgm#5_MAJoeyQgf!0pM`C4QNs7aY18O-<&EFndB2QwE zcEC;&$@>8#4r*2Q1EjB2l3C!uK7b7mQtiuPcp~ITp^HjomsLeDlrR8FYuEf3!Wh=C z6$%3Q4Eku6w=67FN#G>9qj?<=To|GSng-Ev@HqG{kSc%%DUEUqqG>Z;00>_`rAo<4 zNvaw^3AnKTmDVrYhz(fRqt)Kh+6rY@2LJ}Vt)(Bp=Cu@zqB+oL*)QJRKce)h2K>Su zq-WDBYz@;wW!l<~0Fg%7#W)ExtVgxpJ>_Pt`Fs>7SbP z8LX+O6-=#iymC%`e_58avtbj+H;Bf3$2jPj7Y(wbJDOngPt~Lx`p3G#ezlFT(Vgzs zm%RZcUbq7H0O#ol$PyzU%G~vg!kML>2*QQ%;Kp?MzPXcAGw+ zN@F*ADn>xZ;}o}0OZAX*f|Y^}`1Ig4fpd5H+4|=8+{U^)-n64T(DOTUKZ(Z{ticKC z|4o|`Fwba544Um^XVSjt&+u2905L`-xzADW;Z1rydi6HNwkP2%=_~ZJ>06Ho&{!vaC;o6atcei zbe8H!e_Frp0=0>`EiM0s~r#lJuO$nKJoBHizT*}`a>+F7b= zr_P`?uNnZ>A43KM5UhlCjULX@y`vq2KKR)zggUJL219+-pb{ojZfTIqxCo z=K{OyTud|ai8Tk5Y3B*M8E<@}2r_A<&<2O%B5!q2*WPw)aEgI{DtZ@Z&yr5Ny&X#+p z{c1S5PF_hNm}#`E5CM0fSqV)rynRQ z4BH9Mw}vkzdj@>ruPUw0X*_%<)zUi}39PmKJeu^)Y`61I-IE!csk7;XeRl1#B$9sT zxiENI1nTVX_vJ3rB52p`K9(kxI?!xe51!6jOB}((!G(CM z>|l>#x4EOWoo}5)-%U+CHD063)c-S(?2m6=UeB4cAt`ly9;`WtyS*p>O)hd|e%?or zI5y=|Mp`;H_9D8AZTA7atLy|zJo{L3d{Q|(#tl{njS>imGEe}GWKvq5(()n%A@7cw z)9hFIbGm8uUwL>s$JKIXU){B|PM!_7iRZxA98H_6sGNO`B~Xl4=5401PSpLFo30M~ zS=OCYWhPcYo;$3d|Az4o#sKx}IT|4pj76n>L zM1Jd_XbQ@#wc1U-r=*8E$~L*k++31qd`4Xd?imEqJtDAvrsD-&=9PoVp}ZPZz>*7% zb9JYt0rZvUS>#`U8yEw%I!WFBsi|zwvTBQW(S#gNLj7x}b1k=)+2i=~_$d%DL~FCw ztGBBTvCd?aGuv6ekCAft;_+(0b;||YeZ2ZQ(E7HrAiPU!Dw{MLE#WHhCCkChY_U-{ z=9YP$B$g(Zg@j-T{hmMF&eYLb_*|r$`T0Th{(G&M$rn@`G>*Zd3ad_lkmUSsx&8+( zP99L@g95i7)gsK=X?R-?kF*LxYsu(zt^J)Teg1RK1#S9fd?tRlzfnP%M(N9#gvmm9 zAQhk-*_P#LM+Bfz>>B_T+;A`e5mcXn;5kL1BA`;@F}}?&pJu}c>!|@~71r?_pMbsd zxNo#8egM@6e?y=$835Mg+-8NW6bO;VgURKgH?v2^;>L_e%S(S z)I`xc2FU+FuK0%x@d%Ssk($(b|GT>Iv5mzP=QC`d3I76Qy6OZvc4ljji+`$HHBeEf z#Jfxf`}`ZOvwbRstgnlooc&Xs!1K03%gTGIebN5{bFLvhuAAZ!-*f&{SEsqMvw64X zHN)7(KcF52Og_W6zZ~m$89`U^U%vZiZQ%PsASMSNTm1R7_7* z2yun~{!`t_YQmu8sHmW0+n-()VSavr`7GDBX@fu2{SO8y6|~-xwvBarhg=I#72ND% zhLYy)K|Sh`8B4!2BW~fGf4K>MK)u6#!F|m+QHRqN!HYW2nd)M^e@mIC@z}bo7$-d3 zlZelybvawQ;4w`|Wm|9b?a?xCXS$ujJvXjSi%;sgw>b1+Zaz3%G0(jM94gxMBnl?{eA_LtE_2x( zGw7D$c2Z~^8#aP%W=XS;Zx z{e>S+hj_yyhO2W$HOg7=X}3buu~DC^4acV&nfv9`G7}UgcD^~KDASLTim>^scAmP4 zJr{^aU%&gVcR!~D1XR;YM3?oHWc7Y|usy$l@O?zWG2ta_?q>(sN6jGaD{g&qG`cpA z?&bEher^h|o6~+ian_Z8N#NT3AQn7d=0LdejrO)ghHG5q-4)_%@~qU;b29t4Nc?ju^syVwTP5$_}SDdW<3FE1;@9NaS4o zWu8`g;qCa1JO2Gnr$u;;#-2zE-nlu=cc)Df+AM~gi?CPLgT0-QRb<#;jU9X;uut{@ zd?WR`mwQ$~Nqmi>)|Cb&9HP{nRdg+xYN zU7vTFS5BVIxRDGB!`H?#ZEr@J^apv)d;chKBmY=gBz~Y2tb5Or8xg0r zSrvkRZgW5k3Knzwu7FmjlKpATM1DjGu%7AWjuCrcf2PNk+lyY973T$l86+}`Ok9iK z`x3kN#I%JbpQ&%UsWSAEK?LbR>;P|d`C6NDqD9VA=`iQ?*UbaHO1k8oNaFi)j$4|$ z&MZz`0Y%=aL0lv;zGN3;%>A@*=f03R+EWP%0%%!*1pgvze8Y7HxW2!{Qp#sV>4&iN z33h#jUU`oQic#6KYvqxH!$3^i#o-9{c1|0W2v*TSO!9T*5S(X)fZM{i9nRJimwd_v za>!FMs`Bu`DHdjlXRT2lh?m`xy8@ji5#Mw*(#(_@sQ6-Nm+l&P*vv4L4I4OW2sPU@ z_Ghq~U)zrt_p*Sz){~*{zo8lMGR)45Q!O6mcP~StqWr;pUe7r};o%Mxr+4ePm;V^l zurLqxMSr8-@$-_eOk0{u$@46So*RMwh;W<4?X{(H&gV9dxRL_03$Cx|I_At1Z+*C9 zYU<9}O4OQ;D&H@M`*8!@W8%a|m?rzUJeNdhJGbp)b#dj9a6mC5q}cLpj!ds@MX7(Y zbpr*ndW6ZGzC zzO>F#*cMq0E-*ihNqlz6rud+&+HivBlDAEksuNc`6644KY)6Vxgl>UVRJnn!uXklu zSp3q8&V-0Ew`=CfRj+D@d%@P$0cQzh$X8bb%kS2e+#d$T z*L{KY$D)aRW|d+`$?E)Z-@ZSPFn1DXc(WedR|7!Cm?4=clE?3)2Cul$P(quW_~!}& zEvnsC+Fq(}e(Eo0+vwlHnHH~J1kdvgV`5{VJRyz(+zpN&6$X-+O+qA1(ZWw;%E!_l zZW=x>mGgq11Q!crU6zEOM+Wxz)(hs_!~D5Zo~l>-+NdzpHATjyPRSwRQBscQm9JVL zwsjyECXJt{#P7=jfdOaJD?vUQ+c7=gnGkEuExnWlg}=yp%eT=t(zgkBf!>#+D>%(< zNe`!^*P#@i20QniWy@COoV+spm(i}CzXr+BvW znZ!~IwNgwyAyvwkBxCY`dxe02#L$j`xmv8v@ z?L!dIT+;kGn&u*cCQ=MQb0;K+83J?$R;F~oPUYT!GE+9dBwhdjmb#DmZc$YP+i?Q; zuBE#~>VS)%QfYty>1WS)=nMe5WMcp!Su9za3ovD)t(k!J?|J!8|Nl-B+kDheXWiS| z+np2{l&_}R;8K>JQSad3@Thx58s$`ITg|307PA$GqgQkx0%OyF^bH)^tk&9MO76?(X zAZ9C<3Gkj(?&)@utvB_jWmZ7=Bwbw1h0zx$>Jq_18i08wOdzc0vdu3psw4X^_i3O- z;4krY@&z?X>-t`$#YUZYKWF;R+VJHI@pyz{LU}p7#sLydJ0iJ`~~K=*d_ z@y&2nvC9TNEygvFtTq_5?R}(MtZ*nX+M>@ZD6NJ(r7M*#+z=B#t%eb3aElH9axKsyR`d$ zB8R8`R&x@QWm+#u=+?Q0Ur6S$Ch0A0dGWvA{f*Z;u4UDD6ZRl4iiJAe3TDSxdM9KYZ z?kO5*i?LOv<^_>+fdx;zaPzO@)vv@!&7`86?_lo%CjJZ>;9%iK=PPQu=xKp~z)NxV znbZN=g#e(@5Gx~0%`YQE0FkAWmv-Se_|%Myk;v0!q}uX;1Hc!ael~Z!X1%D{mCG!4 z<~X?jPt<#{t{1YcrYg2h)q3pyd?~AEO6mtCL5&@`#+rK z#)@g%-I}&PC_GxtInI7VvAzrFBUqbxt?9O}YpUgkxBH7BmZ{*WiqRc!=zNwoXJM(R zNYD+v!vgyQBd>}dzp$P4G96iYk)h+EPDG5|i^1m3!;*>58rupvQ&(3yv%WD0sTgAw zM{dZ)Y1BT_6K^Dp=^f2vT#kzxtThuWM0PE4yesuD!&Q!qHc@ysr!0Fm%N7z8c`Y4m zik4#mg0Oe8fpQK~E_4IzfxA^! z(@(GOsE3Vs&yH+?5ga&EW=5R8LQ7Lrr-4q?rO@S4#@nTZlEkqHzUZNmT9U^s6q#F8 zY!vOX;Pco#?Dv+AP;xC1ry~TObDzy`Dsk(7?lqf#*;`6Q?co;<0pr6k-DvB-SqF20qmg?phg{DScP# zZ<9dFLX#H>Z!;&Lyt|jOIpOVR!CW3Py+Y2_lILHP>a9E}iI@DkBZP8y`KgiT0`huD z^L(OiVE8NKW3Swx%L374gvK+p;PU)4aM3eCaJl6UEQLhQuVNK;7OIsjOJ(f*Swik3 zV$_R!pAP)2JhaAD?=Pa>DHuoP_l)^f`d9B&`wz^VZlM|+lGQR@xoj3xa{G6i+3M(J zjUWSfU1@OVUtMu`4^>wbc^ytsS-7Yh zf_B$Kr4h1W?Q5@^^a`@L`hN@rO)*bd2Ll$?R_pz1?;BAqp{C&T7 zxNBYl%e6Cc9^F&uhJ3WlyTj&+#FE82N!Xqh$j{{iPSRH4nIAe_S|6(Bg>?BBqu;;? ze|>(A;mUop>)j@QduVl6c4LF?YLO*3vPu|9rN)BmiE&!;mFSVE3dhOG4-(F7$(!Pe zQ|Vq#B0~3F+{&IsI4O-tU|U*M%4c=~CA9`WwE{EoSc`8PxQe`rT2ckRxDCG47(#AX zWIxY0H`!rHe6!m)w4=N4R8{d!)F+_-0=fqZUAvMMY6;(c2P0h2kk#1l0MCte9TH!a zH+f?n%iKy#<)$(6$a1`HX*96JtOJ(MxLg!&kpOzl!=R~a0KlvDAwC{9-g>1#uuzCis z4@p(g*n-%0`l88eP8!WNLn%fIiMz}e$3`VH78dO}I@c{HrtZ3Kv za9;I^m08+O475l)@e(9R!}24uWr$fdwwt`RCw{QhrP!Sjeq5Sjbnaoyi?G|6fIp?xj?A$hF4m3MWifQji1JgfTLDM zE?5(r@Dr< zuI)x$br6&Adsgsxu?T7A8h1XegzrVqLA8I5PNh{VYV3VUk`ZcnD)R!R*ThR??c#Gn z@x)z>;@10`PfpukM|kPXVNiMaCTeVFlhlOTQ`D8S-6R_{4rAB7?uEH(Ormm{lM)xL zaLt|`DuP)*eYE%QT9<2QmussU{&<}7W8@>sxoN56M36m{E$NI}JbenRpB;U~jYX9o zc~V+F2W)l`4{;aWCS;FrCXEfFimnC0Os-$veo=XgO_Q(>3|#%4teAD{JqeE{6|#n9*393PYskB`SXYr1yWJjA<_$du>wTi>@RjXhgvf*T zFI`;SVi4U@AV$`kHK5YO0?c$1W*$F6+$S?bd4f(r2lTeiqV?w<509R0Hg5TOC|fGa zTHJ8{T@kB6L4qn)zO+Ypu&UBy{zX|xEz5VePEG!jn$WUPg(qhw!igG#81gUcWUTa- z`r_AT;)0n{+g=A}HcO&ZUK^edOSq-pVxR#Wg&YNLiynQwfB1c_RHNkdhw;~>i2!4l z%)jwEhhu@BismHkv&B!-J9x5#;44Anm%1`9HTn+ANFMhCcs!FA?Rv=N?*O%QG02vt z1N>}2PnC|F04ZY-pl&SId93N04Ye}C(0p}2ma_Ut?G8TPPdx5Cf-%(cC5rCVFKX?O zjebBai5}!%3_tEVqNp_vn3=Kl`mt0Ib$F4E;mPBEg3ky*C~}<&3ptiDqIMldN$>fO zYY#bUJK`oT#KLeaRd5MF7hcl3dR%Hb>QWg{lWRODkEL{}?TEhd8_#1C`>1_NQw2hb z|5z&ZBLHN5l_Os$;Q{Dev&$|zMx(wQG)6xWg{z5&arK6tG zayFVVgREH)G*tdo`j|pi-`jHTmK$fd=$FlUe18&6<--4S`U|R#1F-Zf z(_G!#yf;iGoBw&P{9j}gg&(z{u9!|LV#0_C-f%K!iX literal 0 HcmV?d00001 diff --git a/docs/modelarts/umn/en-us_image_0000001157080843.png b/docs/modelarts/umn/en-us_image_0000001157080843.png new file mode 100644 index 0000000000000000000000000000000000000000..9372035335f43a9461375801315260372ff0c859 GIT binary patch literal 10762 zcmeHtc{tSH+dq;KfmYq=Wn~N$3L!fpELLSKKJ@M_c;;gkcPWeGX|M?~dx8{$$-YvT?McM3U<9u%#LH#KX6Mey%>q0B9;(y97*o3yAlTo=!J{!>T|umkNo@61(dW=)rSj<)jU8vz zhq|g3>bUmJEsVwL4cmG3qrzP6JeuW(`U`LD_$V;(!aqu8#!#zCkvi;0zE(NW7wZ>P zze`=L_oXX4L*QCi@NJ?ja!EPhw`nYXS|-=vX1_BZQtiBD7}z&Etj*z>3Bq^2Y^vPc ze^kKmfw8S`_y>=fuD8(%A-L$i3(^ez6bilJN4BSH^;WICN9%CI^ZkZM<*@;?PLa<% z=qtV6eD*zd**oE_pRt3kC5oSez8>$rYp%1>MYz_($JEv?_M8x0sb+l1-BIVw+HE`v z88ja{$zg;(sa|94H(l2};Cz^&{a^v9BbxX>=-&+&!%DyNO36taruURof8-~11Z{c> z@w8V@9?!S7^<7)OZfPv1tFXeL_m-ryfddIC`)D z`7MKO3A;N+9CX-P?;SGKTko9ToA1+j&7lTZ$GbysYva2spP*CPBc>M5JKjY^#4Tj2 zSN&diX}x*rEPFgR{GD6$#wN9fJiLhULS((=QHv+ zRBs$o%mubhYxr1T_uJIv5t1< z2V$v(ney)5yoR8~5$});!>Lfy6|i1_}{DFP$$ZmG_a4(EhyEt9m0asJjr%bN6U=pI575 zi9&JX`i7E+m^~-t3$6h(rpQ*z9*IIjq#<+Yf0WUw4pHgT&_@%rJNhzgwK@_+)|SH&+J26F|Bke(DufJ zvfiv-px+_?78_glM~Cu`O<07+EzMYzRiiE4BCB6?*p5E_-YDeY-?x=b%W(CTV}!vV zzwln<=!ssRU&rh8f0}4`PPRRvmoB$!37vSp<*+odH@S3z?`pQ~s7b>e2iidWI@m8u z0&xr|+{JogA;1>fyg>NPSE>yJi%G+g*cS|U|jf8RT5EBB2=;7sAf0^W04mdpyRgH;);4k z+P7xL;N@lQIEMLN^Z1)yzY@)MQ-g*RN2-V!JS(FN+z|>s%x0pk&R^nGO|2p;_9RQa ztn>g(A}(;m1+E@0L2XrNuSG>O{7*rk4}nkMsZ`>UZtY17YTul%Alu)&)_x+d{QWZI zf#=cvdy|t&u%kVeYc}fiK}`+H(h6K}t$9=Yt%S#LJ2_6+gmRvIF6t?fq4MvqBEpK@A_%D$Rrex9Pq z2*2rkerhHyW!)-5#-)n53yl!--k?9J0G_)5ktRl5A{=f&ImiFYxzM*TYmgpdcegg> zOEyjY3?jb7A0HNg{8yLDWe;-Q>Xg!5zbt=DvH!Van=qR-b<-t-{rQW@7g}rJ{3WU$ z5hItI@1N_5ppyNkw1nXFP1o4mpy$0yz9}lloU2OZE^g{1dT_gA56(>G3sW;S4z+$h zRk!x8X#zc5>Jk5)5i5=o+tqf$Cz`d8ITN0HT6tS-5bho^gA!%qXu`OIe{@Z_XprFM z6GL(94bF|FVq_0#Yi$*#>YgUIDQO)NKmGb5BI0O+ZsXppjY8UD^3}1}Z!24S0urJF zHtK|`Kdj!#D~ly2ZK2!wZjbi9bK*np7TMroX!0%4q_g?^Sv6?HD|saNK1PL=pv0_} z!9=#YW{kWXBb-@+IoU&SoN?F_3~z&R3-KaDZ4MbTM`n_ec(ePtTnsMyj-`OjPZQ(6 zBQ{o|Gn=#B-|}3)Jp@O*WWvv^ECMhXTtJfafZ=+)!b*~O?h=c5<|sv9ACbypE%dh* zVgQ@c#Q47>b|4%3mvSZ-_E0Vuy#9THEF;j_5AgCL59s%7|GOz$eT0bHE7||vBGAtN zD^}dQefRF&<*GN@?^1ty#!7HZ-q-FA0(gS&Rj)JGxW%i=c_4$g)UUGMpy&j>pq z98&AH^B=Q&1-Kxi0uFG^`E3^{TWBHJ>_YcEqs3pQqxWmkmN!SYFk zZ=*g+_%`1&Z9cP#2bQBWxYL2Z74E(1t}TSTuC;$sXF(9>J1l+19H`58XAvTU3%-o! zDwM!RHf#Is(So!n!X&HWVeqS80256EpR&Pkd;rUvI}6~@@>sBZtvv)ch_0!s*jQ?<@FQ1IE77t&b^$BhLjW`e4y(ye z9{P4orNgrB4T!@d=wQL7WiR45ZG>Fj;h#+kqz#^!5kcmG>^A@Qb%v&eAwr|hU0OFP zwe6^$xX-WP^t+{V#e?OM0Fo|D#hWVqZqLxEpdiqt#L(0Gk^v_5eXczaRa}@k{o^E% z)Ri+zBO_nbvjVe3-WtYkZFJbFFKp?_)iqzVvg@{UZ`^BFsSM21g%BaDr?>H$<&{l7 zbrUlGax_~d&LG@paB#Tzl756*i$4DeIg0u=zYUx4bNu-Q&gYst;7(qBAVW;be)zq( zQ5xVPghZy#l$X?0kv12?#?~ENBK7&L&LYz_vBKwCwnk^eaD9SO2i&$zbVREU``FvK zJ1Unnxjrw|Ng%E7_;XXHOsqeXkGhaII8 zt*lG!-SyjC1UN(&Z{HuEK--a(y5^t_!(RGe4R(keNG;iH!7Obo+fjdHdJRqKjbLN7 z33s(ckfSVOQqL=^EJJbJ!V|38!-j9ZU^_4Qx62<9x!y8$>6%SojBZwq$}a;OwwCYt zIV~+`;hwgqzYL1B<{ca_-b_qH_)Zo3*m*q~OR{mZPH287gsK{2(-)C8d6r_>6s`I> z-YMF*eAdMD&pyhvHNV& zG4VRu%Vxbp0p7jfkP*Zs&1a-kkxLp)Y7L@W^)~KJ*lQ|+d|@PZf4Sv!rE>ilmss2h z+qo};N9gkKiNs`uio@*A_#Nvz zBoF5^YLQ!?CEqC<0)oJbn3AN-gyDR$$L+{7;N13tBp$fhNZq23G)0dsti>(eHrx~Q z5M!$oVWT7>KFW?G7k*MGxrK{;Oswn;Jy=*H8fOKdu^PQWD9_h1DD}= z=S-24D&x5`S|!OVFSahXRJT%5HfwFoIdcW%I^#o=z`-ncy4=&&bBn^(8{0}n$jk8| zmp=8bCXSXq@Mu-EqtVe5ziPC_6Y^GdY)RJ3_ADHn2W0=@Od#Ga27>KHD~V zE^#ocjo?)n00F|Aii{lK-n|R%t~E78o-YTPMu--r&(G7i*;;dVNWi?Zw3?bHanQq4 zEKcT_Ga_O|BNFZyJ~PwL2Zl_pNjy3@Y#F*wiR>xUj1I!<!G3A;xE-;*vmop-`9`{d3FZb#5Bn2MUx6Y)&y# z?>=AF!{^R4-(!2EAK7esjI0QiGBa65aWeZe@q6+Op!*l?$b`U{<4%zGJ&Deb(MX9Y zPEg~fu-6#y=9mzpY&{Kf)x3NF)Cpx$lP2K9dsCjlsEFQPRDJSUqqBWyVmBa5emln~ zinmrD-MaS?e4w&i-3I!=vqVPa`aU!GeZGJVl&FIJ14a=Emf!bLmS9jY^-CT@2u~lyko4tJ^2@TT5Tg$niBKCN_$ixW%v7E6 zSG1AM!|m;OjUVY+-fkG*?GQ6ZD*GrxylHtL5@TU=!S5kE00+C_cV*r5xZ7AgYsFHg z#KIl<|1}S71nBV~0EBxoTbLt#S5^jSU`(+Dp9k>ppl;${=sLhwXY6`@?r%3E3jo>k z{)Zuu4>tG_;L(8>12T*O`x#CcJZtsB0p{p7fR!Ile>7x*+$ert+n<7*Y>=$wr{;;mwi4r5B+F#$m3 z$L1*J=vnc7h64WYZm{xg?Wa#CyeHdI=QI4~7wc=qj_ha)PHhiH#565dQlz1+4NC^6 ze4A=d-*@4X6eN4!*?-KIUb`j*nSG6m37|5C3V%KXZj!{yaTbtMnPk<>J)Zjm0j{#* z8Us47#cyYLn?^^tNNiQ3bH>rI=fF&k#SGtRc6nW2xnmfBA)AJ22n?AsgCfK8DkaP9 zXzGpFYdauypJI|j1&FfTWsT0doJy2nU>8pTMyO1xaUs9>06Zvg(n%d!F3((k$@14Y zcQIJ5p-Bk=%a{4U@{v@G2W_szJJ}&{%~!A2&fAXKWohoZV)MA*y=sgw**s|dM`H@Q za`CfLpn~#F5SYrv^ZOZ*EC&Lx+LGvO%@U4V@3wLn(%D=cv($*4fj=;wHDniUy$wYT z7G2DU8c2xcbdc5wnXQRoV)+=44LzW)wN-IfsCyAC2x&k|0?D=nT7^2~8RL+|%3XIG z_gdY((UD=7aznq**)QbrKDSch!i-^8Y{w$b3i~4SU0!((t;MR^!+wh%Scr`J$Zgem zCQ75p$}Ljf!2Q*9yB1m7|EmgIaTZ7($@yy+L-GSaa@|Nwe6XawDoXLzdw+NGnTNQN8TE7D#u|n;Ex=fe$DSb{vVP;WAq(G+Odp``Slf z!osiu$b@;&1`4_0e7p0iCqvfH|M^n86GNYlGho1+D)PJhpm?Nt4 zD<4t%;pcD4WuPdrwaz448GHDZDa{ye9gVSVp+LLH&uN%i3UE{2_JbinPde$&Kb+oM z7vIoaS!odu->Qt~cet5RDPu5Bd6R4^Vx&}5#Vu`JW1CETDQ4-StVi8uw5IXkOWclY ztqVAsRbFes*Z)Z`T6>Y}9dQj)Dk0;^$*frl60B0m+NsHdtpa{EfxdFt+|vE_Wl9f` zYwsFRMpTu9gI$ez zdg|Fw3&9=#bZHMzl+|6-ambyH?O}vn&qm$Qj3~P+RinzY0RhmYk8*;bYQUD|E+#9a|S^rr!vv@7<``+USg&}5YC*uRhnI9PF&3cYyzhX9tUAIbt z_BR7#WHD^@zW2G%aV%ehy^&D`A21maBRd<`A%(1GoqGuV;4|}p=v1?JMVT@~B*qX~K>Wvx_ z+M2rbM@GqH>cSa^KAGMLeR=xCEk>E@f!tXL*|RBn;f)j!wO2S<4cfD#4l~bHk@bI@ zfFF3idVVVu619hy`GZxqJ@4Vr5B8-eF@6952y~de;82zWt?y>N9R;oDU>;^4PVdcu zjMgijMS+(8s$(YUM?;A>B>`PsIQsx3=r1-Vefu|8HyJ=L5qW!wK}Rs;8ArGeK80-g zmE+x2$d;E6F*ck#Ui|zOP|#10$^obv2Tw5Bc-*->g`Axs8eWn z@4Ha3t71k6#Y@xGy9rdDx%nr26<29uS z(BJMqCJfOdm~=SqMz2>PpnnWFCJxabGU+_3H+E190R7X@F zQ2HC^83Dy*kIvo$!sE4AO*P>7A(oBL4D_3;=fQoC&1@h;{9h~AO&}AtERrGt6WW0f z3MnBc7*puNJO8E$JRT5mvt)q1ewEcs9D`}#@BzVx)@UPff76Ijw=8qwjqB#4-?m1BGtLezYc1Y_0EBo z8QbOKcH%WWv$a5<>t(C|sACZ220AT@+}o=h6E)boa;E$fc6udv~aRewZug z{3$UUJg?g`wbUSBeiyp16LR%mP8(@E+wJfOH}g`I*{q#oS21#yjOY!F3;*Q~?jQGu zMh314TeC;1JTutGr!9S7!G;jcttE0{X_=tu)WZpP`qRa{&k^7<_%QA8aXLTp2xvx} zG+_@{32(qS`Gq|3iAGtcdX%hRy>66F*eoYM!eJW|;2sYL-K`f&}pZddzwCVCL0B zXaQ6vbtGu(Docw;)1yp6{@GzJc-ot{rx;1NSTgR!a|!Jh3_oi0ut}wo%?lUyC>NPQ z1<>tiy&pfg>{*mbYSC<6AGn*2A-u)kgKC4mU+A7}aWXsf955MX&Wv~PE z!INb=4$HSqLSo0(6}Fl7yk9(w?>g7T0eliP5%nccozk;4nErADdCexskK6U;k>;Ur zM-N-XZ@rTEmD`m&;m7xUWVY=O@PGMs&MTQDT{)i5n>?tR(r_opT2N(O>fiu*N5)P0 z{GLC9Vl-=ni2NH}Ciioabp}6?&CmGqef{LzJ{iP?W1g7S_~&)>a!a3APKHUN7N&Qk zRCX_G=ytHMa5h430W5o(%a7YX!ihc}V4krAb{a@Fk32mj;TF9@ujS2bB}~w`5hCY2 zY(&gFN>s+3Qv>y^^FF`BSe!b)kD-eD`p3a}PV`kMq!*AvMZpPf*J)SAfR(KqoZxOJ zh%&FY*>8h8gn#W2c0qs_nREVjI50Ng^hOGt#Quk>B+s1%XRkAFrJxyr0z$!Q>%Tw9 zG|~U>Dplk4p}=GB#%|K#6%JM7K{V@YylPjEUDp6l1dy4!Lqnb&I)TN`yqk+Umtnaw zlK9-LR|>(8+1!@`R>^t%W=ZTgn9EgxGfqi}qq?iV9u7Ih>P+4+`?8G=6*t?(vQoom(b|ew^}v+kD1-j5L_2 zppw6bdz}I=QliP94pPqPBlwUZIGv3!bn*_+c6gYwNBW!tleoT>SLCAtyd1ZvGY~1V z69%t~z^K#*m4<3ZfQA8QArgjHhCWu;ytbv3E5*xJ7vZb#XbS#;Pr3Y-$aM3z2OnC? z|H5LP!~8sa#i4{QDi8V;doN_xhO!NUw*XZuN%N2N^a7k4I&>GNeS=<6-f9K)=uk;| z*s0ClEgiy~@)mi%zP$5H(eXxIF_GD)62*&#eR(yT>+>%nm9A{qf)@%y=pfo)`nnV5 z=;zhc!(wY%@Hu7&qb&?MT43gEpg z{Xs#t=t?ioK6+rrs-Cx)y-q}L-=(CH>qjCAqO8ID7x(V*@21LgqgbP7rKG$H!IFR_ z!EfN%g8yapybleDN%;YDcf#`NbNzt{9j=Q{SDgGfUi=)c@sPnsrCyvtf<$BXJeP;f zlzpA``NUMI3SqWCepWR@F#>dfO|IF6;%bIHMu=ZL1ClbAu$)7`E?Pdq!S>Nv;-p?| zl`DKP_z-S$vOuxk7@g1oqMiJnis)?2LFT;QX|P$s!M`q>|xwoC_0LmWouR4 zL22!Ui9@#l6dz^0i%MEoss*?Lwy<$v-x=HkUm?)Bg?!~+{uCsSwHS$LCp}7a)e*Ld zdr$Yu!Fc@S)I+~DkPP?WGg>D~)wO<;hxglaqy;X7uLzYtmci%hBwk3l4A=?Vdb3+R zlirFi(eU!q=)sPdiy)AueK<;Wm&Ec#PLMNF*0Sa)AiK=9Ftc zQ#wL?>PyBP_0sz<*)gS4JsTnb!gk}lw;%9x%Ir$z6dQA2{}nr7zwg7z z>rUBF%8N?Xn{hJ-884$MgKGP$vu4Ygx-eqbJ!ksmvbSFSfvC?4ugx>(#r+pf1X%<| z23^wI81wtUd)ZJQVQpw?&9--Y2EAD(3_gN2AD~(2J}?%JwuYL}=27n38#}zXrBfQH zj#3|D!2V;B2!A1pCx9p#1-t)36f96EwkUxKPT8LV%YBW&GC(K$zX}cmlSqC8FA{Dc zbGI>Jh~s{cqV_AIphGc0I+FnDcwXE87wM1ywjtd-WXkM79|Mmro(hndgZ~tsiDt^~ zW4_}r?_kRIuZqR2?q5NZd$K{3IbiSM<3;v#S}-~rpxXi-T^xh=urg))5L3Igvpj_!m2 literal 0 HcmV?d00001 diff --git a/docs/modelarts/umn/en-us_image_0000001157080915.png b/docs/modelarts/umn/en-us_image_0000001157080915.png new file mode 100644 index 0000000000000000000000000000000000000000..39ab8f1459138b75530c1ba2ffd6bdd5a1505db6 GIT binary patch literal 16516 zcmeIabyU;u|L{-96a)iA8VtHeN~0)>q*Bt&fH9C1kQksKEzM{|LQ=XJxk;BWV1x>c z24Uo=d#LaD{qE1_-}`*;bMAAl!x`6JS3R$KJma+x4YkJ1iEV4X55n~{NU2e$=YI4Z<9(MM)sHrQkp&s8j@S6P1@3e_ zyP_)ZKXdEo3P6HONaPmVewHEyH9H&iCoV+og*VUaFGXL!@RL-BB7z}+QT(}aZ?R{s zbqP!qESK7}C;58sR8MlP3Dw?PbasrXKNSYA_d&$ZJbHcx0|W_(|4tUbm*^MN&#r2; z6Ogz3m0kb@Gm8i+h_3weT(%IDkfipRzXhhJmj+-;@H-KQ5_Uq{{wmhel22|bPw1fa zEooP1^(`SS^jB(un#%(-|87xGkeirR1Ij>s&$_AC%j{20_!en3$T|P%LwGj<;Knya zl1M#)3gbUDK?%t@0XK-r{zJFMdW!y!cB_BNJ>~o6OHtDW|M-i)y64|+P4G(|4lpj+ zwM>N*R@vYAITT9~XNOImXD22pVKf5U!2l)zFGIq0Qn7LzsrXzJO>xuC#ZuPa;_`1d zHMJ$Sqd;(Ec$+&+)wZ(GyV8EPf1!B60eJtV@Ttepq_W@7r$79EJ${cIkRX=nzCeYQ zXolG+x%9ki;bXt#fB-3i1x{LdLN5NArf-6@Ezokix^2=609Wq{y_8H-c_pFXXv=)6 z4g5^n;U^84Nju1AMnL?OGM|mGx|M{yrNvATN;pl4eP(c-0nEUYSp;QPe3`8l!+kzQ zMFRLKB$y{Z4rxde8fiF`H{8a`xW(O)}){)mEoUU%^-Zf`GUoWPT|1L8U zc)N?p=U2Woe(nv-W$k!kEg%*oanu?CIFI{G#Y6M}(@c_I>q1Aq=8gY%UHMs2kXoA1 z>qXfxI94CfG@FhV&Qn{5>Fx(BT}m{Owo+?4aB%doIzi*=NYd;X3fBx}E!5H^sn7Zi z`Nt>ZKzm4MU)oFY_iUpR`b+>ds3Fvn^1d>~u z@N4%zSbHW6V-$VOvV$sH3!U2$D=C?=Li;T7WR}XfA3CJwi5A5?@Zd`G}1I{ zvREhjFVGp!N(b{|E~j<9!6aTAA;HWrx{~p;eyDpN#-E)cjD9e^-U5Bo(iDc(v63H1 zl>+T?cXhIJ*IXr*^!lM^LQJnR^GT+n24zH8TxH-`2t!U*H@^YZsLRaI%?M+hk*Nu} zb_Q~& zQ%(sWe5Isc9nbSp@37MX@Zr0J4|o!c*(F1jlFP44*7Ll!GWqEH6IBK>j&KX|>0I9` zQn6i(L-NHu+j3h}%&lHesIy%y+VNAFEHSoytxngip(dQ!9QBcRwg!giw0zo~oY884 zv7SIBgs9CPB2_*f_iQ}7`Q5Ol6tUUB4t8lxh9k7b3ajni!a;ePz{JMp@rwmfKg0?> zYsW+YJzH&yDgvF$V)l*$(0b~e&%^n%Aoi~eYIW0II}=U)d>$a_dH1!n*Vc}a20yRD zt@|A7x@}jz#v6@=*-Lzu6CA7cd6062j#ppD##O1(8P}Xd%Ya%ICMBaNYMnLHCVXo) z%_jzP&Z9ml}C>mI># zw^wj=C!G?{j-Ij{fj871s&#gS85+z~+gg>Yi=azT2y&Z!9WUfkJn1Yu*0iy)Ybn9W zB9~NSalHJ4?M}^0qq>sFsicRRcY_|(@PT8+{WkQqCd>tM7~w?v-p`)K+wMOO|IEzKSB;RZ?*v z#?;(`2!$S5Q46xR#i{f>v?qSW*2>%gwRUrl%?BGewrwzPI@}Ip-jH;aIypMHbl>`2 zB_TD$Ct#)eb}4El4#s6rx)PS}sFSgI)z8>CUHsA`1L3rrs;Tw+*WF|vRd&?29Um!8l&28u!G2JgV6VzrVANb}+uTW(R3y5vKlX#Yk1ZX9a8{JXoQRcdzYswhK6P_+xopKMvcBW6%U$eaJ zL-b`I4Vm`RY$z-&N_$v@UhdgUdX$!bnPL_41#MfJt(nV>Gy-K6!;R|_8XLV$%=$l; zx}|`cQS?vJsvub*vmufl7dR$mPEsOX4dj} zA(W;PR+mmfyCEJ&Z^J#%hOqPkiB;*5gx)?_kF1&9-JDo`@#Vh6V+(WB3Ljz@;=QjHt>Jqq1)WoVRv`;~7sx4~FSA}ZyYu#0vAaNQ-R z)H2{Xeh=rop3b?QqLVkP%hPV0fqGcBTEb`I$?vw5pk2pb9PKNB@E{AE-tJlLJ5mC( zK~&M)5URdBHAQDtP0#Ws9S2AX<3Qm~Up2hiLOw4^((yt7tcTL~g-myUpK6+TaO+ek zY>oBk2cv=U!H<|J=79{tw04_XF+KH zy|(?M-HND^#&mRsti(L&jKk_!A(h?!!7@NAJWELSc`$8_!gVK#z<54Ii_b<<${|)K8`w|X*p?4U3ss!$RnJ~fYa02?)IZmePWQon zZm>m)SjaaN2Cj@*3MTpPma1%jufRF0*Eui9Ea{m^Nlunxy?&hNt(|^yh6$U5p*6i39p5UMfwcq03Wo`&Ab|n>QMuXC>1} zJ^0?E^HRqm`DpR%a|K}jM1gkSJ1t)|Zsmizo5ojxM4p~82aOXN1CLh3v?jS0-As6V z*dRQW@0&xcZje6`Ug^(x?!XD}i7<3YCTF!^P6|JXk((UkD|`-bNdKCV69nQ9mTY6S zmz-0k*D&KS9-ylj9}x5`-UdHGCG1D1-{XwP$?(*cv#m+PiI~fWeKc0;ht@S~e0#Zz}NYykH}apdGY4MhK~X6;8zjJIG%Yb1p9TCuD;a3Bih# zC&o0~bW~xy9VQ`93L6$IXu_l1%sdPx{Rul$wazvlDJ1ifFTJ$(h55EZV@zvqml^q% zIV;l=)9V!G$+)*3j0bI%siEHCl}d9w^5eBVQunl&s}TRkwBruHwS)vc34=YubJc=v z$o|!btg~f5TBv`_xndYcjaOly)EboXFoFYeD!jrO+(`g%RqjQW7bf+j4v^vXm0;>` z`HAzt6v)*syn;#GO}3OE*h04C1u%=bdqPAk_&EZv4%Kvc^Hv1Zt<@Dl%ZchETI0b8 zt#j>IUhdyI%!&VdR)4W`#a57?7_UD8S$ay$`CVksl_kn?B*ofN#0>u1)&74q#7Sy9 zO+~pL_>6X(`CNC~gr9-2T^t!jQysh&M98|}8Y0ld=kO;Q*QqXu&+Uhd8RG&`>!f=- zqCOP(&Z&{}LgXEf2R74MKiB9TDg?$5f<^f55;Y~=s-?@I{;f%}{?SLzHAyeph}?en zi(yH(j2(r8a|q}1rneMkhhE|PmC{B4*okV9X73HQs?Qgvn*koB)V~ur9}fYs=(*xG z=<&J{a0z{h@pm5nuYxf2*W~=Iwws!2;h!WB=YFX$kjWXPC-FyvuSA{?I+XBlD*)90 z8C?sMXez%2s0?vTc$zT;`d#b)tMV@s$p27Tey${J!cB;_v(TS!ffjo}IIyE9$OUjB zp5pX_fWEJX08she^qwP!3nL*RU@VZ3yw_`|{XUZT1m%9lR5>f*D0&i!Up3`-i19mx zl9)DWl(oE!oap@Ph)&YoUL+~N^-x9Sgoatbz4P$6%8QmmcB>z z|7ljjqg+4uq54_XwBKsWHOLe4Dl>T#Pt=;P{*>}p&XTd0D{Yh7Xe*%DB zKl7~>0r7~~#T^s$$zw+2(o=97R1)iX)YJo02OC4^#@?8>4B}`0+ckXGigbw#MQF|k zHv8eEZ>w;9Q;dA6oF}96P`*k(e~`w{-H)wz&uEfvBSYz7ZE7)UJ|Oi2Fn=X2&P!Kk=8&Uz z`qdQJ*CQI*AQYi{@5)$J-$hYDJ@4pycS|qQT0Ec027P~hT_+HQIDHmB08jKpT%3h0a8iWEL`5%bOfF@63jyq|}pj0p_8Z z+E?}YR?Cb0m7LgYT$U>*+<3CWI;lf+)3!}t6-2YtpM7N6xYxqHS07FcD*_YO^Fy+k zR+$v=QY0}XWsQ=(=^)c%uMs|O!O^yEm&QrvQV{t5Deu6F!9kYcr-72%71legru!2W zmR zH8b0qAapIYOO$JtlcUS+j}x%dyKW^ADGOw71lwyFRfa(Wf#YsO;EG&B=_V*eXDC|8 zlv#!W0HGBT%Kx$FQK%27AxHe=4&n8F1rj5^4-$FA9+_?AL@y zGZJ$L&R%gpF$TB*Us?=u#stuR7LwQe#zp+*!cB<$0&P>dS&dT4PPgq$b!CBbFiiF* z7W>hsq$y@lYq;loqKg9hs%sP0u%{JJazSB8*72yd`y;ul!C1@3hZ!*y(>IZ}R*cJA zywV^Gw~{3J*@>+N`?HW!$(v71YGsQ$R6`yOBb~jAd38(Tve?-?=_p@Qi}w$HaqnW< z)?-0Uvs*QARx>E=tlMVZA;!Xew-4mzL5eh-N zTHlvFTUBp<5UHxLQj$Nfv&{@6T&33_)$xTso!-0WnZY4Adxh(9Wb}Z~`yQvPi26Rt z3sL~B{%D*66K-md@IW(1^1JQ6`evM96> z`i4)s=S9rH+@bOGy-5-Ngmt+P4t<*C%M2A3a-H#t)FU6&bD~Cp^@dz`2MCkZmp22_ zJQkImBN;g|u;ZM(N%DCf_FY;MY?i^@+fj+FKEx+e-J`9A}) zM_hHQ$$m24R*`feL@s@EG$LDd^{f>qfME);J&n{_%cG6ew&%f#BnwN3b?9*2ox3xZ zhrLfX`<8gZ*muuDx0@!Rd!Ty2rFov74gQ4u{k98na+Se7fl`?UX09mI7=i(RHsNn9 z4&t7RQPN+IT8$Vg>7*|TS=z;G(be6^x1XjY-{CuO{`T~U?OOcNg3EfXRk56uP6b3QOoB_}_FG9c+~`y@7xYZ zzfAJ7f#@heqPhVbc@lO;=tnv}1ijKhdcPtnXMJ3@(m%R3+b>ug!ebsrPvxvKZxE#? zlkipIbHlN4=JY`x4c8j&VBOtKxO4d+7th{kk&xF9_eLJJEyKtj^!L?xi#R3MsRNIh z?>zj1u050hHVi`gE#dTmlHUlxRR&J7N%e_5S7a`>wFuzA_)~~8kgkGQ;L5mbV;l^S ze<9h`Sd#sWF4tA<$&OW9NNdkVX_aVdB3;?(94kT2YGO4GT(zzQY4ZNI>ZathQuTGx zXkrhciP?DX0mtqbU?P5w=u$E&HB7VyoEo)WU98<~ zZv1Ips){~!f!nP7o4D-JA(|{RYuIY6YlO}Ed)=DRx?|&Z~2$ucmjvn?Q-{At> zsQA3qa~yUd@oK+}*+!G5^-}rvPk+zTtV+IONv&$1#mYKzO~KLj7I(U&b|t&8@cl1J%+t?jkU*`x$%8X=Z>+F(b?;+dAtFkl3(_$h1+D1M2 z(tQk<@$J(NYIhQL@0Jnb`QNIl0MSy)pOkTel){{x^eKP10B>I%Sr-xv&chba=ADU^ zs_S-=E%aOGP!QvG=d926UcWv=C0C>$h95!bOynZJ7tryad!&j|@nE_3v9dqp7aNYj ziXrv}RDhioo=wR)+K73`K4WjVKK8J?fk#~keTfpKVjAQ>frm+JaAiN^&^cR4R#Zhm z!0pg-p+MUpLfUZxfS@3RffiS~eu`F%t?5UF6HY@vR`)}P#E8VmS@GX0h25;nJe%Br z--My(#VqiCvZA6^0`l+F;5rl|sNQzhD$%>eO_>j&?~1>c4C^Ov!Oh zjwnhndTUimW5gvtNyIJ;=PqFXKDN}HV?D&R09*#dvX0=Qe?+zAy>lcM5g^2JV#UPt}UF(b*k_*x39 z)IKy_(EKe)ZsH;D8AoOx3@H|(!YR>wePr?tp^^JSg*0699mBHVq#EV#7GILvU_JLV zL5=*Weq>ZnWgXkzSjY!o*lhy}!&H9EZoIwo#P=Xo;=>+`MAPi$E^ywaP^^XH?nX^3 zk}10z>o#kp=6qyT;J~^W&~qC}U_~)td|lTyTrUF%cT{6eQ;E|1E-C4B-TDebrPKSV z{cG&8rZuYM&FR5#Yj>{tAmjo-3=Q>NpIL#tru!J!3~-?fEWBJjN#1Pb&Ot?`3i51R zS+n6iBG0Ft$R2>9zItwWvFh!n888okh$zTKI-ufSz{n!i*|+XYO+1~oMG!}rQG=rp z5U6N<1iQf(7nPedpo|ho_B$x?W8OKKSfwbVAifOzOgIN}p;$s~h@Kre>h?z59dJ`w z;+8x-w3wcN&tO!n%xD|XZQO11{4O6IMO$I&5PNW-d*tXl3g|29lP56I^zmw0!=gGu znUmBBz8|S08*M&~Bi)qN{h}dt@A5Fv|~d z+6~s$YKXbRQ}Ssl1yS|N94&S-<-zAn@m;$+4*m8pG{T|A$MU}IwH35XF^ld`RCK(a zvhyPb+$6%C+^T_mrz80qgJt7_#Y%1+>nqiryHYFsAu2)YlOdIib9=#CEY-`j(cq*> zD~rdzsnfa$j)BZ5;evkJiZx%8{f)Pc6rQ{jLJ7bO9$^KE0RzP7?S+1JihR)wzxtmc zhox}oG108Ihn37b`}@ZtQVt%2tQt25>(SrG40FDvj>=exR@l9WzaRgNG0;gx&|I=9 zgm?dFpjie-H)46uFPCj@dU{5!-o3m&Xfn8xDx4ju-(+t-5gtwJo^@vPg8!1LMEz7 zQaiS$5mjO!`Lm z0GL(`OvVpUcl~%-iCtZAn7>KiKms0n;99P1bsEHL=2G`+7K zmtAS|%biP@SS9z~c1U;#-ia#uUX%)#hmz|LxAjEJ$gSexIKp{ zyb5pV)4sYiJ*RzZv(C+NK#3GtqeL0*7<+X_?C|3pWjOsx=6#(->3YwAW2w=Ee(#P5YuyNa5FI zl|0(7f*~6c#>mgoC8H&t5!IbYy0(gvJ)9ldi&pOFIM5oj=<9O4*tx=viLD)P?-eYw z-nGbABGwnqn)Fc<&M4F)q%fJ}BF`0zo+dB>pc^Z8>zLUgoLKpdyr`O5Mb*d{+J@xCQ&V+v@7jW*~c-i}ly6)j65_UfHyG(Q`yr zDOt=OD49Hs@feCw$M;+)(?Nwrm}72z-#Jyc`%I`~NL4sfd29dEgjBVK`-U9Ecx=H` zHUyj;tT-zUmQN!+Sp`*Yx^BWH_w4rG-fi0E+U$H!fzua;#yh$q;*r}2M42UoyFuFh zua3gxWukF*fQs>ijLq2lhD=;&vE8Zn^IAs@q|Z1Q90e_mVqAX+ThI>PuhTktP-mor z9(fqWR+qk!6eDm&GEgtv1om_Ju~U$E=Iz`kD$a~6WkwyT2jTU;7oM8iVY-sl#NODbgW1FR>Z4Ri%8OKT;E#_+0AK05<)jZbhu+NKmXCz{ z@?=0(_j^ciJF$1tZqbFON&B%`mye|<<9)o3bR=0@(73hg_iv{7_$1iL`!yVmO`htP zl-LlQNX$H(u~KVZ@VoM9(-6IINN*`JMmz42r&Vy@g(MEcQwY=JtvlI>%C)AmpayZ zyR}1m*cMqHWlDH%C|^ai941Ce5!4oRT*7rL%oARpBr>n7)TxRixDbbJHAx znGE6PdS`c{o5b|iP2YY#A)J&ftYiV0nUVrZOoYx3#se{%>ZjvtX-Dy%yST){hQQ#E zN9brgwMFtCyl;IGokp74a7qCycqviZk?j*rdjydk(6_=`1O@Bm~* z^VDT+HvDX0a4O8>O^c);5tTSmUUgO4!6^7vejT1J9@I>LYvv(z)x5XpZ{FE?sD92D zv@!nbeapy=umPWLY{um#-_`rrdGu6yaL?^J#_A$=@}<}CgK;`!@+U*f6w!=oUMS<= zWq*O*BU+m_D8KW4#vX$6|9>a{U3_3%z0SZ%kvK28HSTvz+Ic|OrC2;_9xhtEDpnkXK^E*6Ht~8wTU$#?yO=I` zcAPbdsxQkuVOZ5m7R(&GXWZli;=Vo` zT`r+NgVCqIQ7p%&G_9FuEPkb_Cv|>L`LQmgY3q+0C4+V+mChDEiqc6lj*hVDt`1l8 zVIoG}*`VDKw|Z;v%uMKE+^Xl_2v1q|k`VDo&KPOtS8>kL#WU3=q=v1A2!hSte6c`T z!iOVz%x4R#bG>;_g*LEdHT8rK!;2)vjAN6&O?r}dlqA>$uu}+rZai#MZTNZJ_x@0aOP*{sOs{|26{Rj zt}42ST@-n4D%eEIdIUwKs|P1n9GM=g$<^YZgV>z7+lCd+#>)V!`}pgqYZeV(bBrCU z$%z|>IqWBjTfnw$t42MQmE#nF9o@4`VhVkO))^9xs=dT2d~LhCbZOJYOHreDqPxNMBUnJTy(T8I7h+Puv<#W@E= zZRLv(k4N6B_xhbIl$DV3_2%tn!r_Z~6HXVIkDHiS0$v~X=3T~&8Sc`4jconuFOekD z6#9vLJ;#*XBH&Vx^M{y7b%O5_npXD&LY%a%jvi8mXM$^3>QZyntG z1Le+J`aR&yb(&(1%^u$$E>bFAdowoR8>iRgemvw0ml7@Er=FXNLT^}VAdHM3t9x_* zh>Cn%-_$5a@H$DocQI(89#4l!>R{fp;}nKE(zJH(N$uh9fwP|hTI9NPja6B$$d36R zfiO7%wreSZte)fGgbNj`%}a5SSg!-{J?FKgqYrVij(c^INqb2Ux_d)$t1gwkSxmUg z`sZ|Y+ik0!Xv_`1<$LhrIZ8y_Z`Y!`dVX;Un9%B z*Q?BMxx3u6)Wn2|>0^}@yN3R5n&#wip?*e;i$~dfi!r0t+kj`WkaekN3({2z>6b5A zi4)rjbzkk5pRHZ4G(dblo>WZzIO&wL@$`+;`|jO2HD*|sNz;%Daz?~=S8H}dce9F6 zdnnW@BTcFsp1H=J~|EXEin6zdZ}M3-N<(J zYj0UsHl9qGpfBu!?+J#V20q=BDZB)%{USbRbgjni&4nNRtL#(;kA&&d851n}PTnd4u&Z#99cV{m`vOm>GWK|bGPN^7> zvvGId5J&nM7de>AhJ?xVRtiNXi)4yY6-+ORVC#~p2am^MU7N48NP6^~+M{o1F2nmu zD6Bd^#*;pTnS2_kLb6kUmO?FW^vgNpidV#!^3qc%h#>5mLNQ_o@v+0-R`9spen_i9 z1VOxbTSzgR(HYubHHJ)OsF)zL*$L4Hp5nSU4n-SMOU@MtL1xS;H$&6P9qGPqzb5)hR4&rkTB7?x?AR+!b)q<%t!@z zZaDq*Y%$03g3$-g{w?-xewS^&GnXfx1|xOV>s1fu1~waXruWL5n%7jM@(P>T{d{NG z0~CK;;uBL1bC&Kz*8vrmwphYJVov091>0C_{X#XFZqs_76_;bt2-?zZq>G)?eN^Ys z#N19ARvng>#)SQ%tlK+LGms2Qmo{tmVZl(jTUU$(DE2XE0JE^k1M&2O#Qj7!2|G!4 zc}aKtX;o$nk$ds2dyjBI7T-zmtoQ`b4|koyZ-EC3j{E*|n62%eCwJ%5ZDgXTy8pPyYc>HPh z+3#x4>-;Y^+y6VO^#5%||9+L?Z@J+=V64txw>bzV;)MQ_E^|T&jpTW+JR-ay|5tip zOh~@8P5+#TER*1`)IvmNDo=AQOyL;;iN;?kK^L#>GzP`}t3J-pP$Jqvl8zr5|I{b% zzd+pr_D_8TWW1U9h&95pdpG{6Ps7xcL@=`l{Dk!Xo-V$uCG&RcGpuv%G(1{) znR0)fQ?Qf>uYDTSIKX&xOOo)fLGkZ`dOlliUHvmqF-8AUUt8bfX{L&b3JdqI8vSw|GEvGluGl0)9f@7o2UsfvK=>ZFUHRwaNLl zP8ILX`e0sl5+?SMR{04@nDNs*UW$TCxx6ak8GT*j*^0dF!?}r{Z+e}Uh+Vi(Yut5< zQ|0Zl9I9B1KnotJNx0_*C&;vV^HO(4(dsEjYfqP4rnab5J&fe=)2OhkB$x?iV8&i! zGMKjq0#Com9V!mn_&j)YnFqg6U*KOOOJvDNQZmEsRO=IURZGmhDDlvV2@=kj93Pv+ zN@3e(-O-v-F%Z1mR!ik?@?7~k^^{ldbLs;6v*}w#V_j1`F&_++!sJlOraWXtZ}E%u zetj!Lb8S7BmA>XLK#reAz`BNA5lx6U+C#MkcHS}zAityOQEHV!RnXpqTGT)SaLPCluJkuy08-KdbPJ@i!MNuywl;1ZJ z-pDbPMUJLS>J;Gj%Qa7hmjW4w;f2XE0Jkx_TqzS0H!B;qtd0xU$#c{sWeMNpA>zJ^ zw0>EkHpqAqaGiG*k8oUB;-Ll8-@P-S0a@CGoqE%^pRFvKocN~n&Xe9VQN$TmLkBoN z>yhCBxVdfPRIn?Is!DKU=dMiWz5TgrDY`hP1l09z-MR4_3x`gHKcEhRn>4xv*t%Nl z)Uq-Bv2bxYotk#Acjr)X63XAzLb9J^IP43VvTu7ow|+gwRBaqd;(Et9@1o}Z0qOoD zu5(+n&&erV#@gVDlhRWTi3WvLn#t>Zxj@;n<=^BRO2qI%-r*F}kblf5gBU*KkGw~< zjMM_fKaW5aNz-KPB<)z`Znqi5-BHX*_N;kNguhoSyoudz8%Z2!;Ln;Xcl=~(Ip)4v zBpF^SExL9A`$?D6YV>*OorJD7wo6ra%;XjohilqrZVC*z3DsH(heT1#5BJ^^XP(M_ zG@ij$<`Jo`(_9>n)u!CZ+gbY5J;z68A4B9bm+KNC6tu9FU_-E8yMjp8Dq(rG{YB+; zCBj-#hy5+i9qz?q{$4CQ&Ar^rm&Boo#@F(c5NmnxDaWH6*ETelG{AY;aZ-bTSn`S7 zF72HZ@&`oufVoNT2czDTij$HLh?apR)sdF>d{UgVVKFL_{G{I~!<7xE<04PBO)#c`+jJ*DegN!MO&d)$6}i1H|^ zPZ3(L5i0-+2slg0Dyg7i1177PMN3Y1ZdX)bp@*%j=f>iS$3U@~5BTMknO;C*(4>0N z@l$yP--Vp|){&rc^Wp9bA9_o_E>2oFHc2VyvSz>7U-(6;Jk+(i5cUeDQw5~TWFO9J z=jf&HVk(`RR)G~2S!IzthmIwUCd*ZaUTdDrX(^U-RMm}=X)Z6KsG^Oj;+hK3ePtOs z?VYOZKhVc>_6oTs{gVnCEz6AkH2Tda8^5rES89_9dvzE?2~3z-RK|-g+&ThxvH22y zN?;6y93IO0cN&~pp>xyG)PrmVYjm2wVH7M^0eTM1GHIkFtzRQ7Nw z8U8S*btFUw)pFdWF^O>bwk z&uvi&kG>oDWi@e|KQj-MC0SDuP!_vb*r77Xe()`$Ki;C;U}nR+PKKFMzR`-G6p?Tn z`6lf0?h1}o@5S9D26;a<6~)b?422ab6tj`C$inXY;%ar`dM7lP7))UO%Tt|#hMH;DJdz;R)ucg zs7Z$S8E$6|<9{TWQ4v4)?k2vU)q~0f5&ZceCHeKwc&{ z6}R8M&;D!7G9E6$-V`DfBshG*J+2A_`XtnuIwrnGRtV*=bta~NB%AH;faV9E-t4b!MzG+AQ&9k*Q1O2u5Uw3_LUlMQ#G!0?)2X3q}kXiFK z5-&_hWW^WNXJ|5x;uhs-yiUdxFdT49!U=8<$cGO(ovC;0p+n;D+hnWLqQlC6F5Gio z`S$Y$zHa7p>967czz?TjCI0-+qRx^2H>`ZZcQ&uPUG3j_S?7mJG3P?o-#|n_rlO7C z$lA>AAO4j`UX0&GHjlXey6exIHZ8z6@@e}s)xSZ<=KO}_dFuIZ(9y&xy@zckBRqEet7o>xy#d;1Pc({J-7zf;GRH&y9Rgn-~k%9V8KJM;O_43(zrG5(8x5u_r7;# z)?I7Xoo{_>&VPMQbywA{s=c56?B{gECuJG*H$-m$006q2tfVRc00#pAz$_snLhopL zN=*O&6aYC%@h=`3N2}htiDu6)=PB!kx!+M3!*?qu-*E~3Vopm18Cq9Xxy)kxYzjS* z2N`JUsMDES(YxY2yeX{HE=)G3Tg)xitbAWA`TgLFRhAe6W>7Bs_jb|HYBVIN&wEck zLOug4Vc>_ATi?-p+DgiRFX%I+B~?e#h2oXMZfnzKw>SJ3f}6N)J^tv{vnLSs(aZaP}>;2 zfw`ufj;B1|!umS6rCgeI^H_^rS!3pGpENPmp9SBlI;&YaW1f=?LCM?=QaIl;y{jTL z-EyVNADSEI%*0NmUBH^;D)mi#0&|pXr`VQh_;PFBS<%Z?147PNNJz+}YhPMnUq>{o zitUg$EWeXbpBFB?BR(-fd3AgKivwT|r19;iXTo0aZzgAJzk=5znsqAua0DG`p>)x# zEbK&ug@iOBZ|7tsaMbMLg0+%wmqw>9ySEdjFxcadjF_aPq@A3btBJ$m<|lnntbS_} zr?IDR53K6&v#+ozef2k|ccL|3-?f|#c8VcU9v_@?#HsO#xUC}2+vq(^?D%rZ{EU~j z5%SFqDmuzz%s?4`D&Y;UsO@m+001D+mEb9y_{ochMiivo4qc{!vfmH9h(*kX3xb0? zV{aZU;PYB5c=G@7w;2hHlcJm757^BYeFWT6tZ6zu{o(ShO&mSx9C=75A0^qi8qdZ) z-*8i{&Pj7O==z8_*|&%eY7gP?+I697 z1ScA>{oU(8fIgc0O95S|#+@(i^c@g;W&`Zb3tFrZ-z7N2ckONZC;@r{55QS)hUZZl z#ZP`%%aXL4H5G0@s;H2cHzse>Qy(|nGvFMqD9+L#mVa48?@G0S&7qH}s$8o?a9+HB zYLI?0%k&GN?XUip-g@1yjd3%N)E{GfWqOX-B2=m;)=wa5kGJAAj)+xiS{&Y#9CWDM z7Zi8?-2_&p_N2y_ZURpbi;KZOPuopIgtwa%njd0*OL8f@BXM2cf5+Ki6tH;6s0Y1Z zsNr$5uk|Y!NwDMTj*LE;;8b^G%Z{*5?8b2@`n8~<11_c_vdP82d?zHTHQPtsLb20n zcj+XoY$0*1zAZWaCEI zO^oUB#h3afUDj$Xx=0(Z#qE~bRF$pe#7YkCmln=n5&@M-+b)*VjkVo$*+tA}ca7Gc z#IqGh&Q|#ucBEj9w%F)v5gq?OeVhMbj3Z$fd{r%OF2o|E4Q7K23>+f*vpHS}47+b_ zYESdWdk;DI&y=HDFEm^lXVY0D)awUrz%Fvj)pCwqq z5woX7Q#&Dxa|A+urb~Y|KR?P$=@C5G!4Kz28$XKtkP>7L2$@l_s2*Ss{~bdW{>IVh z8p1*HyY$bos8fqSA30mvC^yI@_t}J$&XtDe$M0#smg~>(Q7B}l4%yKbF%zdb{ffhc;y;u zj;~H>Mb?T3WuZND6jAiMuWrD@-4%zEz-*6W2j3B?x$qiZ0HXVC7sx48m0e`%ygCqdm5wCqYWYQ_V++3FEX2eMQ?#*I00%B`jiwqhdbfGuYd)(&hh}tc=%S>qzb%Bs!%Vk1`~~<97{eBxQ)38rWMC3~ zWVDu*URm$y>s`%AWQ~n^L+-G>i?B)=utU{!o&M#d3L?y42@$?AZO)n!t9ime%8<4m1@s6!B#_7>yecv&246rva( z2{gFD2f665kP$S`(2KPFDlI*>QE2*ZOw#=JPE!F}R8RZx{-wz8Ilf+0rEmu4s5(LG z+>+6jyt(~LMdxT~iu&V6zOXDo_D>p;3F4y{H7dRvgL8htss{avx7~ zd!p;V&hJjLhmLuMIp*O76_GzJ>o4wpKqVM#NL*~WUV`mw<;3=PhZ}|M(%d(wZ=k23 z4%{cqj-PQuwufIKk_ZGexjPlAr7xF1)WbrH#dY}zIi+?_R@Y=$+V9yC@^L?pJgd-+`ZQt96rn2F!?BviH(d)F zCyZ@f@6mgBep4X=QMCd>Fi1wttX#CCpRCRX*6DDc&BgTBXrcX>4&PMs@pUA=RavTW zXnGiudff)>^04I*Eom&U$_|Qmns&~eFU9A2&bTtiO?%$060aeP$ z0Ixx*fB_{oy>Co-@i8@i6zZ*^P@_`X3yFlJR2bfU?Q@L&+E~*A+p8lG|2L8~HTFVHK|ICg?Y9*OIqh9+^s^0v&f96#)FRuWyO2B)e&9pVhM1GY(D1MZEObw~ z^)EKId3EkIefqyZpF=b-VFP+iA(6E>k+NlSKX=9#X4Agv@Uh`LtDn@O%R)ELHW6b2 zylMPPUc?ZYqN`80OX!mcOfkN^;z;<;H5RID*#WW?qq765vwFT^+b(XwS#rH-pWV~J z{rGjpdwwRigtE_U>!;wDN_%8AlbB@fOUX^+K8*dC$!5`KE}U|SQs}Ix>i7c)hwHK% zQq{cr3Nl=r>W^v1=YDDpO8Qtf&7DVlVKu?O+{B1gyN}lxOixkU@wDGcpwhbJ$1{;D{>i>Kajc9`uYeX`YZ*M=bicnb+&v`3X{U z05uUo<2@k3c$$EgywYsEk1D6b6+`5}%+ZOqXM)xs0s&jpu;D=`XE$jl|4%eJ)oj@s zZPhl>%UuaUu$G6&+=t>5VmOY1oPoM^ezmDLTiH3|$8!aGMEtn(Nlm2Mw|fl-JXX%$#Uchk3}k zA9XdK{LqlfJC$S8HE&Vom@a+kv9dW~$U7d7!xXm(d1(7hv_)dOaV)BN;0BhuG9vh+ zwD4mXJaPT$Gm61OD@d{K0ZG;Blc%*72V1(c6i$2z2s12MQ&S~KIyVI{S|p-x9;=4D zs!k#7CZe3USK6tX@9@G5W}r)3Inrc;D}i62drStMIacwPNnx0jDITYDK@AcEVpNcC zIBnTQ67ikL&@{fW=DG7C)0K4A14TxTExMSm&jUf_k?6>I*J9bSlLd9hB%FaLCbLNE znI$|Q(fN%?jHSPTp|MUz;JWB-ejqh>Lo@y}y;3Ttm1*-1-YzzGh+e{zgCl&)Bl1zY zm{sRQ#&QN1zt0&nA5T7Ld(?+-qHLy#eol(BpDAMxBA};#_I1GWu)p*=ox?WcBZTdv z2NNYeJEsWG&q;S!zR5j**a11=PY?QVuWM-C;dMM!m9pXq>Y7O7Ya|{Zf573tOg2t? zq(PLBD)D}M7e$T((uuZI9ZV>`Bw6^I8qTK1AODrTSfb=m%59fGhS_|s9ed~p_H}f* zK7(-#ryINdk`11n&?gtTll@AuJinGU?lAs1%BvL=-z+qfp#`!zN}-PALCmcf;;odM zYTzOpi52`+`VB5Qh7(g!)gQ}K+ZsCBXs@%GfpVdS?w)s|)a@0(?WBQP8d4^YrB5I^ z4c#!P50m|>l>m^CK}#t7T-0_Q1T}l*nx}Agxj*9?n>Ip-MNUI;R$6|7e~=0eKe%U^>}4loi#DTQlW^VpJa#`~ z#HJ_!WdLmLscbRWl-;gF9Vk)!-tMUf15XxiRKW;#oHpzP$JIgnMN?^5%|sxSJD*}@&~ zJ>Q`2<2FUy$FpKQB?fD9tq~ryh0Dj zp0W4P$QhrP2e-JmXDx{KO3g3|LWF=o0pSGe3OidQ3e~J;0art46df2It~mY#{Zcux zj11Dr;Y?0l_gleVpnvyqXpjBc>;Jj4|HgR!UpH~J76il%nPxQ(^x10BoC>RG?@B=# zmW_Q)Bq*Z;C3do5T8TGy{}iId`aZ)#Zpdk#7Es2rgptI$L($K^ttGnIo^~`o+t!l2 zLX#EbL)|8q165ljO5@?=V+$=3=Xi>C>8q2afaFp18Q0S8-WCNC5p$#Ea_1g8Qx47s z6ezJ$dhB358z_o#IX&KdFj4+NUq7p%N%OX!@KOs%BDYoV>+dkWnrS)RWX2-%%52pt z$px(Zt+NqT7`e5$>z^pBvD`3qZiJI@zBlYe`3`AXvZZoZkLf;oausb8Zg%uGNpVWy zcU@B4yfGD6TdbSI4$Rkjel%#VxO0frDVupc*6KtOl)U*PT==g;p3lC1FAqpA6_$Hc z*=vYRAGcM;xC2cGV8NwN6o%lu5G*C~?iOwN?%94NAW9tE+alkcJsnP)b;mxx{BEhe zrQ>pJn%&-^B6?bA021NB0R$`)H?J>O6YI2r*`3i&U+4!Zx3FK4Q$$#U3ACtq@;@4%037K~8K4HncTC_oTGL zUz&XhVkLnc2<6(4FRZ^X!Yrhwe&v4oXy-Qx?vkV!sK6qwH#?v2SY~2aji$WZB!m@v zWAFKrelvd|3X2!R9nst^WMeV1Ye-WdbI5c>HNf|)aAY~Xsw%j;NkYV?MHUa zH@N07V7~KYqHg@Aap!&07pTRY6iS|_16yADpybml2encpY$$eaO$$&}gC>_n?(>LJ z6E#a z$65rv8LlO%e7{7cL7!WQyc4@Cxjbj)v&-QMt8Qe`gUuu!DsiBxpt`3g8=J~hOy=*n zreP3yM;wcyFDG(dy{K?@#^{4nov4Kc%`K8t)raPmcnH>hYcfkQEu9(dndgdF78)Z5 z%W_XH3g-0eGW3V_4L@6zyFOGB^kL}rCD*E*-?DI3r5+bM?GywW*=BcYu{6M2aW=Sc%O<1_6erbP@1SC32jW?TR(hBN(epak5QSHRkMJ zhpaq0r7-cgnmNNkIn|1$6C7&;qi?z7OxRFZOFnRlvsW#G`U;i?&u(4MQ{zRA&n7ie zo66b6*(*QfAyxC(nQNG0V$+EZ&P-F^b8=+eX=J3{J>!jNkZRpS&#;CIj>Ff6CT?ag zr-`qXjiO^!&rccyq^)n%R{6bgkU~YwmGD?q~O`!6)jW-g(9NuZZ z(rDu=8aysIYUq7o-bRwy`;Zh8vJ=mJDz*0KmN`9sYf)r6I>e~tuGR1)8OJuY%8#r1 z`_X0AJE%LqpZSJ8Q6R(XYaRK zj5P0C=Vs2>QCKtjxCOuVVD7IJN~?;VW)rPsc!mRKd4TM3kF`J@n%NE;{lLM*5#<*R z=*NSHf2izDRIe$FaVG{G-Xp$ET@C60fIZysPs^`#>Pz;6&yzn-KL6h9Ij5H4*b7t; zBWR|AQ&_cdG}fbWZt3!Dt2icZ(`&jlSx*5k{)CH_J6clg6Pn4J;mqtWH|-6NhphJ4 zOW_kdB45;IB{r{^<&WPY@#z|^&|#~)n_c}SYv6-*zr)h}k{QuJs~kmWoT)L+8_@A2 zxS;h(gWNvPl@uDH(j`ZqH5c?+_x+uCiwQf@R&5}8sIATK^ha$YVk$)pJHUZ3E8Yu| zhjNB5Zal<~$TyBKhg;MA0|#Gl`mMRy8y7vAW_Qjw6ehvT8AJAeXp6{2%sZ!#txjE4 zwj4N_NkQb$LnM8Q&HHxGtm?xdF4++{W^x(Shh9Bz0CoC&Cgl(8iCnwY+%^lpE)bYZ zoc+0J9HwN1d^SO7=*XL$FM1<$*uwrgbfQr&G&m|Ei!pCbjepMvQK~$rEcEE|5%kL= zVtdFONICgM8yzBp9VqhLkt53)?FI&UriSt|TseUk-gObha8Ia9Qn+AfTx}GeJ(>2S&YQNVw!05Z4m5$b7MtNpy_bSDeW(upC>;Vc1Bt0(kK>T@ z(@7i&Pt+aK-2riMcJ%7%I*?))Z$-DoM-G1205exh35kw32^zDp(qnrA^XAgoWYBE7 z03c~8jc0pjC&K`HYH2hZSWSN>3LeTxXgfW#QM=cT?;@u-IhsrF_LC20d)r(~1rZFR z#Zap2X>tW$W9^O9ehdjxD`IK~ug_R=y?%hk{ogK&Lknx0M$zam6YtgA<>aa4YA4R5H>p`nW8$r1A^0cs%2|AS? z+Sa-B^^8dXpTC>%)IsTAHh5tf=dU*8eD+EI0itb~TpM7xROep@0T!P8!WrdjPo6Gw zFak*T;`+zghywHN(cUK~P&~2y1}Cr!T4mci8T-gfC=y)qBqVyimMM`DoCk^;$A?nr`}I zfF|-kD^LZ6MXVTR+BY~pLTlqoII&p6U?aMvNB+#w7Y>QMIy7G+p-$T}3Eo#q{PP#) zb4xt5au$>~gyU*!Ux}9}>g!mR$E%=aHgUh`{nc6+mUQCZV9l^d&lEnNk*+$uy>O)g zV^as5=BL{Ak4VpaBd=ITqInJ)m+^!-_M`5ZWq|qT%Wa|!P|MWbSaWcFW+(Qqcrs~1$!oF{qq!*H0bJEpjd3Nfh7`(8{~#t zWTBiT2`PB}Ks^OYEOPyF&m%g3X~2C&Sy0F#C+e2F&+!YLn5QGpQ*EiXHBZ_LU}vc% zbN^(Fx&*=?(d}L6qzkplkX)Yya^gT8O6SaD+Vr4ODbTvNwc3T&z$z5FNuG&3ym9?d zG&8`UP^5U|K3RA4&kvjL z;?hS+10~DROOM~bW(kefoFuQL7?e}h{qW-QTCsz6o5S$$eBu9;ll;HV0scJ}#nmMY zv3w!0AA7aew51aZG|74w$FytuyZaY}_;R9^wU!jhpUN`G!)}MxKwFrL%){RHS>-xi z4)OWM$vMX`LT4983=D-EL}!?Hl0yktxmqEf->C?!pZ_OCOL?VRozR6Riay(hL77OW zf6Cpwk&n6;)niT!?R8LEBQ8C&PY0Sw3|x%Ir#uqgGh~aq*`2%Wt<9X1g-O)U@$EYS z!^6KMv`(?4Y@$SIz0!Ne~696!^-%{Ote<2c0#a0pu7RH0Crm}9BD5;qu=P{de zd-tJ89>N*kz4UJ_03VLXRakb3)=eS4q1jN9G#!=ON4bE3+T@hT`!9Ao!x4#znLWL@ zI1Mi4lFN?1;R0pk1wiZWP0KwIYX2SGiz?g1YHwUggidCE4^enOJ};$i#Ptch1qTHB z9EK7@uxoflK2!8p$!$7!y@Y_7-kg>;$|o;%2Dh#jXIy6<{k?>+kBdlmKHA8_txpF6 z>j-!Scv&@O7xy0iQL7Kr)u%r3R`ddmayz{$d#Z~om9q&mPnXVoHBEU}S1W9TF`!Yi z_CJnJ!aZ~y2F6ClLQy}F+l5dB%N*ANb6f?lwQeJ*^JI$q{D|{H(%^7&pvi2<%u^BO4IPQJ72E(vOdBlSN-@cwL^&

tlJGiAgt?IbXN#Q9 ztutAIjNdfs0o%QAZ|%W#-b%s!aGe3J9;(&1hVFbBc`kZiS1V{2e-N*Ca4ry_5OlgX zx%z*0N_Xn?g+=otAxdQ7!*BCLP(%sfT@i%yYU~`3`2fPKT772w4+d3sYnLsD?&A3 zg87R?uihT;vJHRt9)9bNKp7A_d=jJI0h)&fh_`DGZ63_$8w-nQ$52=%7@jj-=fHox5ogxvA2qmmx$|??cjj;=2q22t!TV&w@*_;QkcWw|1nAx!N;!DoWyOR6 zc<;<-cY4WmFhNHF{XE^5;2WS)o^Xv1{1@73AIlHW_8w2P|i9W9MZl#%ei)i z#lA$lI4Uc;dD9ms>=hL1+MmTWKM zohB|!t43#!=I?g2&7sDHvsEUC8BDr&N2lbwL5+rM2Z{t&N!S1^L02hZ7?~ULzLGuJ z7Lu|G;u>`0{jC!IOqd#Y;C~f5u&T-G3!Gn4hs(47XXpgOIS4qyOykiz(+mfsfbv?d zO5?H%RV*g$I|Oq`JRZKTd359roo8|RjxC@1hcOHXXmq@#i@D0HG#NaY= z-DYJZsz;VA zb`R2}CTF?NSg>zra@U`DS?T|cC1KSvrchK~y0<@7%-vD#@ln{G01F2Eg8q^MxJK#< zu-F(24M4dO>@*QBO$lm3=%{}QHR)EsZ*mvr_HX&cn}bku3KNFWVT&?hNKwP4S#D2o zRMy2#{lc(ra8_SL-Y^oJCziyv@a3)hrQey)28u1}w^OA4=>VSfefTt)PUGM;qlxA> zUkm!gODx{hwNl<;hV8r{lxB0)Tvy+`nLu1|rliJ^Euj!0zHV6f(Y&rnggSJAhFFpI zR}`n}85N_uvuOP}RvG3mOD~Z^oZ{UFn3FW4=;>)pqOoAOgwkqaB5_f37UTNz_bBWS zH``Yky`|syXK1+|6_ylUtCK{*Qx(k| zrRp|_c=%19&$`}tYjXBH|8^LH(}c%GbD0Wv%eQP$U^gFt(~k517=ue-SORp z(wq^C)FZ3G%Qma5LVy-+VXYq`C#G0Jz^3K3cE{@1aC#G@(oA&v3WxBO!8e;@ zJkbG9%ZoN=TP)<@9fWKVX=Q(-O^;%lj-EO`xK1_isSd1viZ~|hW&zGloHCE!&+XyY zzmMl_q40lh?;2nEk)iyIqBagmJF)}_on2q9<4pUVf%~8W7koP`5#K1yHog(ZCe3zy z?fRWD>30>*HTWXYfG8+YZqfVfvj6FN4Y@IEiZ26AVe6r~re|i-B(O!X&W|s$SfTz5 zQv97T&&FplnJJ{cMM4vP;+0ce`cn6O1-387_s9HNYVTme+XX*fi*CQGW1TX=7>YMf{ z+enlc!v#gK zm9GNY#{;}HPr2T=)KZO;lBFB9=C)B|C$oj_Nvo@-kY^~uz45x{oa)Fi>YROH3Lj4n zezVC6zQXvXim9tDmM3>+BPCHJltZ~rjJY~A^5?`vMru3RxmN|rYU0N4BVB8E|7ADI zaaDFb9UI*>Y{ozO9XdK7ve`oTCn)LZb3bGObnG!U?`1+5j?(OS_^w3{G&8o*+Z;2P zue)~JE+;) z+;H9Q$PW{5bN0>|fhs;B5}xzqXfdiqHmCX1T6cCL`M7A)gZu!9dw?}g2Xa}=$43nC z?A~Mb-{y>xOFh{2fvQaiIMWLEgIECt$Uhi zFI&#?$J%A^r>5jY7Qv=#e9juw}hzOtaWeT42V@5q6f7d(-B%+cG5 z&rm48iI3&&h#(AOt~zf^Gw=Uaq#UZ=vOV!m?B6V`KCX|8a&r$7Y+}D1T_5(+R(bcl z@t+!`G>-WG^k$=nE%r}mEGh0;&dE*O4P_5;u-tZsh_?upkM_a}<9Eku|9cg0bLcRZ zSWIu-1&?1$cm)~xYAc$SLui&dH5uq&T!hj@Ov#J7qBBuN7Ms*Hl!P+TDj*DFMtH!x z|6XnWr-w%d6mo9eF`hg2jr{RZYHUm-d7iFS=(pBMldvEWaHnHyY#|O9-h$60@2osB zBi)kaYe$Ud%f=c*NtvOjvTOn-;I{vdx+xDDZY2ySwVlK@8EsfXqVo0jN_S|EN??UZvb`E3Vj&k11fA@pue2qS&$dk#5#s0Oz6qAi<`z$de< z&(I@+m+lC67Z`@J=0Dc-(z1uQ61IWvPl!QG7G{rH$FPoT{X~nP2z#j*o?)p6a)w>8 zEHx#T>><+?$6o*5nIYEWgtOPX&+PWE_->iN^U?! zW`76Sy_hG~b>b&>h&)68l^uzned+z0w zWsjf^v+kJ*{CxZ4#myl|2Gs16)=dB9m<$7}(QLNSMHYt3t4vsT$e3<)5qKG^;Rzn!iRowiDib&@>OXX}4-Zxg&~`K02lc<* zPm_CWLC(b6^!d4V8hXGmDYcCq-NR!bTR)n}PqxllAxqBpg3BcdH-2>CBmSAvZVh#3H|>^d6sO!lqN52@Nb#`n%s|F zS}FR@1@xnsl#k$=pG;@c#e%-18 zJQU%}+(%l4IV)FP{zDfNh8$DAE8yQ#$v&7#8RMmvK05z+NhBgVo5}H{J?pc9YqvFe z{#>ScF#bmvRBmL@?m=}{>Mc9!s_gNlCJtmJ)ROj(Zjvlobgp){*1tiqv}Q=msjcR-bz=8fAs#{)dwlZzr>KO+h` z$&k_wVDW~~_Gc$w_8MfvmB{XMH)X0nhIdj^iElB0k>(jH2(Gxeq*4Re7FpR`-<1In z^^47BOSF59_ifkm%qf{&RqlD=f>wcES@1#-l~8X%xW#;*2aM04BBRa#3y|aEC5nAU zV9hAjh>?_Ey)Xkg*-3nZTGTZ^f%Sfy}WegTYg&Dxnj+-)HkC#!}T_oJ<2 z5RGoYG_U6DjVjO=B=cZR>?CDmY`)?cRo3bg{Op*&(>O8ME}T3($K2Zf(JPAjK}){2 zOx5O?Dcv?F7`k98`~%dJTW4}VsJ6pyq8zlwt6qL?xY;9$;>(LgavrlYqR3JSZBuOJ z_T%buh)4_-Z771`)y-S6kH;nvvg!4WSF%}EY^Hd|jczQw*#*6@Cx&Z1qz8R%9XcWB z>>nZC@i(q~8wLe1pMw=WdzPp*J8zb(wsy#dOKExPuu4zQhe7RxqO6n`TOQ!b8^7Sa zK4GyEr?KC*eMVr9hgqwevUc18!fgCxJZ7~yVon5yVp9@DwBM6erw+GaZ?71#w*%L z*@c^n%$+uGB{I%ivynXnI4*@vDjbG2iW8w3_q~?bk3_`dVqoSwrbxPL(nJ}r2)yB#5Z zw2lRRCEgbkGY(|ufNaOfIUH2LP5hh#+UZ#2?kc5Sk1v@8DNtVtJ=O<(dN3CXFhx_-wg1o^=@)z)Xgbas!RF*rQx*PMjcf0mF1H7;ct%*O!gh-T83i!B z-v8HAk8CbA{lrrWZRuk{LsJm{Txu@;?cRz-T&a*`LAFKMh&%hU!DNBTiZB_K&mpx3 zG@$PPsjYSzX6T-S99V7E)6#MTJx^0b0MYWimpJEWBz8cwA&k#9b=Uk~3wRj57Sf2> zQI+k_bOQLcmYb=$!9aT^IKGC7$~=0A-&3o{An=$SFxKid%>ha{@zpj0=$;&g@7a6b zkq*c%4d#qto_&J0_t*r32eVq+;k*}T?#tAq}@K zPP*yQ^4(1=_bU#uK_PwOHg}8no$j13=%AQikzCygg5k(c%1%H*A9RI-$6jM{7xk@g z#a6F>a0AU!W<--j_Y5<7w-#wwznEpk+(4IP2($}qExmNGl!2IAl9W!hYLXtT4nKtK zI9YfLO=HGm{BaOayh#`-AZf%xz#9YTBx9AXK0-b2Mt0*8qh6Z6+%?lIa;6gFF z(?smi){>Gm?%F!Vnt^5Ovf8sZ-i=_h*qW`kYvPz%w=zOAU_C7L-#@dytBsHd!_d?JMV-O#_O2R1_aPMfX-VQ;8?~^Yw~jZmo=?dP z-E@5X!2VqGFJ*k6uwc%WB0v6CmAVB+!63bM({-FL4z#PiDQl!mENF~YI7vt3<8H_D zH9?!>Boh6~HREgjZB8*dU)%5*EsI8%rpqN2U5=tFP!0Dbi`*8EmNgzR7#>J$&ff;E zaxNsqe?HjLnsob`Tji#n}u;EwkjBfQsD>v4Pb=Obh|#+dXbVaWzZZnkE%tG{!v=FlO{&+kA?=JNUKly z109010i58=*{*)}a6hr>OC2yIaxN}j>+ZwpASr+}^zsmASd-vhh-tiOieukWT4VtQ zKLzlSPFwS0@mmR=; zAz_`)92@2|3)x|)lXXMxoy0xb!nPq*b2dlxa4oA$Uu0U+I1X(cI@>{w@NuNIAadKz ziKF)Xy>;F4fh#l0MOI5FFt)c>ZTQC8)YN=)t*dXeH7$ZULSK55d{K(NyzW8Drp{N# zb|pi-z(ngnolLHKx}+yPub7_f9%*n4M*e1I$0GL5@!hbWkffVYcuMQI9x}l#cCw0y z+0f;|L|4j&KQYk0dKd)G$||E+g-s*msizheh8tI>y*WNUK5Hd?PfyFLQCQ}iC@x(n zn5F-fsZbNaHdh|9r~O-7u&i92WraW6x2;cPoYBhSYgN1-_RA!?PJ1cMrVd-c zz#Hr|Ngd!=Aq2qVI!7&H_YFaH&N}>3c*JPqjP7WEh%wrPEpi-Z4cRWTBFU@SIxu86 zba3t`b#GEm8>40!V*aw(GPKK~w1h}FM=AEfaoAo8G8BItd>fl`<_pX0UuyIa<;m9$ z!G@Vs!xqgr0y9PcQE!@3O1$+2`RWbQTe}mw#sz7U?ZC|_t zq`w7&gu45}YC1Gx)-XcosMTs|85XeiNjzQg(t-YXB4=}fd2#OiCI!@`Dw1wiR5?Z) zRD<^GsI8X`j>G$XUE^_wh z)`CN0rQ)~J4;E-b={8&)#+Ee*)^Z+rIJe@@`ogaGOE^=2gzCIK>@Of$BH~C$Ul|4u zb@5vF+~q^S&RcLB3qM~+K<5UD#hALf1|f#X)eV17lfCNM$%f9@Y<8q3>r8KL8Y&_) z$%DD?XN^ivYi_o-R-go4$NCTJ4RkfAU5BbU7n|VLHGV{7@3^o2 z>X9$Mw|18iZ892z5acxKExV|3V?BnW!p2wQ4od3(vcd7aCSMZ#Ttz~~!4a)Yx(v5= zbrPqft4Ou$LjfAe%D&aRp|!b8tvirxrjF>*LEZhdIacCL);8*EM(6h&*>YKXol0ou zUeI?E=gciOMmV;oQTGe_zFrnhZ+Yd346`YXo=8+jwVq&x|8wUO;^ipKv!(}$6ZC9p z@baQz7(RQ#ozt%2BZf6O@DN)^R;}LH)}OxtnuRQDjzM@2tihE~d_{mGQ>WrS1e6?T zHH6V{Upa)?eaF+2;=YYpX?|G3T!X0K%R8^Cu{CzIEF&!v+Zu+h!^LQu@RHpMiHZC( z^78V6OcEg%#`1|!it~#Yu~i_v-WAi_pVOI{)k70?>r=mmeZkUN(80}|RkcBhos@kq z-l|QcPf4@(VOuXtF-;Pzyew8PvZNTLNev~@{jgX zxtza&|nZt>BVOj6?hvR9|~bMaao?E1}B%|X<>=RuvLr}5Jy>FM*K zzN#y^ae#B(Koi-PiuimDoW)P5i2kT1WPjfRBkmxDvC!B1X`Y9HmL2f94-&4^J1i!t zBn<&6MMWmaQ6}=b1U_yT9hOET9-{gl9|8}N;!}Bk7gQ+=kH(Ogc4nE;Gd4BC7^0dy z8>rxos8FmfE3ODT2FC%P_)`_UVK6eJX@&xx7T3(tR0XF;l{(257nv-V@qn4q2-AQupbaJXb3q%ID zFLAgWD2=RQ8d*+}#npqxZ&L7M8y10-G#CmxG7Q?Q(r-O8$xF;>W9>f}{>bV|=q5>- zpgzqK*BYWSS5oP;W1sS#ct_S$J{aqx08ud%x_x2IPBBn6mB@@ZbNn2f-E`Yf(Obuf zC2EUkb`4<=fzdpX72?eUBy%XTw{{F;syn}cr@z^%{j^u!ejdt4gsc*rkQ1zV{2U2b zPUW~l#}i5u`%#~E_Zd&I007gY5=HNPu86(Hg(YG~mE3{$Cx#ynl|UppH3pySGEIJ0 zL^P+@H`|Zpbc@AqxDdSXZL4n1pH#)x({!o$77UM;E&Ev4JEExhx;}LJ7Ia%-h4gBa ziheGz_AMfj!Dhf?LiZQy;H`%Dw%pqZ7hyX`C9cldN4~^d{`ILL{)&d4Pt$=}9u)Am zY>vd*BZ80hT;bj(Ffl4Boo@-oWh0C4nYxvht3rIcj?^JFF?hGrbvqvwWhqg~(U|+Z4G~;?v*7-m&hRVXirhWl3B|@3JS@JZL8!9$`tQtf}TkObWRgmHpG}Z}r>9W&Hsr<7QAtIX#@((33 z)pJ4#A?{(kUcx4;bONi7u!Fov0QHNx(tOSfzs(A}`8S#T?n_T3cIlCUI*{>lrP5XW zv6_5HWyYZWRgD(P!EhBygDY$@>hT~cs30L8r{frh<@oXUiZdv?RB*29!r|R5=Bck8 z=Q~T9c+0`P^FYY)1pehqYqGD=v~1J-xw9$t_}+soH4)0&#rXVE z{txEfI;yQ`+ZVygSDG-uUuY#?Ia=Yp*reT)+97vXlYI+drCH(WRUumnB55`(LKxcb>w{ zuf6)5>^vqjMSynF>L2rkgye~2hDAo^f61EYh-c(3%*EZV;fme@ipw55m02D?kqBJ# z|F{_Ni?gDfn6mcH7-dXZf8Tb%vnviI^le=6r>NTbb=i|0FQZ!-f-z(6RZrPyqh(qopv3=PwZ69HsAlkKjsm`Awq53`vAC_z@i+k!6YW0rtTf) z&XT9U=~Z6x`h>6FnHU4~xi4pd-BCM10b`P9h@GQRM9@(ydSbQVgwA}j#C^(n(O?n1kcId>_7t_ngO^q;t9tC!P$c` zJ7sMw5f$+VAUb=p{Cx=2tuxE#)FwTn|8<%0nCg9`hv~ExzX^#U6}fzN#f@njgo@*x z_V@rttWlB1xX602(0uDFZJOe%ECv==_9K2%$2uS$;p)-RWg!RGqQLn`AW$QH{*PFi z+4LCvOLaiAAC$!FhKaP};+ZvINqzzK0|`4UAx}x9J^Vq{uSIV@UyR>ZF3ujKdcnoE zZ7o-qWm0RCH>k})13gvc(N2RF8vEjvwGV;QU8io$FEQEIGKVtCZ;!f7Xn)TQU2GqLjXz^#U+1vLvykRV=hl7((oUJOY z=Q?ZJ?C3hog$@pWlec=2bZuTQy{W}S??hGU99u8mFUr%^NOl~eJoaWrRCA{u@) zs@c&U$#5o*if<wKu;QEAq|TaTpYE3U=J`&4^%GZY9=FeS>d?FLnKAXw6Q-nAcCPObRW0+0T z$u3urZ*YS^PJ4pDr%?1T+Ue|&rB<)w)y7N5tkI{tfC^lPQ|;JaFh_Y8p-AF*A;BWh zjnLA(*D6epCnd9h-6`!La;Kb{*@oRWmRASSuj$|d^z@+>8fl%lu{6~dx@FK>1LTkM z648;>5m|H2BI3sq7!(p8tx@fdW44DE5@fUZ5-)*kuU4(-n3j7w+H|Nol~a{Z-|=uw zD=M{)3RL|WUzHTklDz!6gvYKsA(E~6iY>5e#M@0fU|FFCxH)pvY9xm#Wgzq`MPqp7 zMb~xb>x73(lLvtIC(mZ%`$Perr)_(x@nqv$4Ghh86T0lL7>JVqod*3BJ+5Q!(Ol#^Dphp*M6Fi!Y5**K3Oj1dE=)b$8<2= zLE=08pfLB~m;Zwdf-O5d%=Vy52eL0{In5T#khSY_i6>>eIy`~rS1JHp#jWLGSX~=F z-}DgJvXP`te(OeG_u3V2%CBVfuHQCN%-1{+5AKy^voR-yk7;jP^p@PJEIL&EvR#wh z=8Z4>^zjeIF4Lr`0-b+!ypjYcrS1m{`tkO|BR>DV;IhcUg$zzg+nT{?S(4AX^u^H8 zFk4$CM?V6_yjIZ>?GYL8aU+na39}QPA9%ve?(s4%6?AUHJ*yNmK+D$0Zv$Zg=l+H` zs=dnGIioc`ZZkoSt}%UazD?QpBjfM~mXtPVW}b~YjhZ}LoVeP|y5i2_zQ=23W*&RC z{CBJGP4gO3cT=TYpEXPtkBwh5`>!UtHfO7+pK{pM1$Z3m?8XR4T=T|fQwKHI`x+^l z-y?eZJceT)-X4LEq8iNG#rnz5+>`P@IaA0A+qbm^cf>NqWhBbXRqT{4KPeP3+fE^G zcGe>rc9!a4wmxW2rgT$UEA6g1WZN0mI;6o@zBZp8=hanB@;!dD_G$jLB|H2!9)2q=C7Pgcc@#U< z*wEB}NklDQ*wzL;F*3cK;HJmo1M$y()o#6M-<8tlT_DTt-;#`6XTYz0@Zp~Q z*)8g8=R<4xffe=z4){8Bnl1CyiCdN+IBvlL@uT7Mb2>6!XBliJ4u3hP8o8<{8VP&! zDV~tj)BA0*z*oXJ@_PKB`mPR9Cq*tvO#X>NZszp(00xR#Fp8d;k_$aG+VFxE~7+#kNBqo26Vr*`G8%`zG{e?pXui9|DQvPlg z*JC5*Io{mDj{DR8iI(q#B1422An0wsj-J9SY(i@VfU>XX4nQ|}&BQ;Zs?I@dXlMpG zarB*iBFu*5m>yDJYwZSRIvIi86t^JEORNBBHcpv{$Dv zU_3kGW}A+Sxw#BEj6l~oF?h)zv<--P^(Rqhavb8Qe+7$dwVl#nee5HdOo%iSg!=gS zGBPqwFiu){xt+<&1U+C!G7eyn_dF%2uC~|LJaYZn?U?(w#}&qzuh=p=8hnne3-m$u zG@ndV#hbALR>8x(eOy*m!_A5|tyh@=jn>(Apw`D`!)7~sS^-G&Jj@9fCaou)v=D2g zpxiYoIhI5EOs&b|7(bjo`NZT5-O71*l}z^dMv>d9xP0|d&$$+n`7?OSHS)-bNoy%q z*sHs`#ObYd_mn+12U`p6Cjkd%dLgotC8FITo zex>=%jMY;$Jt?7pK7Q<+u>jP$MJqwtP{+pQK@#PjJhx=T8I@Vb(|{l*o|6f3G`#0R zZ{qtABfY|7?;$|a5Gi3Ly3Ind@W-<=E^{-X>nOuZ^n`FrGYv$BO?It|l&eL#A5G=! ztIYm<-!FkD+s~ZE=MOTdc+$=fqno|(8?=hpID&dXZ5e+?GP@dfPTl(r&$N-?c-Cz+5d`=OKAN zM9g^ZA4%eA^gp?#a1N|E6Bzfrt1G0`p(4Q;`qYz+;ul7hGFJCNah>Nm&(Oz1GD^-A zy$@s8D#|>M=xB6Kc*CeFIKPZ)PexADgRw&$7I?+`>#=U$FXNa$ z=E(q$?YvBt_OJ}77nfBbrchUxR(?C4x53-YB%3ip+8gC~G3E&oqKlT|z;qDmf#a;F z&|l|(v!Pfc4m)v%tXKu9kXcn{YX}tnsNo0%OduHAeGbo<1xMrZYRhLqfXmv9njp+T z{O42bC@s>QxA#vK(cs(ySN8ILZb{%vh^0n}2y6v(W7^63SR{(1PFwKimTzBv$BXHQ zFDUTeEGus3P!zVUJ87dAB3D=^A~Vv}vUGla-0mo92f`fQOnkgvz7y;b$=LyYeG!4a ztS}p}O>fA=J)kf#P@lMndi-s-DI^s?uKN4Qnxyc!vH+L#paEPsx;`ejns&@lQTJ)} zb4S5gQG~`%g%}pLS77cT%IDJ!L5AxNJ@P!mZ>rJVRC#&}}Q^CixfyjVCGHE1FB8?D$Q^Qu+$ z**y0ih=jY`znjcYCCFMqFRo82xoSh-L*u9MVPQ~W6=Z2CUy!tGVSUn7bp(u3|9g(g zUgF+7-_p+t)auIV?i!fKPbJi_qU;MFe&&mM2Sb?`$3C zfhn9l=m7!Mk~E6uwvS#KSHs|i~xNW_LJEW@jVwapCu}yK*vGSLAc+F+%DNE zRj?L9FAvBT2XWy)X~+5s_J`(4S?N`=jxP=15K_g!6FD*#R^ZW^ilAc-||y4M3U29jNl ze5`j%2?9Q7i^C71q0dJ%6?cb1h`DmU-)O2y!CNd}q4vc_8(zhO#>7Qa$Oh8G+q^JM zSt3*U#GbTv%ba~sao=xgP;K97h~9upM|v1fD-gvjb6rQ7FK@*VMkJ4jnR4DE=l;bj zplJ7yy8fSw0~Yd%sNR|5Il6zx#GUHzT38}DvwXjVqHHn`Bs9A-Jv{?q(+UvE8({t( zT+xNQ;6;Ox{=J85>6gdy<~=>B7rt%hR%pKoUUQY#+Xtz^l-qq1Jd7SF*;13@(s{(}RAZgcPGzmkU8!;Fz{=xi%3Yi60zl(8 zyGU?ht{OzBkmq<5-OhMmN1dV0QY*Sj3vze=A_VD&)dw4W|P4V00q1_l$_J(?q zb?ei*CxwuH9=#NPE63Hp4v{+ygJy;=c_`-lHN?zH8TG~d7 zX*QDOKrAE<2Srjc;~!0wz@cKKW$U5i&qA^;#!{CE)x|hoNsr3P`v#1?6(Lv9U=r`K zeH3o_I>Rr-merBk_T%Xo&HRVY#fu_P9w|MpI2a8$<68MXAX zY!RP{tx>!uiZF&1W}3ck(0wRc=7!ERYAbqaAiB%+%61|iV;+(=x>8+bIDuB;Cl)H6iZsgOjKI0*| z(Dk;FgiiO^?|j{3K5lqs{u%uhh*;+nukPV;xxqtwl5gA0gD+ulnF zsmq}#FgDkGCE^HIX4SO+Ik|F_zA&uk+PWDTAm{}7YbLWDF^|fc(qEA>CWuyidpjsK$si;I z@-TkodVu+sp)gLAod}JYQvhkk-83g^1c8u`WKm5+V`-8eoL`Mv4|1N3o5b&}#u%=$ z=AZKM>@GDcV1FO7yirNAx{?8PlJ@o?jW-UH2z^ZZvL*FwX`*6jX-?o5ZrNB!znW(< z{@;2hUzL%;mSk+RVa&2LUl7WZ~EGFXF z$0+Vnb)}^bJl*Y-i9|#a&c;|tJEjhT2`R+<&3brvm~V^AaJp8G*h>BuPM$$~g6b4d z6W_rb?1PRN>M8UzK7VW6o6$^yzGzE@i`2ShaR!a?aonL~YN;CTz&naW<3TsJA9q;r zAsG&l`AL=3E2Fms!773YD^zG)K#sg7j6oT6gGQlKYfk=fcC?l!?{Be@On<-t&*LaI zr|MU=r`Mue|ELqlD}ReI^yJXorvA^~S!_3MnFxQ}7e5L1^>Wh*9u^ZV)F2LcL2bCH zyaw#p<|`Fw+1t}>Xma(|TkT|Qu-yN}4x#Pne2l~)ba=?wC!T9SP zkWksSs2%8EW`qyparZyo*_sWoI_|uma1emAYj<#o8n%O{~C-VDu z46}DEuMgp1s(<=POwTWaB8P3vgaH8zAg9{sIc8!R_kE8`{8tl=o(MB06>omdqizf$ zv+_Ri)2B?grTUfCTvYD%OtpYF=-Dy_qJl&)E9v|2$qM?`7EJ&1^$EHPPa&kwg3i(R zm06_q$2ea>1*Bq`2%#bTE4}oahRBuo&o4+-^Y`=bdH??nw;aq(zj$srL8z@Ot^PrB zxu|qn+vk=j=j6>yO@-81Ohu1SUflnsGyggx`g-4B|072E->cRC*DcUYl;->wFMI|& ze1hkGQv{*3XV_h@1 z6&B}NGmyRYdU3CBqGI{-eHkORg!F>+Hd;PXe_O&JFSsuUa>=7R>^O{Te^(uBv73`& zP&IzNJG<4?N7+C)V&y^k1U0MGe>`ryZAgR>@m&D%PCSzJl?ScgfU7~bHGxQI;!KRO zNMLI`e#my5T&+O0jK%Q&#J8bRQ{FDcaK-n^_0E#&%f#F;ym~aE9nL6`Lif{`FL1Gn zhFlBW7<2mS#J=R?@lJ1C7=sp{X-IrjeXG|AI$7m1nHnX3aYE zxZv6F-G$M};75!){rKm`AM#qds~7xFP*9}u?9-*5rI{DDTx&aBxy3`4lu9T~%sCM8 zb$(~BfYOAf4aVl0N+Q*&KM=oZ_$zgKA?SQD3x`%kE3Ki&XC)cTcJ$^29e&5w%cWWcUWYEvvWq#a?ej` zc-Jhen}G0*s@Q+xc)?_?iiW|4zhlzUfdgOlzHqGqS}g7G`acZ<@cE$+kb zmLi5GmV?F@!w=goRe8IaIfCcE#lbE_Y$|JvX*i>D zgP>dLek>Du9YO5ij{K_m>~sBRjQnG?Bejbi;D~MmMn51v;g%g^iFSfDu>Mooy#XcR zZ7kiNM7~;;M_p1d5Ad^wvD3Eo)1H01_EvJOvh2+&3bjHT(+L1!vo-jn@(lyEw3y#@3BEXJqGScuEZ&VoJf2d z*p_8gE6MpDnx~y2E~y!+&i-NYo;&N|gv)w_39kA@p5IrVLOAgeS^ zjrOQ`sa~+J_x(x@XQdk;uRQAeMb^c(<`4rv5pd^XH{=rLZnf!3E;oM#4OHBQ-4WZY z^h3tMZ!*$`7PmCZ6jsCWyQV@r1O$p3$?k402O+92tvg;0i7oK^f*ZW}uE^~2W;(ID z#l76muGpN;d{eQYCyteCx3OJk@10#?=(m!$7NPjB$hf`OoiCOD6$?Yfs?pZKx5zi#nM^(z)yBQ3?17Z`2DkabG7rasHYWw zG?tZZp9KHl~IxEXGy? z2_8xWsx7XbF~Jwc&vl<9phS}L|HZS2xx1C`ge@A+PkdqL<7fKU4Qd36=? zUm!u?@@$e%*bN7x7npy-8HfH@CsjCw zNRqz)5xMbCr_uBWAkSb+Ph5M>d`j3CpaYCO=t86_IYn?nkNN-)ZBO_+vvl>*)tt`JCvYwx*GjT4K1h{c|jh~5vKA@kQRUrKZCuSTAi`Yg5C1Cr_&<~VdhZw#I! z4R%8|(nTL~c~nXWz$i~pPbw3(CO&#u*CLK$xyosd6#DdlV)vHm9=-#xIa;NCZ7?r~ z$ZvL=tKOYoRQIT3j*rv?(pBl_vk5bjTECE5BlBLDTG7UU)Y4jQL)|wjtTm4d-09kO zVCs<3)X~WP?-L)zwljw7*oQlyc|j3b3!|}6eUee(+ z0p4#IODWIH0g`mPxn#UkCr))&3|ctm`=<8$eIs1*`Uo`?>BO23ZE2IlrhRHPZ zx^fdvJvpxD%iw%MxJDe&=p%KS`?91HJg{qjjPYG(cAv1R133ZKbt)RyVj*@u*r!H? zWpg;9YC!KKga8AnV zq|?IVZ6m!7a^ZmXd8wP}mr+Q$P(t-N0OLp0iZdX7B#r#c+&=oCLSH^3<1f>C)$eV~ zOL%mv!f(O3q3-7N42XWGE0Qc?|JHOmjZw5=;3AF76HPtV&RR&-M(@F#e<6iJ#FS*C z^Ss|-t6bs1jlvqf_i69B_nrOtbz2R-Ch%Y~EIeqbV~XpgzcK+m=8d_}A-`VsQInB* z*TpAW!2D9XXuJ1Q`v$uYlGmq|lu*&=ZZc!Cs+5z$8S%i}#+cWB(@dmdwY?Yc-$5Q0 zbFO!MzB})8&Sk>&%wym3S-Yj6tssI*mVGX=-Rc&sZd*DV+taeeU!Qq z`YF}~I!@{h^qJnDWv4!ZDj|1j4#jQXjtB&tYHvqY+ZG@4?!)&`@;d?jFiSql z-Y{L+Y=7}+5w1pB2}sysQyCpqPuh*_K%PdzPAMC>*aFMIQk1Lo~a zwukhyKMlLtgvu4y0U7D0X!BAs4=-_!bv>U?_4}{3WY_c>;Ng{Y)>sy{v?U(C6uHw0 z7Mh}!$V{raJnOC`GfP0Vc|k&3(!E$tfGS1h@hwQ-;5p`z@7t3O405X?Q7!ZVr1V+e zp_{6`9c3*Nx9eK`g=OZae=N=VrnE+&NCLfHH8F`cfc0n;Q&3M|y0VlLD~;1&P)$$H zl%4$1%Ro@;!_7qn#vNsm_`Y8IBi{Ie;yWeX=rudYYQbY3{C5E?q&#p_Lhbz0wyl#2 zO4DjjFs3dHc>&vaOw_b~>8CW23Iku{LybT5Zl#d1QyHfT&Ho0_*VVWH_`NT#1Pm*o zG~gX|p%~-89r258?{+NzWouXZ_UC00b(H#sI`F&IS?Nn!+)oo<)$?u-04{NV{etTQTvto_-5~(7(>N7nzqs-jHu>Df~m{2QLY`)5SDzynN8#29cD|UwcCGoz}l& zaQ^<%*NfB(P9I82L>H%3Es`rQaI9+-s*j*Tf75Qx!xcfd!O?raPn zvf>uufgsX=jc;J}<6y9-ikRL%f^hR|?b;8~&A2BL%<^PJ-c#x@AT#2*glE_N8o{878ox7z zvsLFgP)CJ5z#OWfXc{y^m(SOranzhvxmBaLGB4Ghc3v>;}Q#E~sYv ziUnp`Ua3|bykCoCLmdv7D_GB!%L=A2 zKlhIu+}1esTG6yyV&rn&a(iu|LhQZc!nqtJSFYJ;llk9NHH4`;pE33j4dSa%>h*^% z+kXXl&SHOL&ja!Lv4$QxC?t?)_WkUCy21&ybvs=e8PRKW7_HAl#cr`A30hwrQ8GYe z(bJ#3lr~ZCY7Jt#j2v`0sZ^BCr^-{UwWr)TR;ix=`kL+442DHIURS1wTl+5m(DupA zTu_$W{+NS=4%9xMr;-p_9dUutJm;Y7u0#YbBa2v2VB_Y}olD7BX<+yOqST6v{KhoD*!APuHG_ zFRIvqJfRnB&RITy#}65AXDx7H@`TOjb>^!`0THOgxfK(;t0LtH3&b26F-~GKVg3g| znR4`;12xJNqWhg-hv{3#^(sty@BNiB;Enp~bXkblEupR@2> z3LggQA)Jmu^fHj^Km|IA{R4KgYv%+(d=gaZi&zNZte(b8L6@`VDD><;DCgnf=85HD z9T#G#oSO9HaEEunm3YiVbi!G82L2P0Yh0kAb(89=1B*cJ*}LszZCM=5ec6X13o}2} zK3r0NYIA}%ZOpBJ0PYUrIayfZW88fYvc{R>kY3wXNHMXu`{BHY zX}R!bSXC;4r`>u_PaVF{TCh(glzOJ%r+JumNIbX1hd*4{z3~ocE*s=s;aV`tt)mk1 z)Ayugk3gGWa27o@#wT!8zp;at&FSj6s<8%3Wz{|PEFx;>kWPCsaveQ%yPnc$}{DJ^kD>9_1<$gT+}BwwKb4YGkBwF$G954syjke#)hMbur_J z!RJ{R9sALys3z*8FwZhppr>S4;xLIQ;p!dFi%-!A0XN1X(gmuc5p~kodt$e=Jtx1e?$6V|D0DXP zzyNZ_VBG{+v?7tN5$0VLSl`iD_@TnCdjCZKVN7cQN@kJkIFLnTQ%+hsM0i#YbD>;D ztfjRli}qPd!M?ZO%8exT)?`KThsTP8Ed^-^a5eAqqE{e~kLi-Qxu*e#I)m?*&%0U; zd#V%H)e;r@>ew%-b`{JYU<~Tbk4ULC$2o2HSQvlmmJP2*bxs^LB61V684Kwv@ z*+YFdlz%)$b}jU*5WS-G1}wfj0>9Y8-bAsq)Liie>*U~GKV4c@dcEhFn%ispH%iCu=9r@aVf@h>3*%?x=HF;TSVg|4Tue;IA z3dzn-uP=~IPB;aBI%JopnDzYr^Fgu?UFRRju@I3u)Ngsx zq8QcYXp!#Aw`f*sI6J7F%4ugJGJm|8e4liLx!S@|!FnhCGQ;IJuV7=u;hCpfXUdYR zYWtz~LUa>*G=|I3v9M$kjiYWOImK!(=-mTL7pQ^naU3wuE#hv@7hSK8Vmt&1e} zCf@`)WAK3=b5~{@3L=C76&|Y^{B#?)KC6LrSBYWLCpD>8*ub=AU?Z2qy z|8$cf1Ei6^{1^VRf1<7b4#@ehp_0qtK28{KVNp|AUg=+=dqyVPz*$;-Bcu5k3DlW| zTmIJZMS7s%7Q#RN!GHgaa{qf})A0XmLi|7fNZ8foiR7|M_nQ4uQfEp_%a$F-*2zh~ z|GDpXNNWMLH50NsGmglZ7)D## zx01w}?Tp159|B638$S077ih%~S!TaQI$Z1ij)TfcmaC}Of(3F-gvl$I7M^U2c(j>> zGh9OnSI#B1uDYc@k;-Q)(Y<4usD4N^*KDbeX13YndiHq5^>J^{6^W81MU($}OC@ws z<%t3Lus{NqCZsM>7abU{48Gi%oxU^JC~m{c8d0LZE;mbN*BzV-b*8vE78Qpl0L0h2 zHt5B@tx3aooMoVj$`PHbTMu^uKjo16#@(Oi#I|D|Q~`xhs<&+C@c4L~=}puf^zBTz z?8KdAS|Z*tnM3cC0fMtT^%X}n0Uf#-2O+Zokloe-_LFy_hYGhRz$Bjb!m1SzjZ3-p z(pVtohuBAM^UD#tfQjp>4~EE5bm5IgI${$t?j6VU;s)e_(M)sPX)L2Sg4c>~RYLj+ zh1mi=E#VjZPW0VuK|}Oe;n5xs!UNt%uJsac=pO$6Q%ILz+TQ}RwDla>B z#ck99Q~2kkOL5(;mLH3*M$Esc+TU=IYKQPP>TA=Ypy(`xYG|1wPfQ-`KvIogTWO$i_{=}NN&Np4fMZ?xuK}Ev{^0VMYksWq zHbn3J0dov-UFwB`uG)KcU2@k@0GKB((7B`AX&GCILF3Jfk)fFH88l}W8+gY8-G4{o!H$OQ+Qr3`fODM!(T?8 z7RB3Wlfsnsr@gl_ib67W2p~g#gS9stBP@nloIHmo4yeF%DWG}mEu5A!#w*^uqZS{C z>PZB6o3^C?;v(<~UEmPhyfH?k){V#4hBhZ6$xrL!0<(BNou>c?BdI~VK%!kc)ZzS! z-W$`!^&K%h(s;tfx7ylD@ys7}!sU}?IM3siy>UL7Bvt**jv7ehk6?9Y7AV9 z;m9}`Z4DKkP4QUZ)xcG6Uf*pEVV7DL?E|e{Uyz!6elr1@LuvI}3WRQAUmhsXP5_f& zvjHJ-i3RBO?-CCv6OLvxDISj8t=8XeJ|{)_UqpC2X<`GpZl}2P^kl?;@yzjOOTzl~ zw(;@gClzF&Vm@vl%rpuM)|>`XNXl+cM7d;{#W}nk;o){KqFN2;WL`Z*ecKQZd9?R< zTY&pb?aTh^(oOX;DbI7uO>-OssIM%aV7Ry_n|tA-A2!ajUvA3m^@-I8Yj*)3j5B_B zL`QKnV7<#gQKA4GRDG=0Q2qTV))LS$K+$EgV%w#_=6eA zY$eh79PWC$xN>cN@weG~#8j(r9u+OjrsZ^vjKd^z!yi@m^E zxO^O|I0_cw1^^3MAA8n8;$KaH1u%${nBqQ8g-ZbzRx>yIP0KQfX4kbRDRu*yS9Q>I zbQ1!ga#L(#os34nXH@MvvnE`$lt4Vxv>Ry>Mpv%XUqygGMGB`eRN}_Dd8d6cI0neS z1>O73SNKnhR_94x`7tdZhVI#qtFx(~ilX}nOrN@mQImDt3NrcmC5(w8X~<~D&=DFNu9Goe0qz^33~e}253>EorXT!tODYTAi*|3%!!?|pzd)sKpk2eT zaP#n}hcAUzmLfp@bco%dp!L4FC-xGP0ads1+PR?8!lCwq1ug^XG4OHiy*vY|m}iR= zSqzoqs?Eyf8Ag*7njQIm%cecg4U*eoV|V$z!K~h^BQWGjT!XAEF8w$AuP(5Cy#`4W zxw;N2BQ`wO*7WeO&!jCn8zu%Xz5x<=5PD{hoS^&IDC0j#?^h>}IRGC&b0B8lu_F=8 zUN$l$N(c+7NqdP%-$8%W*~f0q_VyI514i0pa|=OAe`|$NfB09j;WTnG|9<`->4$$W z{2v5n|HlyE|DHek2aZfrsk+s6+d2DHaoPkPb@%e(qHCkgeXOt?O#7SHsq=uukxwlD z9gh5)B?>#^5)3BB4E_``Yo+%E2|b_j^ZSn%)Qh4 zR-Bohj0CwCP%)l_3)*>?=HQP~(Xm_d2}v)Pe=Sd0Fs8EKVI=bCi0CBIwXt{X*tq@x3Y2EwYa_RYSdsd=nb_Xy}XLG1D)eg!RZ4<foKPhUs$%e`~}i^pqes&iS>MLQw^%z zULuXe$J*KZKketJQ?Q15XQFxywe*;=FQw(zt5fVVgR$o)9 zYmN_t06(Y0WHwrFi7t%hS(r&-$WwG@s+tCNkb4~$lH$vHm})Eqk=Od$*8c;L%FoWa zUl$O5ABZwPtU)u+s`pa6+Go#8FLQ2Z&}2h$#lxPe)lN{`w(EJY!b>;G$V*ghLe^~M zKOqT6_cvb#FJBy`3bQ0Bi+d+Gr83!0RpHkrVAuEYK}DK%b-^aQ3A^??Q!D{SM@P~J z%%WcEk5IAdw2zcO5c~pbjWLfX4`a-egBSY)P+|1~E!JVA9zMP!d?>T~!n?nmhMU)%<;ZaCRkedz*14 z)S)gFu-CR-*#YG+aBWC`XUalZB3f~ZH;0(2!yHl5AIm%FpBoCtdezaH_#BwS;0L_| zl2k~vH(Ruq9IDCjov6p!+Pct|fO8AjOF!$w9$LTHW_>AIYu*-hEB?ryrBbveRG8Q@ zop-EFRy^p&TXG~<3@sm)L$neff+^*SAvpE!A(~h6|_U2J#=P>H4Trulmrm0yy{JXJ3k}G&_gd)98xOFP}S^Lhp{TB|eq+GU-+5%tUW&52G$(DJL*i|o?Rk$zf5F}hAkwsFf3va!g1 z*aO(Uda6@;CWCu#l--Kzq8>}`L%tq5ye!!2q>T<=%RxB$Pvf3_TH?6@|X|r3H(sQfl*v26XP=aM}gZdS!_YE^|7zLiham zOU5Zqt2YHIQ3qCe7anpGg`Ew;XIhv4=rntdSldzXa@-eJ+9TT5N{@pNH44AV-4vCq zzb3fe`uk2$ebwU1OjE6g(-*(&GF0u~ahKj8PS2G-VPA;RtE2TTx~(ZCi3iQSF&F@x zE7T^X5$Y(Pi5liH!GuT^p7hYmdV0Q^#!|-1c%eLVL{5vTz-1(Rz`!@2^Z4i>DoY*B ztB*NV-F6tet&EP%d|+-yHEtU;d3Qghr>!WVR-zVCB@=(CZ>noV^31^VgHWiGF>06$ zuknHHhC&V{jd|LzkaZ!P+o3+@utj0UJec5qkzmku+86%v`=|I8hy=t$$)8i~V99n- z{0Ob$%heZZ&0Tl;3`5Kjs$a$}-AdM24l_!{;>4r8Cp=sr)};gw#Cu^`=jgugsD{=)!TrrC_gLAN%LhPwIgU*+;yJ}$^hUe!-0GH6}YWhUWX7jhok zxopg|qoN_*9u7c_%9g(hvu!?HQsA?F?!nK35vBQ`3gv4Bs`)g?NIU68v}yB zF*CeV|Kmu!vUp{l+NAsFoRMXo62yllDMfDM*$u+>|@v& zfrn%f`K*#^(iauaxVx_lA{*Otm2|`8%PIP>6t{R9ofF@!4%xn&hN%<(iw}b{(nr%q z^Caq14ol67uS{M4(7oSl7XDgT?%e69%qDwl!9iWfx>VyRuydn3QH~>}hfkuJnst3F z44Kzu5F3l(oBbGfTRC0y4X#`qbkZ*m(j|};V%1bPpcBR;n%PxR(UQKMq_0vn<+fM( z<9U2+5kz5Tq+>6s!1T&-k)FVb+0KC6E$=7Izua1)bA3<;L@n}nMtZ5J z^xhOYcfQFpP!~KJ0ldDUsq^XZ3Lzl%=?plWwLPta!S>z0^Z4rEGO|L5FTjl!3gMEi zjRuIS3&#BMLZ~gJ>K8SqvC`12xYM&<$83;lXJb_f;^ps8lX1E=;hm)Wb*xWEV?N*b zlDt@w;>TBk@3T~Yo|Y2vw0$+TE)C>}ip4P`=UQ(FJeWVgUdpHK0)|oT&h$TJL2bO&EkL3(ND`*4f~F;hfG1o>4-VCqY`P#sG1HMYdb!HJBW%_ zMITY_v}3PEzMQ+ldO7zGXy8&{f7eYGwu~@)&pH?MB2&iw~2gThJG+E;qhnwea7 zKfa_F^_x4?ta2l=7X=L@FN>$9b{A~Amw8z3n@yN|oz+r#f1q9Nu=BzE6szH3R1m&2 z3jbKQg!e2%zen%DBTYDa^QXg_druOC2(mq3WypUhn4Bl3ns4-z5ZA}SIMw_6fvn%p z*i!`t*$x~|+gDy{km^Ur6BeG}Y|S?lj@gPzGNXNte6zP`J=)&`RDY4Csaq*W3^z~+ zHVMz!VsE@=TbJ`av!lthq!*tIah&udRiF~jmp50XV5GLSnY?uW`1#B1AGhCyhS2Dw z`87J<{bz~Flac+njt{;akk`qdV8zXcIF$^qe#O4d(_^)l6qRzOEk?f;j#9X8`+YnO z9^y}%Vw^DU?`)VpF<5k?#0K07CHr-R+&#!G2UUQdYFuXjy5He-8V$V%V4}^3WBFjF z8$qs||3`V}8P>%1_G=WSNE1*%Y0_cS1?f#d5Tt|BJ5r_h8WCwq5k#7V-h1y5r1#!C zLSY8f+#e~u}1~)urDAk$Mw|#Qi zLd=12-jpx&i>w>LP>FHhEY2;}(flG#hrV7ea!Af#{f#aL4xK|X$L&sP1=J|!7R_V$ zQjA&l&gU`Cs@Mh(axzK%?n^tffk9CAI(&-iX|%~Lx{trs9+&h8XnK{04==wYsl)>> zTzgX&kz4iUwXCoSW@1UOBQ`z$3tB1aBqI%`ZRRH;agQ@&4~k^9YXH zzn}7v!rX%R$?wfh3Wm=ZbQoLE$sY+kOv#BYMd8C)+!^e)wK?Z2U}!=ISl>sP6xekH1$`JFE8HL2f5Su;~M(#5)eiUI!+y5 z9iE#zn-i>KJ`(NrT#ENpTp`YQyzNCn3AGklF@(^9&s%F;cbCk5y0aZG4&5AF)mv`H z8b85M(F>}>d`lS>5fT5#xe3vu^LPj?&3x|MTx0Z0I~$tl17}Xgb&zdb)*4TS5 z!*Zo8X+J5dYLDIHo4Syo@~+IK;pCoAA9auKfza)o{n8HR+AZZRvFnCSe2{>QbA+pE2jX+;{8JZArJKuE*Rjrh*I1YzW%&~J zkl_?nz>+UI65_n+sImi{l(1?!PYE(~PG)7LH86r8_Z=MwO&yRPxwd?&jD%p-%K+Hv zV>^im3qr)%r6+J!>^&`HPdWakEwSV`Q#aCxn6XS?XM4$Q2`kMRR9Wex^{YWCM?iSx zah*fY&zGd1+1q6`ULC&Z|3&%z{0&I zoj%>1$(zi2wi7w>O=G9n=a{W(ock|rt2m9q7U@fns<5%|QN#>`{o%}i44he7b;>0v z6$-E2*>fBatP6*Ow(tsenSL$EGhNOA)F-~^zMYDav_eNm)||$hP<(GnG@)6>I&Hx~ z@YFGF?P5BVCA8AF&qx_JQw>SrX2VP;2+q6Q~yjTYx=sIJ$ zoI~ALe^N^CoH`QI`6BbDWOZ~+b55#D*IL;@^X}|Kq+E^nuuq9mS(@4|Eb-fwCrfoR zs8UARwbf(B?tSgCE%s&EUz6H`aT?ci`2tpnQ|F|F056z2)X5$<0@{bqSwyw;y7gtw z2kJmZWeA99M$L2IU_G%cHR@f&iqZVaat~Euz^oo+^eI~0g;Cu0# zs|J?NX0Pe5(?kaLB{?NX<7+rUDJ*bC@bT29@JN9mL<`fsbOli`Er;o&HRTwKJ&#@> z`W$zq{d&jFie~1zNd~R6M3`f&RoZ&wkVR=Vwftf;u+k48hMqJ{lCQt>k=eMvu9H#z z{u`6{9}EUEDIUJN1^~$cPpMzyx@ULAK3~u1)!MgOD{B1H5!ETn0eV;p6hh<=?4yj| zzcVM~WI@WC;K49R_`@zN6;3cSf_3hEbCg!$)7qGW3Rm6l(#7mt6M6t~s~%(*a^ed) zo`<1d&ucmsSVmXZ`sZT#(;8f(==agk?k_thg@#^;NTB*cTy0gRllJ_o#;eyjxlw(v|j)X*Us;XBzzAp{2^2?sPW;HEi4;2XtUv-l- zkUTl?B4!(@EcFz0ob&Et`y2UA$Mt$-VpU^ zk?G=yLG&4+ko&mYj;foUu-p@y(wIaZlYescUF#nnJSi}8Yb0fib+=1xluwxmMOtfA zca2;L0g%qKXKe_*Rc-NJqs{t2-E;Apd|nb0!xefn4&2)?I1X$#3eOGI;dQL| z_%o+0->s^ym)<47dTm-Ik=w@+okBtQ)#^GNxK-!Y;xl&*+Ki@TRT}xC)v*E^?3omx z;Nzel5#EI-PgQ86p-KyF6fehjt9jf^tEOYVre$=|zFkeSVo`HhY(b8C&bK1eHJg)m zirqOt6c?xruE$Y=F<*C8_?D>Bjj-|4W~BTQ80rHKZ0ZAed03SDR?mu=-?21(Ncs9< zq)NcEp?hGpf&D(@SWicY7}WF`?#X}E|>NV@-?h(U;b=Lu`s4j7{^xT5r;gOv! z{>Od%Ds3matLJY>G>K=^@58u4LsvRqyD`(syBOul(WnT zLtHIdOrJ*U!znoA;FvlT>c&Lr0uvsniz9&)Ajf>%DCb-!LZXlEaoXZ?b^DK#tz zT?E&5bvZ8GeK73S5|}))AoSXb!FAIwmgDv8_agiF&a3qQ{rN#G&volHR^zO4@8t^;j zKE0lTE3(39IbLq(Bc1|(8M~60di2s^M4F|^x$Gf7X2Z)}XG~|!#lude@Gj??o*s`c zZB&1U_)I@o^LiZ-fYp4Ql2_b~u}HiRvJt=hRJTLLslbmPF-y5=!c&`q|Xn7}lC4{Q+1B zAIqrZ924&%K-Zw-{Jg9{m1_fPUm^R^2K_}}M%m4b!+BQ7^$|%5Uf_wVas-Gm&W@5{ zaG>RDSVw;n;Y+9Zi7{7#?&rvt>)&mdt>%aem!mW37kZg#UpV~-$dRkcZ|3d9Hq}F4 zaITK^qN8^ETE-v~Jnj14?$(qqfxN5*j&?9Jae_u3ILCd;Umj>))sb$hoYe0*sqrAY zXpP%04mA3Ov22RbRO(ToLmfEo5M!kS1WYNNTF-JrOXT=s>P|^GY0;)7)Q=joF z)nD_q_gn_CXeCwtK$liw37^|g>?q~7cj0djYnOON?svXv>7$rH#=V)V@!2gfe|4gv z2*IaM3mWHF)6*zG2BwNEi#Y|R{8S^`UtV}`aDmLg9ZYxr5cDJvs9!;+%`{5R2NwJads8M^pJ091 z^+Jb3xQu+f#Y6mI>u@f%;R@Yg$#WM3(pu7s;e9G||A3*@24-Z<4{0za=J48c9G#V* zou^Lz-jpI`%Rf&kWq|a2Km*0UaLMVJE5i}zCgbjeqtE6w@00E4Z5Ge9%unhxJ$=FubPq<$L%lO866a4j@~b&U+ck%_h?)oGkzH zC21?8e##R&BjY2|at`Ak+IAD8Oq=le1d(w~r)g5gpN>o!wgqpx*SXg5Y1Xi6Hdd?G zkKa+CimiUgO5bPY4m?*8Qn;>;_D+Ljia%!nSd}(ie16+oJVrWB(w-@$!SL47>aJMG zF0JzeQS_$cykadCR+K zE~_qP|C9L{tB@IAEp*R7&wc0zK;$qNm*SL{ZwM;6=u_k(Rp)d67PEO^r7*SVHk*W zb?NYO#10B0b737TohouB_U#aa1D(>timH7kO9SPW<{MvFTJIYFKD$1pHhp}ZFil#f z3JuQ`$jzh7Soz^(KBU>}YVuxKj3RKy*U1p8klQg3Hf7SYLXD!G_Lsj>VasvwWClRZ z;zlyjudP6xD}+zs3SV@9VNK=w_U8+iL{2>Edc#$xtJ`|oeuIU&b)wwTi{PJ%0PzX? zKC$wmE3l3ymr~jo10#=+Ivu&KrHc{P=EF7WGj(>}Ol_x$RVCegoAIL)Fx0tu#%O1J zlJ2{VFB_X`j;WJZ47rO?GJ&@%izi&XSO)fHzNIH=CS9;O-h(rF`=X5dX0jHdU;bfB z2Dtx+!09+>0ep+j&qmlaegq;U1G)xbQ!{hy9eV#<3mp|lv~dn5-R!AzNtNu@dKBV} z-0Cz`9@Jm&e9f@M^M~Ta{2^H(bPIu<)CG`hh>RW}4^yHfr5jzGw3`eR0$US9#&Bp| z43yuEK|0q47?c;kpA?@03=1iIEq4hCL1{+|E=Vwt87%Psrd$v8COu`(%fg>(ys(}{KpuEzEiWTl&B9MvQ~SlL9<#ZmW~ufWIlN_y26&GQ`a5n0f+`v z2HuiJAK2j14c;y8lDFP!sQmiI9s`4xPeJy%*4w0eDa)PsC;TN()|wQr1hz{h+fRtN zBJ}*mR2_11)IhrvEjjzghA45`gZV4Zzd08X=VTbVgZ$gOv);DUue6#MrytYN#@2Fg za_NS(t1i_@m$G8>#&{GrNTg4hfPbMhbyX)cFwxvY1&K|$MY-cSCp_WH&gUUI(Znb8 zH@-4wURsS9YUL53N~xeuzfug#kejFV@wEmyJ>@mcz$RGk?dR|OCg(9f?-Ze!D%ZjK z?G?^i0)Q+NXKbpZheeF|^$V`rq3=-}%Yl&Z|3!?H7Ib~ze`v7TLTp<&R*xSTr~z-Q z*T$D`oWk3{V?PhZXGaCIu4up9+rkCVH!H9_OoGCq;z|cUBe%JI^=!}L?NW2T}j8{|70Z9Du1D< zHU?~V1TMSHIY_$8Zi9@Gf{i8cl+<*z<^(OQ4cHp>DgFUt2hHwU0Mk1n*7Lu3d9wIn zOwGcq!`H(@WTESuB(V5{v+c^OM3NeT1sA`>lG{f66Gk;jEF{V|*V;bcw-3DHdX_e~ zEy(A#_dG*JxUw3}Fmq)`YLAyt7x1ODlpYmrTjbW>*8^sp%;MCIB5t|5F-0Vo{ldUr zioK3#O4=^uy}LgR+=sLZuRmYtGRd)@Wb zOY+c!5$HFlTI}xf!CGCk$k`sz4VrR+Id=F;l#)!6XAN)$LBI0~*S$Yox&6S*JCu!d z)idREk2L6L$DCa`U4U@X_#(2;vSB9jMZKO7xtWP9EYxS4sx}s{tr_#r zcL+_cEtdA*KRyIFi(G~eEcSRd2hcH8fXx(*9l7XAN5gEIg|Yln_MDwMX7TzJJFQv+kp#WEY_Bwa#-0;*PJ|438$Q2~8K*(`7G&Q#e(1E(Bk%J!uoBTO zO-gjyxE^ zIWWeeA$J*l*sdquRCS<0L*}w>{9b9(0jcdJm8ZlXdWJs(gqmUHt2OmYhy-S zqw|}6LkpVTj#amB=jS>Z5eO-i?T>|u3=D89Q2cGTh`|aOs`QQn zqWPMej3AfgFSjp|;vjp{bV-80(0^+c9)u0W?xK_Y+GlYZR)ietK2}RN_{LLYDGjhL z`fTA%+95xEMk#RPLPt(*>uQ_D)e; z2{KcT;6tvNhI=)qdBHo0D4}1EMU7hk-{o`vl(ym-t)tOST_d_+Mc7RZETnZg{X!A4 zt?4eV2KE{P3u;SBFNy&`r@QvpqY&{ZTkaO$RiIUJ$z;T8@hccE<+Hhw3X{L?lPQc( z;Dt+geb5|+D)>-CJKnR&aj^Mk=RT^2&_=NoWW{U_V2H$PU&dV6V5~7-4n|}(*w0dq z%-a>&^eRsE13S>TDk^>z(SA#bn-!+!z=1&=ZpqQYnS$%4Ij_VgY_&gWj8!_VZkV8O zUhH(8=2)wKv?`sd8gb{zW?!(yT=AsAtxOHJLc{IB8NjJvnBpW4B-@jR_JQ`x2Dm|)1%7AgNb>z+)xBuXt;Vzft;Y{dvD*+==!IpUmQ+o zy8t>C7h3K_f8`N#^2+}{*{gu<&}JdNY{f_Gq0mNaNe}@b%ow8Se4n!5lV;+5uZc?W z^K3}<_PEW|Bh!#)?c`UN*(}l1hXbQ%bNu`B7aoVNNKdr_wLRx}yMNRayTL|a{@~`| zpDEQ7PH*esoBD0SZs{uR3SK^RxbnL~5BDe@BhQKH-HyW4cyqype`Vz&=;1k~p6z`F z#?NDU+lY*5=ONW!5v#T+N|Z8vXOMy6s! z?7`?rUo(Y9dEadz!@l^)CbqIe7YA&fBir=ec0{(-Ohw(n<84!q;v}+9^tW@Cu!xev zX>cUNub&l#9vD?~lKOs{Rnt@6eGP**!_xCUpBPbhqeD3LMny#Do%m^2VdwS-f zNMJo%M##M{p*DEJ^%W7EMO8Wzyy1fngPVm^$=fN8Nxt=8v$>bIc(GOO(y>`F?cT$e z#a`?H#Y{u7)~B0R!jnioLyu{wq~5Qs3$}h45MB4hmlNoTU_1=y`T4Tv`i9mp`Dda_ z646LE6rt-h^E(pf@c9(}SU4Gi%4dZ!7gf9JbVo2mt|j0j#Vw$gT~P!H-Z8g4p#lwG z#(h10_%nWfTOc9ZHVUOc+n)UK<2MOT1I-|txB z^DEy%6KJJ5xaKXb$k#EXz@d7qx)o_*E3Gfg6&dFPjxIa8Z`jaIoxRoz-9Fwj!qY6~ z^DCY5+OcpWeH?`G&!Tfy%ZQ)@)aO2No7I)&6RGsAs;`goGEA>2pmzt zV)}SB`8V;9%P0d zDRcvKx8zfU5|E6Y{R3y>TixG{=>{_Za;7M>O?+#oGw52_?jvyf!~sgQ|2MYBnOMt~ z1)oCbiR)?eag}4jSSolBp>_my8a>Ox?To-a=T?K)4mtXynvrNKu$JY{e;WbyF`(5b zM|+S(>KCUH``9wRUBtydewybdQK1QqiV~xRofFyV`38T2bi0@J_OV({vfjq!CMq;*MNjE@=KK?$&?wA6(I;dOV0}qsS0%b zKxnPj_^-8LQ(e`{v;_;2-zDhOn$3rgMQ45;%8pO4xx`v^AOiSKh8jy|?L__tC{EtpOMHwzqv;^Y)%9jD1eup)+>N z4jm8v%c^n7gK`=he=u1n7A`PIrDxT#8Bw%knYG#Hwf=!xCQA*tg>n3Ej>z!x_J4Fl z@M+Sfs!aA7m%W}wa@P3s@CGS&!F7(uM~6~IPL{2NPd`Xh^97ybf1;^4>Qo}FO1KQR zu6jfz1oHCAf!XZ%efr$)4aL=-e>MA+vM6{iAF2~|qLut9T#f_rBhvJUIG5!zT%wUL zl953B{lD;G3t?lifKTNo9%Kzfdy+SXdaBJ5ku0C)6H^^N24LS0-`G&@q#l?6rV75@ z?Y1%|KjI&?kyb`00eSi>;Pm+!53lc7fNg1%AMq;K;UC$dD=;D_85DN?4*-Edvi#+K)PH5VX%VpC*nqhs^nq z=B{|aFKW*^f8ca4@VHvXCtd4iic@cen^!!z{_1g8IG>k4^Ze2Nb-sFpJ18$sqfV4m z+RGPp{RR#Jt-%u3IB{68L~GbTw-D{op4DQm8Qb$V`d9ch^L6-hcTuPxH-@)NL{-50 z)*Mfq+0J-^ez*sU9NWnpbq+};3v7=uy zbv)YUQvSS+Z{IwfIcQ(E@~y_1dZrg`pZm*%6xX zr+Yav+@pC9kldF3Td>~){~reX8RMJhMd*`=?g2UV03m6vPM50?imgd{^+VS4w z($+}l&Nbz44}GkymUg)Q%Y{^bM`mC;?7VMix{I}ELCqp_)q2Jnw%PRBIKsi330x;i zYYvtd#iyb4PZ2pK$(1Fo@{DvcYS|WWF#CMPL}*NSQ9v9Ss{brExE}9D=%~1PITgpF z^E`iES;>ZP@zR(3!|Nr^t;zeL$y=?4z#?-dsuTA}T8j{eK0hu^9yM{9a7swajO-8h zcxE+S;VekjJ2h1i$38!{i3Ycng;wpfM4!3omb>oF+M3g~a^K*;ciy1voNe=HkiISt z+{0>ztZBSgMV~V2(9Xv!0$Am*SSckm|ii} zE$wIg`@Okv+z0F%iG7V)z@FES@+9z=bf&qB*7LjG+PBZJ=n`8;GOxp0Wfff0P6B`r zv>h|9SuFZAp=cY6YUu5QL1mddRhFh+6|3vft;IBjj_S+^%~EcK!-0)b$nNK{jrbq% zp+tj(Vsumf+y>V~2x`&l5#=YPW|!Xn_!Kv{b2k@M&FGwgoWrM+IWW>c@OYhFylh|t z-1E85Y;4VfoBN7UvU-GRDJ|pPxP0|vZWxmigvRI zMJW-l*u+72^Fr^XTqB`w8(%y|cSJh9pNyhD++WdF3=^H4IV3Bz(WeX*7uZ5pNaq;8 zDZNROwq-yHeW&qyNwMIcSO=&Ig2RO^p9g{5CIhu!7B?1mUo8eY19sDZj#byI{GD$a zg+MMB$a^{02&Qe;${`zJ({fDJR9xCpJLKRidUlBN>N=s_As8U3se8fpxBQW_vc{ph zMEJ}eDAUg))kh#LQT~?>8ROT5((Ib3JX_ap`kH;dP~c_$>AcX__6BqbbA8FFtE&r6 zB;G*46qT*f-+qbph6E7SUB03J004GjN+F4nUo>TqfI2m0^b)SSApkY`3x{0et&y-k zzIT8(LiMr}v!WCM$Qt>jUC6~JCraL1Z9xp=rPV#p< zg;mVFLEhBiO`Ko;MhgLyibuga;8!xyerOodXkRlw%Du6_0Z9F_yL|tZQ}6ESBlW*d zt$UW;;ee$}WN110U!UgzpGzzFo1PVGR}0&L9JN#_Avv0WIqG1Bqn05~WCgE9$aB4P&AR zg(^Le^e$h=kbY}V<#ZnQkznttxgKTW)Oww0b)y&PYw`(YNy#KYPG(=%l~(RqPj)Xh zxF9D}q>(C3=tR3kqW&rvgNoCcvqWh*_W$ta$ue|#BR9D4A>;lvS zMOQDc_}@Fl23U5FEt0JdxfsqCyZ`)YW&QTk(_e#Cq^!crPk)6ownV{8k8%+)$(Ey6 z0Ya(SQF6)L(PG^DHBuLzUni?4E@B$_XyO;7Vvmv0E5 z8yUQZ{e-zoX`ILXLGPbtgz9MUW?)sm+&C#ijoT(CBP<&n!v5o=q-}$0i}b$1ilj}R zR$BPmh*=D|d;Jw;x0k27=c)jprWdk=VF89aLHrO2{elV|hQxe|z> zcHFCS({{t;Z@~g*vGa|xGJ~)<^&c$9>6O2p*&}I6W`C7WjDVWfEMUc zPQV_8+;K#(o9#1AT#m+~(E+Z{6A6A_)!KL+g(+9_vKjn@Hg@RX|9nPJGt& zkl(<==#z>rwT15DTU&95Ck`!Wa3WiUki;P_yb{@aH11s*HEFZE=(tGwYFwNl!ZDNh zv#I1W&FLaxvX zsGQG2Gy-=}3@jA?tcW7O?{z4&8nfqH>1wz9mlZtV?hXt^{GyfCjl_S#{09qV%VhpTI#rBDoT~lz)`^AP z#uY2Yf7}Af3oF>?r}sR9t4c3qG!CCH$h5E^o_6I-j?Lc^atw3$9a`~-Gi3SNn`KuV z)6d-~d7k@bL82Yxg`gwRjIFqdoD~*n`4$&wb&lEf>tYru6uRqYD*rFXLq~e9_J)=7 zS1W1zV%j~Y*bJ-7eVrz&Q-zY&R$}6ke82vZk}OEWfHiI1$qZ#5E-y9pznqz5NjQSn z>d?103N`!Dlt@UT2})|zJ73(HtGgZN+M7ibw1pniG|R}0^gzr@r$eF-j3{WQ$PkJ3 z-nZtNdRe&7_W4jM^n~nIB2Rwk8b_tt@o$JY@*TBpwKQRUkp8Fctlj)o0bx zg(5xj64&%#2WZ=I;sdgIPkQTv$(!$kaV7Ojzi8`((pN-p2p92ZbXTNVMlZ(*m4fA~ zQ6MMoF^M$4v44UzUV6(q&KpU#7cAcuIJu$ac6L2LbA$J_nzPW>#n>IQZp=&ae6NK~ z2?3_~$wo32B*rBmRE-`;kddi$3W~d-n))r$t2=db|J1N83~`W8m5i`*1a*|m{=(gu z(rP+7b>AyeEm32#b;%47n1f%N-U}7P4TdYy`)1HbAId@$E_%%Q+Q{e?8oP* zKE?iMUz$42*F{ULOZIi9{%vi=;m=P>()$Z=3lB&5V<)r%ry4l3F1H@j_%Eq=;+WHa z33E0?;G+i@F!o6pA^yx9G_EQP8?-`oufb)1s{ErpR*8LwtWaeu%CVak57baji8ALr z!9okC7L~ol57m^<^RyVCWA_|~xNeu5PgQr#J_`&%+M6f$)9!?N5Q%elqo> zeBF3q%cJeIA(%?B+U}5&yh6~Al>2HB{VqE7E~Puwz*uUkI`3FmutfbDLq_x(3yGJZlDse_ma!~JF6&h$Uyxb{6c zGO_XYdsVngiv(_)5XC6h5vTu(sBW7%&883mMxfgO7nNOE$ogg%Gu(R*pO0tcA?qsW zX}0*IuH&n3oX9`5_ZpVQ=98Q##Dm34Hm0CQLw1QJf~M#8|Yu&mvX0t0aWsap3(V;{0&a3`f|Zbi;CiqzlWqW0C73sjO*?3E4eu>|Oa zubxsW0kl+_L;ds+O6+6AVH5^M9ghJS%BVUmx#-}cN0z^`D>j4+PFpbqbaQOJw4zBh zKZDtYjg(QjbeU*%yF5wtx7zfpMM(_&AVpEg7XiA+ccXaNbe&+;QXrPLiVn$pWxbUk)5XAX z1_YQbj1pCB(1sL_vnSx}-(MFU17vr%0Tmv{jtkSe+qiTGcjiGy3BHT%xqEHEDk*da z62PV8pN5N@J>F#A-$D1E2mc?tF#c0^{{M5A7uWcP^)Yyg@o5;qkHQNT*-~kf5C02d Cpi0&N literal 0 HcmV?d00001 diff --git a/docs/modelarts/umn/en-us_image_0000001281686748.png b/docs/modelarts/umn/en-us_image_0000001281686748.png new file mode 100644 index 0000000000000000000000000000000000000000..7eee8739227a10fc496c580824b19a163a86d5b6 GIT binary patch literal 25214 zcmcG#2UJsA->-`s1yK+Y6{JN)KmiF=nh>#oNL56nhALg8_e2Cm1SwJi3PkBW(t83T zy?3dR-U);dNWa0o&-cFXcfWJS8RL$dF$gPtt^Zha&SyUJw<7d)HP}w^oMK>LVAFhf z{|N&Fqb+^A;3O0M{>OY&9R1%h&nFsp8A|&2*60t8+pFoQF))eu}X(gW5j>(LxE4*K34-Ci>`^PK#9lfe@AoblK{+ff3} zm;c#(J!r#r+$}ZUi|EGb)x#ryCaOLtb2)E={m<*_VR8fgox@(+(~88YpHhKA@YLJE zx>>%Tbxr0Q_`ZBau_!*jctEkf98Am-3TjZe523RPri9AmrgKs z0=7HMy9M|;N;N-sxULZVq_r@xyyDczk2dsGEc2gzu?drt)RwvQBW&(knuWH*%CgbO z`o7|o1aqfp_AFjFNQ4b2jE_V|HUemo>QoTr(Ys88Sb4d#vQZv8G_ z1xw$eEU1zdbz>s*2U4JDaAxyi=1q&8-xClgp#54XP;v8^3W;(_<(KAa36eJe&q%~h z@e?Tr{99D!3Lhh#F4e;)KPjV`qcN2yn5;0qzM6{!&->IrJ82p9o$1lYhrl*RrC<|N zD;VcH6t5U%;uJ(&L=X6+tR=ocoyJ)vEdt5`#{)j7d1gK!In%`F{bK|6Ja**`CN&jQ zq^>+eVVl$bNj%Ly+hgExH`$hq&Y!1AjflOxO z)Zc@_pL<2jU~4%i11~c*_2ZnoIFa0b`bY}gIv3btihP7I4rbn&((r;l(@}bf-R|19 zF|nBta2a0ToWUK)1O%C`y~29&yqE?nDYZXejMPMfA;(A$ef_YVCuqI(n#9tv*TL^_ z%A{u4)u{dFve8IY;v7bl!m3B%=9G5q2VEHNBpTqWYYG=tNzFRM-!42r1%Hy1E7bcL zSyc;MJd?}*k;UErZW}p%EZ|_TrW<6Z>>QNk@exT>i@$O8pB8lSIb&yT^kfwOzH$6_fpI`q@RYz~> zownl%A`#XeUbK4xUn7r^FfmWf(epM(POeAr zmf$6>s+g{@+}n-^llDcBB4)s8@o1z4vIv%sf|RkuWHnrl9RqAs?A6HfK&DZ9p8+Iz}$vHx2$0yzOhk)5bGUcAmRZV!S~^_+2$Kt-*5Y=Llc=gDn)Ed zX1Ll+k=A$8H!`jW-;Zd}0Q5Mc_E_Rik~UJRl%MNsGn~*R++Ax=gu}iD6E$qjyo?rtg83GXAgx%4;PWyE;`tqb5NqGiS?C1_#15Fjm z)+gr)CI+OO_+Q;lsN8p@Z*h#)dmS0jrfK$)K!3}A#%Fs3z&kiVI{iTLmKvP)d4PQ_ z`d;Q}=1i;CcQFpmFnBo&+L-7>Dy0-hNk=gwr*MX!lM)s@r7R`bVtI?(y=a(8xx3HG zdxG9?nAZcNs)PJ5Td2+pcP%a$n#61T#x*zKmEn^4A>@_Bm29sd?AxIjF%a-B4YbYG z&a9WdY(DZsos7Sf3CP@doC7D$b#QBAo2bXE!uMIlKnz2fv1E5Nd_g1G3r6j$u>T5-1;_sfcKVfKo` z&6T(Z0=DsodS6cF?708bAj(*m7gkWd{$lCy{Vi}i6TPJS&SzTsBlNpZbq)A3yz^6g zpTw|^>U?rL%S<(V5ZsXYO=rFjQmGTOCq0k7fobwnz+WlecJAj}yPtrWV25Oo+Tt2K zy?nRq^B2}O4QXY!TYsy>e-Y7Ma~nfnMd4MgxUn6DTWDpQX%oc>zbIy zz)Lqeg)K?8q^@r8EHp(TQgVTMhk7CSPd%E{tl7kcWqAm;_mH+~Z}aU-a0@oU0x?5F z#3~P4DXsBTZ9DG%_lhNRns1caj6N2w=7AWqx`k>8$50;{ygml0)cWXb+G;n|8&A;& z;YLI6MZx>lahrAXCXh9C9^r18#^|NKMICR+ZoRm*-|;|EN-}oOT0oPG4eee-KY}8H zb87ZwhWiF>BZ~oh!>2_Xit5jdKem#2+q+(tv?VH%h{*G8r-YERi~b&M%)T=T2>kqw zU*4s#pf^QQdW@_1YRuSnw0moYF%1{8(-v&o+U_apoq0UxT{~u*C`>Ml_%dA#Vr0UDFzd zxJ6E@VD86@0sIc!UFLBiXI9nIpfk9%gFW9>9|(vCaE!g8zm6(qbCzWHE|E6H1Qqd! z#Re?!S<@f`R8EEg`LR0_$F|5+E|p>Bc#Q+cs`-Y`36^(#9xAF=3t)!X8KE(S+Z62% zq`C|J7e7*y<#q-ggo4=yh>eh6QA)JBwLpxUu!hauWmkC14(J4BeEiiObV^|r&)a$h zk8yiIU}W$r3i|bl7 z>wq;)+VX8(r}2;CwDaCYEk4lC_%)W#ho%>y(Yo*v48V^1m$7g;ZUazXkqov;ih;c+ zH4;`>Rj+tv!^1aj!KUs6q#Zo8)utrlJ&B8@xJgL2zLt&kU46cLo@=?bocPb-Tvb1$ z-_6?*XX!lxG{^13^_Ih9ZyrJMM>ekciVx(xG0=67yjiv zX(?g=?>%lC;!jVOu1wAF?C!hv^1l*h?GO(+4sAkQ{cu2UCI9q;nvBgEP|* zfq_YLfmCWMYl@^L(L#@ZR-e}LZvYr@~Eo4OZ5mXT%5%2qS~wVm=^4)^kNUe znaj&frmJz5TP56=Z%6YDco2i`Z~A(MTbOo6R|A-5@%(+DMbu$OQbpR#110%DxIp%} z3Y|0D*8(GIz}g+uLvMFmYKswM-4CqC%G7&XUQ6ZV(g#PBD#e-%&I>-V#AVA_re3IlgLT9pA}ZXWl;h+W@jwn5NjKoWY@aA=VWiY1SoVSN&Q__H=mWslZvx$N#Y)<fw*g0499 zD`nF+^MOjf*7fuj#;xI8S}FS)70-B6m+h4g1bB1j_$$mqoDJwq!BGX&ZV`yGc`3Z-%Di|^7M%uVZye8z2IrI^W#iD2|kZc-A^Tq%#reOAWk0ruvbUb zG>{0Ft~z{0gh3zKuJ~ZbcvOB(_taYN|o_7#fP zLPI!Dccyohh@^rC6goI*{Su_1#5&SCB*yy=Hx&oSxW*iS|fHW35Scr3l-?# z5jZILgs?3`gEzgmaOc;bkdpiZWw_eip4WZoHHh06WS5cN17;infeTMM17ML%Q&3LT zPNx@aCoY}Ty89tMm+;VdZb1DD2n~g+jxa%cI%A9+w+dZO{91X1zE9O2?K}wIp#CCX z`VzcWvk8o)g2u1c%BfXjXm{*^wD2knRM!`P7z9;rTqq$^_V8lT-)Ai@y4GrO^nq+?#7d;} znD-dVBIp{)dj&rCNyQdS44%X^P}D~dR++)}ykzJ^Kp-13se>x+#9d+$@ zJ4wF*K0x_KKZ7n*fuKaEfqgL{&~7QG6oc8uW4%DtAvmOlt{fI+Xb)z~531h=6PWCU z%~0g&M?R=W3y-OMuk}C%w+7w!++G?@&qboY9eU<|388nzH1Xh_J)W6S+l7U`DQs=V zVIOVNXbPv`7P_92?($aQIqKNmtuO8?04#XoPJ2aKPxag3m==E!;GXGpHVBfYL&5J* zH1f!c+wZw6F4|qIsZYW@3B$5aK=UZ!lo%k5h~MZ3*#aP>Q`PABIU?I32ueCNCBFf_ z?I|Lh4Apl6KHxvx=?36Y#!4Qr&G9S`Y_l84ib7smfFU51)-?bgcsoLH7mPa1#_*cM ziY|~+00U4q z*$r~x`Ax)=8!cFw(*dcM8>2Bg09*FfZZv%q`v#>B1aVM9M%q(xemAnMCaX+o z+}4xT@=!NIGEzvR&C$_sKeEXiCHzG+evTBQPw1S9L-a^Lsd{3@*@d>JeZ#xO_v1hz zWami)h<1{+^ULTd>4M{7w^8R;7;FL(v?V5v8Ul%lsoX3-6I8PVgoVr?h03%G{E1s4 z8IT5gf%>Dg5L<=RtNXr$@YGGDYLGO8Wx|pBYFYdhUNKqDA94KmPRNQ8ollRT@-GIb zjOd%$8tR9mogQ+-kedwQCsfZo3u3&n{ho*Zo>xChkKSV~VEO@6z}PeuKAN~8rtul7!pG}6$_Wuto{r~%ybD!h%f~>RhSGRp+T#!KL zDXtU8m%eth?=v@DY2wMd!b5D0^paYLy+w@JD?*=7Q+yc%5EJg}u1i+(oE89yN5jA6 zH*EJ0hi%czck&gv=c>?#k;5k%#DH4JexwA$&1oUT163lZk$lEl+Dxl!> zoY7Ob4>;shui5*eE7z@L5CD=%RB9+wPH{Y=GUPpwgZ&z#i4bdQW+(qZB!VZ6ouSqUqep?;%2gLzO2}>b$Mym zn0)fqxTD>+z?P8C#(vA!!AETcaREz?!kPR7J74qt(>y$i- z5x~-S(_s~DM4Q#1*^8l88x-Ihuub>4r~+z+7xx>ogu0>tuiYfWNE1)8fLKcAV!uk= z%C9QKwx#jB-~Rn6VvKT*v>JTZw#Q)=wyLDpQz7X~U6bnGvtR9bdJ!7er)Rs{<1QGz zB+HZUZl*TS{_Ut8>wEfG7l*A?f*pt)3$?U z0KDxqovDyCNhtNU|3U+&iE;mYeS-Llj~{sFQ!kEZ;ny=>F-*N}c^Ra!L44f1f8ldC z|J}|B=kLQ$P3KkSw0joPH^e;i4brm2;HL-{7=w+jtVf>MQu^*JJM#6ityw%=&UY(!2hbV@KXmnJ1rN zE2|w;_R6<{C#e>MRR_XNY-<)c7y3RQ6&oCei`<5q?=Or3GiW=rz>~kw&3Md-W@`Uq zI%QR@(sV9;pn@+GNqTlQ1DVoe5g0ggAt=XZT=8^2xLI2|E0i)la6JBMG2z$UJ9_FQ1}G!v-w2fmbIH(^ij|ap%4ud zUc{#oXLSY+0->J^hq*9=>4)fpTGR)%rSkNfN^2unaM5_3XkjPD0Z2+a;4Q`0FP_fs zpW7#fTT)JJ3hIK4DE7PQ#NGAw<3x50#fqf-i#pSwHxd5|$K) zu*QpWU2_>NkD_$jMxQgj5~)`vNF(u(Aa;Z|p z#Nn|e_8EW`bYToNMFhnJE0bs+EJ~EHiEdp@`&^>X2aPN#$-k#(Gxq!aj@32%;xnRd zlOLqZQQ{F0;mk!|b<**#Z{#FfM_&O4UAun=&6j(dQ0{_@-Q}NwV5Q`&iJDRxwm3`!ZuLnnslv7`V}b1<8s3stY0PhB^cu^bNOh!2XSRdP+l^ zQnL=R|rp)l8 z#>-*sCH&^+PmDW1#GVvJwl+rjL{Gs%ApMl(n=zNZ-%gPzYPn|aSUi;B1B92OKlKOi zifSlb=~-BDuycl6cm~8N?VRdCu0H)Bd;BgHo29k=X{KD!iQ}O4qAPmX_PL(Xb3f|< zH$8>-wH3qHH|Pz_k7SXS--m&l<`(<7<4!P0Dg@cS0J=@FR%7Phm!l}B4vXEU-nyBa zZ4FVKM@3EU2PyCEVDZp5>^ z=5+oln}_u&+~s+GYw}5uJGG`E?Td9P3%fyFv%=nHH#jo>U=^FCw8)-)an8sV;+Xk~ zJ?G-?I3jCYWBqdV6iRaVJG4c|>kx6Ug+646g%z zk)3CeLP&gfvMPR93^s&97|nY;LKa@VcVpItcFI6qzTr!$_tc!2GZ8rKn>}8-nS#>Z z-|560N!rhpPz3ck?R)Sc?({gS8p4f$a8NYv6Arif?P(kY_HWRGeJB!Za1gNpOp#vQ z8h_>+Re<^q*Yr@;_ARjvOU zY^=lcpV&8No$~$~MNIl~U1n>FhoT>}Ve{Ps@dd=dz{=(=^{Jhq-2?;QS? zm}fWo%`RF(K5Sa^1>nciwzx|sgL514U49z=9L2m8%i^|&AVLV@f>g3vrk zu><^=Qt!Uaq1+zbHP1<^!HeB4piLIt$X-WudEMMs|7fL|4?-b4Fx{dw=IISS^c%E% z`F5<_T~Gd7U>|i{9EHTr2RX4jXc#~rFTlTbxaE!=b~CoYxWUm<^s#gShuKmgZnHr# z^If?apTGwW6q(*T^aE>E$rvZyBs@mqsm>W(6R>r97j=_bF9w><<;z&z8UK>sTv*NH z@e86%CGnS{&z1iOJwxDV78YXU7T?H&=Dor`ip0l5+Y#0yJ#;VkF2s8unG)3-##(Qq zagT;1dJ?QGx7a#iFO^APQ^JFl#Dn3anuqF!(8o%wY7nlyPR4G$IVoaoG z0=ZV>06T%Ibp;n)duqglRY>=CHl?|@KfN!6DIN~M2B)KUdBrMPKVIJYX_ski{&a2E zB{(>8^0SSl4ce)A_)+<<4|2SB)~@GVL46QI3MEMoJQ|2)^_iD3F&6l_meI^08O^Ps z9_FS3ot?WUWRi;O{S4HhjZr)kXuTMNRFJ2-rRl!iyb(kvdidkMiirebwwq;w)Y>Jp z0w!L})Pmj^_kV2R?{~#HPQ8e1NXQtHk1l5+Dq=YwL!;2DWMb<9Dw;lW!33Y18kT)6-0!F0 zES)eNhSAO%>Yuj^CucX7sGS?&3Hc2Tss}htUq%PC3iThj7W7D252aJM!Qz_hNTk>R z-bX`%H+=R@qpa`Z?LA_ijg{l*aC|Y87fv0|wwQW^#9u7#&n~Mva3lrFclzq@ud*(6Sbv@O2CIuB4qyqrYmqsq^#ey)t4tTkB%(vn$1mTHO`ZkkDwoAUzK)q(_h48zfqA7cCGBmL4f`-+8t< zMzk-xa-QuShBR!JR^z9aWu22LFBz{{hs__0l`R1Hl{kLiRH(hje>lKOFx6tROhSXczdY*8Z{Iyq>f_pC{XJC-&F*$O>z8&;n4!iQ{Z0 zTg3w;tnWkUn6r7}=%Q~0jPkM~PbAp9h%7;0Z}#+yM~)ApM3ge7`_+GGrEv*icpyTYR1#xv9v&AgsRSPg_ed**`{s{KpK7x>r|o6RIcryZdmYn`K+fyAWm zFE@YGm=;#}S%Ate55WrMAfdkU$&fs&?*!3jgWvFaAL;A%f*I3mRgT_LdsZ2}@5s)Z zVLg+0bBKyyABf4lcvi-OaF4hV;7DJAH^?F%Exbn_+j{DFqM=a_u@`vt3`DD2n+Su^m9H15W z3mdl}xp3z#wB&qyzi{8)9?}lH;lz)30j824%Y1SZ@KuCH3*gPXUColK1>8A{JXG|V+1?;|)KtV6;n4+srqw8uWvnk_UwAMo~9!fUT5fF*(-^^~Za%$pXBedXJD>e{r%gNZi>OCDL}> zNt@$S3_8m`0W+=ST3MNPT#t~ys}X~*&tWBh-gP1CH+*j`5;TNfq5PDqTnCy_u{`~U z?LcVAv=Aq!@m{u|6@eQ}^(GrwRB?lOykGAwJ-1{_U^1M2DQCgh{juOYI29xTsD3Eb z7gm2`=%&(TpIBq1I+w23`{b~w$;t=m;%;#kioL9vYK!b8Ci8bYm`!^tyo1lDOqc9P z2M{*K6ER%(m)IE;PP$M>aX?jCSF-B-zfhN2O-EhTRe;LQ%eFDsaMXBE{E#M0e$V|S zW$#sO-4~1YGJC4hXp0C+51EJ4qV5wf$Y3P5i8hoL{M<0mj5bTOY9Km#3wC!rfpRP@ zRD<9YjYaT!y(0NT4{(c7a8agvd{a4iDR-N594u`&Foh`I<~D>{QPxM{b;2Xm-~|AR z%p?7=NXPkS1wfA)HASVI_c0vFPfxOt5N&)y*dQ!pJak2mn`x6i0|}zFznmmSo`ZG2$@x9$r}bi zWwAT0(2eq&1^bIPXbLVwHr|DndgiDrXpY728wg^VDrrHLsMknl>M9i0n1RIbaDxR* zib51QSGXK2i-Vo-K)6~UTar?u&P|UD)W6Jt9<21AzN*Tnz)Ft>&|!6*dH<@_1mWH% z>jcl3?WID75i{pNo%cmR4Zn| z>qS9=MY@J)`05TAF?)SGSRroZAdYgMcQhwi0CQj~fTAqNfw6~)R5k47%8E!_e2{qer_CC zKo$BQS(aDu;eADJ#$S+){t4*6W32Yse_<@mtW2XbHR+v1t{z0iJO%NxF#luVQe+4E zVZ3{#n>;tz$udY$;yL5+rB`3@VAR)_$W!_~?O24#A z5s1@ka&(C7|GS8e6#@6}j@<$a8pS_Kd;4y5%_G#zjDPjv-Pz}@lr1%*GGZSn3-Gt- zJL|e0xYeQl$D+PP?GcUey~H@ww2{b&_Aon9L_oGwuVdJ4cs=0ml;h{b%ynrR70&rP z+J}z0rJp`u`D17u>{PT+TAkKX*LT;T=IfQki_UI?0n6=EqzXzSc5o1DdKBqIgN|D|qIfj6BH1E+EBalFl$7Dn~^)AJ}gGyeA)Tf$JJxx+qF-Bai4A6RRy|6_s7Cbx} z>s(RQdiW!unH&6mw<}lLY&A~yV_;^Nyv zi+Y}K?{hDEAbb5IUb+0q0Ewc7Q5uPqS_Otl!B)1`}76-;2@{R{1D z<9doqkp(tV>5U6ai0)@iV&W5&b3Uolmzjfn^yFOPZpj}Gh<^PT^7jM-uzYw@7vDm*{XinAi`W4CIeN$DFUohn>hH&_Cnl1`4&A#Wac zdLNZeWWhs4T>~jB4F};{z}v{3MJSo6*k!FMo7lX4+Lb|I6ctX?^Q8}%Z`art4>P3m z&jmckvLzvG2Z*N=QKw{l&ZpHZqV9KhTC2_$@D)Z~hx$@mcJ;Wyr)Il$m%g>}B+q}2 zWx2OjnMZf2{Lietq)Xa(G-Ph1?T)g>r!%u(JY=dcDs zlDFfD96FS`vbALbR+_QC@qxQyD@p{>r2KYW2to(lbE0%=Wn14kov?@O6cF>xd7s&BCfC> zGl)p7l7OyKPm)eY18);}$1j{$zREZPFQ(ysReJ_K9_%*MKw@<+)f}OHp8OTg`0y#y z3hE`jJ6-W9HFW6=#PYqNpIJv8;!pZvJJmD<=-(tq^p832pAN2UbJM>fYT9+7%#hVI z6!=U`S{RA$AM`2zEn_>3jB*9(Ke=#n1E#J>tNSq%vG`&DKl~YuBfT))Gp+qS$VWJ6Bq6T7Xk@x}(e)X}fkRAx(~I#0zSry}JAso`6^{u#9v2oY zCI(CQMBV7KlTs1Ls7G&0`uNIzvZD-wyI6-2+S9{woP=ebcaDuK9PvmMh5V30f+l=XlRnh{)F z#HC?IrW!W6RZdC-_0K!O@ZfTUDsaAq8WHn($5YNf^^QQWW=w>810gExYTYmh5;}H} zv?T_gzJ#9Q+_sb0-w<~j8^2Cl9i|24z;0g^WV)FGUS=^MDqlxw(09D|4ozRD{7vq^ zTgo>S0IKK?Fw=wbb~~n&j%(pqc$QePp~=Yhk_~-USSujF_ipJl9-fX}DD6(=-o*Io zTso9PPn@jvDQ0&GJPxut znu;Y`({UE;vh-s0ghbRIob^eylIsg2N%gOMw*R7YKb#N%Hf^-rBpCH+K%3d&PUEX9 z&zC~;A58Cyt=%hn68;@hKrTD;6j+$NWvoF=3@qR1Det=U;?iWvpY#}m&)zG{g<5u&3SKCoWpyHRdmyFke89`+HN;&O4c@L(0e$EJ|s(4ow zOj{EJdBba{pKgTCLR`Ju^N6Q6j1bfjxo>!|#Jw3R8SC63TXw6MV;FLqAoG&MZh#!e z6)SOrpS6ddDra>aKRP9753SjPoVjtleA|1SboaXcYOSu36TtN!?1|}py({F`@|>|f zb)Iy_vlH%7c+02o^J9%uWo~9aU3ULO6JEElwZVQjeP@bR99>{7H64wr9sS@+zxauM zqkmjM^8ZM~_pIOabsPCj&9{a8lNR3a7+4A?GX6em$u@G9O(nSL@f&r(J5;$~SJ%F4 zbe6)hC6GncK5I`Ew%&I`TQ5p)Kk~%jg#0%hzgK*Ib~Gf zjtA4f$oD**d|&(*`Kn~)LA9*HuIE0>=zpXPStj#(e^4ZUjMzTCUde+$U!^PXHX{O)y zU8aY1R?|IyJ0*d^{_d)b?C)jNkKAr9-R;^ZEXYRzrI36E|H=v+3&OlfGdXB=o*Wkm zC429w^3Xycr!!JUoF%7M7sDC8B(M%y3ae%&dT*HuU2WV??4JiWn;bwG7|uR7!fX)R zue_Nl&HM%hFC5#3?#nNl*XfWtQN@fte_PwVf48=oe5n7`RL4co9wD^wJ zw|fJLZ~oom{=>ejkPSDFq`FRiCBukv7`&@Hq4|2C3T>$7nl36zZ5`NZOcjM<4m(9q z=_Nfz)uSVAjo1f^Zdsi*lhwRYz)MlUc2?)Fl6{4^-FB@C(z>E+5V`cQ*DO0$4GtZVF1#>iu0-TfocNF49Ln1QfRcHC(UE zTr{1`3g62w(yFg0|H|~>Z}_F9weibPbiXX`9W(xaE8fK6ZKaKVDC)8$luA`RQoIcp zMcV#Kvy*x~dxMCG4Z0y7UZk94$O~Fj=!zlrZ#Ueym#YFkWR|k1m1ub@kVsX@2$Eqc zIEo&$EQlcfK=2+L@sJpm{4;s*u>7ju`trZmVJT@IV0hwsnEp^JONWxwzye_Bhb zda^kCq(lrQdAJmIE33S-WJ#SYW!}HY$+kSYea^ghiSS;j@s{~;$xR+-HTU4QuWUIQ zn!~Ich~|;n|C)O!$cZWa^6#zoUW-`WwAXbH3WI(U%W_;1FD0z2QO2*`ynT}}tNtbZ z1MTKhB`<&|Tz9)%u7N6*1=<*WZ0EgxMx_%XZ!!{(lzs+*(HE2z8#GS`5Vvx#@x2Iu zh6RVM11$mW#M*UG z7Fh1t!8*+$(@<5kB&B)8aO%Is55%`0rfX4e5T%U!L@tX+h#L$IVk56Z(TA?wRF(y; zHbW@njL+ZItr51(I2U-+DWT;-cp`UNk7Z)optePd*xAo>(z!2`YC{WU{X*)SM-tJ7 z5EuZsFD!(vtl6GGC3U3EAg0$#V zB>~Tv{1*7?e-^yMf!P;`+2#R>baDCpNOT*5>I5#P{w*$HATMYb<-}19+PqJ+zIJ;} z{8INw>x(P#TAJm<+xQs=m3+R=vW=yq1@|0QGw3(0(a4wBh|bnxodAXIdW)q_Z8Jh( zyW71TI`&I%6Zb-PmpbTIkJN~!7_mR<)09*s#VVB(oqFkq%#ki-U&uW=db(};qMb;r z>%>?h?R@Xc4;5M(e2cs9#5X_6a%acylW`+A*Ml-UMU2?G`|7tT46oHJ!BzqQD_Sg} z{v_=hxx3rPo$TWA8Gdk+1XH8Hu!Z1_ei_`yotq`2fo&Ur|8m?>WL3K3250)X#lP5V z4t=<>A<@`_iMaf~Bi#M|bn?>ne35_9*3T0reKA4r_7_$TqE`HWQoz2Pu1%|AAUXVM zQuKh1ZnaowfGYYZZG+vt4()}T6fx8+`|XYI9vm|>}|`PA2BV5@>`ouCEZHRd}r3`FvXYl z{>w~On$PFzfw%orFT}2UFRf&<21%Hs!0DHkUDrDuYfsS1l;K^yYj$>*{QFXdW7pQD z6WG4+UWuT&L_d&SPZEZg^1~m~N4=)YJ8>U#Yu&7Wf$TJf?ou@`D^<*-{gA)ud&}V< z5euxO81eM?H>~bz(*p)~yuP!;$qpDxL2{-ZZA}~KAEwQE{ZM|RjWY-5s}G3@n!|nZ zN{sowwDzHHYC79!!aGS;%x^trpM5G^5A~ipZuBQ~5CeAlPc;kANO^nart^UZt+IBC zbAe_1UJF*o+^|3B!T~6&SICvQ)c|taCk2Cd@NG%3sx{z~60`VaJr-AhiiOO7Hu*c6 zU{79zq3&#diijv848?s9ilP^~|KDKto7)MaYOe>GqODvjG5q$=E87I#@Mk~}kgAFM zUdad_u|lp-u<@1wTgx?B|xS zRMI<`&i|H6ph+$T>Yxikcb9%IwA38{G>byq*hIwb1L+9@K0X7Zt+bsQ$^S5~ZcI_Y zX5Ka)h2`_;ugMr>f`+nnb-~|(gSK7q^YN4j`3U0J^lt()X52f9 zqWVtVoItU>3~g7>`Y2X>JpO@41GTO?b+Hu(whBuD5~!nbC% zJ@%RY&=SHeZ|VAphoa#0gC|C+3{AGzlElq$0NB~8R@MF)z>7|fX6V>2Ak zs>cQS-nfZ#U;jz@ZZPXd{ou3TS!B83Pbgf{d35H-Wk3CEAXx=V>NnC-ICEtTMQ0{s zX7S76%xE3<7(?4%8|%R`mw3U^f2cJx@uyw~aTVpV#Fb_=+iyI7^}oz4&JawYFH@}M zD??7tpfyMpPmGX*R%}mY2HI|WgH^MAR7tn0Pwd?U1;4O}UNA1Ky2?@cS2LTi*tK2d zPM(Ot2w3zJ9zw4BFxTpjz*e~>H_hZi)?rzdtpT+m6r}f>GvD2GFT6#VT@BlhM5sUs z0FbzEXTham9vamEJp#nEV@e-Ooh2Dq--b~1CdWL%@zsEyd)4+S~vTO2#^PJ`aF^IL;!eaydZ_ z_?1#xGWo4V03R$=stHp~O{KsZ$xkhESk)&s2)Q<)Z@4MaO!=tu_Jl(6O1)mbG9vq1 z_uBsyGx)0wcFu_2cr8Ty)9{b`&P4%9bcfS-2j*}v>N(kA|GPCRK}^S$VCT;7)X$aZ z?X+BXa_Z}f-c;L|^-7-{JqOz`^SQWgl})|FD|k79Y_9m|#$LaYEH(Q0%1pM8NF@5J zt+9-x)#B+iFS{_~^?|g4&4x6bO29fB64%X08 zO4D74PJ!g=>=9l|SpqorxyoW5G&;-7YR6kS%mDpPmjG+&}m`b0Cqx zmzhOuIA@dO!g{F!2&Gn%=|ki!Sqgbu|39?s={wx{s|}OzRk2rhjceu1v=?)7h~l!1 z&5PPYKZbj_Wx^iQk#Lde8&*Pf^?<`Wz_l(Z+uiAz!dJpPl^~%q(GlgZK$MGj(VCdA z@Ni+`=%SsU(R^b5t2C$ZqKK;h!%lvm>$wig0b>K&1ptL2=U6H!(T2mnyoIg!uF0kQ z$XM!o4f6E`!MqKqs0P)Ao@#{O53frdc(;4|`;aEvLu;}#E1gD_I;75GQIqCFU0MhI zS187aXwgCg-&IY#GMEsZh!z504?g2I^d=oby772B_v)tLQ|K|$`{8~=nsGxey}nZa z60y^?04!*irBdE|>Io&*Jyk3Z><%wdcBU;!py(G4_RDKSCA~*){WlpKsVu$v3Hkw| zowUc&7J>ac3JVgaU28g&K0&vV^d!fTj1>wvqEN9mmE(5BZ0cEJMy5ebkNfJ2v?lVB z^2$?-^;tDg)}-g$^;Tr1D%l?dJxcP!6-k0lc+#!(fqR3oC6zZdxZWIkpXBtXjTo9t zx(titheqd?1TjPj%2D)=0tY^S{3{m@!Exjm5#%cr@BfIvf>|BVV#Qy9TSJB9^pkJ) zxxUBb5eN6b{~iIUqdX?-$lpkUo4{Z6D#(|JT_-LaTkju(`ft_`2HZ`H>6*LMXhxTe z=!B4?4Dg2lPH8Fi+Dc7!d;f7Q9fl9E=%0xwWoX=rB$?BuGV%Xu<-DVs*!L}t{U``3 zN);j^f>J`0szjO%q=*uc66w8(h(aJLQk1GRsX>$`QlttYCpb4%Gcan z)6!vjQoN7K`pXCY$JGReJn;wge_U}HTA3mA;^YW0pO_w2tG)W5#}$bq^vi6su=21e z@t9ufXZK-f^+?y5r17!b&vDlT8PS$9^mES~Mz>_Es?h$q!i7v4mw!;wFE^Hc&S^uz zQDMXOWPV}wx-44s=|x)5l5kdl84;N*&!;DNSV+zY=S z)$66nZZxM2g0Tp)t*ckhFyb+&N|z8n`dUIiE3D^9)Y0><3mf=HE0a`LbK20|k&*HW z%kz$(t1tj!*+{4EDej%Sffp_e|83vsoT0l{`Sjui8=UhrOu_8I$y4+ad1@_<%5~yx zzz>31@<6?ZtEUC9Z3{L2M{0uSno>di$e?ALCY9f*K5LeB>9NtJ) zxzm%AsK(+hPNZ^JlhYUEjOMUVaZhCthRI8z5p?kP{((+Yc0^&BYi5j;`vBm^D*F5R z3Us8^IGp$wYA`n+p6J8lLu(@HtI_I-EBT=2JfyK=WZtC1*CM6)a7a4bF{cHJi%RDkrQUl8Dpj!WF!=BKj8twPf zDcb848k|SvPZgiUxDljGjm-wI*G?^@i+0-8i#8#0`148l!L73E^Z z5UMjLWTsW09)CdV_emsEyP_v9+hZB`(i5;3AN5b?jDp2H8(R{#wlW1!fBSpBi7GlH z)nuFx_u?S?!G^<0;YOm%Z#Si3iG7AM*rB2HnSE=S9+r!E&FH`%81GU*I((7Z)QsBK zx4MJJhq)UKr#jWf-kA<7dTjDb^J8-VOdxb-WR8pWb2L_ctNR8|Rxx3ezcd&fMknG1 zp%y9nO95W;M4!p}`wwp@A0;hm;UK>O;H{t~_daFRQO3>si6eg}fe&v>VDA@Gea|VD z*`L>_$Cl<}OzN5mcxXd%>~XEY@9Hn9U%x#j!JK?M>txMh}0{?epoEKW;9kCM(pPRepT8oLD&n# zSxfi*0~gEizpUFXd+d*Lf0Rfr=1Yg?`MiGPQyIo5KGoU8i3Nz72+OFf^%!0gX2`pd zcbY#e{AA}V8wULw75o6UrGf{xRIvM0J6X0c%(`~398p+=mHeXVoRha*)MEPH?SoxQ z7j<2zENmMM_>Q!u8hIkE$n|^FxmE~;G>@lhk2aqLNEA-!zydlKkmh7@tW>)=2iQik zCo~Ry7Iu@2yY(|5vQMiIG*!WY4lpEwJpmI7E$*B1rtpR4C7k!34u`*qDF~O= z5g2)Nhle1&?tB_-6w;HGooi@}E^=4A4WnnEwT zAG{R1mfuwz|5LWi?$-FT{cQlQi|$;z)d;b>Ex>4SHHUR}`g?11%G1&jlkCQjBPW+v z&M)@pC1qjJKcDcls*QcVwJ2{BT8HUy8=XV?1CvMn;o84K>8=*7)xKnS*Ao^cKgKP$MoW#)q*)NUKRo0#A5C+!ugT}U+8!Tj&cZh4Y zjNH|#x>wdDzc))d^xlN6dHba@tyG*#kd+eo*l0`r(7~y}kC}(N5c4fsDj{v|Dy5}s z&jeot-_Lhdt~%6M7|80Dq94%beT02KCVIzRZK~0Myg7uGsLeq65`KYq5|xgKs4erw zJj8KqOTRUz-vI@yixdmTJ^qdf7Pt+qLm&P=Fjhw&WOBLPAPxhB5!BJKL<1n60>Giy{nKDZq z;ggIgl~}mf6Rr`T*}gu>5f{Yu4pjm2m1;G1B>SS$qaabQQ+u-Q6mxs~8VU?YeVYYY zPS9oO1xtv=$8_6jC6ugkCx#O5UzX>J%oy$jxCbCXCc(=!1=IWu`~6Oh{b$ZsOrpcR z8=AA@AM+IssoB!Hex{W3T>uH{#kSV)~X9J61o*g^13a*CVjpHW@+fj9n=Vcd?P zA1^X>bEx{dtof%RgC5la+9v{l7hS8{l%0`|05<_zp<=glx>HJ8f^SL^>E%dsyjx(r zDjXF%Rr1uW*GO$wvboVc7K>@)+)wr>9sTOLfx-t6kjOW=;YTB9-XC7oOmLjQzuwUi zGnaAyoA4g}QF0A@qFM=I3WYqFb#)!Brb>IS)n)^tDU?QHwaMk8H0aq}c&pjZkk>+M!)?8u@t5{d{%FKkI??|&wpQL4?TY+9_v0dxV?svmF> z6#c7bS4scX5(XC_@N?YV44}b~ZHP(^A`fnn(!fH!0X^Hn?u};0NDcLP*AF-S^_H=g zAOKJu;>D>lq7@QeytPK*r%?W=(YNCVje>D{p5Z-MMo0^oxJ~KGh)uM)MlTf(2_1-l z$xcfpzHUuY6=%}AWi-G`2=RrVq31rU5&2z5cM5rKTf!ZqWoNovZ?WR*3>dUU`HsPy zy?Je?F;^GxVgfF}iQSzke}mjzD3KhXFv^8AZ2i`OG?blXf1eV~K)}G$I z5{lST&#QZjhz;&L{%QZhwr6~DZA#mm?Yx8|c2_jkya=W$ z5i2vRCrvR(Ga)s3hfAlmvC4O*!s&)8F^U&hlsqk~;*F#h6fc;=X{l(|kEN7GTINqG zzaanm7@^FEg@s2lMxNpk<7SJk{qEl9avrE#HLO}r7i;sGR-%##K8c@cw!DIxTMO9{ zehNpNRQIL*>O%0lsG39$<3qCnK(BX;zLYfTCP0VA-`-`U_f=6S^nl^4U%!c$<5meWVuU|5Tp>?2Kf%^#uW&?aBoFzC?=6 zO(J(@!=Y&(Lucyrjwv0$YoB^?PTG9WHuunF-wlXo+db-X&F3Sb$4i$whi>g2c2Mb< z_f<#>NiL05gqNMXN$t!5N6o-E3-1GEADQ~}*AKnrg{%;%1r2SgJY9!))TU1qkmcnuz z_#{I3dUHQ&>FzJKN*5X$&$N-F`Y(hKF23i&3K9ga@icr>YhJbGm)@};?pA*2w?DmP}Wm%;5cs3VJWn( z_{^@2nM%Z}AZh+B<9H5z_XM6J0=Apks^Br<`D&{CCP?|x)t0;J6#NW?jqz*^XD9%R zJoiVX!qjc`L#a>-r|Yp$@i6{Og_p@mXQ4S!SMNN$Hi^zIP3O4JCpqtUDO&aVc+tZ> zwL>?!*zGG{TT*aQUsnL7d^brQ<1^RR?48tYd?oyta-fmC*M*o4E!dlHC9~RQxpfC( zd);$_m*C}?fxDl~U`cu65#?gG`6>G69Mz-T!(QaPo63V7gU%AR?P-e{(VJ?3sjSp; z`5O^*n0|Rv(JN>GpnaE&?hkcjs$Y2@b7x27Du-bXTxIxuvEDo~4Btc31hQ;-Z`P#1)%E8dPWOI} z4ul7(UX>glbCg=@A&l>_EsbN3int$`cAlz5Wp4ul{=>7C>hT9vn%^{?V2ZV_*m4D_S&F^fWp#xq4u9~k^=U$_$jmU-L-~|Gmw*1rrWHh z=_)e~I{EMEHto-RSG3z#Vga7@17iKZspns*wqIuU9maBr`*N}>D6YZPw{pG}iaN7v z5b#Tn5l`D#t-_sS1iSDBAxV#qf(4$71Gw`^jL;TG;|;`AIRVc@N(9r5eg&H)_;jv% zF|N%fV6gr5@ICW0<6%$-Nt3sUUepccB$F#*Ca>Rk$y!V^IxvnY!ERsU_6jR)zZzX{tKKYDea|wa_iE~1( z^vtwL<#(pY)th7?T8p9)q3WCANKW!_T(FPEksjEf4n>lR{@sgaPBI5wj z_{hdt;Va_dx4A?bs=3i|mojD;Q9&?Z?R58y-7Lr%GmAh=zFvbcG+1-l0!R&O@KfK1 zlLRrBocT!ANYT<{*p%G7tcw|glige&EdfgCcI&Qdot#{_N%9EESe9Ej(SclHkbvmZ z^jf{TrwQ)6#!szSNVaeX_$GRiF0p(-)e5CnB2p5w^PAWaJV=N^=&D3?yH{(EQb-A9 zpZ$TD+#IAaT{w!f745~CC2$-LRrf5W!2C=HsU>{t#DPg}d=a2hgGVaGB*T-0%h=nP5onYfEzm*sVN^v-kaZusg^lSd|M zYQQwh@wxKBKH#2Xu+N-+q+0|GUyhV>W!>vp*&(1gO~+QrOQyfcL0TkyZ%i=m;OXk+ ze0OPN@xJj``bN(+mVt2 zEn%cwN+WURZ&8b;zn~TfNY%u=Zy4?NX!LlwmyNkCX>(EWIH;GdiJNN9N|r=~gE@Xr zyv^8s*|GRZ=1<_o{~_}BFh^!$%&LyrgSI8H!vUHCxgHny!}tj|Xq@71HjH zBO>jt6?z?{Wy4sv-Ud8`8~J8KL~Wcg?_p;_yI&66!SmC8qSesFow5=RAPX~{MOfM( zL@J>sD88- z_;hrvysg-TEzQ6F2bB_J6_U?T*yp}nVcyH;u>He~0MoayfG*(b zdyhzHy05WbAY<6uChCgMnY=^Z_{J&gEn)bMMgW(lgG?kk5F#-LIc2v$vsP)9od$6f Wex+}|=hgeI=dNn&X%%SPefmGO_MbTb literal 0 HcmV?d00001 diff --git a/docs/modelarts/umn/en-us_image_0000001290603082.png b/docs/modelarts/umn/en-us_image_0000001290603082.png new file mode 100644 index 0000000000000000000000000000000000000000..8df2101cba8f188d9dc71992055e1e8312cd10fe GIT binary patch literal 38137 zcmXt9Wmp_dv&CHl!6iU~ySrO(cMUGV-Q6X@3GNWw-QC?6cXx;7E_v_wgNJ!~XR3R; zPn|kd-4m`LCxMKBhX4i!hWuSpR0#|WLIDg690m>s)be+&lo|97-cC}}5ey8a_u~(I z(6-nZ4D1WocTph~xAfCBS2q=t`>v~tIvI}(zF$FcQ`(1LWQCw;a&yBBbe-?aR&|}1 zo!6Y{tS*YP{Ya%0Js3!HLNkToNT5GSR?hnR>imh2A^(aIaB#!R8^d9C&Ziv5<#_Ru zscurcYz;>egeDvYqul5O0eT_`tvni2;ok+yOoJ0d?j2Q0@C9n`8((%E%K0l%gjJU+ zV(e`Q54VsaTU2IVpXcFbKM7+SGNxMhQuBz5Enk}>p%a3CR!r~CXyO^TP1UB0iLXF@ z?`T@$SvOJDD?MY%)(sz{QDY5XH*rVv=fBaHlgL%<63goX21}3WkCo-2nlt+(u21nS(7Y~iN5`Y|%U!|?0&BivvX59e@nvHK&XOiFf?|vda#ak20^mJ3b=ZE zI>RRqOeR`5=u$47u-&-xyWWU(?V)H*Sk0ubx>}b$;J@M-qP5UDuZ#`Kv5z?b8o|hH zOf>;KgSsg=UfCZE?MGg^W}$7eP&Et8esFh_wRg$FE{?1Dpn?I}72LDt(-I#vzGu|D zot+p6vJ%h*S_lZbx7WN4@_YdxE)b}$_p*_jywvcJ^kM9uhdwmYyxdGtT~rIYx>=Q8 zvqU1t1txA#5Uaac<5~~bX1(OK3spMvKD=Ow6y(@9GyH)A+FVx3j$B9@Ib^L67COEB zJV(6?`mUScEvJiwYo^zU4b%OIGUBt?ec`rv`ZvF+ORXb{Pb zM%6t70}D^af{7C^Xi_f6!1nm|CC^pG-Q3`flXbZtYJHPu z=wxHM;DZx&~ z@N?T0OB>VQ0v2=-n6*N5dN4jf-6W`)@rFZ{9n`DCigvI%qxpgO@txB>p!9DDDbBIQ z&o=TlN3Q6>@@1( z*$J%v|51B+^%o{3LHJP3p9h}pAAP8&Q4;@a`tUX#1EA1BYY z_qjhZejGpJThWQPU{rqB8Zo36@ZaLihZpjdsE%c_!IH42?mm8Tl5Mg5uvBZYb~ZSK zOS#2Z`&mwnx@MF?biV$@$%mgMhno@&HufMl# zeb4l9fC%S5pEC`zzbx137F1^Xx<62F90n&`={t?%c<-^m3J9~)U%Mo}BuO3#r zX8@)=xadfChK8ccLBpy|9JIf!;d;6XSi85Tyd6t>H3QL1<}3PaV*A4)CdO|tKVTFw zz65;{)#3CHADP@ZpR5EX(%VC6b#daKQNDE#5+ZNT+Yn6hZ-9HdNxbUBWT3LE#{GUK zyGb_v#NmK98GRhEHAKqqV~b0}oj=6`{DhrfZW@7NYta9fbS1_;jz|2fLhF4NFDeK` zBtdpd)6DxaOEns#8fSZH$N0kkbE>Po}K{i=Zo8}#xKK_o$!V>|^@t`$_F za&@|7Yw~9<%>)(wzkW3ezvG2mBuNDcyuhok!YDLsyJA@3pvOI;v6o$JBbrIqDZ{YK(Bxm_=oVweRFwpx|mcya1 zzZAp3hn^n6fxt5_(@y^Q^a#WN-P=d#P_$nSnZ(Jqg9bGU_3wyDzUmR3`fx-y5Gait+hGTNY+7Hj$oK#mly08}2nhTpi(qV-1*;#A6ZLyH~MU^JIw_cVQ z$eS-Y4YiX8p;n@8B%ceY z*Yf0P2>L(LgXV?uS^R5M%VDMB1!8JC-!pLiN`UR9{@7LCdIZ@v&BL|&co1^Hxgxn0dWKn z(V!9{Mg5OjnP|rUqgMV3#tso_Gtel7S!40?{E17T#C=$GGM24f~D%Ps2DF1|Qvg7z#rqpaXHz2R%f>tUTBM{~eD`jsKZ% zyJIA;E~o|BBKb22B>fMc&E{VQ8vkq5iuoY@dh#XuNO%x2H%TE|8U1^Ost6AYBM>id zAG67>;DqfPKpv0(Z&19tr-Pu`At;2XyPg1r9Q(vS&4WHs4NtRZ>iD*NQC5qfSTpUB zLaU(S=;KXC%opSqFbzTRx7z@w9^rK@!54;QS2bWB29FJ@nWod#*>tE>;XX%|=H+rq zs2&}mG~tXM5AHz;sI6E@fGt=VOliC4qRE#qN8PyQG~yEZS#DmqKxp}V3j$qF3WcIH z(a6uCEsW*LJ#oZHp`>e3b$He~*l2I$ZH^kw4pu;&QIn)dW;s*~+E8z#1IcuV?#n1v;tK7R{=B`HA%}uJk_ZNxojEqecWUp%c>(T_&iIZ(ZV{OXf zAdbd>Zir0orV_|}(Vi=2x1Xgor~>LH#DsxgZdkpll$VK)!BOv2_=E1#{gz)6&lJ}$ zjS5>zI4ZXC!)R;;8xv@LH(BN9`QLIuexMBl!?xO_dm05z%t8z|3xZ20BM8$&?L;I1eVAq3c~X`ge}$Lf^s#Qmqk8wwG00HNPBg z3G^oP3{C?M2>p7C@RNADcuEvJ3ptzX1LJtoC($?kiKf0m9m{kAfQ#T#;3>)owl+MS zZ-LVzzO!;&$uB;w-pcf}<4QKLP2JKWiK`pL0qtg5;9sA6vy>B7%{DH;yvYodk~j9< zssY}&cE0lF#6zD&K9U5wj39Ge87#^Q#g1SElG9p^ZgaH>kOL7Esrg4@&B@{MUoSDc znbl>ar^iP${ZYX)urLL=wdjeW65^iohm+*yoF7^K#{_}U%u=U=@FbX?ay>b-A3v6> zTk;zxb6vRc^_+dU5lEAj<*VT~(FK{OLl*gPHMKYYk8sY?f`B#=+}(__E+?B(|J+kZf(d>( z_wu7p4=`1R6%0lB`6>$&n91e&=U!%xW=(9v|8eE?SdBEvVh*lok;e0)J%H2ZD~Fs% zSn5d)=gYFV;1nG>=0yXNt7ym4jY7XEZJis(r~MnaH>Ykjn<*J+(U1kr3zj z;buSKH^6P9mUcOQvf!9BvLxyDE{oCVo-~Hr{<*n{n@%QFmm=M@jT9I1+n+$W^0K0@ zLFRgv^{6&?PsW72g44yuxjK;=^=>SkL~%B?DO}YTX7_}ALWD#?r%H9_#BNH~+x-ZObS`gr$2+13#2;dM z^bq z%;A`v#TBp85kL9kwsU@RDDxFOnsfKZUE@GsZN0Mp*Gt@aTbU+!<5C^iDh@0!7Pz1Ety%WacF7idS zJqkDQbqF=vZpo<;G8tV_?`kB`eEvfnVR=l+)!^U$E#CLH^m50YHI=_rNNjGJfT$(^ z{=uz-cs-xCp!-zL#hpx>EAti5<9K9yx${+a1frG2MS+2Ljf8Ye%Ef|6^C{EmkT=E3 z)8%)qe6A2E`|c0Q`1Wn)Kg9xKjC6YF1T* zpV=8P&wlh?>tYjdrX1@dB~qI`v}939h#e35Zl^r6a*06`08$17CUhrVNmw~?P#-pM zXLqN=VK*{GTK626>fT{ZZT0$@AJS4fE-kou9_!$y0fl!Up#);LURrj|MQJz*&h>#! zS~e2HtFKOBWJZf>O42>J&bkyj^hofLx#cC9+w+J^j%7$vjWu59^x1rdzvSc?t}^Y!uYEye<1WHXZfs{gFbgaxr593l43pc`?8; zVKW^`t~9a=B#xx{tqa4nRsDXw!sF;}Y|S`WD`47xKfued9c#KSw4}D#wiE0IqHz|G zV4NP4Ng80-%Y{cFEBVP2pk{0rC%W>@-0%h(eA<=qZl&{jQPr{8_7#Ga8Zl&r8|Q_D zqwbEBf+Ll)axwMo)YOr!L1N(gp}AD6H%|L6&I16zqWv=^x_%Bf zWVY0kEDPrK>NTA>A0UWvFzCZXw};|zV6$3Mp1xt$ z>U&2GQQ~I{ue#vG`aHmh>{+!nL5zbc5P3Dr1cVEeL1;bbq4MBrVOKwoWP;2y_K5@f z1P?v0uuNpnpSh}bC!_}@&lo<66OhAiYg*hGBwC?;egos65KBq^mCF@N&kk}xaQGYV zRuwfQEy#s}%|RD)oc4qaBz=3Sfd2p7$=OtrXZfErKII38#b-m5hXc3E#zFWC3Q$7T z|01|>n5jz*p4|LL#idTK1X>Ex%xqcJ{696IPWKi1e>s3TD0<_H(i7~LhhCj&gNiCz zv77=5qhJmsTG4qX|EYo5za}u5XfcNb5abl>;F#N%jz*rB&l#!#S`nZU5lSg@*Gk?$ zA`UIZy1jN*4HHMSgOta9v6tOTiuun(DRWOvS<-}X*n*Khx^enPPGjC4g;fBXQ!8`+ zPo$6pnVW{ncmIZoDgh-}faDhg>CBrL^)k`>*h294t&h ze@rQA)p+ogAns$e>IyWg8i(8mt!DjbXg|sjqAaCM!sSk+*tyekDkARfH!+L9^g=&g z$i1a&s_DZZ=*5Xx@TLX?VSXcqx}R))H`rUQ$I+Pt?(=eKW$zb+tdByfUuH8-|a&QI!+5}cm8jf zwKJ!KLth-W7>Pm2*~UKvin;|3d>>p%KrDeaXb?_D!-)qoajyD87-1=xrLy+er z@$Qf{L$D}|*t?zW7&gP@y})nct4GhsE%Y?w)COUCq22(}*5!oV8^|+qJvC$__h_go zbGUQVOG&8su{5i_?FJ6m#cbUz1WT^_*1~PXfTa8~yDD15m?WB2hNM&{_Z~^tbhDqqvwYm`jKZVJgRX(Q?*a|6VMgD_E&y3Xhd}WazV2|m`MenKa$81uq zp|!r+^hvE+(2CJxYa8WP&7LRtS)P-2#T@U#6G8SN^!c>0v z@t`j79T__P@TOs20Svapva%+%!+ly?DR^Y3eDH6%+8^`=8#(X(6X|dCNNBzqjCmN1 zU?MuO1!cYc{`x*ojp`@ToScV#XD1=i(N7ABgM88o@(-yn{SR3L{Lsq<6I@9}k(lzr zIu#atAnaA9+NB_yc!4q_!~4%d9^sBL^+%iuoc1k^Xvy1q4o_xTV5X;v^w8Xq(vUE8 zQXjR2?N?_flD@>fq8$ZlkOWEIi_iEUc{Fz5eip9#s($v=-6s6{TchlkK8RL1&qHC5 zl>a$+uNcfP#H}zJ!(W)Y=|<2^{*kXH*y>?gEUym-qZ{&3u`?}OQl!{Ff_gvFyx(dt z!mKKFu{Wk$$l<|$G^c^$VWVuh(T>;{HnV=F(|?X;i{x$ZW%^VL%80NC0`$=7GQ#bW zS|Z+Pfp$!Tmn6I+#MoWdyu;57X0@F?@nK7Xx5?fK2ufz1ldtvK-^XdmDR~imQL(F1 zhE|(XCOdV-k8_j)c1W*WXWxJ}e-W0det;t;6Zdy&^ZIMi^h-3<*e!UusI|nSsSd8V zn$@uBi-;;gO%$_5{LNW_OO|RafokEf20L}t#R{r)V%ZfXg4 zVgAm=tg`1od4>j4S?tIcM?WFu4%Abd(3ee>r_2Fn$}SnyCA3Rp)wuOD1w7aFHV)s@ z(1%V#SUDOvz3!b!6mZWTNX@glFJo*WlVm2x{KjbSF%(BMP6id2dQy+6A8y{#vTAtQ z(*Cfl(hiU=OuU8>v*#<%f0ZaPSUu%%yBUAK+PBK8CfR`!@A~ogpvYEFt--e;;k0H& z`a(Rf2>7Vhk$J^D>6;N!T^Pky3!Es;qT4dJ;49X>ijDfpI)qO##{zcB1r96Eb)bR! zO^Ts7gl8babz}34W8Lg3SVRN+Bis9j8ksHoZA^V{NwZo|4_1jC z;V<6Nc0!$cbdyJciKRnU2SPKyC}Z9)p=;HaaIXAd)7Ia@NiXa|Hr!Bmls?(b3YiQG zb!0L{5`J3;u-d~F38ak!T`-`&$KUzD_y>F<(Fh5P=etDXj(uU`zpx zivyniimofZa>dzQWcrBX;x&+k_?Na>!NJ{w{cQ1ix`n0e)$!z$Tep0AU;NU92o8_o z*KlBFv42Ys?SdCynuJ#6}&w2>AAdzVn6j(}WM& zvB0&|!u-5ZfAre&D%a~$TASV5Q!I<}8v6d*i95Sy%eCZSljG=6*h1-gZd9jp zlDz9GL-y{cOuLjuP!R$oO1C4Fl)BD$pyvj82Lk9yte5M;3s``{4+;!JX}x3ps-pT2 z%o}BZfh4gDma;Ox&$hn3Tal+DWM9(r^Nm%~T&vm(W9AYg?T|q`5!cW~XxICf{cSp4 zp9ZduUfHa}fBq5p-&p|7Thd<32rkYS$A%=Zrca#K5U5U0IoEaJO$}tYyk9=OT>$$_ z`yN8%HL6dw&xzKOX2$Y%=NClQ@sez4$=ViYy*ixuvH)n~=+Qf5-58Kx&LfBtcEn{) z(OR5}e*m#Jeku6EWfe&5_6{^?u(%*#P%f>_kVlWNj}v zz?W_P!|j$pT>V1j#e2i`XB&Jg>`4=;GKSR+V8q#hi&m`Xq{Wz4TWSHZe4efzw*`-j zBeO5f&Hlx%xjm(?k0+B=vQPi}U2L|;mAY4!(dO+IG_9Vh)HWL`i`kXGFvHVv;C@$th@f?GVU3mEm1J#`^t(nhgVD|z4Iuh#;e}! ziS`LEq;ZQfwONAGE1M|x5vqsYYtrPDbet_=7d{?$y@*V4?P4$7%+sK3N$7l-g{dHP zDb51sn38T)%%{76UksypuL0Oc;c9IBK;NfDoAQ%=XlW2!ULbA|$j_ zbkr3wFhb9^`4sH%Dnb8>J}rQ#1D zljZ0VOnbkN`S~X%BJ&W?r@osoH9e!@uVS^ASny*fE+t9H5`DJ}{hM^_3a|2hh8p+eU26gt-;FGiVAnzlFC!zRwCJ=- zp_PBqiPwG}t5;FQQ7myXXZ0&da(3h=E7y}2)Ii(g(k#Fb&|Qtl`w0+Zx2iWU zSChFTk5Xe?Tu@jI(}N;MOv!M1GE}IWK^(Csa>Dn)^y0qyvCUX@&h`V2a9tObMB9L^AW znfA{c@}Tbn8+Q5k6pBYZhY+(_%ONc{W*E>Rg8`_t*xpB!4_$?o(SghaUqHLo5*Lk6 zdPs0S@H<+1`FwA<@6fL8cywrjCU?D(29n(AE44fLMF&fAZz>yO753U|vJu`bRP4rd zH$CocT*lX4|Ty=*uNrM{Ui|5;c zs1IA8yfLat$f;c>56*~drbA`++lebp=J^_H9|td*$P38G;tAX~;dLb|^v7~OTWPL$ zLvKLdXj(CH18NLy(Tjina{Ar9EL*-(?8I+77mj@~A=BoGN#oKXy@381z8ySEVOs>k zPX8A9V%&D}u_7~{?ltpvhXjDNV}D*T@W&wr8*_?{J9FI!uOVc;mnAlu9B3^-xG1({v0>D|d&1Z^yZ)UDTm_Rd|`FcMJAYA|^M$;B+FD@BLxnbE1TcNh# zq7g{jPY0KkEx=F{#3v>gE873i7yAQD37Eb5;bi{P7HNvq_t5^!eg~phfUx@NX z%%)sDuf8^!&?F{OPc>uAxzNTILJ9Si%e3Cm=)GNkBht_aMgOd4c!kt{lJ$D>U|JL0 z&+?T&imf%)F#1qT7!6bJ@g_{$eX|4l^&-vrb)$X5s4uYUQBTM9U4{Gi#Mtu95dHn` z`Xq5tXCw0X&mB$Rv63krj^qJ*r*gfnOdq2S(F2$5>!Scs`-$VHiZe0Kxn0i3SCeGw z+RJoB!E$~Cs^23V_XC&mxPu9qG>@alP9}k75WTiIqd9Gwv~EXwra9fc97*>a8ETTZBz-nbQR2niEl-fPS8)O~R$5{MJGy`i0M`o3b)(?qn>fHLS*cr)QBP(-jw?72}z$`#272ten{SMNrfAY|X z;o39id>Jc2=afK4LZVrNdymKM>X4qr>Yd1YH0QY0Nk6rs?UeS;GW_X_`R;Vi8_;^; z8Mc+-)kP6PM7Wym(D(^ARdkT(4cGC4FG?;#xH4o2Wi8xgdK-Ts# z1*acAg}`V$M_s|QF+J{6u$yaU*IZiM)pY9?Ax=o8^tbs2#+JLVU}f@v)`tSuN}RgH z#1ug8Axj%iw?;{2#lY?-@yU)1xQ0w}K89o?3YEIneV4-ZwcjFpyT~NL+)yY&+;umF*%N_sx-T=eX81jMt+Oob zn2Ga=LnyjOJWiit0nU8WLL)-20T7J%d|@Xuq33JdaR4*QaF0u*Fp>t5Pwk-_J$b0n z6g^^*%L*btz0n|TaK$`3zw}_lHvuOs0B*7ryncRtIY?o*eT8@5T0~G`bUlfEJB;8u zh`60>`1=|!WSi0u^$eN5y5Nijy=4B__Ot7fk^Rp0rwd9++=*g=4DJF9(&U)2g}iT* z1X`u^-m$|0o}mJquqUmjb@glkTG_LG4>R_htJ5-~KlUy}<5MK!Su_(~}j{}VB zVgCR%d*I1IR~!#$SLvffcB@5|*+(3tZ+qTm#7fbzG8Qn|2mv1C)RO9Hb2hWN4WCdk zfJK2!YP(aa!;Zc!Wl-03O_r=L=!_H;MvfyrpYTy-DrFIRwlJ}5KvU-+h5L8u8w z;W!^K5#GCW-Wj)olaP?W=nK?AAy^9SRtOz86)BxZCnOwgvger25n;BhN4mB`E@Llt zFt1i30WRHR`A4pBLC@VoNWy^$P~4B9Z$G#l8a!`;-0ay1?d&8ywG7M$IVEWN^mF90 zwdvjQ9`DVVk@ZSl%*@Q}+}xakwzjm12{bIq>E7Mq`*^axkGQk*Y}pZgkU6xBGl9k} z!r~ng1vhs9I%>NxJy?PD^C^{5ZsOVWG`gs(r18>CWli4Y=|5FIJ8I zdHO89Y@(oJBv4T@wrnO)s5v3HY|T!Kg*X^Xyt8V1;IP@HSEfc|S&!%Hib_hBXXofh z!NU`|vZ93q16is@XLPa7C-cXN;rsUtdwYAKf)>eS?dk0{9Rnv}&<=A26niPOkvAbs zJTyL@j?UfT9rxT4BmQwQi^lXZt40*L3b#4{prerWR8J3zXURmKv9#*)ZM-CA%brm_ z4X~c?p7(%eEOmKTojGc2cA94W&cVEXdatMJz^{5uv=RM%)8!XT`-y9h;(CKSzV8As zV)`k0sjq>aOX)4YQD)b7?n1>eKD!+OWbd5k@O=ebB`oGfK-<=J*6jG?@#v-f5%UXG z2&$#?x%0=n5ux!*)(U1e-Wd2D>!_|S9ZoXw*dHKME{gar`xZW5nyjo=SXQ#Tvm-qi zlYztI!dqSa)34U7NUh3#2Tv}OS4eRhUeEV!BR~Jv?Dh~pEG(>NeqI%!geEGPD5T!v zh*ng(R+A3%TyMNgUEf5`l3TTpal4~L#{1Ca8Af}V|HPfh*PAB;wt>QX+~tRD%0%gV zb7C0F?4k_x6Q)}qGcY{a-9?8nGSlmKfyB)4t~GdKkAS*Q-!(Q%%w8YX7XxYmA$I)-q3d#)s2qdeIs z?#_en`|uhL|F*}+4;@{%kSymIq$}F zQE)OLx>xgT@P%wJD)%Mq{!#ZbbJ`7R(~NI;L3M*^)|Jzn2#C+&aw7EwI~;l_u7$t){ZPC|g?M@!$3uFjjf0Q-A(!BGYm} zLQ00uI4`AxE?x&NPZ4txY278dN=}u1d~yQfvYv%Lhpq6EaD_h4_^?EyY~?U~s?{;*fonS=|bzN=|CmMypZ8!Ag7 z`TIVQs5-FeC}eL;qdCeoORMX3iXQOT)tTa=e<~JxaK06t)=F`(N;Us0vm>yosiE$! z!gMe;tLEZa{qVK)^Xi%xkKJrwtOKJbLQyC%oP)R>Yc11PkM&`)q0->uh@|**_q_Pp z!-UC@fkLKN4xxs4c11NIx!SN-%F^<$Lg{=^u-Nv_iS2`B@U(gM7lGtcGFBYK_;P!L#d_=#9jA`qYDOHeu=k(S1@)^e9^y8-TW zx&+U`@j^BP003|}?cm_y&m?8nge{M;u5V)FlNuuPBw~NX&1C7dD@jV4hek$XqzQ#> z6m#@;Z-X{J>R3;Xe#p5)qb-;77Hd~KjpPUv73|XEW{be!;G+$C`lz6w zgR^TK3vwG8Lh4JCf3-McXtP(6bo}J;xWnCyN_el*VYOQ`VV{N?q7D{Bg>zJ58Sg;W z_**s&oiJ={yjYQ8ym`Usa)8?f~vW>d3SH`ZQsn$(6Fzs4~^h7 zzqwgf@{Bml*BO@r<+R965~;hviM^6quvB)S6g;`rVZo=b%$xqU`P`ImW+Nz`g`q)xRt^N=oOb)N-7dj-|g+})foj|S6RjYN&y=;YalM_T| za;w}THmc{mIvM67CGppZDj*J0B(y6Z>|MKO1&*k%VlcMs7)b8 zgEdq`lY61W`{;zLA-{fIIx?vyelF8##;F~iqiu#Pp2vehY74mxGk@tWPuFvN(j5YK z!4hu$a`0K96(L9BvA_sVRgjR!C4tRKt;W|EuSkxnh!g@c@B>us9tCLV=)rP}kCg)S z^d8)<=i;D93<`;w^_I92F~)JA`BR+!4pJAR*+Pp`t*-iYS{8-p);3y)c%Z#ao#io8 zmZPI1(9#5_rxT2fj8HN#yvwF1Cu0MFK+x3wii;C2lFQH<$5l&q;b3VwGo#hKWRE2T z3gJrB?XY$<#um9!8ZS3@mb0f33+QzJxwDdlTTuF~87ac9Vy!dZfmpL=5^f?%HqSnA2ckw`eS?@7z*16?2#E3a-mGcSKLZyChwb3$8w0NBs zT?V(aNne;%c&vP1Z*T4Q2_z8Ry}m&n&K5B+G1*(Mw`C&ZFDU8i_58Yr=#?!f2mKzzOTUxY0 zVrG4PT_(j2o6qyUdu=UGvnXg`e%}7}5WV9ah|kP2Tdh@YUau7pD1yUojTs1|8z|wJ z?uni2A07^`uNP&}q1Ny&x|iX28{sOGhZn8Mi)vidgKIlUyys zIT@j%sq8ifcd~hef!QvoSI}1#(sV~Y=&<{%dI)I0ytx*xRE2a-pP=dA-wMp?JV><- zFr}8u%$_M6T%06kSiG6c=+}^NU$!W#@PG7d&Qa9oerSnM8u>fcnbjJ2ENj3_^OpT* z^*C>e4Jx2wTOjmgf!=5XlkR`Iwnpv6d)1WasKkg;-vpKS3T|l8*mdw9F1lNstvE3`R1Q^(IKE|V6m9- zPIfM-V2Tolv|6m#KB?^Nv*!2ZuheZt0XR<1Ir0^X-jpzRO}aDwg|D#Er;{Pa*zkTJ z-0*ohYi5z-&WPdnerB|o`I?)XOYdO{I?rc+z5Bi6`Ix4`dO2Xaxa{}N{{BE;I7ahP zVOqZWxN6rAXFdEtfq+rE?1`1@S?-R3H^C}VA@|l zrP2gVw+RpT{nOeF$s1Qh%F=l$Jw1BI{juH6?zbgtc2+!OboAc*m{XUk3?W%e z$Fgf`K4r*BSXe!e~GTsN;>9vUCtie%s?;d?`~tpDTZC;0q8 z+_yXY5dUkcq93-)-NR$1NM4|=&5M$bE~labt9DsNO$|m>RrM=*LuUZYj#e=p11ZM~ zCKjqIK)0^A%CML!?@JL{U?l#z2Vv*Cl~G~KE>ttux6QNg)`pUGA@LD}>mj;|!4rT9 z+RNXlD%A-=@Po)AhK|R8hzJB;KECFi7#_FRtHJP?7@NzB@R?EaD8t?mL(oVkCJJk6 z;!aPkg+)Zd!@|t3Hn^tUP8_p)z~JR@?rWIwy>z>Gq@HdTksKUqBn8@tYllT0ifrnB zu2xLHAfUig9GH{W*U!7~q;kN5*42I*KE{}^P(;a$m#?zICzKwq0DgZ@>kqUzO%RU- zmLBNat%(Edw=q_{=rXem3Jb&>&AkL6ZPTts@ zR?K*Ov(ZdE)w?C1d?!mA9^EjKQc`A%=jtBe$lThnAk=KL;kHL^OBVk|)1H->l;m{D^!j)e-BDLQZ<)$!@#*gFu1upkAUBuPbRzw`fx(yXaApVnCD0k0zL#FH z!Ij93u!smlkOl&Y!9UkEN<~DgmbHp{dLu{5{g{{pW-kp{O@;udqWvkY^uNId6cs*9 zEXbLMo<}K>YS^H3*63s*OQA#sBzFQ^FERr}TT9@aet{QZS!HEVuF zL^voZ`xc6X8V3YuN&9A4qLk}RFli%o0~2=-5`J8V8*dKz=#&}x6?7y4>Qqp6GkW|u zroRt=b&XQOgTssXzR?;OIA~S%x6W6@ zINDUItEh-h?DM$QynWPiQekwuR4vO)Z`Tt9pE7Q`P^KxOr$@wMHen0`yZV%->K2^7 z`jnoox@iR-UUbquA@_7g;Eg<*1})b8>GFi@jj6-46H_SaS?e(ZWjwS?mXjRDa$()Q zIT>Oxoi@86BhKa3Re^pd_~Y|4n}m<-$l6*4lhMHMv@{W|6?%{ycDyz=q@m81(tI1h z7^6IvRMCLfgq1UZGan3!c)!{%@6@fdn9O~@?vC2C9v&+~@m+a1o2-w0O{9Q*itjkb zUw1npQ!7d~Qf)??^5&K2dq8TuzbE&ehHZQ8)@dkA-q|RmGc062zC*SWc1=n{;eG=m zT?s(tO%L@1BXZeqLxDJ!h|h!3rbI)4k!qLP?DFO?Wf!7Qtf)7=W-7C-s$u9 z$M`+wZ|TzH3|A{04r3fJqt{}cxgrLzmZsTW9gsk^7=LQy`z`bYw&J&0mnWpVQ$;Zbu7JMtx;w~Up zxAk9s0iAV`x^-O#$-|bz?1-^3CooVT1!WRe3)R>l&|0sx1cSzBz>LdarJtFZiA2PU zdj_hA?EmSw;8@r+~X`2=R+oA%>tdny5|nB$_6@79Gp zQm@sb!@k!go#s5Xlg}Bh$0|nK<+gy^a|LGG!%wrxb6U&4%#XmG$|ZZ&q_nI@{MnFi zhD_?$FS-%*73)Yx>Khl$A~m(OpiBrK3>+LhihwIFJG)Rfi1zE(piDl`?W3dcc}tV! z8WR_>KOInf%{`YWl1=J^ID*Pz}niqFK?~S<0c2Lb_SPf%_zXZ{rd*koxV==2M19WiGu{NsiT$E zU?~afzE&ac71|r$vAKS8XshXk!O0eLh#pQ-b->*>aHO_(A0kf!O8(>7zAWw`5gK)e z&D~i*N?d?`HczWBe@{z<7j+allJi{(0>YA(b)WN;Xr{Y4gnx*)#!bx-&$dPq=%7&o z_~aWn7F|0I2zv4YZ=&D7!-MIzIFWEQTP(=5{u0m20ezBIrrj7*Rz}Or!UCEy#5{2> zqh}*NiTkp>QrjeDg-?$p`w(quPx`6v3sTpcOGyTKsM&824U`=42I1Q2LuIX`17%#@w@TH&WdvxF&C2 z_^x&t-;>g-AcVF6ynV#A%Yt2`b){&y*m9N^`+Mp6mlfxbY8yj1hZfjD0{#RaJ-pkut<1Rn&2SGW};rhD3GPf7H z26AkLN=8yHWjV6KM*z<6mVJh{TQLY+_eG1YUwyNKTsnUZCr9(p<}>#U0@Bfy$&J$6 zBE4VacGS52o^A-Zym!G`<94H1%E_1vVFeB-eweCAYg*7z@90z2BO8sfbPCj{u`kr@ z#7jK6jhbAKlq&uZcJLrIqZZiO(R3lueuvERt@fXbgJFe0FT}lDTIFH#ZHii2%-Vk| zH>0Ao>f5dRRv693$_ZXvLett4?mcLZcLRpONPhCD$8Jha#!^{S)`VAZ^}S}#(ArC~ z|F}_as|cf6#Y(1G>vzNvP;@MvyrwGn!dwa+2hJ&8umICS{dmUHoxG%|mywC7cV+~} zU(l~rV$$j-JCtinyE{R;ilz>@vw3Bobr1 zCP`vZ^-C8tCIZ_?h{r^x^nRO4Omyl|v`j;|l^7ENr+08m+>NzfVu&PV-|#uAw1YY- z;IC9ujW6oald3{nR#qo}sK#lq{SKE4@$y8X2R7+o@Rpi)MoZdcZFR3QtvI@46lA|b zMy7aAB9d0NhVO=wzW!u)Ague-Uh>nVFR*tY8lTJkdY4YW6&w;~s#3qx)5ZI@WM2?G zhMk?AVBR#S{x()^G^o+(!}kt+=+x}+;sVveUF#z)*hXL+l(#t@q=a`JH|!tHY2V?J z6`XIfOTWh0su6Ad2xnpNZ4Z2Gr*Ug&q-x&q9hYw&fD#j%Ny(BGUr_vnVm|4C|9>={ zV{~0zxVFQ_wrwYkZ5xf**tYE^Y3#}1Wk9^Ci!%tgya zJxP0LHS1a!O!m*DSp6MywYA~eKxd`X35O$*hmo>SNMClMGJhQImK!)v}pv=&yXvx10yT3n!x^ z)?XO5BZ)befnuaa_i0cb=aCeXId7Wg;q$u@p#)^xj`?C0!liW0MVx}gvcGYo#7Muv zV~ZK>virT5A_xd{i%B%Ax>m>>DmFu9U1S z85>(tAUIUL#Vjd40dH^AA8BE6KSapx&rVi8Xrv#!h0&NO?XFFa({R;^x{D#R3qz9x zi$vfFVKjG1n*M$i=4^(I{t&{$H0TEHy~*PeU?p0ze-T zePLh_3Twp4lH|INQz05!8c0ZHf|gz}{&@?%OFS3PfuF2@}Y0t{3`gJbki!>g*LBX92C zW4FF<@r&IZ;@A3|#rIORa*Z!=X1{mxpkSy5&%1AodNMu7W^wJJ0rZCSU(Y*wW>1bn z^Q@p;{1QuVze_ebIjl&Fy-#nRH!n4J!! zQkV?>0;R#=x_wV<@?zJvA|^D%^V{3q7*%eq*Mn6Gy_SfkCU$XgF?6U1un8jKtk=S3j%4O`3xBx{+8Lh*R#hSAORY3}TnZ8KUN6ti221tw zGes@W*Fgi+D(M@c|`>ZETmKsrlzLF#D0zb z6geGTMb-S?)poS=@!xvebu&WGgE~8M(HXv!Ei@1F35cz>3GB#FsJpFaPW!H#b;(2c z^e@bgfOFd-K2e+$rMbBoo=6}o5{;^v&dAEj>Q859*6^@IOl)kz#9>uA3p#A* z%Gw$rJ6K4>6Uqw+y(sN6AOv2o@hQPEjhMy_&4qEl z-QKLMeRr_DvvUHm2f?xzkqaTWSsy;N(DA~$G&Yx@bWJp>^5T`_d%x*!&I$I8+oP%w z#d%_Pgogf>moS&*6IFrWaB}E05_AiKx9PbXdkcZ*imE$C;RtB4J`eMyF8;x;BoY3cR#`38bWra!|FU17>B0m*MwRIK;rOEx}{ z;Vt7g>F6>boNW1g%{?!CR~OF4rhk`1wvFPnA*t1gXN70QWY=mQHV9%Rf<^N4+S?O{ z!@LH+eZ$6){w2e}bDwr`8gB@Kx2X{gLr-(XGrc_gB4wv=l^ z4B)&pz>3spxqV;1pDovvR97R`Y07zca5b1s^-o<)ixgMomzPKV*WFsqQv!wL_2c6a zkb->fXC)^yIpb#wBu#!UG&9L1NQr%Ye$+KJnSQ)J;B)`M==(TKHDKoCq~hc2L}1r8 z#`lUnDD%fPYC0qiJxmFSfiL6}!0=Y28x)%9WO^BXl=O?>o^qYMIp@r2Qw_G~+lE-) zQkL_+$EEOFvnQ$OHqm4rfZ0X~uH}GYCbPKzPOh=V>1IoR!uAPjbtr&s5|TYlOgVWy zIIw_Rmq3%0V-OlUq5W85sv6Btl7xAq^qXO(cOjgxXm~K(T3rJ^^EvE6C{tW@t-<16WtPoXGL~}^tG*U5JwB86D6G!)!VGuN*0&^B^r@K2UDu)FKGw63_ z(SVH>&dX%G*07k}UC9^^|FavsnnnC8l3BUr@wRmprZ!};f0S)B<+HT8c&?GK%a_u^ z!i>i%Nn@=pIT;yNz`c+#QT1=$op?(Cbo3_y8Ka)MdS6-6jV@QNh^;LG5V)V;pl1Dd z3@fX=hmQgSYfaQf@5+1o#WJN%f)ezn;!!N@R4u-n|5XDK5$X3hNEmS~7Sq?)A1+tl zo>?d~yxa)o;_YbkiaS522XX~iZ;~RT+4D63jU4$Z3bXs9fe@XF#FfT==inekht>Jb z2S}i|cXuzR6*W>aG9tiDW;_UeG?^Zpr1~cC0`v~&+da7mcpS)tgn$_DTw7a9OhpyW zZoTv?C1uzoo_}~IUv!-z^jB`~bQl8ON{2fWh_J9Q-@}5chlj`g=o+kEg8Qv?j^_b* zlE&VzcAMd|JlT(9XXM>)3IvAjZr>M{m33WW_s+y2 z!r>`ct>{%O-%{IoTfX}1OzK)Yyp7DS#7xC;EG^ApU|`Jq3vtrY{vjaP*WZ9WYY!Nj zZ|X92LA+O82^3FMD*unxXaWZb`SO0qG~GnwIah!eBet{5UWH5$L5HEecodiS=TFeb+oPeWsl3OaLriSP9^EAu zkZD&McO?P2Cymt>aZBJcEyId!_D??;B2m=tN&+}GHa0T9lSoJ08t}1PP6r_1-w@5t zS7+F(wGnpvRJv<5Ms& zBzxXrZkbEEZz7eU$o<>X^Ds3v1uV=U)e;3@<4uLYB-OUG^js*&sjFjzBN0jDD-owo zq;or)btF+G%ZSU!z^=F0m;9*Fm&@e1(71I$qoSfZ`kOs*x>&K=>B;8QhHqf7?RJ5w zrK59qJ%Ds~*>UCc@#?y_zaJSD)juTOLu^SmjF}VTduwVFZl*N106+P^u#^zbEdfkbS9Vy-upOlgS;GqNjEM?^*jE-LEY z%l*8fX6G}`Qq@~zd%Ix86FDU%;K?HNMxc;E!&-e&@sp7St5;|knVJsWpUl@acYo;D zR#ea-@!umG*QL3hEh+21OX7?z?Vp`_Ph3(kFin5KQXo&g0Bk*^B^2yFCp%4>4|i}Z-qOt}o&RC1fA{3!1P+0Fpn^n^N@8Zx z&}DY!N|OoMKv?{z>MewJ8K@4gWPh0091vwLGBD0#S7r3=+%l_ml(pid$VC{xKHP&7 z5>SeZ%@!6P;XA@9*|SFNzxaZ)dpp1(xOo#5wgby+YL54YT!MmL(J(pG%teg1{W^l| zd-TeZtWRe7V`F0jn!C$@Dh${+PP_e{9QIp-c=ny*HTqw7)w~WnAOP(J1qB7z6zdxp z?n#}Ud;-rq;DstQcSlnLLqmsgPfbeHgX`;=z+BtiMa%Z>;tzv5fXQ-O6pa)k4MQXt z2b|Ydj~i0&$8)2(LTR7tKJW9bu8OIJ2!Mu_u{Y5on>)&?dJHg61$f1`JI00@p2HO!`|FJntLp)@cQWw8M6d-y{V|` z3;_|{X1hJdL=uVBagIJnRQu`8l%_=@-{N5WQ1<@hO>n=deGM1(*7@doCU@{X1;m

&g~fEKaupf^Y;fZLBy62msO5)m?L~ z!NA149mca)0vyyNMS;kc7WW2&a-gU5`*>xsUZU5i)a>Kj_89=|xAh}zprJxUM7%#+ zj*rG-a=KV!2ZDuG^SM2}{=3f|+P@QrxhUa0P^8WKeHZLy*jvxzY)-Bmhw~+q;o;%Q zoQ?pafj^glNCkG>Zui8QoS#x(fx$SOez+-4bnqEkotm1OsQgY~17ssgzj7xOtu<;* z=W!<|Cr=hT4ymYpUlENr73m^ZyC7v{t+VttN~^2)I_w3(d*?eKdv=~sNJ*ewSK#iX z;LtsJKQ6JLAzBy)(O^T-G4)NpKCx36ePQV6&JgAQkfTa>xZ_^_unIM_0IO%LY%#Pd zuTli6Dqy#Sm*l*{p;0Nk&SUujkq9^TTb*W1Ma5a}>2K{QKyt0DjA80|PfSci)%Up( z7ZK6crz1s;jEIRTtgB0y7W~2odJ00oE8+MO!ok7WJw1&|PhX8Kn1nE!&Z%Oeydtl! zZ_@Mdb#=HW$b|H@;Vs_mXfKDYlZufd?;n-wKcJcSd|kz=t*f*CI>?@UI1_)r-{*u! zh&Nsmh11bR_H(Xd!o=G-Ih@#6?Cf=?6Wkq5_~7bjB}>w3v>=QW>)%&oG#cXk+U;`J zmtbruZhgi@jSLx6HABN*DdA_TnJ~12&W*GE;^SosEKy&c<378`E#)OikizD)LtR?p z@VOZpQBzlUbCdG2BNpNe3b+s^N8LR-`U@=o2$Eo>(fIGO7I_Iqy>^HIUB;d><@L=O zi|=Y`YS;6Mf{r(P;yIp2IQ{+oYv@mDEGC+*4kUoagwNv=QC!^IlFh}%WoS~s(AoCF z^KkDIcP#7)Q4% z#!MklP0dly>poAQv;iV(!tHJSUqopn;-v5P^((6$9<6sDPlDI|aI%+c%Bi_ERd(N5 zScDnv*I#7hpR>?mkxB7uC3jDxdU>}kj9da6cBMz-4KOa-7WR?estNn>^wr5Nbw zjdYa!7eL)9*Z%xx-);0KCg0PBf#=tNA~gKDnXQ_8ewlP%GF>V*mj!f6%2dS1_h2A8 zZEnKBwNZpZx;vJ4BY*{hsH(C}PB!JLMh;@15hmtsiIs)L*gem8eHh=Fw7Y;o?i?IU z7Q_n-0s8_7ReC*Wi;&DL0lyHHQo2%CR#x^77dwC3T*NA1iHZ(z5ZD6+o8wL|C@Lx{ zK!MWF&dw63V6pnXx`2U!t#o;FVlnB9RcW+%A*!m*>(rD1)#fJ4i6w9Qd4B`@qE(YC z1n`|!Zx7Oh`^tjts9RQCyi-Rv$E#m2laqyM8EMKYbL|ABZ-HJEK2G%?cU+mlrdS;a zygu^3Q&MCAc{Gm!`D(s|$f=DYJw270!8WkAp8(j{`-71Xm6eq^0({yA6SI-jY7SMn z8`^FQ+eK}wRaB^v+M6%MDrFI|v8LNC*F9^ny5G5;p5SK{wzHwdIYCtrzCV=OSp5Nv^d=cdiwmWUj128X1Zoau zA!6h|N-mSfB2~%-OQq`+Ar;4$@Q$FW1lkasFKjk%ez5VXnVi;|`^ zG&GgkO;I~LhQ$imUXoQM&3ijLy_4x|0BVH+5E-Upi4;|@K^nA89QXl#eY0ce8o@GE zctlJ!c1tJAxBz0wm4hugR}-0*k#1y0Nx^*rAt8Yp78WnWXrC?~q((cSOM5ss^nEz) zt)ax)x~!7A1c)M3$IWoor)|6{WSbc`@9Jzwr3xuW3o6Re4P%a}5Y^j|ld+l*9-9kWRFrIP?ytlBC}<3PEH)u~u{h)| zX#zIUwh$uGnykVjhVCZcI)0K!J~7!o7#Jv^z_fVgFHNz3lYy zOXUHIt;_nz+k>?*k2_SO?TX~dW(SkWh{E7TC%7)_!@phsNO3(qg5>0?30NSsv;mnP z3W=~VfA{f%1OV9RHEVaOVbp-1sH>|3DzB}_>0}v-;BN?4O@>DD@>2|l56$Cj|4y)J zaT{5`weR#|U|-F%p_PU_i)w0X-Sr+o)ipT_QKRpj9z_CfJ;mX7AYb+W?KYL%B$J3a zmYb7qw*LxUbtW7dFMklX0)*MS<7p?-8|aU3A;zYg0&t}{5rNPmym@AiHYn5!?Xr1@ zx?u@Zzh~WEx_<#eltzudw7NP4z^#ldEnVhtPheqT8CkD4@DN`@Qhq~T5>-(0_4SR6 zjLZk{jrVt-1uNI+B6t9V^1N(2{tqxyO=toBE*6#qKv>IfSFUiRTCieIPENkRSWg9f z0pp+d#~6)rl-91k;>yaiReB6Wxx5v)%gf6Jv!xGzulPg8C2WsX70Cq!!)N*i>?joT zF3{njq4!P?WfvA(e6UtU?8br{us!Ru=S<`m_x&zfjT8fCAg|^f(FrD_w{PsCH4u(X2&d&k;#*IE(9=?&MU_la1qccA0W- zrsU2?LN&@LqKS0;E7-U|{K@`!>b=yujP{uk>^U#~XGVn{yOs2%r~r zew?J4oSI@`VYgoN5y?lUGXJ5k`HH(74L3I0Dr9QPoq_o-zPir4dw~|9wie2O(++~$ zsPgT?G!aS4)x)6WVDzII=66Cu$nfwtU|VJ0oAU$#u%eT51Gaz<1UOXl2I~G%`Hn)V z8ZAJDu5@~$8TdXNUKL^hJ?GezZ8q}Shms-G5h!nU%G@s?^NFg8Fzmmr77D z7l22AQ;s$n*|z^m6wEq&zI!GnHL0}s_5DX=0WT>To+&H?9|dFM{wl<-;$q6E=;(*1 zPkyngKuQ7%B|ETErqFk9AA-m_7EDa1?B+{Jc)X#fZQtP>J)MXs!tlsQ8p-`fXUZ4! zB+6+kRf2hW8=H%|S*$>Jt@BBj_0X0)NW9w?gpr}4Ip}>gth2rW4wtyr{CW zbU7Y>9>U5wdGq1n;q7!>055fdD%WT7a+CL~P!nrP4@MF&n{Z$etOT!K5<4=($WTsAse&4ZY;y z>+*fa5j5h7jCAyQy;;wjHMFvlP`l>}3J%{&LPg@g5G4o6MSQNJu7YwV`*k|n-%Mdj z+ajO${th-e**W-dHgs@&yt}uz!~oGxOFO>W+2Q_Q zZ3i?#AZdJ_uoJ00f9BIV$RHQ5jlg`kiW|sIuz|j7APiw2m=Ve^EDU)#TfPGvD7T5F zFTj5KkB70cq7}*i6%hfQm7R?O8w!uj{OgB-Z!7|y-|oQ}nr@dD28g@6`{&iymRWB= zv->!Z#;jEvCBgQ)u0EQY3jtNe!D7==qeY06#mr*?+vme8A=?+HNjo+;7>zKedvlNT z)NgPQB**%NWR4(m0?giev(4~8BS@C5qyhMEzab$~0j;;XdNQC<42_LJy1U1FqnLyP z10Vg>dh2ao53E4>g@nX}k8k&-prbje)q1TAEX~yAW(QD;P;hfgkD$04z=!Ot1;>5F z#>8N={)8@3%%ucc@{kaS%Bre#c56%xjrMI{Te_yECVW1RUkp0UAUj7#5@KS546ZKV z;NbsFNo;m-0-Dp{;Q8+J&AKerzmslV>cB1j)=YhJ)Stz*F8{{@?4Iw1HOO*Skj%_8 zj^pNOn`2=z`uvWF*v03^U8a#t-b6`eXXJG-hs&J7Ci|IIeR(40XnW(2HHwz03Kz)ISiUbxFQ2?}L-^j?ZBIFGvWii0- zM9s|f4UQ0z@$z;+@+iye%Ep_#P2+e7eh)SJM8wW3t|w03Jc}od_W}+)u5NU9*IP|f zoa8jPvn~VS3%}`)G15%K7Se+mL9(Hwq#?5D*L_q#{ z#4b}Pa>!UB23y+52(+xsTuPoFxEl#m8H?Dg7Bn|D_}T~EEVZ<7>gwLOd|%05UfQUW zW#AEFBjN}}fms4=^a7x}0A4_!psH8JscQ=p_9tgI_(|tK8v_xM5{O=a96GbIA_YLE zz`6taJBpxS4?>7NqKlO0MT^aN@jC#VK{5Z|Md6c>|F(b;#$eFa$ny`=w^vxuYtn+E-woSiKWwWRtISwg2!Wz zK%r!{l6B^4-Bs<5rUSPJh0Bo5i~0GhE0_JOf(8$J@av(*a6O)=o^~wfSONwG+O+lY zR@8#2!1cD!wxEWe+I8XIKkKrpz`24F;AE|V-SbB(CEttQy66+o?MEzOV&t8F|0d5@ z@9OI7Nd-QkfmH`2$h531`Y`xNVp=P7N=hDPw{`^Dar(m>u(}^ z#n?1fKr(0i`(aJYl?lRk{b)D!HA`aQ+wLelSbUc}xw7 z$F4{*f`iDSOI871GZ~?Mq=6t!iNGgS?#SD1?>X$=GKI1g+9;W#fIr>i#yyvYPeYtb zGSxIw|Dr&QjpQ-gAa`bE9b?7Z(kYbaELt>syU z)~d_aLd-}pK(%LeKEekBf2*~e=c(4CBc?lpU9Q1${(R>ymiIS?q8b-1x$4rx_-UwLW}wknwFfm*7#1LiGH@G{tOG31W{KXJCFECc9)3#!$pyxBFl+iBm-*~Q_k@B=Yw~A{cSJ% zxZR{@*)w10xZ?Ahru}60@a-!d6tuT16rO))M8dY_ksNkZtXRyS@{s%e9%E!K6iC== zb8_eRXHeE^>N5Ij3Cmw1DbmRMf@$L7ThbcckElIaG{p)ga&`}%l?vn}2x;=NhqjEV zffAcO%2tww`2~zce*yP7DfaA#s*)Q(8Hk5r03PHw{c(a03;7?8O4fPW;X^*{!wU<)M@OZA-1u@cLIlVS@IWkCSXxq1QWEm< z;R9;6Kp}_a<`bay`7J{M$V~5gBbMKBflmMn{Es@_ydhiyc!otQz;=)+x=j&}5h5n0 zrjB^IJKn##+P0eRL@Xc<7Ht#fiuF%bzmi`ftc#3c26s~NHxliPL7DzY#WI61R`!W8 z-8UdU8rh5?WF>mqmM0$^rUO~-7cu6zIK0jrXmhy#k>%T8R>3RZ=Xz1PxA5@ZS&_A+ znsPf1$5vOfHCTUhh1TQCX(^ghkh%a79-d=9>#n1fPD8=!1OEYDd8rbuCcLn1thM$A zThsMP36(TI_Dv1TB$A(o@}RE9^KgvhyRFvXAQz~p_RSJGR#3hYCybg%^%C@AJ(PW^ z0nkJ!{8c|awHg~AU!YRb!)62bU~2!kJMVzMOwm@>)+4s>J1!9yCjfd~6B83c9|y$1 zz%I=OGmMuPPr%};wO-}`Y^$kGPdFeCU-6g4mRD9Dyj}I|4n~G-^9wi~Z~^bRtYReN z_u_#CbU$D;pr-2)7m%=kq3B3p1Q-bEu;OR+Bd@QR9$*`JWo2dR&b(Y)%s}$|Q&7M= z2v)4e6~@W}{n_Nrod9z+Z-4HipZ)fo#Ft{h)I3o=Lt?3eZRP%QiVAy2$1^$6$>A&5 zIHXr3PA8W8+qes({upV?3bdy`6g=aXy_}vKr52+weHsspASdsRB_@AEWa?kt{6?@} zU@tfLQxL5Z8VWoVMQF7>`!qI0q~R~S=fp-a(QqZQi=~0D=X5_i+HOX#h2CvUuMZkx zyuU29P57%+f@LJFI|fz_P1YLFsADCjrskEJw67B8lTo1mtNAHPU?*K$J6=^`Coi0z zSgOjSFkr>+FL;6KjH%gvTg>Mj1sH!Eo1WeuK-#AG?gFh^JlQ)uT)bHv1tf#hQ(6EJ zr~~c_p!5d;9UeyIlI(05XjOA!*VEKjE& zpaaI;#>=~#gIStml?I6~_x;lzk%JK--=qqieiny8LxOx;Ssi8Wyg16j%U5Q6$L|S_ zjW!UfY&6mM5>jEhdN<%BWg#sJ3VZ$bo0i>Jo}p<`9vZ?pL?o{+N}i}kGHeuIR+c3_ zawI|)Cw@AD_d)|==fELznE{ndh1SwS&1y={x}wzj(A%82{i#3Y2@Nc8M}u9rrG;e9 z;?ta99vn$&0hikTpWma%nF=pgV0)w-rOD^Jct{Am+1n$7Zg-Zjt1HIMjbmYP4&XwR z-Cqep>nVDjKHpTkGm3^gf~j_HkZuz z?(FQmzO@?8=@>B%2ra@xtkybTYs^Nz{5?} zxwL(I;!abT6YjV50o|xLVs@W28TmN4fs7oHu}fF@xoo#B@ssCcG6`ako6T?*4-nytvl{uRLGR6~EbK4ffs^ zbzrVOz`CoRSveozdyijn5zjZ`GmTKj_ba&K0+vSO5Nnq|fBgeXPq4EV=K*J9qXEzo zA%<10AEi}Qmah9sKZL3@tO*2Zn~eWbQf^+cf0wjH>m-B%MC@%ojp4QJoNBR-*ZPll zFwavO^SwP|Zb#BUpiTiuiIVSNiV&*c* zt3KUy52xE{C-%kcF9$(7e%iDp=5>iRsLB0AV(h|oyj-)(1otCPI3T}{EMJv!ZhM|E zxWCKi^^%I*K>w14M##!%MI~B+yWgCoo-kU9)bNJ?Fg}1p}@Z1tgv+V*4bHpzkmge$9@^-Y)!lnY%!~T#F7k)GS zXg%%_oyP_VS2h_!*#QO2d^!ubVsIde)0fsHt~HS^;A{khyWA~8 ztofV$97jhg!*SVCLPAV{{R;Hy@{W!qK$%TLJxayJ74z#?AK*Q0HVKqBHik5qCjquH zGEjPh8=9FZ0B-?&0vs;p=i#UVI_H#WLW9B2?aKikE@tCXRX$F4L~z~9rt{#S$q?6f zlU2*&>aR4aflglPK-%Mj>D=oh5^s$QFL0{-2-}nE!92utT}4h|PdlB@OWyTC5p9!8LKG$SNFDFYBX^Wx7#0vlb#L=|0GvK@XG z)FerSofgm5(cTOawj-5`{ZJUf7_V2OYJ!>pep-C0JIb)w%+SU)BpO4T!|}->qWynu z*=VcFmV-I@Nrc7=Lfk`TmRu6C=7Fowl#n2;ELN`t3mN#7HFdB>30NkWQ{r4JaH?|G zaT#Nz)>w^)$|^qLvF6pO{3=58n$a_B;4S(&5MnW=M+*{L?=*2sJ{huSLPc{?Gh5bP zO`D1>>xS0H0W}6puyA)}RaHjoORm`jDjf+r+9;^FTdPJcS_i}J>}Y`O_G9#oQlkCmb!1^SFa!l8^oXOsrp z(B8N+RK$W&e}baK;9>@GOjqnJ`>u0mXhelZBP_zKYx%Bat9NYmYMU$+bY9>8Yr}i# zss4FGcJ=F+Cx~VD6FM5P-pgJ9sk1fX3G<ll)Rp)nK@Wz5l0da}J$YgTGCWf$#?_IzPQ7*BOlhdpmoqoG_G&K(H4W+=UlkMbXU8W0K))*?IPSrV zuYKLH>rDulMPhS~F^Ep23fR-Gii>4j@rRTESeBmgbwJq|Ua)ljw=MX4ZSC|gV#!p5 z7-?nIg@dExmE#y%fHG9NMeaD1Al@d0Qe+$Qk<$~)(@)XKrKPyhQ43u?{)lMa(VP7n z0N-?cy_c@9uO>|zLxzT+Mu(M=i2^DzVQO?#vuzWflSu;6h)e6K^z*X? zbp4%`HTiP|mFt&t*3d#a{|1vQpZ_reAzvpBb?JjSPS3;ILhqKAgicb zXXTe~^CS8)V0oyVs9c|>IdyW!^Y(s6gZV#YKGTFSh<4U%+!L%(|K5E)!EQ<5*t{>h z5Bl5?P!)WJmMB8p8QdH42r0u8ece$XC$W{{>W)<-cuSAL{@m?|{&yiBK63CHe$J_iwnKdRZ#9U2M+CZs>@I*W=h06*mC zOrDaBtt|>H@3wgj84E|Z@}aU%dQw~RcX5F!10rwaw#cxpV)m@|cStIxMK*pb(%{K> zrC%vxrhbr!qG5eyxGV<$HLv!W&4U&qUkq- z>M}5yZ(MLhF}-4%;U7)p#yvbjGM7*x_pSvk^&x}cH0J^Yk|{F5qQPLMyXuLuXGROb z^yw^^&P_~Bsfn$>O0zYrAd(*1U4hgZ@-2V6}9fxWMUYWGT% zmp5@K;m6{8nzweMtLhjZ`A=qHVc(S=!LR~JBxTLgCT1Xm@Nkdx1nval`KMD)}wisK> zd6(0v2WWFv(W_4{^YE7!5fv3^Z2BLF7q_&T>hbb}+v;zvz_+SD-dI2On+|?(g8j;E zPCTg=^}Jcconzu(l`kaD+?!()^+ONoPM5**~#}lG}@8(^K4$N`r3oH490$ z(bpdRG4pSBUSHR=9b-R>VBsB`fCF-iNe#K@rY6Mxz2eay0lpRndvG+pz_F>`__ETE z{D|P7i91*;4jG(`<`!o9)*%lbesCpR4A}--c5Vl`e<~&+&KLP+DfZdJbN}!Z`k;vq zy+xcJIN9@Sr85vGLGX`6OkEoWVLYQrN(;AOn}^fKjJsHvZ+zq=Ze$s9kxZb>R`?A{ z&LD=3ndJ&o1OhoWtQxev3$i+0W4D-z%Nbc!N%S9$?@H?W3b7)Q-$V;jSSRHuB9xSM zwB@w2k@IdyzSgZ- z1gta}A?wYC8|hGb<*Bc68){Z_Xe_&i{OgSUSxvc6Z+4

ua%QMkAoik}?YRUG~anQEc@Q+$l(3S5<8PA{9P@v>8gibvm2u%FI?-r zWd!CWTQ?xF^B~sezE5T@3bL?^H(5D++ju)3N#!Yg<+!{3wuB~q02ZpTj# z>6#Ze{{m&@U!AebDfATThtWMPWMNVGiAi+W&-XnqJ7ogr2CT6~>r5t>7$Tb6pK@s* z%xEH;ED%bIiJcd|Deu1EVY5iAzWv0ntDZZ{RLF4B92JD#VNZ|TH|lX5fncK>^;x0z zl_dAuZS;N&(Sgw_QE>=Txidm(ijBSKq=mB{3;XHMj(uNfuI(8qhrn>?mNnTOPFBH&S`Vm@L7k<6)N5(XPOf( zt>me}EtyKW`_N6VOGR{AuB^pX<;I+X)v5SdDkU%NSTXp|r$D`s9J3ho8I>UcX3Ye<|fyc)QI&~jB zrUEXLSlPQRZzf9)37Y4tea-dH-*m}ToVDDav2SyZZhDSFA?$zA?iJ@_amixZ`h}m|_X(_g(?!DC7!)^XE@J z*0_(nsz6$1!=2v_KJ>(AjEY#$iwns=Yri@e3bR+D8`R`yAz94ggt>TyaX?C*uS}vf z!Ku$=xl`Q;PihZioI{G8Kty9=;8CLcCe6VV<@5$RiJoft-LM532I@NJ)7J@%6=_*0 z>iwbZRQ?5s-!VAR3KuL*!LC(DSIs^AJAsZta4`>usLcb|%zNJdR#ydK*_xLZmOndA zD8#mg<^3a>6NZhJio}iREv^jbW0LCKSs%OJY|ruR>q%)fRiEI-2wqIgiAzZXTTV^}iM(y)h%uNk8i-jzE6E(b@B!+SLhO zZ%B4>nGhclbT-O$f3sZ9Bw9ux*9#sun`VCaH;X`N{QUOoaoYL2x}U|y&#@_qMZpBcEwVBT@(asqO3Ir7DhOa?Z1fC= zX%We4slQd!YlM8AE2Sb5(&1&M-(AyZbBc$*G*Jb{>o@%e$>dDVC+OfGl~-Ine78yr?-YzdWf}t2lxLE z+h-;a%o6RMSh635F}Ghd8+v(Ri-^Ktx*@)|@d<~NC4(2c)8$~|8qzoQX}*pKBAeLR z0xI?LP}tE*`@3I=5_#WvU+B8tT(4$W_RUpl@)NqGK$gMXiA`YYT| zmnQd;9>AitCFqCh#^)@}Y=e582#KNfN4y8rtb)I^0 z>G&K%1y`~KS{LUkP(D1d>86pbT>D2uX`7wznGTM zDz3uH2;XTliAp5n8JUqqN>%UnGstRS$6Z$RczW!CO$YNEIDLXAu;ToEnyPX=$47EK zrCG@!ef*Va897&93=MM7?*_)e7f;gs_zf5>E+spO|M;=%#m4i;>;4~XB=!+xlYh>$ z?&wT@jIgpi@C(`hCq`AWyLMjDOdlMpuXClR!Ucp{+y5C>G>>S(Pf!xpKlNLxSyTM9 z&oQQVHym8h3VC|$g%+y%^Xx-!&3Q#y0(R~t2@IB^dv&WSV0L`o(LfeVLHp(j5-d6z z7$m=bUHtdJd-czeo2TD$UXUWFQV}Ge`6_@RQ-Upk<$N#iy zrl_m^gcJisv7AH8xK9aD0!P&jKvp3hTi7O|vO(6O^>-o)=G;_9I8J$GgHR2Yw>>k%`ady_n^|xk(j&weT3}36kmo;4H?BSX(aIk9z zq;{Tt&=!qLO73vU{?>jpMJAWB5=el%1MHc~)Ai)r{y(C#z4yrP8j~+6@Z(I8tZxvt zH$o^(U)yXhmYR6Lqq79??l29^35k0a0{lIcU?RTxt^M(53G_*bM@{Rh`#iB zNOM6z(7mO_gw@PpR-aLm9jJmO@5$=im?=wQlRCPOqH9#Sd=Mk#j7QlDT01y|SiCtu zMjP!@ohb#gmY($Vqq2dsW1X5`xGmONNRCyGZx=Vvz7wegH(FYcJ;LORKjp)BbRZaY zgOsFwP9sTiTNS=twHsDy;`Q1~;|jCT13 zJkIm`9qY$ZYBcrFiB8K&%}(5&<{H+EX+zeeket7ku6}wHXCVc2_)!HP^f!n(J&5LRhqJl4z~+huxIW&OwL$!ezOhVXKr>k zH6cUpzkV;$RHVHRUO`L0u^~y>^ZDSG@FIZyH73`R?6?%^9Qv(UgyF=Ja}Yc zSgUwG+-RT^@Jl?V-)Sn^8ZRXVKM(W$4HqnAwH3pQ^1cWYjl$HUiOUWfCHzz`hyw?U6PGPQ!N_!O^FFf?eeKEB{ndl-)_vf5xM)9^G7!nWq$;eBB8%K}-4>!_ zi)5GM8Xb5Xx%67&&dnR9geC)W;4N_F?}UR}OO|5sB2ygKoWTu!Ck+#O&i61)SiRhW zR@~`o)zUjBLuSSedo7p)=Es=gmAM>w^a5b^{SCAz>AOE6zCog(ql*RPCL=_buv)({ zGPd{f=Fu?uyp5J|?0)2PwG7KQ8Rs@eBBs@0?R+JBr*XBeEvx79hxAE_(N`N@bDOocE5zS% zb&owt6pKJ4Z@n)!%**g>?&Y5BC?~a@Lng@IGm{qA=OJdC*)tA%)*-w84xDuHxkFZv z4oZo+(rY<<`1)0r`0izQ7koGW9lm6qn3MB2(bD9^@bAwjMJgJaz&*BbZo8@xv@iYs zjL@&$wj7jRJ`UT(i&^!ozM#}YdMH+`dZQ^!6)OMnC*2nmM+&s<7o@U|=319Wk>-o$ zIdYP3cZcCpRtx#k7fL=Faz=YG$v@AfOlK?Yxg6>Vr}llhnosDOwckte2xUGW5&mCW zR~`@5`~77nyDXEf`6Lpeh{BL9m1Sfn*-7>#gKWuIQquCl*talt*|LqDl!TaSY-8*V z!`NdC-#hgA{q>u_X5ROC-uK)y=e*Bz=XIYs?-w)8)fNf5L;%8IGC!>;|_<`Nkm5s#^3VEEupK`NppVJXQ{;fULe^52)&N47d!U+;6%7F z-#Kg)Y5R-RZkQ8Cv6_BzTSf{JWqfPKsh-3zIDcObX) zuxi7q)CL6Vi|5=w(VO`Qo;KJ>T*c){#HDqpBJ_M5Fv}C9-LTq*2vN=Z4AWN|pDMwl ze;@EmwtKuPyc0{?N^%3q9N0_Ooz>>-NpequJ5C%jLBqmXN^NDgmDwMfcrgFq z8Q1SXEW{}*t!L{4`xTJfA$Z$W3HZxJu4n5K1ic3%ZF?REa$ld{&QX7@mZ8+!n0GTx zJg~rc+%r_|Hxnky=CleG?#A=~p_5xUPcU+u$jJYqWs%>9JYbmb91}kfTo%i)rk&VZZ`2W&&p5b^zHNH5Sd2XoUKQff z#+Ncs5l=t5nbLt`g2o}YLX8I8qc(dpIa>;SBIuMAqF;*DX7X=v7Nr&{0rX|ZJKU)`zw z5xvTBh`Gy;JvpIBI-ju{f!Cu8N@2bwjEtqw*h88-t@q?~@rAEhvs*iB)_94#WaHlz zfr5d|;WL8|)|@}&gEIHG&W)vh%b;_Y-91n&727$tq@=j!^Tg{4E_*XKS>%tY-4zpX ziB&F7Z&}kJH9n8^KnmZ_Eu6`o&E5bLxj}tFUh-nJHVKEReXYj;B(-8K7euW>Twx^xMvT+7=I;Db^X zF`IIlj;zspd`)=LVHzQ#5oGn0PD;4ULimm!R_E1CrU8R9ogo6#C}!&giSW{uIbn?E zn7VR8^T+pAWe5H^`H=dx^Q0NNcD^Ix^|x>H8aa&zrG6Fl-V7)Uq)}M7)eDMaGMcS` z4~-|BrKy`eqeZVVq5pV66JQ3HJWW0OVPHwHG?0z;pzF~DIHT9J>UXD6RgkN-(HG98 zcR;OcxPeGebfF6F1^LW0QeRrmAd%iaVu8CH+PGPTfx8F9+f2gnd^2Va0YR{~aOs{f z5uo3)ONXi~5r#5mycZ_V6icr-T+9GgDHvkT4Y6wwv|d=F>EGH zzIpOT8BpQjgDg63#lt4@J5OSm<<8h4(HVm_@~()?epNiK--Q#I^tRr~_=kTzYOy8} z!G43UB9j~`CM$n0Q%qQu{aZ3>FVU+qJa#eOew6FFZONGUV{>itR1MKB{*oV6;ZAHE zl?9AZ%-fj6)NW3LrsI$snNDY^Qpzoq2^|v92pi3~(_h_=GkRf}*r?08A8y~MwEd%r z?Ps|yHEpoZ*{FD_Skb7yXqQD@yB;uq1~QNfToQZQtO^5BWV;)8KVdkhqEqh4QVS+F z-OfZnF;#_Id*bfpu`g;8OyG&}Od8TRe5OVmUh<=bP{-Dpz#i`7M)szwKEh6X$4A_^ z?fbT-V_(O=jC@Od*VW~UI@wLlxSj);G!t5^M8Zjoz?#;84yi*@eFFPtQFUYR21_kk zoJpBu{C8+2k5xhB zCk58YQT!9)%4q`j(oS$!|Cix3vVr-y9i(frE%oN^u~Ci@mjyqMP)4Wpyibl^|5vf8 z;n;?++IzI_8jQQJMPTR5aZpJs)Bb*_5*U%ogQzuTzL=YOYA~FOBd^K-I$CN#Bks!$ zE2>65NMHmHsPG@}D2R?(k1~B%e$Ii`^gu51a`VoZjb7lTZUN|l?noxjpBY)y z+8%Q^^5H`1l4F2IyAR^n{{#oG9w7Or{=WjvK(7E4Xu!wT{0LUJ!&+0F>8^ni2$&_ylO})~Hx3~NG`T_JpAHkyIh(&9 zRV_Vr^R}Rw`c2R0AuGqqD`G+@|HodBp-(y{@gveWh6*w>UF>w-l9pSc9k)5{Y*LA} z*B?FlRA$6IOK)kXb^4k`V8rE@HpTV#y7;oZ3a%Dp$9#||L3!mg4favh5%Hhyo02@i zvK5m2tg;azU-gchcYx(`y)J~^s)mHHzwO9BdOKvNTzr}kkkmibAl5Ru9r3WuH8xS9 z3;M|!8^(m}B8lYF5 zO}vo?DQ<1@NzZ!Tw`N?ftH1EHBWH?Z;MTT_X{cp$B1kXBw%#^tash5cl9B6_5gD1s z=o0Y3M5mpj?G|%U(mbd#IkS4jJVdXjeq+T(ENO(!x+-~1WXfebsM2G1+Xzz`TGr}} z5i&6Cvv%8+>G$$Sk{7=0loE8!T@(e-{{A)2AC*Qo@x}k(vp`kE4)^1DGBad$53l%j zP8kdbvL;xL&V~?5PzA!-)yydx9lLD+;2I{Z|0#zCJtvTv!sZ6tc^rdsB69L21;?O<8qk}0Al~zJa{k{zS`5v{sa=+HW?>Bt zL@(gsYlmY~C)I;3rEK{EoMYw(4B)erSs-ECNR7~8`UuJ7n>TA zHaIp59)6s+nLZU3f6Zj9I?<7`sx!O2(+yFj#)BPw*g_Qq||p$|JVhm_S5?g<+D)71ZV~0&kAbZ^jYKPEK}kcxyC>=|ZvP5p5%ai+{Z>TIvDZhWFlBq2 zbN`6h>+EjE)f=J}Bp*Nu|HuZ_oE~L*rf6(8kK0%zH!5%}FbqmZ3Kf}jAPqKw3t zB3_MN5L&h5+QBNUPbRWWDypm{-Wdv^ey5OplvV5GeU`2AF+A=j0i$#IDe58MqkPq& z+MUXp#ASMdqn$QyXbb2Yk7ku(%M{4!R>uTHHu#0o##9b{%P$+UG@?ZtBKFkm9?@DE zDe(S~&AN%UHAGNl+>GGpXx6!3{qvxzka(NRy0&9kqKf?9E#fLDMj<+XPu5G?DVdzr z3xlt2+xPnhMda_3LdsXX1B@%vQy&p~cD^{TrBg+C6hKIa*4t?E5-!WjA=7E zw;qM}*>gOb%A%HSPEdwMCW4aL7SmWRr}PI9E>mKgW5&;q8A^GhaURG7O0LKQi`hoF74W*FZJ<@5VfXBR7!3A@ literal 0 HcmV?d00001 diff --git a/docs/modelarts/umn/en-us_image_0000001340184197.png b/docs/modelarts/umn/en-us_image_0000001340184197.png new file mode 100644 index 0000000000000000000000000000000000000000..49333286f2e0992779e7e269326160ba26dec2e7 GIT binary patch literal 144308 zcmdSAbyU=A`!`A`Arb=8f`|fwgpvY+A`()f(%lWx9ZD&!fOMCV($XN(QjUOhGjt7| z?=^cr&pGFx-#Y7E?;mHKz1DWYF!QbZzOGMQ^WlYp6yX(`D;O9Ugfh|+FEKE%y)ZB^ zujApsE3ZpKb1^V(V#r85dF`6CHsRuVU3v0wyYClAwfXI=NG!%nkvihy_?T9&3G|E) zkfW`otHG-p6b$MUUPC`J*J{{`F8@)Et|%bazJ~cEFFlCn-D7E1#+dR%4+n?MBTp7-26pW+IVPt>@5l zu|TCGG9OpLm51ex!~Hz*EeR#<-HO>yMkfZ8Hu#5dFYgx-7`6F@4vhYS*uQ_XaLN>s z8Cxe`hC7^{%`mr4#%BCU{pqvlf1ZX}`|TyNf1k#CXSsL(_sjFi@Uvi_6+dnMP_h1n zVm!ueVmYJn1=R|>qJN&L&OeV1Ig+<$`S&|tVQ-_gGi?-$)>vRQseJs=k1~v3-itn~ zr`kR`uNtpgBOT50Y58qurT3NWQHq!(h`R|J+Y?leZpDN%rduaHg>wnOmCTAD8Yx`@ z!|y6qzEqIV9BE*=;Y)pRM&mm?iwX8gS0<=R(}WO^WlWb$V}q(saZ#%c=z>r{_d96X*ukN_!wi8 zmZm>cP%n<0zjG>r~KEMH2isYz&TIHIX{-<{9t6eu_0 z9}9@*5?ST*71p*-xbQ$JW1Meo89H8F|GQ>vITl&@Vd~MDgx|7iC{4?S=-IuuZ{POx zuDVPWXN{`&?2%;;uB=|~?&uKp@?lbYCvfo+nc(YiHvfG5oD{nEf+f zZEN4mn{VzD#Ga?Eje>8uqqe@iu}Sg7uA0~(c=ztzm%O~$WS?Hsovc;q_nSDURlo2V ze;)l_d6N27mFrElhVV$434da|kdsB3CVt{v(_8GdiP~679oM&MX*b!}+4pX~!ei{^ zlSdrdt(x3ox;5f*b1u|RlsKyMXR*!Ev;3i`*mqvI#D9)FcDFv7M)xj7zpgAC@t?3k z9yRtZwyowwCjG6$&D8t}Q*TM?ObcUAZIxci@92`9c0pnufS(J`Hy>V1om@$JcGW2&z&#`EXT)xN)BlnSS7HsdvT5KndA;L@qZph*BU|@_4#vPv))ZOrlxIuOoinze@E_snH8^CYin!W!K4Q{ z1qCS;Rfo3A#4}IN(|9^gySGYD&+xq+d9!_V+&32$)BS!&3>0kZIKW~N>r7AE^?{F>ei1RM-#;i=}+HqmtBYk=}w(s;BQZ;?X_4q zuTS(GvbXOv|G4~p!iK`DKAOc;q|LYCtk2?%GM_L}0F&E7?Rh3lE$k;+u+UFr|bO#T}W+ zTNd@0%$-Gu&36~ak_=Z9^gY@W`zg`>0(M$m#H8tJ?|PiPiHXP|Eh#P$Gy4f^D#y9N z*~#ugi;veYRNv_b(R%mP1$t6(Ewo5+Kaxw>xyCjs#X2TP*tmW z02BKgr)Hpv(m+A^>DgKJX2VGwk0}vX-~! z3JwhkE313x!*Vqo-YO|6`PtUSujW17^mcw>q2$esINOT#T<~>vR(AHei}TG3*M%6> z?|VaV-tbzc1{1{6tSHVW1mS%*rP60O z{jqu)iEOLttr5EXR+Fi!ulsMqsMNIfQ#(~5!E1ZS1mBI=4^$q{^iJHx0uU5$M7F4R ziroq1aIiL|+y1h@BU04F8}&3xW7BE_ zNl*V${gTmUjjs-!Yc69wU1*PWY!W~PnC>ajY0kM|)b4Mj+ZIpmt`7zgH8<0B_F4?t z>vCNLUa`E)(1^lhXyDT&K3vg+{oKXgr7pPnRMh9fP+>8}=;(kwp(0#1pvQ!X9mz@& zayg_`T%}jZs&G7S&Qa)S386^EV^2;n>76RB5DEU-s1vFuh#8Hx_pm zPrchvKzVKEnp2MEJ&I3cLEkK==xyTdndXXVx@G&$17cIoLg?wvzu7&SE2>Ml@T_@% zpGj(aJujhjuP4|RpXkbBX4T}}NbY%1U#zE$x%piT>Ui8Y)mrzG4$?1qiod6LxSFO& zNRWvbuipLD$Fj2!YiDFK+wto*{IndOs+e2tpyXro)iOAP_3-FuFe^WAcVi-Pd|XRQ z%!M*6JUj=QBF66e7-dk)6CEA;<>h59=MiJT)7i~U#lDnBBlR9GQLBl9cDFu${CIeB zLPkd?XJf-jC}m;*=6{zXEb;A$Mxz(MmC+X$2zn5 zae+UF@J|!2;AwWns45&KCce4B)f%tvi&DeCZkwAPqpIX}lc6=uV2Lz%)%9+87@vSp zcsU((%lmCEKOdwE1b)$DyQrrf-hQdoE<2#bryb8h=AIdNSx+pmgdP&uTVOEtjq!FooPm^k%q!fm()TP_GJ$jo{ixA z`vSkQTTr#^Qrh7IWmO2HwrxHYXhW6nM`8+pYS?A^PA8$VkZU|w#c=_^2gV)L52CrICpZ>Hb zrj>*$?3}UrBiAilT?K(Q=6?C|K0@^PeV^z_=xT8zeeJ~L%`l$M+(C{~F#ZrdynH9ZD;w9p@{Y&eGBQ*XfgRvG7KInj zNA%Q6e&ppZoIGMm{U!2CdGBh=^$7ygUy(e=B}gwTzO&nEiP2BTsTCB$8efq_W#F7)3^mZUUxXP>)HAL5W_~*h9CKMvl~U#&iyvW}Xu@qV7iB&i2#`+0>yb z7Y1gM&XVywS+9bN zmScY97ZOzL$=_zhPewt})sF^U)qdJe z9gpY=5bis2)_QM{kAiz<;dhohX%Y&H$-jNZqIVzF#_j!- zlg!OpK}=p8K7BxZ_Q$M0H!Y3zd6ZMq6F$8M@hjyk>g_{XCzm3E*ZXu^N?ct0eQ2oB)^zjQ-x9x9xfV8YNeZUP zj2(q4mSrdsYU&RoRwkR1^=1i@3JOH(>XTDZckkX!PfL4z!7E{6!iA3dyTqwH=4C|o=q2j;&%KRzbL?r$;YgA zQMD>@Hv#pVeq;!-gGyEIrha5~J2NnwgogCFnKY|myH@9tKGRWO-U5bv__>;fMpu!S zhxC$$%j4Xc&Zp{C=GeflLHIl%G67wAm-XIsdto9b-qowOMT^)It881>onO7y`nkAH z=(h*DCXkuA#A4k7nxaB(lv5%-#2+x+8QSuYGh3Ui(=i zPrIUpJ8mP8KU9>XIqaG z_bm=Owtu0np3kaR-uLp@V`OGaAg1unY8+fFA7kXOcs57m8KdQgRlV&~ft7!uE(*-d#J=<_F1bL(Lo9Uc8jc|_~i>^opK z|5?+9@$TIZMcTCwTsE}m-R3`vyHFQ|*_fJDTQHyM@EqOJ7-_2!y|#LazMp1VFL zQtudUJdOP!HYf7ZzxF%n7q^4rwSZ0hqwd=V5h7uWj&ofcEyvqL?>1<|X$&x{0{0NR zvXwE!7FgBmwJx6$U4n4*;nCGT3;XYl?8pq;!4mP!f*U18jWHLNF z#r>REgMSbX%g-R5>2j}6j^S?|ptW@zue12-dUS}~neDopb;Gz?XASf2$Qg>SXGd z-hj_%dt~t@6Y}tWjq``O&tC=X5PPro((Gpk=Ui>}QLyt2LGF0v}glrb!B&v|6s_pKYRQmo3HJVZ+; ze~Tn7+auqRu1W2FrhaWL@_c)jYg9AEWp<=|U?xYiUe$kEL<}D}yTyLic zk^b#Fhe(xpexblU?5Yum3rQJsnPlQp579r9_sf}lTo3Q3cI!MCJ&b$O(>@|#Bab?= zbB+KbsIGPzdrjFHd?@q%_7N_b<_D3XV9?&x(FKkEe6;ocuKlcB=B;Ai}1p%d2lCkLVFfY#@Y=!#k>+&gYIKV+F)kiUd@)*=SHxPP;c+@}(ul>G%yFAR{E||E1 z<+~qbf49=Oe&BIotk^v|eo=J^PyV~FMwkRo=Y7l6tZc>iA>rk_U?sURHz%V#-rn3H z*x~;%h2Il=y5t4E>HBrxtKCfCXqHE^sVXvZN|M5HKP|Ljei>-CPA2@vHce-pHZ@z- zf+P4|oxk;e`Nimi|MQE4Ff-idzb&m^ zSXJuSbv(wOm%$sVE-ZU91deTNfrA&%ZGRH_n8B?v<8Q*ng$}=uM_YO#NyHC->(JZC z94MflAsqpC^SGt{i`SY+2{=Nw^gmYftcW1PTiw2I343Sz?{ONq-Z?|$^=glp!MlfUfkbBy_!0W0>T@$FPg%k61!+Xp-g&+Ybq z#PeXjk9#zRv;~vhwr(3|t}Tq|?%l8I%QX`=yz-VxI{$+&JPEeQr1y!n^XBA)-sE?? zIker;=Y9iyLc)}$&60(0M-nYZKF8^c7g6p*m%Kagr?HW4!Dcl}Q6WB-t9O6fJviE# z!dEu(>(AF$+Xn}>F}qiBu!+S^xiQq$)jdJuiX1OxfEd>NCX(DdDo)x;snTvFfH?pZ&&cRvsIhb$(kCO~O zT<3;b8py;2Kj?H9p?LfD?KjXr$w){Z%gEppP;$<9d}IQGDaTC4BpEt=yjpB)PYbl^ z;;>zeircbs4Q*42=nfze z*GfyEJ>0wMH$waGN{N^<>?*(3)2<91AG2+@ZNdP8sPmll1S$4)v^JKgY9i3*`IWA| zx0;6k@xZoFC@}L;Bc`d_ATE|LwejJqz<>es4K4RYB-znUx8MUF@K{LUCvaSH z!Fc)7gLjBQ{;avLub5d|IooA(a(;1f_PDy%%+Znek^3%2n(b7BARW=p&W^OABC@ob zlZls?9Jqq}{@U`McI`2|kqfzO$N~9lk6;j+vxZ(>Z`*Jbm@ONdQnq`r zHqyf7b0kaezQkBsR@OJ-{xcx+&eS&9tQI>5$3itS#S-;)n={%I&+fj(;DB)9l(%e? zvg)-rgUa9xRJL}Htvge2YIXdI=ZSylz?|y2gRhc*J*Z`1y6XSo477#O?k(_J{q@aD z8Oc#EQ~`f-(Zo?|@6LkFY|1qsy1S=#uKtT+G8mUWv86Breu z?E%WK$yX0HC#ngX_=TFv@-rZFaqC%4GKEhpRV)qQ`6-lQ$bLbc!hsI46TTOtFC|TW>JZ!>uY7@Z77YXn3%!YQ`bMmESAHa;#vh7$bkF*lk9$? zlo0bPM?R4iP6-}ryBTk(VLOpZp~k{q6@Ow_Y3T!~VOUQsuf0LJ_y^g&w!QN*ku3c} ziL^JNvp`HU(&!@=oWbq7rFRv_Z+MC=`1cY&bdyI=^vLaY1{T;Mxd^o32aW=U;3zb?-Dlz)QIMDS!;Cj0kBum(1sP=psS;c&TDrhWH~ZnD z7GiZDcpkpV$jn?7aVjnD@kP^zO+O23Am%wMaB3HXwRNK+hW4t}v`YZ0f zpQ6}VbQrCG{d5!N%U7?e*DJ@M?VdMpdLIarl9JXfvWSUIt&0AIjlke(!srzK-&uey zOq;62XCV~Z^HGZS(4sJAXJ^ro%XTZ}J?$ED@F&a75pn$1G_b}{=LTgX-^SMOsh9fO zOw^zs&aGRwde89Re28jNOp>)6Do6({D3>It)QH3b!{>N+`E;+?#~7diDhp(BcPy7F zbkM)+&M98|>VZTzo`WMxKyh#S1CR}alEVU7b~d&+0oxRBTSs{2=-Yu}Pn$A*bh_1Y zY$-|);8O(4krkw!&ugxEP%Y1scX=6ct$@dxOmk zQ5zhTUbD&ugb$Bl+f`CBG61<`VV9lZ#`Bn!6^j~|&0a!*@&2@wt5>fY!9zXWaA|DO z@h0DBo?_NP~#R`B*R(W}}dYQ?6NPnk3Pb4KFX|k@GXgfVZ0X}do4;N2SKdKK74FzPr6w>-K*D`m)#&l!68hisi2ookf zxWJMTt#e*TmYa$f6bJm2Gjy>)oD=|kK!b#d-eTbz<1voJ#fB zsK$MKq4Skx$uEJk0`T{q)_0Psu^w=Wdm+Hz`xv5HsL_g06zf5k5mL>}xM- zU8iMeHR|8=Fwv14JPUyrh)gyl_-^3h_80&Qm}4QfJixIO1|QpkJBuy2hmI9O3y^kQfHva9&o{ZPCMN z?B58)(Mo94q8n6U> zg#PrWF=;^#0! zWF9o20-bsqWo5hC3D3Pj)zcN##vfBtx<2kY9vZTTU5e{P2^9xVjEu7S%Au;EH;>!3 z-*6WK)^FJP=}stv7`<<26wB%=ke^#DEV<8)uA53>i4BZGQVrY5-`}4%4&0F7jHS(f zCqOxYCo5AFH$mNg+c_flEZ!vA;gH z4A7fAHCQzETKSHU5Dma|iDCQSGSi63mMaOvwHEOHs|=!An+s?j2}pYp85@iHM=QUm zZs+-Hev3zc2^YUQD!5-UP9p0o{c!@VN zLv2oJQG?Z!Dwg`=hd6X%pm4|p%OW6>#JPxNr^p<5>cA;lp)~i*{cwqGwrp&H8d8L0 z571&mSOcA*3xLNrfi_&!>ln;%S8x)Kru~S&LvL~gr9v&{T{mprv%oIau>Zs`FCU$4 zd_h#O-M;M)p-t7Xu)ypunQ0b{DB#)P8pG}taHU$euLkx{SfeT4^Y=2KofY?ls)gm-QJ1&Rxx~6#(4*bw<=eEx=B9+EU7Qizq^0IbC&pHNWpQWy37;{%!Y(k=dqZvhWi4n|Wl1>RLSt&hgU>bQSE|9^gVG8|7y zL(|36CxV#GeGVgSP*YAR%nV}QT0WF5DEVI4#+l|c4<9~+0U7N6Tt$WlJUkB_mm*UX z)~DMu6;nmmrW#Yx;*&4cVoPg2BskbN0rTrGlQPkpF-VWZgq-EZy}Hd+YYkb~?g->^ zU6mIGpMVk|1)KqHlqBWu1&;g6X&(~d&wRW>)C243Ah&eARCI7;#}BLq%sH!Cqz zsM)?Un7{XA9?&38*o70;Dr~qY71UUIY3Zl$-@gkXyaux&Q$}}1{T7`1Ne7{107YeG zNNBF>poa5RxrpclA4Ap*WC$SXQ)()4(>MJVwR(5HT!myTixn}&gu%hV=$IHxJ}1pa z?=!U$gMf4VZ+UraP)ePhot~hCk8`P#6j%~+jtW@N985-57BO7wvZ(_r&}Vjv%h=0e zzDy=wPJ2Vhak^%;Xbz%lNWXaC4uB?M|GD}NEh_JO>|5!$EnIJlU{E~Rn7~~6olSt& z=RT)%40eKapszRTPzqicWD&i|!Va+XTfcbvAaVQ<9=?bM-TU{IbdhKt1VSkuXq$G8 zqdp`%0AX0bFF}Vqc6R1xjzV*D;1zEBBc`iz1o-&r@RZcYRyo7~{0sWLFu1{YJO}OL z0F!R^AcIyB*PqP<(%MV|a2k)6n_s$giQ8?*h*>_tu`=jc64UQl7LBK|oZ4}ZJoup& z=)qhN0c`@kaijh~$BX-6RHp_yYfxC~jU#XA2m=8P(c(S0dQm7APazUbim`0g2=ecq{jd*x$b16RxU*(9%jX6=jPm2zUD4XHHt+s z%NxMs|C0ASRn+S+_2MWMXU}Lo=nypUVr=8-9rU$u#)d#wL@i}TbP0RRzRNY6q&LJH z=q2=`tbEizR83+ngWyF2cv~nnVpg#f&yLaVyC<$UvL^x6D~z1fv;sEs z0J}hZz)xsATen>vSKo+kZB?vs53Y=iiD@;Ac(mv=EyVaU;2D^CKkpbq%(@evU{-Z| ziwvnLE{+AGHs-5VU>W^$M3}Jkhf_gb81}EFCTzgpG?n#46OU1X<6qBoJv24F(BUT7 zh2ppV-woBBP8>$g$ABlM7w2c8)I2yoJ{KyAQw#PVKYwn6=7>i`v%d9=of9V+q~lVnoVE=0`m zD6$GbF22X(Y&_lJd7jJBcebCE-F{+F=%A5NQMs)Zd&kZWJybg?Eh&l9{h0$K9A6yL zxwenZ8v6Ocl#ilV^B zx)#|9m;qF;cdrg<#t^?}<-4*@g#}DR%=76m1lkNZ9TS^sB~3N{tBs z##n>j4Vc~L=8gpNLX$uETtkBf<|F`nfaRk(px@lW8^OQuUT6gHDm<01U%#TMKl}}i z*)U_#5y|+owH5Zx?+inyt?4J>3}Qva#lUi5xT?1EY(@sPyNt-n&R(W$Vtg9d56Vqw z?8S>0O|TbX6TqW}TZ79Y41bFhgtsYaXuy1y-9qB>h8Z=x$8}3WWm243j!9U%muSFyG<4t-U=eI(nY35$*`XsrQjD zba&L=u&!a%*z%vRWZ?ILfAc9mUPeytD)1b%Tmh#QSTtA(01U8zb@U$c2pYN;n9RaC zQTn4?doM39xO?HZ$JIl&5ZEV1Mn?T#U$H@g1uOAy$DhCn6R*nlG75Yb7)Y#QAAuU~S&5`4$j zZAwKJkVq6j;WY}1p(WAx@855!^TS-b*APpJ{oH`sSqupp z$A9=o!vujLy@axNaH!ttV&6ug3@2)w=tMo^U^SsJz~;~g$UskVS`Kq6q>A=H392`G zKk_=>HGfma0OL8J-TqcsQi7$%2>J|odSq&9u^1yK6rhk$jY5-ZVqPBTl_WmPAy~`l zy|XXfoVB4k_3rw-G@W2lK_h@Yyi7R?!+?Rr3<@t^kbsqO2h>&PK>=I>NCiE(1Ug~o zP>3W93=ONnE_>*{%Zf`(e_K>E1)N;~SHD(+=5^;XyE#cTTeGlWt94!nYry_&Z`klb zyfpBw)T-? z16--AG`x6KV-wRkqn}ci5Ey_e<>)CP$JJ!NbdT`B!#ZG ze_acY!s7ZzBBf*Nj+>Lo?8SATpvj}3w1I&ED=X_X?7mR?M+sY7Z_rK*03+1{qxZu* z^ANJoPT+>cpgVi&o{XBBTC$^OjEmwoEih&b3~m8s9I6ugx6H&+DK(Sl@*B34`}S9d z`;h#{LRCWG)2*F~%RdCq>kcdHd#BYQ*ziR{#Jo)|577|?oEY6F4#4YF^0|>+y?NzW zh;qepEdiNd*L_mX!D5{zu7#nHB=|&4EpFJx1C-cJD2M+37a+ay2?-Zq*I|Gk1TEwJ zh!wM-AeB47&ywSA5AH(M= z%GUwY!Bd94b4-{A!X7eMnAIg!R;C^Ibqj~DC5T97UMm}s8!pm@#N>Ca`qAu>JWx^? z98G_UE9A726-({=_b(`4bl;PZlw4{5;!OiP4Zsk*Hqcx^kAJL|P;ww2wxxJ2Tc;G4 zQc+P!1Mq@-2;ZnkQb1BWsfL7w!JHg9FnD_8mZ6pzNzhuavJtU72}lr#9=M ztKqS{*NlBNbau#X$Y5|yX==mY{}Mb`v=Hom<-n)RpP(sq#I~6d6bOdX&-V7rJT@pY zwZb9Eh^i`2olqx*v*RC#cdLad)>gD$$(kHJwyv(rtINELh80^hA426(L67{i zDFbUf{J8q=Wbb8Ln%cz*6*o%wCfLu*>3MlwmQ_N1b%2;1!p3&Tq%+g!;v8db zvOW=zO){KrWZLy}bTk@Y&vJSf><7S+!ZCNNxx-g!qkA$2Q2y!DkMJA9YL!;hP{+_$ z!1#WFwiEx&W%R-1ck1=AJ^2gsHMxk9dBV7=Q=H3}m0rA+HWHeb z2N6o~!q;jVc3SzE{NrU;_Dys;^dRiU=Hh%=wVPSKP`joVY-X>avYwH)wsPJ^FtD0i z8r!BZp)H2|`2WR|~ zUFCLEz~+X+d;Z;A&9`JKqb=KnK;Sg&%@5c!h2y6rU0D3DnKCN3gK7hXCzYQc;nV zmd0xdBF;rSpqf?T@VDXR^F=pOU5|cWa4H~}N4t3{`ik9}>$EQtdjcC98;3U@gYQMy z>jba}^$YCNX$6cEn!ox+rTnBauli+{1`!yLh$z8wGmVSVpmyLpcS#u zZ}dQ)n`$NL1nvw2(8+^Cv1d?i#H`aSIxNOTmqvJUa>uP?@ZsbN!525hKbnumlB})d zUL6(okvR?(ti4a7baFGu49oNospeN@wKYIu~65a)oLPnK=imEE2+QHO{KhgV@JA}9} z%o;V(y28NTXD2{))hr<=7eQ)MMJta06HhetcI|}C*`aIAk?WlK<^4;-HdUH^icU@y zlgA6zJ|5o0d>j@jEIldKyCU($TyPpFAN$=UrGJM>VSQlqd1$`J9B4l{7qE1z+WrvS z*QFqUij$L*laP@i^b+Nn$$$p^LD^P~;a|A|5MCcs1$qX;>%V3?T*ie;6ssr^5!7J7 zuSDpi(B)}4@>8xwBs!3JGVfL7n4^a^p#A@7X#u%S0HE!THteXb;(*Ak`Sjel-xmi}VF!Y_l#uf=HdYFZTlgV4c_IV%&UzS3XxU*dja6}hsw+?{!SZW< zk(<2^!wH!g8Se;b`P0yTdVjuZv82?OQnh=)6-&T41D4km$@nzk*HdtDOv=<7s|C^K zSsHNTwejjGV2bGUaaDF0<{oasOvfGB&)`eH4-U>KEd>aYdHx(d=>g&9HJDKW-S)zv zEIKJk4m6mq*Ivvyr{tN^B>6Lan0>aIsEL73FKxaFrQr((t^~;&z4ONShfbqg_wRrF z_U)UbrtvC6J}@H$$R!wc&QB%aIAi_;8M2`LJl+e3dv!N1;FnT}3n{pC3K~7;*BRde z34l+-xK^3jz%8#4G#}=sQUe33}jctcAdA)=2=S$VJdt>=xC(e^h$PU(ak@n9ANZB!Yha6QBEn+feQFa%=s7`C9; z`5_Z*L!fH>@e$C7Vc=z{Ki#iy!wTB{>ET?&J)OEZPVE|An%hTt#gDX$ETO`V1qdEx zDkk)pfY=J}F)>8s$RF2%VKZ%CeYjD(NfVt513o6}er;Wn>%0lG`QBjbx7Rj=6!w}I zpxN2+sHCm-TD>`2BNQfjGzk3jRVEL2Nn}xKV3qw4ObCInQm=L9@YRVO0Ept z9xl3SspAFcIOI1S8hRB2G%1>f0AlyluvkQp%O8JzFWcYmqCbXhog6|C(!F5NlfVZv z%6wGKC68ofWuGcVn?y)5j}{N+tCB&L&9jzLhM<)Ctii#bpPzTA{p8SLQBi6$A%vFs zRW=j7W9|<2_MYK5cYjvTQ}Zbox%Ie|3_H;Ou78Dx*#uAtq4~OZ$5B_;^Rvs_!NFHR zM{x*fW5N8&(qr-rFKALf^tjQ}LPP4;W|I)S@;K7f{^+4RPCMX{daFZ)7+0yd(@A`f4e`xi#+}l+x0$Am*K4H5|zB>iqS~2uAUbtJZrR$y>{2me240JSqh*1jImg z%qHob3E|c|GHBOuRCsw3hBHlz>;E{QuMkG$U}S9c{y$@R=-)DKMEx_xN5VsyCGp=` zfdAnw|7S1rR661Ke~()qFA=}cR8}V6YwS}ZzfAt@E6SsRKJBYGn|=CM zD}LKqUio6-SKjA8wt%CRWh!&yaJI8<}}+1I10<}C*ua`I=FpFLB0{YPUg zJrCz>iu1r{INO}YMn022wmV=SmB1T2hn%V&7`Nf&)v_VH4uio@PmaW@H1j)ct;LYr zjczZzQ>?x`B-t~!c??5hndV|h#QN#c6b)xCL2NBEs%XWcpErq&t^*3?9V2Aiq3l)MWPJKGma>e3J|F1Oo3#Vxj||DtU)@ zi_UZQ!)te9yB9{{Srf=Ga&x8^k=B2-2>D+jXv7pP_SgG-Y9|O6+4uIXttAJ2^M0;=)B|RV#O-SDXDV42XJB2=Yz|Fr+(s_MBp81Ev&L>q&Sjjje`K;% zl4SA4DV()}DA<%?FvXdIoL4S*XqULJ=}%&AYLp|FgyAP#5~KTm8G8H)7+)M;XJ(|c z=@r_~Ijb6$2@7hg2h@*#=u9J}q?X()8W2oUP>5B2x$bCR`o;f>=JgrI8!z78^x?tD z#1G00ewlxxtn;yzE3WuEtWI{$IU^B6I%3m)6HzI z$adQaYX9>UrA(b5RaI5Peyzi5Ei+cZP@G?EyP<^bQ++?A4xv&IAW=2g@y+;S=T~+FdyW#(eo>>fWDtF;W2XM zeNmpr#v@Y*^CwMxR!@dBTFw$rCpXVi$GseX3|O^p+?+Y)HFS9}U2f&ZZR^U-zMPQI zi?1`<{f#a6{^udVmmGIXqvuBM`ZrDX#=X89u6jLZbxF%pmsdK$sZ&;9|4{LKqj;xh z*v7%p^XKM8qQGHuP_Fc~j^EL*afS*PDue}&JbVts=Cf_jgnafXr;t8p?H5cJdp;*q z+l^-}7s0W_2Zz}gYmM`M4D~G+dtzq?b$gTArQ5d~i0hul;oP~|$0slSwVTf)>Rm=$ zztbHJQ>s_5SUVh(YD4(HMadRvH#SQPYjByJjhu?m;f3cnJ;7Yf%XxtLk?0F;2{^o zY#)P7iFZXDiSP87{GAa?t7e2|yOTgUCNJKd??nzN1#IGWm7D?m(;7iEcYoK1PGjp^ z|4n7H8Sm%_@&D?cT|qiW4Icnk#r$WEvXHA zqQ-{O(T@q5^SK(D3N@=pWO}>BE(ka3JcCkw_LQA-m`svyI}*n2|2S>DXcj(Q6+32X zxM(~0+I3z0Zr1BB=M(YFlrzB<;r8`u=97y|8B)Ylt;AC|vwGltZsL8G5@(bGj3z4RIaSS>(QI1gJrQAlgTY~j^j3jlfNt71; z7_N1G1anOO9A75sMoukJ;{tleZu`3>>!_=5Z}#z<{C3(N$4pwEwWGxw!c3XQ9=c42 zGjS!0lBs8+-2IcqzAFF4oBx zm6Dm)BHcad7-Xx()iLGMp~szJsk%nU(z*-GtR*WAY;M$e?p(hMsHu?^uy=+tc;yAGS`n`YwDAymwWToD+tLrA|>kyTo?O zkp~e9t_MDcwj@n68T_LJL+B-Eu8D9a-yFNnlUeQK_&k!DYhq+>ab5e@?E9K~ zZn@X%c7Hf+h`IjEl0H0+)iJU?^{-;+Jv{fjPLiBAyx)HN`Szq=Cf*W$-R-uz&Dcw{ zzP>c9R@PO9+=sr8%;N=R*s&%aGfI*AHz|LPdi&Jm?nZ~zji}PdFXD60W$TOj4FgEt z)e!JjJbB)H%Ogna%%O8&z;J_Z?Ey8h%qM>d@z2cjUuOq8KQVZ2ksvD5_@1TkP=^WD zc}LA5F8kek%6u1<8D#X;dR%az+HnKTNE>?H4|A5q9?#&~WGH7*GI9BXfQZ83EXVtwUflQ}`gL^9ij z+t8H`QRU_R#8v#rtX1oDszS%kvNA*b$*UZR$a>CqWX+xnA9MU&>^yA|yHF2<_+ z)Exd1f2~CO{2y<1`pjLEX2g?Fk+k~wtZ|uw^qS|zrti{4$gk0(^~@y!Y*NKO8d83| zJEdN!WdS6d+)0f7igoGNkxE^QWu{`s_t#JlyWPq-Ne%+B7;XvTX%Xd#_Z8vW;pCCt z1^X{0>L;YiXA4=hNx9K5se97qOpUafYY%8;*xKSo1JcUX0Q=u2uTKrxoPX zpIOL>xb!{cMn!Xb{B3@R*16+AvugsJR~S-1Mv6SjnUxcv-JJbom~;{&=wdbZ^C%;J z^jjnH-XN*Ob1V|_s2?}|jeH(ttP0Whs+uM5{4QMOU5vgbuhc^wm1g{!*--_X!jP~n zp=0qN{3FS&%%y-If_?{#MPic(MJp^!L*Z~m!gle<=VsPgJgd~@y_wlRr31E75WRM# zZ1GF}#r5{dZ{K8%`08oYq)}~XAr~wSb!zV$^G-B-^x2DMcJN?yl^PxH5Q#sp5z6gw zkVJ$^h~H#;CS-|^`SOn2NoTN!%jV7?zLuhE5 z{ucnuKr_EO&${j?1}T}e%pN_UYC3$gVqbgy*2{M&q~kB&zQ^%w8K0RubM6QNuItE0 z1C)|fm8Wx_wO&CBp>+%kg;*L&8?gP9_N4~_Divg2P%j+TTSO*MBaK}ek{}poiq?5* zAJBzH7LslS=F|}CKspk%U0_)SWD@W}qLs%8%^**i7AejLTnMbHn)e>8d2%vG5Xh2* zo!u$7Zr$d_joU2OYvzk3tM!UiSrKf@jqxt8?Ijpv`1u<-cMcCYX*8$HoOJI7PZlkv zX;4zK+l>g`Gs-m18wj59WCvq9yfa&effSG@iZn@4Dy~r8TYm8VLr#_@5XcgZmXf`_ zeSYOz-{SB5_J6>&JFkM$oYWqg5~RY1$lfm=f5>}3{625I`8Fpf$FyCC_nss#fQ?#6 zAq7|)fppikSZmQrF&>W?4@a1;Wis8v7|WekUSnOBbYaGz$my(MwOXT}qn`Sr+7;-# zvMc)x7G;lu5d231zl+C2&$4|q=#$&?15!sx9~&Ul^L`t~4nV3XvEgj%+(*@P%DwHxp#J{qWeBDO^c_!!yar!^%d zK6tbgsO|1O_OA;LE#h+PLm)WtQlN;etn5FRJ&P2=2JM~Q=b7}fm)`b(euL{n*synj z^Op~ZGv;DL#qr6&-etCB5F&!Y1dk7bAi!IJcX3l~oks+R4*`!P!(*&r;XoisghVDD z93TWrwr=o;xo&eHz>dwop)rPsBw2ESX0}2SEs1NxAChyA)!}w)r5) zvkVs``_mCwqESi_bjCNn@lC$_y+7wbW zc&zC#&S9KG4R4xxP1HZ?s2U ztvFd3@?wOLnyxcwtvTGEQdKQ}U31(t_{MQKR_soONS#m=6V_$NWSGVFwuX~s%YzT^ za=16aR%`BDKjdIiFdY`$-kcQidNNVH^X zK`M!H9@ED6k5rQ214pfk<55XuAqiceJ8F@IWU5leiRYmdWLlDhfLTS>SR?{Zpea3a zC|MhUog0)LviItML~Eqdgb*;!v8rm0PFIvo8$p*+?C(zb`j=ki&Yc$-rhgZoQGxsZQj3om(_Af@E$28iB@zbkmxkNFTBIr4rdL{S(Fgb z;jw=}k)^b4OHmYCB7em|?!bAx>U21-U{ z&!B$PXz)I4J*E%=tqb9TWU#09xGa3p`3R!u*@^f#V}ESE8+zckfqxr-_i;ckT=s%| ziWS>{>JZ|_IrxAJy8nFyDs}_bLi9`c-bdjbtZhRkB%?$Sl!^eZk16#+g4_Vr9{62o zV1IZ05RM@59580XW%pn1<4@11w17}z>od7{5!R7mjz z?3h$LUb{KvU}u8yk~ANX-p&ajuw0iUNy7NF9 z$B74{0;vXPp(!0`fgWfQ75zA>Q0$Er?b6|UHd;mf1&gZTcv-S6ORV!q5uISe!H}=L za-DzY8?W+}U;cGeJL50^>`zd^b8CMW0yJ&IFTQ$>+c&P$2FY;$7TwMszC7ky+mcDx z9Tz-4TAK-C&`2RDQq8!?D2gG! z{Hx#Qmwx%zxbwF9F?atp3LT)+<%{Ue)K-a$0xL1yyqlMv0b-azn!zufoNlF zy_AwXO&IL#FdYr4nvTh6M3!X?hn_UcvDToIl)7`IX^K)3*@p!bMUIq`(Qx#WTC#pB z?F#f=*_C}p3xo@aB|gf;hT%K6iQa6eKOPH;rx*+Wsb!+u^E_uW8+u8{R#zEf5*5Pb z_xQ9>QwVh?E`0{;g>8)6Rxa4Hdmr6g2|tld@Y77pr?begZ6YWGQpT7d@o}i{!Z`#! z2>iKz3_SqaUW1DXR{S}LumOCVXB5BtGLOSZYzfj!Q-5b%(1aZ5rL~B zZiJ-_)NRK?CJePiNP*B2fk4WD6t0(1f>&|lv1P==xp+z0gG$eZ1Tg1isGH6F$KAw6 z8Y(uJE%YStr9bWO)6uiJvX37F@K2HI{v7A}$?b~}|9Y9pU};St;7D{_S(KD06=R{4 zlv_DwqN6u|vvw*eD2gGOvUpR|8iO$w>pV@@VQt1RO^`y8rzy@uXFcn>-6pLPLTnux zvX3FsN}^D4j)UDD;%z&uDzII{ zVzGp_BTaLHObEcDZt1MW1xf3mu3P33c88Wn3yTk)s;(&OoK;m*KR%^MC5v@Q)ig*c z$kQ~&W?Dzpb*!rjnHH3d;r)j*@?tn2B_KM+g3#-^4d z|GWJBuly#HoqhiJ|M*ALbkv@=ouzfNx{ER|yFOosDbT*S`8Se)-paoiBgo z7s!hdq&b*|`IGy+fA<~_?tZ{*cEWO1VyxM!M4WYaXK~KbbscR}(U~aO&eDY8Xux1F zWM?vEIvF!4a(?jMU54W+N-4AyNUdXdfshnKxdn<+DvY&MRgFy?hx@yq^Gfy8ZC9Z0 z%C79QT6mVt+9dI~XyQ_4?s;-VPcP5h0FzC!md&3Vu()Z%7L|+33pJ-vjh<$D*qJ;F)2*Xc;2R9%jL?9Mi z*uHq2i|2S=X334k9Hp1E;His09fKGA87;NRzUe;#JO}SDK1gvn3%t$xaLI8YB6D^A zyWRz?wYYwBU|gWJ2Ga?!o}~YQBW}b}AUMywteIw##A;F@KuVko2u07BNrLlurFzEi zVm}7ww`~j?iu9?(FFYmfiE`0%BF(7O>S0^|ldTLkK3I84f=?NuoH}~^W9HW9IbNQL z|V~Q+A1jDQjn6~1>l50bn9pmb!y?LU>yo3>hNF4Os4!Q5b7^P)gJ!m*O898t_=!!NN$a)=U4E9BjBxhH)BKv;W}$g@e0c8>NP>Ji5^VE@j{tD) zOscv0+GV8{F4)uk{=CzNk2!vURRiY&=SapFC)5gRTO&fiJ5WladpYZ7OwX#Co#B8& z2ZV449W!4BCrK0tafZ%(e=)hak9b4RwyjV6vH^aV%Ze|~*UN6xZ|2RG?FO~^IkCM6 zg~1|wrZrBh%b#@wbrIlk-obG`X739Q_<&v^u=zU>V&+YE4)7&fpGlPo7#l%bqBXRZ zJWaTDZI`{t5Y@|sjdiT5hILf~aIk;C(a8w}Po5-5C8^pNXXl*d`u-uK$u9HR5!SRM zT2VF)-UmiSOf;b~PogzirsS!nHI76pY~%a2AaGK)q>1h~_?G>>DUTjJj`810Ba|k{ zBqk#1Dum7mvZl3>bdwWu%08}kcbGDn6s%gw-T8_SPD+07dmr!{U%Ad8_smWgc$r}3 z0NN6ybfkRwg&cffxo9~Ffo>f$3H#Qe(}dj=W``N(qyraY z`n-(}z9zuTc)XGfTy#p{9Xc22Oi&q#7O}h_NLu5u9&`xYKW)*8Aj=bec<%|P zr)&Pb)r?fa^v*SoXD#cp+|DN#pfZk=bwk-2JRPY#E0H96N17C||D6!rzIlt^_?`cl zZ+`O^*%@XWKYqj?|LI%&^reLl6mF0%_N!A)$^^DOxI61^aulmzb(8@l2r z1WrGxZQ3S9%+~nREo0j-t#Lyd+_U%Mdp1rG=u6q*jYB4}j-h`J3m7_qNaV#LhrJo+ zVocg4*W3R7ETc;_(I-IJIs>*X{MfrZJB2{?H3x+dIO8rbi`K;qmYuVNAn|Qvi>1-=1skHA>M7880l4|ico!Fd-!nj1%9tJW z{O?1!ls$GKpu<`0pzD9L8NaTV&mlt({CbkqZQ629whdc=caaC@na{rHCN?!6H%zXO zz3f-?Su!m`2vV(jd2Q@()5rUH56;0LP3Y2;5MY=k$+u8b&RK;SN8WwrQ^xb6Chyfj;p?pj($Q{=bb`?nntH*kG_+ku)j1Z+66Znn$zvKPIn5KSb>vzj zk^TYOuj$OnisjxPF79*__qth;`!$BB9Wd#mchkJoCyh zEfUf^rC3(9Hqc4MBv)VztFlDrDW+{1PbR1&!oYH)7yk=Ro{NWECQ}~X(NzS#yYwS&OREVH7OA>va_#1Ae)#VD{Nexer@Z<0+cZr} zl4zt(42l9NCCm8<>-7pDJjG~C zQVb9}&eN)@$(5k9j?Ow(t2O%vhm1x;f@!&Z>lW*(X0=>#bo7{$*)dI1e=>>x7tF3e z-<4h2-%#7g&U`N?`$+5Azh`QyU8YZiXB40U&;gwINRsK}8u5`wH|)!%-p=zI^*OPO zpQByY+?Y}Ftf;!*#^LoF|EFFy)?1Q3(ZF+wwh$09f+X)FnA)6gBXJ8ZY+1hZ;8Xy?7qMMsWT@VR!F+ zpA|EDMgwlC;_Fl z2%S=-DZ^~@E)?7O76Lq&tqGHW1C|zKA_+No;jqGhhw6h8J=VyP#AT9|RCw)iC%t^p z$D~t>&LR^@sszJ<#M(eI0xYcN7L`ecLLqb@%>+V1rWHHWJ=V(wgDe(hcP_>@O95?2 z_@Dpe0mDH72%NEWCaxRq-1{L-C8w>xRW-g_qY)f0B{nqNymm;obDiVaBUXzh#;Phs z2%a>REE+@YH5!4}1C&}v>AMspS~46R@H_v}-{)K3{x)yFe~*9iyZ@X=PmUs(R*JmH zan4bdB~`ViYg^i`#kLJrNb)q})t6r-F9vw2nNBCX@WKmGCM^VA*KxYgtm=+DOF_oA z)m4qPmU6X3YsGT0q!<)Tc6S+#h9p|^-iP;j`0xQwPL3(d6=iEs0)|;is(*4$`JYz1 z0)1C@Wq(a31fR<57kV2zis%z1gt&QiAz*!oDb8z4 zyNy#5bVPP94WM1X%QJv{TvqH1N)i)M&{+W}*2|WAvkyp;ge*;PeL@!#Gyz2?6iz}aJPt?I z8U`jNFH%Y%5|1=0j<0?#NOYV(X&>aO)0ECTq#n`A0dL-4vNJK%bu0yu+Z9c@=GIuD zrDm}#d1)tQ(dNuf*8~$-I!$LCd79u{bd+SN#uMWs6_&&O9d@QW)OAfDa5G6r(+s5~ znHw-24Oy&eRFZJ(`ZY|~;7}|lW3&<+92`)09_1UHnX{-ZUMRd1gw7G9XLomscQN@$ znq)Lhg|QMJBwf=XRKoGvQ=1Bvjmb#p$~DHcbjET#ZqkG##Ct7Q)b0kVKC^CvHBTFR9s*Nl- zXY0RnaZWhzSu9T3ALpD}!MgPfMkDUrzR8Wf0khK;^U~0D0i|Fx%y{|s0lFA-_j_;B z8V9;V>x3-Lm`&o;-QN(a{mdCr{{%=?laQq>u;^Yvv{b>=%1kea5ap-<4h2=V+HDaES|1 zpFHJbO!dEY=hFv%%!V=$B7?yhFxk93KC%e65HYSSxF}uQFetrA0zt~>!@_gLr~~RO z*6>`*&j0c(7U$ymlvG$}FZ@yn$@5A}L=XagBXR5(`;Rjp&Vb*iv9n^|l8v16*}LfP zZ;Pq7khXt0pY>w!F;;?J!WLvtvu$4HHuUw%o&|OrD$e`x%=6qZZQJI3gjv5SB!5Pf z3B-l65mE?*k_e$dsQwvyoVNs_2twe4r?Zx}^LQ(8NP>|#-)-yOg%EHyAhkq^D5-O{ zqi#DMCW=uaDP)1N9#5#u?_h0L~nX71F4LC zx~K!I8j3977nUSTD2kLURU|4fo#xEu74_Va><$3X`lxyO^}5b%s@A7-)+z zElygz4>X-cX_yolPu6n~OVTW6uyi(-7*SeB6@jzVb%P6v^xenX_(|^!!aK&pA$Y^0 z?wG9{rHwMlT*nx4LF?>BM%Wh9nKf*=VAz6s3N0dHHnCNj=i#`RI_jMw)we&NmmfAHoB07;r7 zghMGwnkc+aP*R;`eC6W%AhLl0yHS9Iz_PMse;;ggnwh>}Xn z51;jM=z++lHea}?M+`A92rk4j02{!!xqiGx%zpMXh`9hXefUJ3)K3qs?h!Fq%B9_xH8a&4g#0vo`gFuvduhY zHS5KK&IaUi#yj^OvpijKP`0cZi?^0}?YTdzIhf?&JXsQW>;4Lz3I=(}LkJq3I+| z*I=zdppaO02MMp=7%@l`_fKn%mMuGl#yW|Tkr6!@=LBzQt;aaQ?kM4h4_E9K8kP|h@Fm{( zi$9~afz}0t(y^3`X$kUdi@TA;TVt?oltxZ>hE!EcWg5aGNi76pr5Oz*sh2ElN9!%B z^XyAOZ2~8a$F2J@88&MQeay3nS*pvAjsHIXHX z!ElG&$%Hh`nC|YfzqdzOmWU+ht#{t$wHI$w#B6|I&-Oj13Fl$sk7pYe@;Nv3(U=qHOCtPr$=|m9)Nd5d z#mf0h?qT!2%g%F|ZQ{&)0E9RLS|8CQ-^*P?5EoPXhv$R6jSTc+*2@i`l-qiL=WQcD zR*6I^g-{t%Y?5Zg=iq|J8?eUF_6c0P@HEa++d$hYoahL_k@g!C0I%X*TIWzWq%+jj znt3^5FV7fiMN1%cAY}wvfq=&&gG6kc72#}Dx-DoGXJdcMcAguxJ@5;gn7~aupA-b= zxqq_e!{as16M>%ej0Z!qK4wXn4%dN{DLT!_iRl=~yDAa;MUPRfUZ9#a}wI$KugJ3#J z84VJe&SNsk!8k=oNuI`7${&0<$B7i<0>OJ48%rwq0J_hRlX`hb%NFD5xQEj zbL|DHs^RY4dyIz#lR+LGO~K>C*$U^qM=8CH@sd)pGfYU8KqM)ISj*7XhIB7sXE3B) zHYnrBa!Dv8^~}-<5Lxt)1#`COcE%xtWSl6h5{!~S(2}vr2+~0!*qaRSKJekmk|%4! z+Yja}Yr{^VdTD)J=l3TA4u_K9J@d+9og>wnri*HrI6VX&ffo|%JZV2%y|Dx(5IVkR zs;Xp|Xa-q=mT`@&o0i35$#^1}oi3T3uDAWJ$l(H)E z-ZLBwxOw9yK}ZrM8IC6mr~63PGMP-++uP;n2_*R-X0WvuohAg8(OFAbRkU5lYPBX1 zD65*bX%I@YS}nQ%@FB;u6Y9F6=`6u}#)CY@>0X$sPdnB`FAD!_+AI7N+7;-#vMc)> z?Eg=X{uvm08k(Igd>3N1L|+j76Et~KdR+D|n#d4+TJPCeoZAM7TsZm(0_=W#j`8Dc zBcnanS04dTFDNbhn4Ra@Sp~kcwwnNnjR{6V=<6^-ls^)SyKm0DVL&&lMZ#?U&h$nLewkW;D$h z94mC9E}Z+yete5H$|PB;nGSPK%ZAyyrfxg-#shXoMU)2n$k=RRF`W;ji6YNZ8XGgMIg8U%(kw$M#c)uNv3MPZUa@b z#`GC^6BVC9V?BwAh-}rAV2OrGgjp6d@IlPFswZC}4k- z^V0PRlR-+_*R6w?+$GOaCWDNsZTYM3{0m-t{Y(7d-MhTHulVZi*SLOgz`AZwQqp#g zB24DaLs^519g_Ldbv~dyKb~t%;IHZ3$YUT;Qa0$U?<6GX;@E2a|6+ zd*dPNvZl0w{c*h&qF-#WxSI^}4-;@jgd}zCx%i`HHHuFKSDh^B3Ofyn$y~` zt~*Am;qjyUG0v_mSuB?Z@kF<;UV)U zkN-dR-mF=YE4$MB&g4@Jcghi&018!w0*hoB)syK#FM5#vfgbcfWwP!Uwai*ln`C4+ z$!4)w0}9AQCUUrAOg^4T562^J+#G;H711J9cblc}abHiBgQl>+-E<^Ue3- zjlSvB=l72?JpKPBZb9E|+{X7Da?Hy8u4B)Z!6#$_XTb;LJ7@~~rqb7@?Kf+E`>_>A z5qrk&?~M@$?QA1rQa4lI8Sss(6C!aKqq}B3Zhwc6J9F>1cjFr>e80zN{$`6;3<2H% z$8S<7g@6zoHz=7g(MUqFg?{(OMiwPPAcIJ~aiif|aDgD@zCb3B@R55Pr`-HwtWs_F zHH@?+C9eB8eM~<#j)3u|(R#WL5|T{SFERQB`j}K)w-XH)2fWmPz`0>}45R^R1-Vuz zFHs^OVgZ#QVCh-l?SKQclBo&?$Waxu4Lae#=xDm0g&kN74r4SpM>GcOlEE>I!Y)yw zL+YX0T|=X#gHYE}DBpHfY)!c9#_{FFhDFoT$4IU<8jF#VGS7%ziyuK zawCaxpzbVlEdliw{W$?+iBNyMik#miluW)}T8)+_FBk=b6NNN z=U>0z>C1E6c;EhaKfc3X+%X&-Rebg8l8ZV}m6ESsU((bKSId@rvpk6lnVg8^aH2V! zmCUM~x^sMT|AeQfOKgWrC^r9tejByD;CfGo~&8$@h2Z}GAmHFMhKmNlTj2#qD(d#o=3D+D3$sVz4zF` zbG2#Gco_wv)p#SQqem;~x`2#FIk zTdj{6C6Gmi5~CoM(&S@8?55&!AVx;; z?%z4)@uNo^9369TaL9)rK4xC!^dA1xU;j172M64p6=ZqFd|Dz+K_n^dwwtuB|K?{u zLro9pzj{Iw0#CmBiZ8zSoU7%6w(C<*|9HP|lO}RhaNW*9wlv!Pr}LKLDTKhscmw1q zDc(u~-o`EHyN%oUUPFjP!T65f0>ln{ybJK{uRp~%^U2~!d69@+Z&P~zUF*F5JLMX$ z1!f2#r3=4fr0pe88s4b(`SAY6nm5I0-d^V)8sA=?`KGPN-+^?x586bq4ANl4^|!vM z?ZA)9o#7ocv?LgQCsEM#1KulyfVQ`^gF}hH;cUvJ$}xV!XAfrl zyDUeB9%}D9KMhy3ZG{E-2*$>nyhK5Jx$9j}MZj z^JFHZXK+kvymt>PN*Qq0Q5ebd(*=vVCs!nsa~o;gfDftSf1)z3nw~CrloF^?Vb4aN zJDDh+q7bZyz`<;aUo4mvhD^e$b1a5Hr3GJ}t{GxvUAO$@FTNz|jN|Eq&%Zk3-n|o6 zZpr0pNg*O$Kb!GF1i2z?&GA;GJahVgG~U; zdVFxORf7v*w3z_6d+3Z6-VoT={FLv2kV%E`y2`hucD`+|kpP$%aVJiaJp0Cd_u$lc ze8&y?kZ7SwgB&XpF-n{VC1Nrhh6Mb4+zGHc8wp-wgeRz!ffh1Zg1Aw+E!ZCCB2sId zhTBB=&|0TO_px^QDEmMA3a1!_|?-_c%R1A z#p@SbzJB$dJ$ilP)#?5VsW_bE-FpVN!TDrCF;N3%mJ@6OP%$(J5y_0^(Lvzk za7ry5cXPvMkMA&1P%qDkT|?9px@JvuZ2~4Sve;}WTwpRi;N<9#-V01o5@SFF&)f+9 z&1XNvGv0PW;867K3Phx_1G*lsi!q__r+0FmUG@wia=NV1SRI~xrD?+(uKL3!-YDo-%JXhR1s5oo~?o16At6}ub zYq&SbK;~px@pRG9nU;s2OprnnqNN@@pTAn7mE!1N_7;WQ$5G&FRKFB4rgP^&W*LQs zOld-lBO@UR+$?Ly^R?xx%bL0y#)K+2*6pmts{{)kfAkTbefB4mRmDg54*0u&cuihb z#25&D&3d!J96ew%D+slH(n#9#WsqO}^uw61r+ zzu|KWNwj7=rtj~lBvJXp*nlDcEh2VrDFK*qx!bsoThMnKxA9Ljwv|5dt)tfaW}0zt ziSG{h;JT+z?EuI(R}{Z7-b3LG@3Ur*lK6Xo!R@YYX;<&7)s1dcL`R)7R zVOMnvkQ5!iRzn2=%@D9{%cdLHxWL5ZI0D0X{EW(wG4=N-sc_OI zaSH)m5S+FheeV&$Cvt4?bX~{!W=&BR6uF@a4oM^*0qxk%2Lc=s?Z@-GQ}OiIK$p=% zK?sDFs1Q*x(DjzHbJmo#n95aoEceM_iBZ@!W6ijkv(18f~w)tiw?k#0IA z#aOje)V)P3#o;95>~ew0GunQ^=g(jB+h=FoIjCsR)FQB~J#P7&xsg0NEO zO?_S>^77RsNAr>o9~@)5maFB0VHj{e@Zw@Y;{<>I^pZ#SPWba5oG{HbDyyibGcv94 z-G+c*(==SITMiEnn9b(AK0Bi-3u27CeEEVZldK!ZqP5ud1yj#E-_mSqLICGH zxsqhXga;2E;gH5iq&n)$0|{&M&Dq7hEhBOv?i6J$W&q83Io(2kNgb`N;u5{x&J&#Dd|5N8DvC^El)z|#HJXrRoMeKb?^s#ERU6O(tPKb$()H*d7f@9> zS)SuvB+GMLh+M8}3R!`OI4O9!>}jIF`DB^#)OCDh3S{tv;90en$4YTHn^CW?sEUjt zGiluO9`+2zBZ;!VwY3Rjb^F7kjHOp&AE97cD0&w-TMewbp05{{vvo(i*E_Jiw*=Rf zP$iR4S5h_B93CFf2@Uy-+1&?BtAf5>adCOU)ndhwPPQIa=Ul97x>v9G>es*I>C-P+ zt`;d-I^NrPX2$B}T*j8oJZxuVG((9?qHQ<9JN47uzQBv-zhnmsq=X=uw z_~#k7pzk(r5wXBDyQ zecbNCb{QhUd!#UUnW#OZ{6zwJVWj)|h>w1pyhb7tZ{3dGX{}FHXC!A`3xngbvD9{8 zRtyZmGinMDVxY=0PRa?2>mpd^92ZT;McpKirR{Nv5S^uOYu0teqR7!gB*jKjF;EBy zHNn-zkf8!XNaOtj(ja>bX68ju`MkC+rQ)?G*MEFuJQp~#g$1&b7E zQmu_bRVA4S7-#WXfRaQd2qDo1OBd2aaQl1SxJP!^#chAdH-?rHV=~}owmuhcFXz@m=&5|zIwrzS8Fc%4d-8N0Ju6^rE0VfyUFnJ z#PGp!#nB|=!O@JP*@Q{1$g>QjqONPSfIQEMI;S5jei-Nm%V8xruX9ct#|K5myqfaS z2M=hwmLGlkQ<}?Hbc17MCC)o+gubo$`0fEmCwEw1T_A}x?S@S|P*ep!`tTut_vDI+ z8R*-Z2eU)w^Nea%BC-nC8oIW@>jEjRXzaj|Ry@3W!i%#D9^X5m6epZ7)@UV>0{Q@* z9WXk_lm)JDFvcNh5qOHMWHz1PbV0pNAY-+>LW@A5;nkw!^Op;roUV~d@zMQs9*%W7 zua_5BJUd_W@TlbBVMeATZm{%D(G8YqnepM>IYB5cH~nZ?ki@kdl8uKb4U^n3bPgXC zt8HRX2!w=rCV6-~Wn~4Yixu^HjZli!5ctp-{>2Yw3|-H&(-luw4(kOcnPO5Z9?T5q zZ4%7ddeAAQ5eTgFBc1)`GeW7<@2a)NNI`Es4p=rl%XKoT&y=Sf0vDT}uNMQ)u39b| z%iyk)mdWQfjgf+g81E4uA}+)vm;%J$sd9rs(XB6#wxy{(SL>R~^Rwjp7a|vn1)u-w zA9(WgDRD^EJm-Asw=)Uo_5Fa0FgQnui5;`OM~RbFZsRt-?}%?t^5HrIwO>37ahz_9uZ0l5L#1!OH}CDiPl)}`zgz6U ziV!Jx@`&0?wa+Lh=kzqeam{i;?gK8smQga@E)mD zIv;Xd2OUlABV)f$%IIztX|+~F1pVOf&fx~f#Sr=8azp1MCzA>R1eN-drbWgi&ne3s zDFQY+1`nB%NTXBsSsT0}RVPIf=m=hrXL?sv+#dP;tMJX!dW-qqAA;?z7>WCISw_%5 z4mV0BWkV%WRmge2tBm^?QA%Nq9t9rK_j?!lFTZ(;);jg~g+QhRnK5bd8$Es$#t1PI zgQv(crqe0+9zJ4y_5v>y$9M0uT3%ta;)4$!^NYV<;e1FGVy?#^LBt2orfJDD!!Mp+ z(VZ^2IQIx6P?ca(O4^mj4p4=3ok=D5;8DqE4-ZCwC7X;2|F&jPD@H%<4ny6ZxgbnTK)Mn@#_LUKlr8Eub zMI`;`Jzi7lmy(c!P3pQ%xqGphaar0cDs-wiAUXljp14b&X)%W_WFqwSr+7d&l!_s2ge%YK1(hftOQDkb91)&`RyFN z#lHu95^zMYa3f>g+2{X6#-W8CIr06vfk6dpiv1yLCQ~vbz<%UC=+e92^dum#xXsU% z%Vc9m(80bSAi&VcKQ%c8=YAC>#@2Bp40BcS=~A;tTZ*oMLla}ik*OqG;x7C(hLeg} z3QA_(*|eTqPK9Sw=U0!h^(3v&%aR?_JhzsEGfMjsunOU$&Aa<=GMBEqrO((%a%&sr z8#%up@a7y-Dr_yZa$Z{umYrwLQJs4qK&--~KAth&qOWnBHJD4qzLcx1L72)GHe$Gg zC97V0fgDBkVIAaK!txTQx@+*3LEGDZ$IP>J6yJ{LPzw;9|M z=~;}x2qffFJoJK1v#pC9OV#N+$|-!a^`to3C`xYg!mIvmWcBwO_i79bv+0&Vqn^vB^h#rcU_|XB9b|ZJ;H?=n|XiG^zNs#WKTY_2_9c zOj*{XaE9+ADy8;%)4(ia$i^kCd)B98#x+7QI*Zca>DxDr8J(T^%?XOB%nV9HI=dy* z;L*%ny19usi!*+=^k0@RBSWa6qr4fI7_FjREd^X&(EL|*(%vyAfA?+&_WbWvw=Am0 zrgRa4PJ9kBaM@)$sPW9#^m>$BP_h(-{8ste!nalFi#%m0d1%zIlqIBm(Su;o`JYe2 zQJuL|)3RZxBR;N_gbyD9dyfanfp=fN(NX4q;Whyy`@Jw97a9FB({pHU@O9MHL6NCw zgvlmW{*<_U5!i)EZe%2G7h7L263URf+*Jr>Y(Y}^A*~Gcti5t#AgNlkMf18s;tEfa zrmShIm~6sAws)zwlYp_RmBmsl38J~q-Ni?8;Oj-~x!F)iF&xZNa#VcHce&ha5%VLR zqnmvKVUBM-`bc(s;{^d;GP|DYSw$1B%HcGaPd=k2og(a}{_pV2{5R?^+TX(;MzfWO zfWNY5BmHM6Qml-%5qBoKQ^>1H6*f;{s8dgEOVHiK*|NE%4x4v2T%U%c+;mfG8I>YOf)};6mF{!e~9J|0u zDWq(%bC9OGirXs>G)}U|QHX9IS8SaGEil`Y#aIebaTqJB;Hnkk60-+~e}1(ug}O2x zcK;rU-#3{pakL;)4TtMwEyQWqkYaq>7NjmM`K_i}^c(4OPlsDjNH!bDrX7Kj1mw%l z^;k#k)NilZ1mkb}W>OukhzkRC@m>nIZ zj5&^Ot5g&ICG%3K9?kP?QZ|w~F!nQ+r1h67JTa8!Q_F!5Ra5PzeZg)w4u}?n!L!rb z+di$C#5rbmMr2YHa1*il-UJ+D`tOSiW@W@pcP!6t&lP8lFMiKX2~RAly&vek`2}M5 zC2U)dEkFrl%dyGsHQ->z^zP=c5z8g0R@zXe&yj596);A2;6ca__r5o{?rR0G@0eX{S z&(qudxJCA@fIhigmX2BsP8;e?@ zl1l?eQVM9Gh? z?(cX;@G2ri6N<3x&UPKI+Awmr5226>Pc8UQ*(eDMFXF&*N%}}fYRgW$M$=RSI=f90 zFv;SEiosd`o1$6&cF;H|A-Jo(V!FYK@9)ttjNfLQY*{xXlnXS`c3LGs=QYR0DZKv5GNe{J1A`+EuN4QNe1k>g zc%W`e>arWhj))~;7ODUbufrf&4BMKE}*q=4fsgmpL94{ zFjZW`F0h2$zjd3g3a+_CZ*_&_5G(mZ!kryrEy_*dbE2`M%uQ&nBO2Out+RsUo}rB; z3;$an>U)|l>m*WX5(A&~&IbMf$lw}*6uwAld?zzDqu%ho|L;~{jA#1sL3CfOvje?p zYcKKAd#e16SEWhTP46~N$!et|f0<<(<~`s3XT0U9LU;Jx27c$DM>H#mNmu_OyPNtf z7C3d?shAjG8MLSPG>-k>YZ2j&P@15>8Yh$leZ~RwXYB{*`hK&I_dEw=e;(u*((f2g z^X0}ndRa#OG)UursrJ(o?%dfW@3n2;##zJ0a()Tq!vrj_X>^JCu#8>Kjfc{*zB<1_ zejGStHEwb*S_LaCi&xgOCp%kuWcpC2_6WhSQWcxOjddck#?#ra^#zeqy*Y@g(sg5Z zqWzQ)hE2*DFcna*{C%o}C`ct@53vaNEuWZEg17k?Zz01`_ruo)|E{t4hQBU5-UsLQ zCtP}S-D${q?iYyX`W(98{h!3e;g1gkAWUlLZyXj}G?s;hZ~}cLO)C~&QVBc6LXO&p z5uX`)GELZ;Y$s_AOg6cVOXr~*KlD~s*#nrl^D=YQV?QU3XrXTkyEJQ z(CYY%oA0Vxi?6N7x0)P+Tl5+mp6o%tQyr8@(28rAP3WTyS1?hN8RCh|>rTD7ghaF7 z%O+duC5#Am{>7<#5;g&cOk4-HfBJKq@&ikz|IWPkr>-@(+oMCE?{I~SoRaVCO~hKL znpxw~)P+T74=y}I$X)VeE6_gJ>1$=HZ=x=MAHaIF{qCRD&Cjj<%3*{~+Ku{70{>mR zs)t%%@iJ_wV6nw<+oWdqbHzpIq`eQND=3{Nqr#H1y{elN*Frl>bbW)orK5#Jp`R=F z3<%YPnwklZO0x5r>X}x1B7@!>1ij;!##+H)kY{N?iCV~*@)~*O$|D#W=@%+U2_r`0 zoJHij8Zedz@I{R#-q#+s+{0X9Ve5DFha>v^2hEmK3M>^+;eMv)t#n^2&kqI z)tp%h7M^YCuy`3DtB)ybe}mD)U)0Q5y+jlhlN4kf{=Qn*GkP_3iN#^fK;9f7)o)#Y9`=ayD8-rz8p@Q&yfj88?GdU}?YMXCLN{L5 zJ_%jX1^)SL`}56d`!eZ`aNOQ*Lxnsfdif6%a1gUA?mww)0+BVz=clUt?qi()gXJKN z9h#`@6L!N^{Rf~DcCZ-(2F-C1tl!x2Fa5M!>*8eY6EXQ;Tg1mZ!9ejM+fHFS;>8hX zl<9bs-)F<_8q6c@iVCrNca!ywsXTDxXl$_?^D8z3)=|yL;1SaMVL#3UV-%(qeX9h> zp9$+Pk!&wfb)?cPw^;$u+f%XKqhOv1U%aIa+D=7TR|8PPLdxjC4?(d+MBi#$;u$*l zVs9Vsu}hpu(-GUzLAPL5+?={nXnkVRa0%}!?qwNW%msQPbu_k+RrFelzEe9wQ%maIT8M~k5oiPbs7z$n~rHn#k{uy+Q zd#VPPCZ~*!IccJ7a|OQ{oF~e-8k-_d-Tq&YbopP?ILK$!aEhnfQq%Hh8UvdWw|6ui zN(_gYrW>mWuu{y`aw1YdSGoyFR%eX*8{Sa#FkrNXZUfm#3Lcu+skm3hA^EInv3k2A9tk6 zG>K(qn{bbAldQD&+%*+>l^X+gV%p&d=ypwXNzTMq2{e3A`&NFsq$ zm8o%!5+&WOHLp$G4(h3WU6{%1eN&CL#XI%oWmENYoGuBx`EM4Kf%8WNn2CBt@7yPs z_Fx~q$J22QCt32ZzFOF1uE7W4%E4#n3gvap-x35KutXhL$m7` z`xGOS?ic#C>v+&$f3ka{AG;NM3e-H5aa0Gl9F@6DI%JLNwjsuo!0b4SBE|jXNm^KP zV#4d>VwU+kb{3f-hbTin;NnOj6k>FOcgLz zJ@3-i1&10(%36rGbNEZttIvhRe1UX1sAeU(wJd?z25Rsa(31;SN}aqr^;j+5N}p<0 zYuf+$qntpAKDX6OBYqZ>2Za^aTgiRw)RJDPzefT(fdv{Qs8x-%B!7@6Mx$WqOEh=F zL+fqMh+@G`B^jiGM~<{ar<5NcoAe#mKsXU92otyW_dFABCxUroUj$FLH%jt5!ipDM z-zA#@T`j1)>MxJ@XvrkqE|0~ReXObmH+IoX7B#Gdkm0jpEX!J9&GU?mxS#d8-`cge z4^`Wa3vtzjrIJOKQh3t;?bR9PqRsU7^l?_@1Ul*mJEE7R_ccNF3fV%RDaDdcMgtrv z3p4x8Ec^{I;r)X1WdENWYL3|@i*qewlJMnf{H=*Dg|$EpxD%MSi-!t-d>hshJ>sg z1EXF08^>B-X>WwJPLq!Q?VeLOa{8cdb{(~aHKM!+4KFJ+vZ?eHcdi3ptm5fmRqM8J z!R>eA&%|B*&tzty1a~PCdq$kuw@ER1vRW=^EG)tOFTz&A6jdddevBGwUyzey^Ogh~X3QOGvSik=S8K1$4 zsiLI+p^hLRme=NjFsGU@8yl?}nL#rY5(DG@bgMADs$0yLL&NkmsMoGtM zYw-ZoVZB&DQKy+W1v48O-knt(!FZPIKLxojE!l(j5jt#JibdiJPD~yQ-tPXJnzQN- zYBm>~)fnK7Sj^^ml)&6!iM8p7T6IVji;DCFC`Zm3s}CS!mr_V-kfupV#Q<}#?Y8iq zM|C}3PIFq8k_bu1c?Wa+eL?VO2lk~P!T+6Uubkk{e>Wav4^-xC4N3^x#l}75_;1&q zjzewS^nu)xvOouDIhP|)nkH^wh>SYwA_c=S1jHI$DF~X&r9?QaL!;nB5QcO@W8!~Z zfa6BuJ^x)4as%|nJnjBB<&D3{A~e89lpfI?0{`najwwbGfEbeAUgp(W7`j9tMxp(` zYmiMO5$+=_eEZB3$Z9&~aD{{-z_+eQY)pbPtz@#2aPZj!7euB@=~Lm3#t1b=bA=kU z>CAypfdk9dgy*vfRHX|-iSO2RuAz{_*I7K2-0R3?qjg~b(EJeOQQulv|8k0%& z^9i02f{xyKkQEKR7OQMJGdy;*>%^|vc!Nu37tQ2HHWnnfj%Je`K?o?zl2Te@w5(2t zQ&5(+>==>}3tdSAQ~OeSQ}pdIHw?}$zC=A0J~#3UNiCbxOMD76ZT&7&U6CCDxZ84Yxh=gVRo zd~SPp`2rGe^vJ-v4j`bElMu@tQeh0Gt+x+7&%L#HxCaWzGOjJA`9#sj)&K{qcOD$>vfh=}f zmw?ZYWCxrI^$PFkNO$h}$8=P1jj}}a)f}XqJ*!$6 z@a85N(}T>4ExbIGRW;2eGxG7-4JuC#8yb-0MXgZz{{e0IOZV(0juRr@=?+p69yhodMV$@S3c2QNfHQ#~X|NE5c1WOBi4%~`qhiUUP zp8+15KK%2?mcLRT?fdi6Hw6r&Z%d`i8r*qRB%n^%vN0>tV$-z=MS~o3V#N5e0I-ZQ z8c=Stc*HxZTkIkPNGm%T;O^|UO9)4+{Ct4)ejw=1Or3+qkJME)57$T7tmb^snS97N zMdVscnO-n3B#D#Yn_Gm>XS9K4>=e#&2JJ;Q&qTX#xirf5nS8R`zTWuTY`N-k|6p^0 zlwPvFXos*^37y8lwT4Xz3M?h&>aWt4r5!RmgT4Uxp~6Fe{S)Odgk6WS0_zb@oJOV%9GDO0d<|)HJ%ugEYBL@jR#a=BeBs1vSpLrI9WIbw{C5v4!+UT$|qF4 zAn9L}S;+{s^Ah`abhOSM1_dQ&P?fT0YQZ#dNY=4UtS*Lnf1e+|Pg6lPR~`B7P^1ml zT0dq&aVQpYRFDQI5#Wtly3QPth(9NB={4p0k6He^FmhpOvCBt%%#q`=S29#gR9$>d zKlT}}g6rwho^L@C`}wGaNaoU|y@VnN6ShAoIcyXVFt#hKN7ek8sp=l!yFCtQCh}Z} z#f7DmdcVkYvwP+JKq3-WjnIjJv7oyG)fl0%;}-$+0I{;=FN z;Qg+@%sxk@Mbg>R@cz|H%%Hsz#;rlXyZ30)TpZMn1Wvz$yA8}|t-|8rJ4-Qa6)Yfu{Q8kCDy&!m8IZGa zza|i~R6`SeI6;K7Z-XJGLOfY_Ti;JVE;ewvB% z4xF-G|D7f3(=%z>&en6Sz_%Ya#6c2?bAb*+3;rBr}z zNq!vuiI6Q{Z9wL=<U#M=kk_g%s{eE^ETW63-I(n^j`cv7a$fsu$yQB?w@Ym{N2EERY4U zPWFY?3{sO4GtT*vJI&UNbKm9Qzo-&)R8+;&?AUqchd%J7+FSX;yxuab1j?1h=ae(CMA1K-i~P7K+3AB08vt`&M&NLlmm&E zXgV%kF|L>h8ZVtJ7inQCA6}0FQr@GkQ)q9x`e&>c&e4Ub~?a{o(q_)d1XH@BO+`K z0{=G@)X*QgY1)?O(9)LIu*-|+z3$q?XSL>2YV)84)+~i{u%`r$!68YTt*}(mAe_1ZH0jp28j-PgrDPLgh64D|iOjlCcwfqq9 z01DxvuCACR!5}S(b~356&G${vw3&LRn1u8i%rXyczd)cQEI2O1znA)q58(o%d%Fc= z0~w(GVF$H|#8_k!crk@+nJKM?YT467UXd%GM&-dp?V(mVr{!f(1~JkHin*lo`2RPP zmI@vgKKWB6k?V%}6I`)(*D8L9mOTJH(z_54MGNCZ@rPF$jyeq%5R*bWMJFdV&IfGv zNyQw=QuLiX&T+G4lHdYdI}hv{_nmZA@FcSUL<9;QzRF))yoQYYeSV#VmxNiGk3zAs zTYym8A}F;lJi|bVEanDqMWD-7#u&2S-?>pO4AJ++Y3T{J{pj`eW?c{OEKl;?%_Ub! z%rs!i1NR>->B3-qkr05zQi#!;A$k1QYB2v&veJzL%|q;3O!iVWR^EeE*4YfJ)Ynn_ zd%u?cP>oexwp)EwdUQKY6L7zms3i0^GXST-=vkj}Fyhf}Q7XxtZ^w2J&Ga6DKo0tH0Q9P`PU;*UK&)P(Q8H9341G#(w{(dY1%eg+1WH>gu;! z?|XR$+thpuw^!0eL=;M<7q!OJ{k|L!=NfNursGB!HmmI+sFj0tSvTwvFQA55c2roy zlWE4l1uq;hQOU*jK7mXjl2sXe|Shiw2$!M>Q;g69v) zF6)Zm(`k1G!5-~8xLFmm@ITFP-s!~}zsrbcPh0GB>ckPvlQ4?PiF{y#vSxXB`3_j7 zwCgo%E;-E&e{xAsi2me2uZI*1W(EJnLE}VfU@2L4zMVPq-MD-JFHAFmt;E_U;33DTxe8xXLE0EoiW=-YG5FMzg(WUnqBXB3uG)~*lFnBm z+OSC4BWejnCk4%)nBzw1&ZJXD1IHSJnMMISm@;= zW`Fokg?SuE_nVL90@)7^c`7v*#`8bOGU1n-e1GSR^zVoRT?aJ;QPH?1U0OzWEd2%X3I*Iy0?37Igt!zp zS&uY^VnKyaydP{p<8jIzfd2ZCgu{wD6>Ki42CjlZ8CfzxwQ;rY{^I*oh6nN*6FeFg z@~rTA@@dG7#I^;~aK8k)&TX9Leo3foYyZ9f%S+Y1j8YA2a5#Q+FF?SOF-VPA^y<4E zV{QI@pOxWJ5oat=zcBPLy+Rmj3(nD%3w5GMI!P{V&skSF067^d+1-OhkickjlzEyi zYE+0Sp_xn-Y)HYA&0j`7tSskKD$)vu*B8h}Rb;}xySJ(NyT$J9LXWzqS6Cc&?m>$i zRT_#6o`ONuE4z&Hdw3tFx+)7T9m#U^Gc^8s*%_M_+w}Bo2%^)BasDC~Qu8F7H~K^^ zIw7gQTrP#(X?E>8E0Zmo|{T9Zia0ERR zW#+7Dom%4&JUz>HjrP%ZMW$d&klA;|-7tgo_C{}zlJ&BHHGU7sEdV>@z$DR+VDl}1 zQdiSY|5q8}2pxV>`o4{JZKJigPz;S#(a_=ky(hFH%}IRBV`ypylB*Uv6R z(gLK8?qt7@%fRC^8%E}q0NaGhwk*U)JBy?Xot!YrO586rtGl`CL5-FhAp_SWV6Lk} zl@eWKpY#P>sMt)87>^)O8q?n-XcC-e1+rH%jE@q4@Nbmhr5@n-b#oI7F+JlGGS@hJ zIZ6v4@%*R-KN(DK1e@OotaYKS$4CIIafjNP)7k0Xf(Pd_q(o4ar{U{+g9~5}zl&=RdyA!y;2c&| zGfFCCi0Obp)-@~3Jk%y z5bmcA;YLJq=y?zIdV`ITw)(7{J`MplR;{UPh7mZSF^EW6&}#S%Es~sf(G$@QF(mUMyphJH ztx4&*cN&}QIZ&HJej9i~%I|8DG?>LX!IibC`Tp8g)ChfwuFeIkSdlW4M5dZd5q}mF z+81s`V+Aj#hOoKHt{m(~(x5vq4O5ALvZa$i{5b;iwEtB z*0VpRILm^D_y^Z9fi4w&yTc=_Dszc1q8Xc6Ad8iQZ3tUU=rc>nRDSAut#5Xy`Zko+ zinX`I`nMTZdFEV{`3y-9Fe+K0g(mOE__HrIud8wS_0V~D1HvEdEjqfD!Ea%@ ze_|EKZI2&iNVRh5hUzGYwO1Ymqb8vEBBQj+m2@Jg^G|QXb_49?79^(?C4sAbbl7Ip z*r?#t=g@}4e~m0w?7j;ME1b9xsYE^8=$E?z5ELeC@_t`*wyvFu2<%SYHD4;+IzU}s zalC65)KU?H_)H?3?FfW;0)W1`&QnvztGS}4-G8ni24ghbSm#Im&qKTT*vq!+S@KM~ zn1<0LD)gAGP_<314Ss0Gqz=Fhet86W@>Nrj4!e$I>mw@RiQ?l~6(q~Zp*Mc!F`UGG zvbI)MN=#SKSZ!ioi#7T)M|X1KIDxEUUE?VhIi#E_iJcjDPYP{1h4C6UF42-wq*>$a z=Xrc|62^F`*oT34Xex{4%#a~6$qqT2C zElfg(JMFJ>%3^N-6;!P&tZ`QC_;TyuqigORqA#zKQJ-u{v2q};HMFkUpX`mE!MA}E zug@_?2VhrN4g3vgXWTaKC4Q9ymN4@y5iKcE!g|>tFp&-w3)T!~DeXuPzhi;IPbm*g z@P^=BSjJ$Nn1$%RKF=Hwi%Ir)+S;`SRl;tD5quPTf!!T*>S}N=DZ3zAt%ch3E#W4g zT0nPluVjlQ9u60~5I_L)>&n5ddDW!nm-URxAl$FE+VdYLF|2%X1eW&+92ATbowp5Z9Kc%#W;n&kML6kxP&q+g##7N5r0uu_ssR?6`uS)%%l;t=2}RT zunt~|Xu858PuRXJ)k^|O^NOIfK~n2HigWuU+d0Rcp7FUr^MVw*SjM*QIvu=rUSsq(HV^)a zPVJkGmF5{7{Ka+#7BiM-+DN6^B{~=t^9Y#fwwXnV0_I58&Quq-6e{ECk&zh{@WJC( z;nfj)1LhZEOZM(HXVvu3jvpfr0iwcaexGsNyts2-PCSRMyS^B;rrE3WJW*e`mL3ZM z+$0(-l2ame%t#IBkf~SgyNkio`doJf4rECj(u-b{Q;+me~NJ*HknQ*CAk=*$Xe9x3~rX> zOt^e#y|Y;g<^U;?-skhGVuBVd_z%-+jB`(_eLml$i?cg#1!x#oVX0c-z#oX*w1m%U z;N~)taWH~e5s+kQ&3{0=kfvDL#oWMVf`73!Bqz}Eh$nsm;p`8AgR~r6ES?3F{Jw!$ zpWi)JnR5FOx*Hs+`BOw5X31B4i|?9uzR?xP{Aow!EOlKxs-haK&rN&i_wI@+|9fW) zMY6i>A%7KjFSG{KB@)Ocep8M>89CwX!jN1fJhnhSn0v=L3S(a3xG# z($2owTkX^cZ}Od6kJmT<+NAL|JFU?R#4Eb&`j__-8CJY03K+I1C-giNh=bC_$n}6@ zi^hG17FAzwyAs561sz3BS<4~>57CnNLYBST3ND2vG=V~jx5UiA$!$YU@9z0Jd|0R?}?spT(%^=0z6b_RXMTdY-9 zoxLsb*LWO+eI{0R6c#DUGgeFJ$g<}u{X<^jy;mU1rrVAuaCGrWtc5O|)l)}g3wsK! z5Sxb~@Jt)Ve$eh z{RqxK?BvZM(bV0;SkvPKkls}_5Zr>k6QfC{^Ta1wlN6)Mk`{}IJZMw#P!P`pjd^Q1 zJ&jLu8bt%loNG%kr$!wpG{I15-CwuRDqCzhrBz|?d88%_Pr$16Qp#ytpK!MF`RYg# zM`1;xzRkt*-Qz@D4Y8?(ekJGNI*F(#{o?) zW(vTgqvC%O6`$v>@~vJbw>ERcpr6e0T6HJnxt9lf=PTpWhV5mbOB zwcf(MMD!XAS{2=Twwfl1!acGkL5^n6MLG1$9#J35y|7yCR0eH5Q6CD-0=Qt{{Ea&W zuF#5Khk&W|+ss}Z?9)@(DE9b$x5pW8WB8st=Z}=B`sTGIs1mxa4`YnU9K}qhk1dis z1^-!n9g={-QS&A|<8CzmQp@ z04*m!2-cNqkvJ>9Ki50_?`s?mmm~R`xS-1M&B z^-tEZz-CYD0f}smOz{d9LvZtOV5)K46-E z@aso}?GhONIUc#_dKYxmn$G)XT_bZ)xU>R#VX+_X{WK+tS6atAGUGcj(-`pW<3cB%g@Tvo$7sWRmI+@oz^tt&@tu_>OEdDJpfuY- zCy+~bbYCbt{mv@1lKwV)co;5&*+5J(`|vQu$+A`w05CU2-2YoDTZ|EB&)9~f^BN#r zdch}o)Ya19Df(7y`de??ksSDs;<5%of_T@Mg>p-9DxnudXnyH5#IPT3XoI~Mp*g}p zQwh${;v1mv|L}8V<-C zsyfv+HZUW@x}2}C-SbnEKAA&WHv~1mBx7;TBpw+B&`Lp1c&#$1$+g!kd`oK>cr1hx zE9n!JKQKTqkFc_)3k>-4Q`+$#-yCLsN=Dl}zi-8N;mLt*vWwlr#b&%LWK^*q2e>*- z5kn-K(Vw5;yB}i^Kp=4~Dh*NFedr~>f8XgJp&98uP=d3!Z%F^W)?c=p4HYZbP2 ztB^xGL?GU8F9Z<8V`Bw@p1&xCc;V%Zx-&CVV%iT+J@&69@4tM%@j{W}0m5-O-?5nc zpMK`Inf-RrZ}~V%@JqfFM>+^tBgPa1Wb-vV9SFFq?b~b0V2$t*aCo!-arMuFEFu@6>DF8>C5%?pgS*V^o9s?%)+M?H;F{C27Hf3-J? zx_%_eb@K{;h(=h>o*Frq$f|PKOV$^UMWu6=e>QIW+UiAm%Y+cx)51-B|WQPX3 zCh8nhc84OcDO5Y}(5m(acY+*SuEn{2Z<`@{q#h*CE!WxYzWVX)_;xjQM zc`v0SK}{JJWUM@@EFH)42O%0M*lvan{UL1w5-n@|7*=Sv5lv>Qcyt>Q&1CIT`JD8) znsHZt@#kmh*l~=WX1gag2g6P=ddlXi{Y}k;>PCyi5`Ex>N-AYa)p12d<}`woB!FlX za|SIcexj`nP0U#vb6bdyA|4~$dZ?w6M~JV`Eq*4TKmw54O0Y_a-zzj{*c;u-nPB01 zZ-^=rZzg+5pptw0@nj2Yo3(-zl_)=M8jhc%zqvlbfs|2qMUrL_Z{E;5S;tO5NT|XM zy`tNKjktulrCFivObcR_cX_75fHze>MDuYS@X{ntwJ_w$EGw-6X)2BLD{e zY|IZx_xKJ(t^E3rNjZZ(6>wi5`9^rj6n1(WXN+p?ma*WzNc$*h~=izb=5~EL6lcH_{mN?B=$o4xZ^<0Cu6}PBT;538oZ&y3Vu6 zvRJzlqG{?-aVl9nS+dMw7=_j^!#Ib2TXB;$%|-$}p)!-d=~I(bG2L9SfwHhnq{>@2 zaYV}lbA8iSmPRL~H~4`$u#uX{aT+f#zyrkK<^@Ay)<2gy5|dbVtR+@T%77O5EAND- zARO8yCojLD>(5|sLJzxKa*b$#1p)t0w!jnNw*>F!;$7)CtAy)X`!Zh?rx=p#>mOlV zdM)Jh3pRJesN{qD7-Wf2d!oAIXHEsoADi__fUjNs`%&ipS?T;u3Psne5))tFoS+vk z+sycca|q|v;*3!%T8O)3Wr1AGZ3^RU$INNjWEqw=NNtCg+>)dp4V9U*^Z}{o=Wu zz4CyM2=C9C?vEv-H~ZMu)Y{SaLSougp9u8A(gB{i8atUokyy4Wv7OzuXXFXf$(d^~ zjTIMrcJqH7M-9$LI%7^OKTJEoU^m;0eO-&INn3l|Uzl`hqnmW01nnvm;!;am!!N~0 zUAcWqtj~P!o;5B)x+P?97cQd@D-ZEeDDCZ&R2eYR_RFfw3r9R{pmB;7rPc9EhW zm_LX4iTz+}p5mJbqdLectN&~k&>r`^Q8HMf&O#+ADdGyZ)*x~FeXO9syf_jyf-|Sc zEdF(ymhAm?8|$D0V9eC+$Zcug$&@<% zw5GEmoqRByz=o%Z!A03h2l?EM7S2#Z>2+tU3lrV#4Gw?Hz)Bp!BQ4CFS>N6?!J&o) zWp1R)1Hya3Dx;ma&ftXn8X_W1l6lB4k1bGJd6IRj8=~0125+vZiXlM`8 z^{c+^Vu1e=SoQM5xB2t+dB>)lnr^rakay~;XL>jI`Oqg_4kB!SEw+CrU?U)fbjEb= z@&Yr_;l754t{P?=^K6ygmkzSZ7$jsK@@$P)*~FmrjW*DpEG)$a*UguarU1xmaQD33 zW_EE7s$6ZJ&dNP7GeoIq)sq~+dI}8X#VYzXf<{9{5Z@ z126p-c$Ic~kz1<R zn=93JcB!vEip;AO$colA;xdxCW*n>czb9j14$pBbeY0Fw9 zyOdSks?~Wp3<0NyP2`6+j^!!$=`>yVMAB}f_V88hmPISVh2{O|mD+tTLuY@_2}Py* z$?|tI-Nr0i!>xi8>ruB=7jx5;ygY-J)XcWB8al-?3?d<)$i}i1Ri~CVFc8hwH-8OX zs1qG5w)}Pv(=Uv^3gB70{kQ2PJ3p)UqB6hIWueWr4f4H*&SibcfFJ(y+$i6WO`Lmt z7Tm23Bb0RsfULEqQlMMPY4xDj2NY8TLjBLt?1O3}($Hm~NS_^xE;Z-<_QnL!Hs9Tu zyYjK3)B`65rC2kYrF;E2c{|%|Msv9crDSv2slpQYF0i`ss68Ow?GaiNuE!fN#5kwq zep}Fr?SifA?ay;OUXT}V(7k8(@%Hae78=sAyH1Ni1C<#wQ9sk=%W+HW=q4}! zf`VTVi3#asdr{Q4#5`qD8Qh*Fa2APOEf*w&^N2JS*9lCaNs02qaAxY_O<4^02 z%>TXOI6^^G;+EhlEbWz)pgFM-u~zCi7y>1=rO`+2%;3P+ft%*V87>_4XHP!co&DqZ z3Gp(AnO_+cL4It-6g=*^dG!9^AgiwQveY~uYVEK4{O1m1eGR)uws3%ytwQ8`PWXb? z@{XhQ(kh3&cA+&b>QKrkYtcF6Mk|>L`o8G_4+Pp_o+K<_Lvww#P@U-htj>32luVpa zGs_CN^QLzT#n?6GnGHv(a|e70~?-ol-N48@4rwRBj=ze#f_$gKCw|%)dlG z?)%gxE!tRnyaaqO_THVD?2>LOyeYn|PV4w)f49O1MuFsR57y&`o!qBhzuZGk|DQer z;l@tUj}2mq_oMO10{@_te^>n@-=Un7n|;4D*_g7^QLYM0UT#$lt>P7tO*W)!zXG|^ z0iEy!QnIYm9Ao@=OY?P6G4!SLHj>T#4wn*IQycQ3b~%)(BqSsYK|p#CQisPibjS>2 z)gAJ``;y~@$=r~XVCQ4p?~Jp)W{c#q%}xrn*B6D9sv@>}Zj>7~-b24k6+lf-+u*qLXZ#mQ%Nf!%zp{__@msw65ID2$g6Rg#2 zz?jXyro(L}+xQvpr+3h9vWyynRgGR00vP0}OUJtCcpJl{czTJOE<^dq@ zQkxK89hhjB`%H@MFO5Tn17kG<#%?% zG^Udn9FUwS7sg3si`2};a&KS8I&KcAfl{fLzQ!CD7~H}JAt{*i3h)l%L@cvPQ;#nL zQz+ofS0G<=RA0*|_Pg3BM;q9;lHX{XE@RiXMf1kimdgpJ?-eo}+&L77 z2FT`*=1yChIm=$|4gQ5_VkpkmBGjuUP2fyf-K)9$G50f3ovxu**0JK;X5xUtjKdCm zwWy`g%Eso7!dssjoc2n=|$Z4aeQ&fd?=M@kSJrZIz>cD*Eep2q zKd}41=0(JR2+roAF9pX(Qo;@WuofsoLZHilTzjp4X2h0c!H?Bhval_lVuscfq$d`_`s@Dy+CU}0?rDjXB8?QPn@3z*;Z1{W zD|}an*vli_QGvCNLe){zc={lQT=cGID*D^TWg0*WK-aqckcUj=|T%^FD z1Po$Du069x;+?^{0J1_T?v1CkUUFl1lNYbuVh~CuF-T@biPfGo)*OyYq%mL&_l^sS zs^Gbs*KtyEIH^dY1SdjmOgJYEs9kKzwxc*;qeG5C5aTY zoz9<^f@#@sd25sDY)Yy$S&|~9U{;q{9~RgrlNr;pWVpS_&Tz;FuV1IEYW7AWO6voG zhE@n4?n_mcZ1#I>Z}yp$HJ3MX91(}}Io1Zvv;D(EVy$S6p{-i_{XSU~QC5LHG)gsX z+hS}3K0;a#jRo(}-XLO)ja(4@?uaL3lJe-_m|0y>wSo3LR(jQYPA97_>^+#ob&udV z88jif8Zy6xSuRo1f(~&Yy8ak`jB`tvwzZbDzsa!QV^-c>I+#KGPjqziqSNwQ6D*fO z2wo@|wv{qS>~PLl6vyZv`D3*I;#X*X>GR+m=GE6x*KeS%UBQe-=#PJt^gsHO#KSGx z*WRG`$NvX*?;hEo`bk6*!_zy6{R#8G|Bo=g{bfYs@PGQ>Uml{ZOHS^5naj664D%}# zhhuEh&=wMJEy~q=`1zaM*d7vT$@jP(zx&pRZ{9r$ZSS5}FCGuHc|zUTyWS@yk%Y|7 z(gE)3>x7gD48nWn^Max+h`r%sPi^tc)oqm0^oARxN>DmQ+~39}mVUa$tU6&@HO%HS z_Nj>@P1UrF3rp)gSBDA5u4ZdfQB54Gf;5tp0~oduk#s>vkWMZ@1TcUSf|uXg=cF*a z{OX%bs+OlWHFZvp1@jG9T8t02w4?fOS{XIGs+>G-xH5RW*^4M3E*-69!oV zLUr?lrI#klER--c^}OcVWz|V47@QLnRf#bc>m`RrN6e>lq!bJ`2Hd)|%|<_EIPB9l z4N}0oY)K+b5=E4a33Y?r^_N0)sG-vXj?gh?J^^6A)co?5Kf!U&(8w9O3^F9UW|cuB zPl>W1ZzHN2*%(|JCK4s!W)hJL$%nJzF{kfCdBP_L8TXo&KuyuS4J&hFbU*hLje2Jv9&9YT86 z_4%G?i_?|@VO?Mj^b(v!C`Hv6nxbLv!4YrWJ3x3xnk2k<^D3Kp!pUsLQ@dMK)^Rc~ z=%pDFPdT4a8OMA+<=*IsE5jjM{XR`&nD~}FOE{WTOr|rodKp{&oRe}+tRt$Xr6_8U zlA@`RO0n6`$s$eFIx;PJ>v%#hjd79WWHuv7a^_9R<(&=kEae;b9&&j&WS|wV-rpmN z6k4SeriM7;=8fy@ITxoYxnLk9*wcy;hZJP`bZVBX=T~q+lD++U=$y^{uKM8 zW8U09VA#(&IyyjDNAJ?*aDTjDW3a*KWW=;7LUuz$SmDF9J@s5l(j+6#5_*v$mXgQ^ zXRrny9*n8minb|nZh$csp=5wJJCBzv2}I7CTf(_7rn_oBQp+DH*M?U*+aB_0Y?|$hk{N7hzrl?Mq*IIXr{?uH{7?TB=^y_wihuaO6Mf)CvVZ4Cnf#rD@A=?x3)g9lJMgvt?8 z;@TRd=4dpcnw^lvn!`!ZTw7e9NU($edSgxrV?*foW^L|Sw*{dKgz#Xy9?$E{v>^!P zq-3L~88~RFgje3$=cOwVAec-?42PQxZwi!3i0y>JwM3?*wT{`mq^}G@WW2R^h_jBe zY$28G%9uP79Qg*MV}@Oa`)^ zK*H--P^QT)un=jrESI>mcaJzp5JHefF;})WxU|*h@n7zD^D=}HLT?(Qc4{Q6Jgyr=!16=B=D6Ho%pVxNnkMz6p8d}=Hw z*79BbgkPH|Mr%QWpV;1DZXO@FK9D8+#LkdSAMvKG_{Ebc_d1=3AKeOd{r8)e&utBf zo#Ho&DZf3LQ!h1d7op;Co(nRjxPt+kCHf{v$uEfHjO(N){Ds8Bx?MPZ7j|L)pV^Z_ z{iomGj2r&^OK+UnMyzf7{%oCy{T$a_i;y{)-QNbJ^zgRc{EuyVLFN9IADPZ{3*`tR_Un`Ocp6r}t@`#Cx;S z>vLfd;k<{oTg2J5Osrp!p9HNxb4IhuI)`2psrTOFt-~ly?1Mg^77F3O3eJ@F34{YL zf^)FY9rM<5T(|uGyy1p7Tsa=|;L(Ix>9{`FMy4@0uIw-x&5=>Y(KOUeDyiuWax!I+ zD(3b3Z(+UX*3KoiH@Dc_*yPTg2jo2krs3ZHBaWxX^rDDv4PGdH+jDv^!OeRx4aP=y&lC%gi z8KI?M)HXbS<2q5Ksb&RxhsVrMM(p%*uI%n`QZx+fmRV5{3D4ot0bY7koN_##aWWcn z^V$`5HaE${9jfU#FrIqP$#lZ?%U4i6McY8XzXgS0|KtE~46#m-Xrd@0N@F6KArK_0 zLV61#Ce9L)IEb##rX|Odqu_X0M^hVw&?vwtg@&aJz_Qb%*Fv-P&@3DPL!K-h*P4PRg@a&DNeCh{24N=P1zwvdXlr-gx zs%g7Sm_%sRWf4~5(d*-=v6jZ#Q^6rbU=H=>H2Ic0JX(H!^3y+q8;!!l{K4nYH?JeQ zOiJ;_mnr}0&%rrk)$$PDBTppmuB0pqS{ET)fYG^XMq?!7$qX_{9`$(Z(TsZsQ?giw zdRGy2T(#7s3bZXL_wR8yt1)ZsmRh0K&P{2#%(h17O=+D4EmOgGGDVn@>z6lq`5O

98Ws@#wgul9D6m*xiWf zJ)QDk-ViBE^vIE2O1QM`(LgT(9e2-~hf(EeZB5;zw8qiTBJwom!Khe{|BF5+rKB;I z##oHC9E|5=Ny=+)?yET z4`(#Sare07l?OAXttD9yv2>68qQ_|+f*Y|8b(Uovh>Ef}BGr<{2rBQ1A|22cQsSk? zM>$2&lGKJ?hxRx;DrleHrAT{EzImLgDmd2)=Xqn^a8so`m&SatgK%zY$)|=Hzj!j? zQ-gfD{YzQG-+cZWua_;yjpfJlO+Gcq`5*QUc(ttg%%IPQvy^e=dBqiMrh>n?wTl9N zbu{bli(O{<=XFaGL#)GCptQy|?XssD7c#XA7=3_wUcfEZ?ZWB1unYS`HUT8^46J6Y zSZ?|9tfTfmQ-JGimZy%xLIU9l1TW{b%XhCuke((?6>@!(PH^0#gvUFBRvPEbvi2!B zZmaW7+qK!>dln~_^S9(2)*@V5r18!{g zDSQi7b93hsbyH)UWiZ^JoXwe46{UAPyS;@l}Pmb`;6DKig6hl`hrXwB5EJ&99X0gy0 zBCv*LiNbind^%@VR~T!Uwhc{diDHfPAgl!)WLLcN_%pR>>-$}Q?^NIT4rFohgiPO3 zZgcG!wa(ERgNQUC2GczwE-cPHbXmC&k8kEp+##c8%vs|Jayw5!{Ly0 z@0+yNb}(2$X*|=qrnZi^kLLXC|LN~jG=GnxHXM&eXe|S(pu1>N$|Z_Jd54Vjs-BuK zK1!|d)*`fqCOF_<`{r%_#_hMzQc$&l;kY#htu_6@0Ow$~|0er&O;uIIaY7zPWICdk zXoh)?Q!%-gq~0(WP)bSBgmbNrYPM>JvoN!c7i)pMEZFHK3}R2L1yTkQlSD&%+^{!p zaNd!o5m~ILt;g7Ij27U83^j(<;=Lp3>L@R5^cf5W%xpw+GG$(d8p?z5oG-s|pHF_| zd8UUGYGVU@U#LJxUlyL~f>_^Y=r^WFWF z-$L>7O)_|SS%ni|MBl*I3#_?dlCti4pm&iFfen4p*gTaupw(L(EN~L*J zPT0))2%TcQ;?f``i9AUvxp{4a=N}D`X~t7GFLN@l_{!~v43fZXILLYk=P2fLjt=)J zn}*g|@;Cu07|o_k=W~)IAx#qo{TL8TX9bNj6xQ?1jcYu0mR$h!&E~G^1==vMeJ@5^Pm*G#!&BDR=JQ=hp5HTm6h^X9uA* zQ5Ge_c=}n4Dds#pJi>X8RuM^@5a~D&R8$Q~lu$MWsm^GdmOKuTZ>3eBK93`WlBB6d zYeiqiL{cz}Jbh{K#^JSMKAlolHQqZW4hjj04`R-0k!iZ8XFBmB z)BT-B=Hk3~BItSL1c!RyxXX+*RuOgeWI?ktWZ{$wu)IZvpz$F9MOmfqM zAv?6naljx=$TH1wEvcNK=l$|JG}a+SfNvg78jOPxhJ)GO^7~pz6rL#7OPDJJq7dOd zD$>LPtnf75XV!x^mMGF8tG;H&I;s|C#u9hX*|IFz7z`+ymfCf-m~o#x5xA(2Fcsc+ zP)6yIu}0@H+8V5Lq@p_}a5QVtfQl4%wB@O$#x@=s<&5Vobu(w9mvY<~?jKk5(wIS> zU^)VnScLk!gKh!|7 zR;{C|Tk5){Zd>qfsUt~O!uIq|%1$Pk7BweROTV?)hlYH68yM5x-J$=iWC|b>rvLyT z07*naR5#M>H-=&@DDrF)Q#i+?_8fAdGj^wLm>S22@{HYR#>iOy?Cv)I= zemLg0#&cdsW3&`Jm&OPu*itLk<4(J(4Xm8!$XK?uW}qZB-TK0rZcJCGZmqF^bb11L zjwd+oi#3h24yj}(rE_5myKwq0?85#4cAg}_leT+*(R+U?+XrxbJmR(2ZZD4ZzA}B^ z|NYmCtmJLl?*wf>y%g0_sADj_Bchhj3l5$BVA; zTgc`2J?ma*Xq$B&>m|4r>lRVcfVZ@Ow;r$^6|@hv4r^YSW#vFF=27*e)vF76g9SOv zBBv>RkOv5nSrPaP*9c>r^SoU)eA5a(_lXzy;iqr%@SFP_jVD-Rh_s-p9CcOVQ^mn( z%%wrjM&4sw22G~jeuk2Q?VTQ<_~3Iyal~M&&;6qbO;b~Q#mRii2cCI`+qd7OvMov} zHu`;vs-|fhR2(s#%_z%)wyx+8HjqNlmjAOkSa~3Bl5uz*EAH<33hvnVZYDz<^Tj-+v!uy=Y0MC0V+*#^MYAh z^Y+mpSGTt)t;M=7^Fx3H8gBzRs%x1}rzoK@wnZw1>h#+dP+&=fL{e3TeqSM!Bui8B z$T1KV&V?G!QBj1x;2e|bn0YZ|s1;sIycdC`RS3kQk1mkLiwELe#tIKS#)v8_qWg8*Vue6@k1-}lH!Sw;9ZCqwlLKn=G)_X()RdDcOy^`T zl#KU}8BZrP^OCa)XI=!D*y5IG8S2)8R7AH?-C+O4uabWD2eJ3=<7QLhkADwAmz(C(#-dE-aV53AJMj{Du$`p#+Is)K&!+Q|Cc+&^sl-DuwK_j*Dxe zWf<$KE<>j!%!&f%J$agg=#jWCF)lT&tr)}sNz$|y>pUrtYr(V$_k0l_a|pZ$kiJGM z4(66igPfoE;ZLyl=$Nm*w#QHX&`0>~uixRT_uoK7QP3pR;kgtlFmcKxCyiPvR&}E5 zJyA<|<+ay%bUfmtFTKEm=~;-7VOx-hwAP`O!a0X=gzTsA^!i8KIo{&`{Ni1{ac{)^ zlR2}hWoOvq+GdYgS%&~dfCPlhTnAB1q@a?DEDh|kvm(GLwN^{rN+kr5mVElfOWZpc z1y)Wk5Vj>v53$qEk#m+$4l@4o&CC3qy+eLw6#CwVlKd|(-s0CLGyc0fkAPK3`AJaE zugpt6GwktPnlRLoObLE#GG|+#p7=iA&-f26Z<7i^iDMuogXn!aVa}qp3PIy;P2f7% z=6VUCAk@r5H?MXX|FiYI-|2SY^j+A6{Q>M$A^h5+1Quw_Rpw{uponhKzl`)Rpn&Uy zC}CP#WS<;@S6{o$Klvv=%ZESm!6zft&l%u;?&p4yfA-J*;W>?8Wvdp@&iA&=d_3EC z-WGDLG|%~fu3u3=FB}@*ef}Pg?ERV!&+7Uwudfixx~U}v`wdi?*#zLLprE z*lf3SBtyl2aKyU5;$7Mv^uW$J7rMiQq;7c6t{lCkG=+7v#xZFddY8BP>@&|Ys0?3t zbdOXEdU+oy6;(SU6CSMY}?(*n6LS_Fij!-}es%qa`;`2 zw}7Y$tS!llf*=SHLNd$>dfhgSb_ZlNp4!uRbLu0`;53&&!i=#cyQExb9u7sEO+yA`J6$D$Y$_@a<=exdK*M3@N|)9Ql|2nE0nYiQU{8;#Z_M*D7{5T1t@D1kBl?x<{y9+_W$ zzC&G~TZgwHXvQHY*Om#DpxJD&vUUQc$2cTuQE_K`!0!HlNnT>A%6Ffb+d3HQuWKU& zLEV{hwnC}^gv1F&5C%kHNS+K3QnKD{V5Gzufs%?i4Cqvbaam(vL~XHQomUMt25X?d zyuyWZCpo#&!5YKsclPOYI)1KHic0(IA_xLTxgnI6W~Asf0ygXWPYQt*Q|~*!pR6>r znla8f(%k!bmX#*SOO${{s4&jaY&5XOAOeLFg0w8L7RsVUU~X~qM#|-_l(O;(@Wm6$ zoLz16r!QWwMRF=?OeP2qH&3-eYhS#ms%n8Z=+A?W~d?Vt?W|Ke&UdS?D%Zhsd&X=>CTSdi3du<+T zH~8bdlq-43N@rd!2?YG@V=FiZzkF+x>qW(*%?7`8=F~mc`CCQhAehd-)brwkNZ#Es zE$mwBym;DQl3(EWN*pD9M{yMIA`W!=4yoYJ@;q_ZgJSVdH-GJcx6cAw?e-7G<4*5#YCe}r`hpnM07?g3FNjR9C!TZYZM7}M9=`tug`LhKVuE9kA1uU~nC zD$D3MdsMn2OL8h}86_!^HZ1pIu$D=hv(kxaN8WVW2#1WCFrHxZ3N15wO5kk8Xq3}y zM%>JEPOhJz-|H|KO$eo8snGpSd6IE9W6*XF90nTKE$lcc<_FE zyM>VNm6%DU)Xn^PQO5sbtgZdId@)QcSfWU3`w%G*&Wo{BX8wFp+iFxrj@`OKJ8rXN za}p^qPBI$qG8vDFRm`3?qDUUSKhH;vsq*uV~H6y}7fN}>Rjl%g3{&b!xtCA>cV4Wi?ebVP}G@=l&G^=U` zceU2@C!O->g+M}C^yoHka%xHN=eMozl3GWo94jk*!tnNNzMk3*NGa&VA=X)vH1pSR zsVOSs)k%%#G|vl|j?79*D_BHWX1V7(FUIY%B9ekLam2Nv7aGy5L%0~R6fA|3?aaUTfo8$e6~yeH!*={t&}_-hDTCjqmCJ=UWX#oLhXKWuy8MI+vH zK&w}pZ`P5!Tgc>|=MSl5-}4$MtAZ$M)Lpg(2xgsVx#>NzcobydW9BAhU zS6d(dBClX}3=R)FT$q!^v_0zTuA;RhmElTSvJ!PTZ{d}fUSv2LAgo}&++~uLNDFJ- z1~v>Ss)F%Iv(j!51)8;9hfXX}vOb=Cf*2)KrRLV=F0b6#=Fa|*I0zt!8Eeh@+8UK{ zthBqBEMa#zW_hW@XaenKi^*h6R%8e%iJ}POKwCk(-KH!xS)R~rwoy{hhyqlga88n@ z8Mik!d?z7LAOzzqCq&Y5meN`#dCpp|%h_I+q^KAb1-ru`lf2+W5Tc@x)#X0Nmiyei zvyDTrHyKmu3asbSNEu>qtTtnuj2Nd0g|4vH%hXrK`0K+-iYkXtk>e0mPFqN9?KbUn zMc|B=^OuUOwA|d-Vlo*M2novjqh)1DT4~xgL_4n_hy!atN{f{e+`+`|&FmcJ&;M2U zW`6gE!XicB?~A)mO{G1jrxAv;!qDCP*J*<0#Ye>Kwwm&8wKimV&c^KrG$k?hkE(yNF@A+1J?Hl{8b*?Rn*cMshZVSo3SQdZcFdl$g~E=@7l(gS}3*uDM3 zPl_?vTX$xE#@!Vj_{6_W7&Z{HUi9h$p(3PIRQZ_Z#Sx*BD5ZGl*pMIZJV2S3Y;N9W zoMdQgdHu#MM&lu!W|K58uv7S_bzC}oiXZ;rAL7h|k8|qGIeJaW?c3XY`3rx*xl`-( zT1|F#hNwowQg?|5FSK~%!H2m2+^hWgv(NL&8#i#m-{01m+4;$x!iWef1ggPnG~92o5ga6ESZ8UD3T|2W19D%Iw~x$|VN zZ~6jURj@Z4Q5i=RND2dSplAk?rKY0Q2x!(Skb#m2U6%ueAS)}1N~4V-FG`mBtNwB` znmAPCwO5>Tj!1cUt@3~Y4?#U-YFhUQbpk1tKDEd}oqByf&-H?Pk0Tn!&t{N88pn&xfwYZlAi4qpLwP9a_0 z6_xWTN)R9@edp8|Y-PDJ$hf>S;C&|qH*dU#6N;9K$g2`lSvpFw)Qf1~m^j03GG?XO zq<5^(@=}vdGbD@xKW*nJgYAUvtv#N;v&XgDn+zvuZ8zZsW~_5;?eDYF>u_Q11Vx!5 zl%U`2((iQ%gODW8sI0d{(7K`#H&|}B7!LQCBq=A3pCAeqYs0UNu!femW->4aiux2zQR(tO0(5wWu;HhZnLqq&yC$3 zlFIWAk1zE}ikw_)ln?|#i*6@oHH_IvDso*>8a>UYaayDE0+c3n&CjMUZapk`8u=iwxWDp|$Uv2+^Li)asA|`6kKT zdu^#=YfhD`dAU=I0Eh45g*ANsZjO$U{wAIx4W{>>nHG%hib7i|U7{;ZWh}C_ME~c0 zl}Tp>hhy!j3W*n)QqHkzEPZ2qL8eA7Ovhhq9VPI~$hRr2^)4fhk+)C3>C>(jp85W7 ze}SL;`tQ;+h6_dMkF&`A;@Lr_BeZK1?mMjY%Ro_MAX>L`mcH;Uw#No>Yz3M<5Q3e} z+hpShpxLF!GXxQ3o*-mMv+31?zyAYIV2ovFcb}>(W@1|4Io-BODYPL90;G}@HPjjv zpo$b>DomMSLy2lNXv7Vqa99*l{A8}wr{^qH5#Cc zgvvM&ibfdFZ#PNH3Mr=oKd!c}kW|K?jTZm{e?4^?F=3#Xq&ZO#&~3MCTLnQ7ga^)- z3<55mTt`VoUi!o;kf0YXh^2PK*nkvcVkI%w5lF{+8&+GAn>&8?n096XCb{3%TE9aW zNTiUyTj@OPT;y{BltpW%KuRQ*aZ=9Uww1QzHLSF>j+e8XbB%}}>o4=$JNulk4cSjb zAwRR)=W>?wQj+n3R%^CQ*{ux!`5U*nTJLwD4Zn1K1HCW@zM2*M*H`awEw3=ZKiwYk zLXr^*!A@xy8pA&e1~UzhGl5RvUyk zf_E;)QPOu5NAWJ>?Ie8fekU5-!NJSAsa?W66vo>i2z9{R`0d7MIO2`V*J$@zoIigb zVHo19VQ+7b%dgzv!G|xhy4q)VXP4){^*m`>@!00}f@({a7Hn?qvSbBSQZO-wrgD@z$68Cb)u0^*C~YWAMv+bE zMKQ;_4Nk6gh#QelJgS0flCpVyo1LWMn>V+3?$#ZQ)(90Kq6T3U&ATOWgQ0VL*RZg_qUi>1d4+ad9 z2`lX`X^}I|3(D<*Pc@ZfZ+{<|B&_v%bUI7Ce(MG%2pHYkqT6b4x~EB!3Fy zME}%%m^^1N9H^<73}65UlSFZCb|J^^t7)`IY#EE0;{NgYE zJkLGzZT{E)lz&n(+CLY7k2KuA%?VCBDW5xxv z68~1``ISb`&Ku6v0tqbUbbc{^;J74CIyf*@3ExB%9;zd>KsrHD8M?7V8p-*7A7>n; zb*#oMwAIA@25SvR-~{8mVw{ZG9*tPrV%RwW2v3mAKs-|EP?h|k2 zT5a~WZn1IYMb4f&MrSfc53T`ioP(X~ud=pwjBzq%lxT9Jc;b=!xxKl;t8d&OsL>h2 zNsc?Vj0w9m;s(uj*K>kRh0Q0lmQS!1o}g5FOv;QMU~jpB?i6Sj69(G=w8MZ?kK9MA zy~Lg7IbJFH47!$$K}uh@*(_Gs?><5nc{`0FUP7K&r6narw~Y)Eu89C;6#2?JE8Q+S z6Xe4IffRH@g$yF7G(lMsMhbi4LAsY5GLCUkOw`#YSn45-=JxJ3H!crRol|hqi;@78jX;m@WraEFvK=v zoR(;imRtQBYsEnA+RinyrWui%pr+e78Hd4hH+MV7YL21X1KIL-PFI=fV)F6lGD& zp)o}PwfZs+g4bR}8H197()l&g+Z!fD&Aw-*u6h(cj*`BkIEr@}f0>=W=|J`7Io~v$ zfG>afYpkzzc;wMXkV=xK319y5pK<5*4nOs&AD<_DZxNm#Z7|00#v4~SfBqEz;2-?Q zltsb+{+s`p@npjOaLj-8U;HODn+?AC>`T1-@@0PNZ+?Ow{=~;f#tDDx-}+Y%Z1Lc* z{N@Ahf%A6QIdtZLK+9q8J?w4b0JqS;$j#nY|D0-|Ej+j26F=t+%=irN>^q)*fA+2~ z(mIW@>Yf$;6pkjWrl>p@Gf)m)mW09*v?9(fwXsUFyPu+Uf>Ir#D8Lpa{ceXi5cJwz zE}m}FSYF0zjmt8!q-1An$j+qTjfv*Pt$nmq1W|*v6DK%#<}}Sl#HcFCk_@8)y1h2r zNzT@V)2y9dCE!m6{-Kl3PnY98#4Qp-e4=p+HEi&tH+wVFE*lSCUlIa<^Fg-ruQ} z%lB?rG415}Q#6Hl8U$J!P~NA@)(RX{CTI;jiYOuDn#Bexaw3AXy zCTwhO@X!NuF(TnKE@*M_?0sx(?NaColzOJ7SQ@1Uz-3rhu)p#o=0pD(2ADD}7t@qo{ET(Crep*x*S~K8 zr=FZY*IRhqTWL{dz-Om_tk=h1CKP}Ct5o_~T%+g5j0M_)wXQDwN^Gd+bD!|sLn$S3 z6cYp$Ld=t?Y2uffnA}P3j0ze`5lX>cQL=AKo*52#Za4%7f3!X4k$z*ZANv$&4cF^i+5?ps^OHaM3C_GjSr{EfHUI!107*naROGU#xN~Ed z^JmW>fH(^I%x8XvFbe5(U*+pxd7iz=9@lT(Mp(gkG`S}adOI}5yHz>oF-`Ce8m`Y? zYaPa#IgifON}dbbo2vOK0Zb+H7Y~lPJ;Q7l&dy+#Fc+0^j@q8V)*ZeX8I!rvD6_`v zY`R&f)CxdM4c}*&GO$*HVgW|#W`ZweR^svkWi(qGDYqvDE(}=hx9GHE2BQR_EDb`E zQAVm1d07!j$7&;FeYJsVHpr8Vaxfq%Emy8@@!Ix~p%ZM6MkKX#`N`8KId$nGouGlX z62QQ+uPfmUg@*cZc^!rQIZg9!(gz_ zU^2p31KPv1hT{=&5YUL5jFT~YqcOrcqDG4_QZyP%L}AEiGQn6&Bqf!zWMxSZh7?84 zxGWgt8P;0HdCqdT$E7o;kfy?x0wJN%Y+$u!H!Wxe0V=XorK1rmR$D8KibD2%2u6~pc0;q5Io8^OpSToqUtjUN z&zJ0Fj!qDhC^1Bp{#UmL@rQMj_U%E%?UYEundGiTZeUqq5vrk&GCShbi_tdc*W$!5O^yM z{Cgd~5a}-5Pq4p(9t{2Kl)@>6`janUUw(-&n;>u`+8@vHl|Hek$R-IVR{FTk8n^Fk zV0FQ+g`~`AAbH}kCs|%z=FY8~+%8M9|M2hn&+rb;qeWtjnIQwLn<{0`5mATndf&@& zl=L0NQM`+oE=1m4wf-(8O*em}UZ2cjCOiT>@%V!%Df#{1|2!|h_!>rQe*7nXl=BzP zyyGJX!hn_a9CbrliA((Qulz&)+yC7w;7T2r$QUkzNaV^~c17Q@X%T^#NJY<23-SD!O$YX#5xS*4hENquL{FFvqy zg*ngRFkXGl%0q=?tn&u#wocZJa~@_Uz}9ZD0>A6&ve4U2cn+qh&E=<%PA7!7Q^*SB zU=OJ)Op)V|EHz^)8KMirWSk<5!{sGHfOdwxp|6HE8v!T!P5LVhlFfbgGtJ)qge!YP zZtf1aHJVU5aK@sIW_e|mCmy*(b!L^*jb&O&@O(1hrL8fyUwMP*#s)hdc#>27Cb4s5 zotO(xK8Qu}fB*59iJj%4i)VTD)(#u{1FSWiUR&eN-X3YGIeX#+NjhOL9FiJC80SQB z#PV{VEX!!M8svFStJOp)g{~@`H5675MKP7u6e1EsOW0H&?zUoOeopS z#_WxUAQUBnNnW5a*sAi%iq7E#6lKmJ88b*bbm9ON$hl&xl{hs`_$xS;C^PUvUm^Gi#luc?6CvmWBq?AB?{-iU+0TDLCtycMY?}7_B))goyjN@bH13vq5MLUAEw&aaH!!Y;97zc2+FX_gT&9Px~;%3tM$2x*u zEVwnYoLQ2DO0YSx>?Dqrmf&PpQX0qQorA~J4B@^dNi!5=)%^KJ0LMF$VeV*#K2a^S zLi#h$sw^ctdeRjBK=NH5*(!cbvxBSuBbuPzWV(GDNx%tfUo3lmvjF83qU? z2^&qEa9FEBDN0?@j2bL0^$3HMPP@rcr^#}Ei6HPD_I4wntt<#ou3~R*hwX8JbOvKI z*PP?{lK~&;wRmQdadUy&^*8!GhSu=(D0$o0Q&t*OYhFB10?w~^72{X8{daM^EBV$P zO=;kRrvo;|hFhb*`XsR_CDM3H3Kd3#D)jtz>0!yHrWq(Hi6hS;%u7v{7oKyd)It%& zM;ItySFstj>4}KqRv`4UU)oX+c0Y%AjzsvhipQTF2yuN*v8?WAAtOY^^ zwBiOLNw7(ZY&MAVvxJ#Ij<<-*fQ)$|_3pSE?hVvleg4}u3uDCW(RyR{us6Sl<0$Dn zilg{0i)mBH!3?^z_~hac_UNPc69kH9zwsPD@Pi-W+`0SyLi-HEfCnyJ;44pmjbHn< z|BWDwF~;)bQx9|hg>$_0`W0Tgaf|C$uM^ah&6)d7@U3rso8SKzpXXsq{>WZ2n-L4CoP{7s;p^=UYC;pUvfI9sdyNe@e@$=nAkeh%q0mb+n=Q{O?qeOD`; z&L)ln_7YxvCoi$;lyzhtqtjZ#`aFa3!4M34~E>{-6P8~w60KL1HduvG*daH?A;ws*KInHO>2}W%os$_%Ezu-jgBl#Sw?Vm^r+A>czJ>F3fnk3%8eX zo})SK`a9>mz>D+AhLCCo_fpy*1hr_4-_F&2gA77W^aS6!0}mfl3^Pb_%Ts3*pMANa z-x92J;prQe(!g?4@UxGHoLrI=r5BQV=7!-b*ECrL-+wmbgZC+3+SVLvDb_l0XXtok z+j3@Eu-1|s?+Cthwc-z6@>$zjTky#TLLOL`l$D?x!xvuDeCdsf)t2N_kA$QZ1OY5X zf@2-QGaH(J^i6+V?(FRF{H-KTKtqmVMrYLrKJi&RL!rp zMiWH7>lg--vhq+xZ|bff%Q@Q!`8%iA`Ssho{BAJ-;A9l=ckWx`o0Dvo^o7C~M$KES z=l4bhJ3%qC)DZ$MtVk}bDqh_#_{cfM4_yfPrOzAoQpcyC2>G379JfY)^=glWb)*Jb zYOXjbgh1jbwIR>_aSI`cYGrJ#EqSHEIka^%u9~r6jG+<&qZCqrahA$jvK$(%`F?{U zWBJPS1mAjz$yR}sf>26M9y><2)uc3zyei2Ybej!YMZrW@L|3mzCzN<-~0F3+TNwnZ1B_*57BHj zSy@|Vb$yw*5fjWX^V915e7lQr(2sA?41PhVg=uB~;QOa0tj^ln%b~UyaPvW3N`vqA z)g1vq;^n!gm>D+}a=~xL`vMBA))-?j&K#^Z)<37+Mpqm0&0%@o56rr`GUUDI``&J) zz##GxJsF|1oH8%K*%{ZQG?rUQNoox3uJ9gQT4S=5?Qse?#zjG-0$Op*D6beLC3bJb z_5C5&_9hInlsrw)RfUieX|0#Jj$3q=yF|T!Kxkf>Y|)4s+!u8DcyE>eerXx`froi= zt;568(U;JZf?m7BD{ovSOHv>{JZmU~O&RZcZ2W?ZUCn$l}` z3Bs6DCr*%LIcb)WW;sI18PriqNwd+I`BmjbiFA%`uZPo`P$){{RXmGEN)=R8Rx=n3 zQO+?;Gm;|bc&CfChW(vK9c@ zKqkNcFhZbaCUgEau)@Pt->Wn~i>!G%%sla6f#Ww-m!19YW`Fyi{vnGmBu&Szz>`Y_`_E#-q^K#_>AHg-XC#eV6ecsWyR*$@}(;kNdX@{8}QQ) zM||z7;q^Vs^&8i@{QMtzdjcT{I&I>~K7-vYqzX87{y`qRc#(Fuhm?xpXo%2zsAdyu zEG;3~+}>vM%4@LpAdz;sa>8i1kI_ShJGW-gL`+2-wgGU?Af=-yO9B;mSZ#p#SxF-dD1-TTy%H$;fnuNypIq)T6@-+)Bh3asz1*iA2tGd;@$@+1 zQ!7gZLU1vNsifuen_Ju}HRGhfHvIRbEo@9IkDXMkwgkt!k~o6nU9YBjye+u8Z+Ps4 z>7^se3$n3UaGjfV>#9leCqL#MgY&cB$}-Oqy$p0D`>(x*JaOZ7kX)T7_=_}*?fJWo$79Nmm# zH|5gZHGn(oW*tA@wetz5lx_x-m4_Umn@wDGg6ihe&*4+a^cd@;Z?O_N+t$7N@MQCY z2N0AMvdqz=0nQktvsgRlB@C=!l$LB|Ik8gojiD+FcGH5S$mbPaW7*3z`;(lTdjm|7 zP#QyO{C-VU`3im*`HrNNG?!L5d+7ql)|Q9`Op=VRZEYi^WV|(^(`^$z^Z>3M@wjlD zkkE``Ry!T`ukQ20)8FFZl^*ZAa1t{~*x1{n6q0V-1}b!=Sv!7`@npnqzD1xEm9aEB z9n!Ku1tD5n_V)L2Kvh*}YilJ-ug0p4#u-Nthg4Ms&Jm=k@7l&Or%xQG9VyaDK{^^! z3UBB3+M^XW*xK7?XK%=8 zut%2VIBSXfYb1F=o)xqMiME!!C=P}FYOR^%8N;HW7dHt)Z_+N`4b{D$cRN3GZ`@6xyFfyzi-Bcj)3FqtKPO`G9aWtmlMW~K5F?aYrT*)q0nVX zqt!yTmN3aar~n%@p~_i*{R*v6(o_QL1eJ)$jl&ccO+dHZqTOs#X-(1U18}FT$era< zGvkouPp-82NVmoB2+6Iz zA^Q0IUY+D%jNl_@0?G=$an11I`vSU=U?Z`#BEk2a3AwRvxUz5g`6nZ84J~VJ$(g?7 zH@{l)gXaT&@O;R>_;yKI!AH-9T;3`8*~en;j0~F-%abPpcE<4NamBy9toiH<6@TMW z#1CEw`M;ilrWC9-B`OTDKw+lWt52e$C?pKjY|iU8BeK%JD9H+h&~*YOyg*iM6p!%t zLuI{?QMUu_D8ThK*X`j>l97U|w>P-GwMkj&*v4<+HB-XAVv*uJ$LySu7GMJv`Gmg~G$T4r|8Q1vB=={aUNh#-VF+ z&Xo7=j2Dp*#=gmSg#>+9<*3vZkpX9fMTVrRGqRmC{V2%MwUY+-?&@dzsw z>zy7e&6u@Tgewh|tSGc)FxY1_8e-~&f{tSHe1J4LTEsQ9O3%BxHDu0OM;HmhT7|R} zk|y9PFYMy9@oYUMvCg5K10~T20^y4nqK10P_o%njbn#djKdxAJZ(=-8?xsbWgPgvD zV~L%?y*!84pVmkUgpx#RZsjtKv~s+@Yxv{?Ay1tQ7*~#)L(A=fZEwj`}^3PE+$Yj*O=G0Yu#<-CD7aC_{by=!g3Fc+-0C70GD zA3WWfqJWVC);d1v1JGvPb;)Zk@svg?n{C3i6Li8RC;Ml(`OF_-s)GJfm%P-JWl5H0 zjI)H@)N+4NvR5Q*y!v%kF1()*xODdW>Hf|CfZ)uTDc-%t-_2EzT|P#HO&kH`e+~r&xaWM`F>q5ZwyyW$w;8Gs*khAmxIM-{lzrc|gj>QD_*9;o83Cbl(t1@W!s6e?D=)Uq*lI zg5uf`@-m>;c06;(@AKx+aC6|eG4RC%fTeDOrEm$U0zxHdhk-YW*OpOQ(d!w&A*Cb? zJWO2RQ7CZ~QdSj()U}n(Vi#OskysjNTe94!7?c%*A>=I&5uIdn?b1LA(ky3lFhMvw zSHkrKb@JR2xP@-I5KyZf3#n=^T!pFaBDAq&l|cwWD7;mQ1v-J?;;B`xZfy8MP7Ud= zs`6g?gAi|P_?sdpjF2-p=;GU~wV3*tj$$51N#9W%#a}`Qd3Y7=?Y;Sshuj=O=vD0? z2ozuZ;-B)xFaGH}+I*|ktodn&bQ})pSj^d-1sQNDa@wcM*Y(W zn5i}#eo($Emln8Wk7eN~jbXpAq)3WVQ)$iE7>cqYsSK^q!*nVjA15Sb#YkIZiqj()4bL(44&+VI-8rPP{9IwlDMoV6%tShf}YK=49Zk^=VP zI-lsb=vd3CND&21UVY&TuRQ%6Cqul@MI45NwVGT}mE7Fj z!eDvdxzh}@luSUUw?vZXRAouvAd&&?Rud^!(OY-uwA)x2Fv${xwFF^^B4TTta(<=9 zavUN9!N&fOjg1Ye@g8v?$+V!=Y%sABp(UY;zd18$8R z&)liVs`+<(nAjE^4##(gJihpO!4JQ`_j1KQf4QX43!Mw=MP$BHH2Yo7bDF4XVJ2H_ zU3-OIr-Kb+WHlyGzW?0rbdXB2+8T4=qyN&M)8Zd9Dsm8#VDlM*H@9JX%LsvweNgfz zUoqU=Q50E$&L?PXNwuQWi;?x17l$D$q2G}OUPv=O*lqEd)fIkYbDQ%G&jmach5X&q z>%5R;d~Scl`r3R6I~{>%V|iJTNP`XM$22J%d%5MOP6mACy5+e$hJWY7G5wQ{KX|1= z${H(De~0Zzu$LOzk({GLpoR_~1k_{|eD-1Qal#!;)&Ru$c*e_bP16qWaNZNxEYp^?(_0HySX@>ANx6pA@4E4FqWSBHW!tY(i- z_j{_>L|9YviWQAW5(fd3q9hW6Mlhc@1LX^!DhMEqd~sMnRTcz+LI_D1MU+m0GnBfT ze-}b&rSXj5WQ-4}Z;3!gqHZC4EP6 z6#qYo4}S1Te)Bi~hqv|Gd*O`l4%coGsJDPU{#D2P@ZEt4*FSgfaZMM6Q@)=N_gJOZ ze9CE8&(62wTuo5-Ks##>b{DL(SU?}L-@8R7rjyuynlZ|AF7}s5jpg=m$S^A?5Ljm! zCMlDmq!k8a)-fqc;y|ITPY^R*krWk5N}^^1VFh7eD2js0I;;#3LeLDoSV&c+bmA7L zS`qgN2#w+LASJIXI#0Rt?bkVRtV=r%SqTFkU2gK|BOiuF$jP+i|7GvJes#_AJFm|Q zPkO_SC!eb7s&`h9Av|L| zW-ylV*m$gNcem}1)m2@mPQ{aUeADwjX(cY!^M-vwh3R%~gTP%P>iC&N5w0X|3hP&Mx<-6P8V656y$-u+RvR zkR+NUO*lH6bFx@aRu!GKbZzWK^u<1SO`vxyoTW1z?peK?*0Eex9LX7|j1YFn1_C7` z6jw@t7NVEy=mVM%5Pnrb`Ty_v`m10?BG(SImK!JABVxZ2@wrk`qWfM-Aq9DwM8;jw z_0aQ8f31`f6+*1;t%+X(h!Dv`o^r`&b-;W0z%Yx8WBJ z)!Cdm3*Nl*Dre_sG6f;&OcdZ=1?pLv zkSVxVMQEY(fqSjtZ|!SZJV%}7R+eBwWJ*5jERR~p^I6JJswZIe4%V^-LI|?07}uptcceuEgXTfcw(zT;C@ zB&U_6NZ`Q??AAJEJedA6M=iU)3N0kgZ-O<~k6%Y*@AM(EVmDo^T6M<6*Ocf3z5cvI zBzJMnQ!PuT+R$^r;5;mvxbxNtItTyC4_)Iw`PHA~AN}#Gymvk)QJN3}nS^4HQJae> z9Ao1#N@7ce2o5a;B8kkdiB2e&J;|64xFGtRa-j5q`w#D-ds3gy7@T($MZsD4VCj?a zl&b%>ttb7Co)&%cpe2VC$0G8bbw$pm3jX{LKa5P|CWK7GF`J%-8 zfKU>tG|mLNq+yt9o*SlI8|I`+g3>g-mf75TzV_-{%sR&xUb@Bj`T?IEYkbwB%NGBz z=H10HnKgW7kh1^j=h!_MvY3=Cy`pYfYU9bYq){4==d3DeI)e?cqcUz?zs~8|ghg30 zC<%~3uBnS#IU~iB|D3a>?ZbA%X zg!6H1idZ@KBK%8Bp8U135NDa_NQ-q(pO%tDDU|T**8!9eG+nF^rAb1NDhfUW-ny-- zYC!Ag?no)oC{pE6@=1qt)&x#U&-Jn3or%Zvu*`QRk>#;AaIc40RtEmyj^Xe8NWxK> zaI^^g*lmReUODpb4*+cU-*^0-A4&LIFDKkNag0*IONWx*`j+|N4FIh(oEb|V zpnmusQV0?)5yAKD#wm5R#P98Lx->NN2ChL62_budkenm+8A5JkZa&C@+()Fh3BiCp zMz|>c?Y&26jdvcYbx&0i*A$e-I!C4yd6sZ$9ABB&bRqD^^O8SZESXxzow`LLm|Dk| zXB9tL47qC>{$#qO^`1AXmZL6SoBNF+*pAAp0&z_xzJ@7~VaPH|J{) z96rP`G^qoB@|NWnKBc+2ulT3081BzJ<4o}CaU55V<^jCN`oPh`^Q}j5Uiin~YWd~Q zB>eKrnzzq94`z<9-nY!#_`SQQjM=((VW8vkFySA9f^_%qHM?@ zFUZoAvov8NF&}OFnUuj}gMLD9FDBv~#yO0&Par5f(HHo5;UzxUE`z>HyR=LDKwC|n zPa{A0%QPP@R^T3w$>O9gHdTRjH7=lfcE1SWjIgh5ZNZg2&IoqV;)3e?Kr{Hj_A;`n zIIcD+gy{FXIzZ1L34He2fGkV7uN?PEM^kk)Qd4?CS$EjZAaP`U@8pVgPXTF`p&gj6 zVNrGDN^_85^=lkTMdP3W4a8C$@Lqf_I3Xl3uVy^5h&zF+`XSXAOOE0qA>v0(91`1F$Ie z9<|P5i_*n=Flz!|dCzih=6N{ZIGV;q$R)&{zLovqT>~ocBiA%9UyI-WKVIp$+xPk1 zIkgPZ&Aie%c>Tmtwwpe`*215>ZFyA2&wcZuMXA6~+|vBgrxLt}JCg_-F5AG@?py9} z9oysIUs+g8F|#w2R3swMx`6Ybg2&WLI-xn5Oc}f|0u+pXk>GED`U+ya2Zbi^7J?71 z1crnE0Q@W>CF}xO{w~n=`lyCs0!k$)CD8g#4 z;v%N?m7~r3?@T=JOjbdE;1Ax2+do+be&cHwzxPkKg7mj1u}@GC$H2Vp*8OHW*>Z0`44Sfb`ix6ZBwC? zLVFOY$2qmG(0lE0-uM0M0jV@aF(3qoOtSU43i{TN?ej;Vyt}|P5a(GR#p?S-SUsk1 zwPHpW-$UokUq`zP`Y!F#{wmq`F*Ef9#PfUDRsi(-@>>4on)ja2lek2o^M3O8&!TN+ z(?1nF0r><$jqlIL_uGvx#%v~GB?K+j@2!9OxL-nSQopc3^^T}SAo?nTCNN4AnS#$= zDY!N?eDPkx-QxwdF?7xnoI&YLokt2mq7+&xw9;7N$kKvQF^+*vY?>rd5~&hG?<^ng zjJTGk9E`>!iN@h6217#oTL=!^HerH(@LNFffP?aT%a*5K4-~(BjMcqH^ST2|7Bq4Z*l8Q9T7!OD6 zEYXIIQ#7AmSMHD<`BMJF$_cFF9|O3P1KSposS?L|wiW-ZR2F zBSWCldJXqlW%B)F41L?PLO_93Bx=_`|yf>sbeVwSoWXEA1LeS{wM{ zTaItsckE=6#s(&JSRemyf2X@pMST4GD=TLc+O{E;fYf;3_C41E*R>>CGbt_I(h{>k+FAjs2sxt&Iueei)W6E^J#o9O*ErnLFodS6x{OmLWSI# zj&eF>R+fx%%{wCCr|%&J6sg46K%PpP&eOSo%7*J_(X|`#_s}f~?q9N@am)7ei}DG= z8$wt-i5MV`LcNuca4r)3jCG2T65Wfko3^9rEU|}P^zdd)KyvkPpI`aumkITRqe;b! zFTc#-;5q*2?PCT-!DnB*N-f^zm5Je%2XorC1*r(m(J4(QQaUA3p(7K4w0|~Q+poVp z@!4p1pR2cCptONWS<#kLyfJjf;`?5^ZIe>sLLk+eozV#AJ?H0VG;M?2*}r%lgxIiz zZdq7iU5V>bKEr=6%HLvyx_b@{O?m#D|SMH%<1B!#asjO&@#y zJYij_!Gj19UKWB_FC1Yd6dnB5zg5r}zw5SWs{1e#kQibx&@O8Tkov&wfnqYw_{PH- z&UD}dsZuB@kpgz}ERs@4jnpZb>$q~Ti_jTT2#Pd`RZS@w<{85@r8S0JYotohGO#q3 zd-IB$g97r5!A?ez7ig`brlYYKA3&!#8(3B~jggebvYZAMWvqVXnWCy&mQ};)bVAv- z?CtJgEJ$xCgl1f%j0QQ=Nki4tc%N|n$}V+d$g+a*XiQl*7}JsE2MC#xcZMX5qU#XJ z36iX!YYm<02vT6I0~smIrIO_NfV!yxuqJxs7D?GO7;Bl#1gTamec-|ADQS|R5=|bm zo@gc`Nm5ScbCiJNvZOHvp(H{CY=3Qh@Kj|PN3?$m>=)L_2O;F@VTgf>ceHiK*{$aNbL8fA?_>2oY|+X~v8nv` z^31VXEA`fABGaW1;8pTOu-#f{pK=`a`=BdhQ1;|F=^S)hD&jtHAt+qg^eO#+%MO;^ z<8by5|NVL~K@!EVJ5NvO6~}blwj@$ewx0Q0CmddX5fSEy?p*-H@^xhW5R#U#_aEcC zuY*2>{U5`h{s!#)DnX49vwsG7#N;1>ZwTrp_zK*NS=k_hBRI>nb_8eopj&{9axSe` zbWUKJmZm2+d8!R#GozD=MkP2I=mh+Si#rT;Mv;ywm#3URc#}8ZdWSzgdk^{ifG*!f zWRV%vYJrylqXa$;2oFJT_9^;a(NLXJe*HIDy!Q>VBqdRjrZWg3xUx59cbN0wWP)`L zC1NExO;b8!XpAM%3Hv*vD7G#eXbmb$$d$nBj00VB`-MYnh)4x%3<81D8W$pA6k??x zshKWnv{qQ#Gh}ua9DntzREM7=yD7M` zH{v_*>~U}QkhZSk`P&wPBcvl-F(NpNB+zsnhyBB(wI+Cr>51FaV2`uLu~?Sas-){$ ztaWQ*mB%HdLWm5_`+K|W4Ranoyw7}5GM~@bKlsGfJ-GNdM#32J1Toy^X^Y{Z_kNRz zyR>K8WzctNm-ao_)5Oo8{<)`rXGOET{d(Y|O;o+L{@CTyg2qU9Eu`KAZxG=h{YQMCCHFq|L3qC+ zdkO3F4o?ZlH;|LZUhn^P*3py|DRA%noUh+M;`X&`e0n$Id?L8J49G;2q-k`}fZL-1 z)3(DCNU%}dJIQ!dR^(cdX^BrXg%Io%1NMhQrd^9pGt#`^-Lq3nJK%h&IWmstu3lj| zU!sR2lu~%VKqZQ0-7?5>Mmu9p&Sy-g9ecYuXN!vIY)){VgS~>Yvk9q47z~G0&f-11 zbo)A!`JAq9$b{gzoqgJ>rfM46&QjMkK6v(uy_lWuN-~`>n@+fK^#!KR;W0RCNRphk zbCl(rG#_w2nUSSKY!Ts5qtS?AQIP6{#c~m=J#JNb^$6!_Z6t!|gkavapcSpPXdp=w ztoO`~p=%m)skt|wGM~@rj6oB~(}KaE0Hs)zQ+yk@w+b#-7X4Lu;cvJ&_`U0}^^#eY zQ5QLbBqdc6=RB%^me#{%^-T58-pAR3cM;+#qSn|p6X%6_la;IY^f_krt#yB|kiBqw zgoUX@Aw^6!C{sZQ9REN64eH^WNEy>towdxXW<%T~AWt=e{hQ=J`tLDp9%AnP5eMfp z{&S=FZA^Heu=Ms@edJ5jO}7aTYtzBi5iX>+%gDCc>{^8%{)_%@g=PkGQ zh6EJaTTagAy!g^{NTGOeHlrIoNAcNT#?Bs5J$#KuM6#fDi-0(!V2GF;Bb_5v&@CI% zVL{hf+GT^+n$r`-C{NhiO&JSE*Ywg9*@#vMPNo@k=a5n`91M_h(f4r6weFpgQ5!er z6-CquwI1Ul3dL>ZKqG8YBJ&;2+m!C!J?`9nlh+?i_=O+&B-fw2#bVjEe5ppe3Ycsjw$s=~i*4p3n=M|`~J&=0c9pzqQy?FV3w z2Z!$$EPmKP@}qw~PKb|77i_lEJDoj(!b>jpzd5@J{(gY=p|6JU5Z>Q$jzQdt39yHN zZo})=KURx}d*AH|9}v$dJ_=%w9ie9vRFXGmHGlZVd+a{=T=D35#&DQW3~~}JDVvsW-@nh{Fy+%X4*2%lkGS`4$-G(;5ZpL8 z;MVn9y!Y@C^SZ&JsJ!FG;XXHSXgXnd{q1}B;F&Ha6xopJe8G*Y*SLN277vb&=!{`n zo#TbT2Td`^IiHsVDH#?8qn!~Cj!vkSOOh-_r;5e0q+CSi!8DVoMAO*@Aq3XMUO=Uj zcrU#t3qX=kwo5u|NvtLx6dcVEyOeEGvQieEQE=dQ3RQM2j0*H&(ben2^ zMXkThF`|+L*VDvr#M~v?TUH$Wfe4waJF$jGnm)KP(|4)wzk=+^lghfKYA=@5T$Yl$ z2G`7S%L)GcE#&bjw>!;3&N=gn;c!G%R|szzKEH!G7u1hl1HbBRjpIfq3EAzJD6Zs4 zCfLPO!y$I%c*`Wh3%v!? z60#gGC06DHBre=oZ&MF<5yKY=;%)rle+A}CAa-Hpqz8|1bHwTQQon)IN@M^%=Gk(Z+-h7)p9}G#UMCn&0v4X zZ@qbk>D3>m`}?0^+Ll}$4fw_z@A11|dz-)c>6{=mQcJu{xiT8^IuK6pFi7{wXA6eO z4iKnYgFT)wS2Lte2+Im*1HrZwnI=y(MVgW&l4;LO8GIxK8s-_MYp6SeN_6b0?xC4V zO007vD&xxTh{9LQE5W35j1rhGTlT7!5Fkl&@+`yHIOZE;P%;wRNGT~YP1`xf;~`J9raktB$@6s;qFUex%>jT_4q53#W zV0(E1iQkgwdZK*xQIFyYLE~-PuuS$AHo*|YipEBWTCdAlQ7rhcmWvvTrke%Fhk$n> zGW0#;Sf!LW7tlh{J(a9L^()UaX!@n-`}>~q9^g@2fANzzwVJGjA@oNivYLx20S>OMdr5a*QnZ-(P%(bHV>vSu(C;<^g(=L>IA@VcBeWt&#d%dl2gG@lPI1nW$b>AFxGpg18iMtNUOc#KJEV~)6$t~@I&gMV zO_Wl^UYo66oYlCJLaaxK^I@GEab1h0pbe6ojZvi~A7JKl>jw@$=Wg5+%`z z>@^P6X5M*<#e2S~u6G_nYru!7^XGkkog;xna26+`yQ7q3I7n$jbdu8~-Yg*mMV4Zn zCrc7S2(-pt&;WDZvuqj)Vx`Pla{)Hrbsfe|5JU)eqXx{5?&G~D{rrz{{mOM7e)0cC z$%6dmCn&C8LnZOJn)#-mOr@Hn*QQhBeOx&u4;Qr>N-0S~ppuGRXF1mi*Pa_uUp+u1 z3OQebfbo^9s3hU&?YGcLigOC%VA*ucr&F|2+>Yu(E3E=S5SpMQN+#?Hg%2>uQl?#lCgHig z7un4M*~t-C27Nhf z1Agk$V~(dKTEMlT;$f=TO&uqd<9IS-v=cvXUFcf`1fAVL3!Srg8zl$o&aOYRG+TYnYZ5Q)}p~=XhcG%u9#7ytB)o7({l?E<)2afw0oW69Vfzi4b(gGsrbcDRy^9 zbk@-tOGRw2P{PDuv!AqDC!i?!p%37`{0b14i9Vh{L*d5)%JHug`vOgNLW0x%3j?e84xV1Cj-RYE(Pw+h~ah#{@r3t0Z*m06^mJ*~w z>YU&m)v{!?x62?;dF|bYjCXb@TT3?HC6SWlbip84mR|7}ci#pBMUkVWq^Ubd6XuJC z!BFB>9}fw%A7CrkURafswBZEdlc zWwKZxw4hwn6azu3C9`q~NQyi~VDL#KTp@IH#xfd@2sUPnhl3$`qVdkfs#ssW$@7e+ zs_427={;JhSdHj=NStkvT4P-xlL=B-n#Lp3f~Hi^+1FI3SyujmAnw} zDXCC8I)>hDz21A7UE5Hn3(m_ZGC#<2^03KJd*e~*_IUkxXQbziC0Plri#RW=9qk^L zX`MhT8J%4fdzrM>cxR|PL!x!mJnJ3tl_OZBNuNIxBuPq^Bq$Z1%~+ zzl-DuXMYWK;~&I()7>S67TgSasRN(9jNJJps9%8_{{ZhMF#mnvE`fm5n$CEnslXd> zfl;2*3eAc04r@BzeEn6fx_jKbc9qsd=HBNI2OK?``3PwFWkDusZBV%_8h+0U4MVO zvb4I}+jl)B&>_$YjEkfyE9*kvU)mE6B?PpH!Kmw&1ZQbl$Nl3|5~T=IqWkf`>JP|L zNtzF6n~G&^sM>CW2?zndMPuT^Gu2%1R@CxThHm^gi0ac``|K!K;3qf zi#a>P0V>hhwxMon7SnjY1Y(_}mAxOZx)f5%jVR)li3|G*Pvppr`*_c*kK1L?cWIaQ z1GFatz6dYu{Z}@vLhlz&iN|1OAF&5;tB-K~TG$d|?U~GEPf)QVgwu88>DuwS7y4em{=NR6+v-0G#Jgv`Ad@L7-)ELC`P!p%-bf@qzsj$C z{zb+`3a&v)!%hhJqlVpsF|9Q`n$PKM6x&_x`?-UC%CZD&k->A)v>c4ajI$J(CJcrd zxn<#3n%@eY$|NnR8*eNSKA)~HgEbsfXei2c1? zCi5wy!GP&}P6!@vKnsn^GiLKybky^VL?-O*?xB^$SwpIKaNaQ(4Ct(oNEjm!I8)(`r5NUTqxvKVJ)ARExuw*2AA>h5=T1l2# zVNDcLS5l%?By$mnp4TPL5mLk|EFqwjL~OD31_X6yD6%A02>k{o>et|~cpromT)%M} zq13v<7isD@&gSaANE@l75-l-SVcG_yL=e!4#@3#1Pv1ju9Z~Z9B$pdv~ z`R02kd_vbeT%K`nQn7P*gY)@>r4Q@`$!kY711%8OYP_>dPfxgW)N*~V;AGiQrdocC#2dsz{YDeq?*D@d^c(1Y6BBUnO zF6u3ACxss)H;K%_fh&7E{Osp$F&^i<`sPD^_vi$E=5vVP5m*ojL8SWvX7kt14Puq2LX@a0n$~ntob{vCZ5%mSP*Wu^~ zJUt062~Sk=R<+4^3{OflecbIb=)1H_`vKYZR3_Lo9S1@$wz_ggwu8E7LlZAE`$#WAuZ#;zwJ zB!U>E3L%g>o^uRDwyToy>dL1aPpsaBzOuJX%5kyEx~hskt-r{Ni?YS851cKMln_}b z2L*Yaabt9q`>o|vWc=idFY)rHUn19@A`R$JBhHRFt%mICl!w!Ej$6-8nlM06*?_l> z*%}$p8eCe4bHb%8+wQNF@%H*xTkQ;=nPy3I(u^Nw+O6C zqBwgG>4YfT@Y0Pv60P{|gE{R>f!F4i@7zBjxCaC&`26QTLEYVA@})mPdT=680I1azym!jdI0#X@@px8JcP0{|czEw*%Ff6! zo1MXJ=zPF9kF*x8)K+g=9CyY#Iu{sa3gJ8^h; zFJjgE&}QqkRq*tYdU~Ex=@SAz43s^hCE`A%SWViiV-nBW{QCG;!E2Vo2^g9084t zdR9J=<_UxG9;xsQt!HKeomR94MglqusnRr^;p+ZAgKUme3HK)pmQBsAAh~&vu&9>k zR3U|?Yb`3xxq9V*x?R$>mV<+R=JP2#qn%i>v=$)*c~-DD9FygWQU+S7P%ieZ1QAd= zp&0Hk9t_ZRP1)3#(jtYVZlcrDbq%Izxq0J&*`fqA-q$S3lEE-%lou?3AY!GZZra!n znX-R=wd`jsDe0_R z&p9hwc*nIYrL1b|rpAX=k||aZAFpOC>YCPt3lrozKoy{sCLavAcJp~|+dC5k7$gi>?rVrs*tJRy*mZ99dPr7o4BrF=Y=2Ud~p}=P7%UT z)h$(BvuHYk(9Ej}R7#e6zH?ObNADQIzZPh`rLH?NEJ_4kd*d$GZ{9*G$*FUsLNGG| z)gQx`U%AIGe(ux!;w{VJ&JbB0adhYFXe~+8EXqRwWox76XJ4K-*3)!mLxd7Mrd?7s zEjwC}Y1#LR3G!6;q$F`gur3B?I!Vyd;k+PKlGe&~z$b`O9VfG@AH$9W!L%~WTF+hz zCzA#9DvmYVvtBxDvCg8Uj3iWvCQ*VAB%QIe#;kkX9!Juw#L5RnMmag-;qip~vz8ls z1xy>b0U#ow&okr=*3XYO7vP;b%Q?7qi^+6GYd_qY_%w_9lrfkVC{is^`GAT*?JY8d zO@eKWBHB975TYo(_kChlu6>;+aa5P~xLpQ)mv(92XZz4G#)aVMqdIy||Fn3<J;X-^$iB?H3MbYk>>CKTpv^x8CQ}-tVjf7nnQcW6^OCat*yLvWyoFuThLfyztyD zgbD;D~|n zY~E7!x_bmT*x#jUJL;xGNNhTQSe$K>_OPlX5@jTbURMXZ&T#eMfbk%Y*TCC2Hw%e$U5`H%QCIK-On8Uz z4vXyd+%}V(l9H^dDU=L%MDYdAc~?BtMakRXGyg}iJHTocXt#K!!G=jB}D5dFa zWC8AjzSVQEN-+7p`nFuw{P{Ouqnw`exfieTZ$5t$WtRNb!*h6f^`KoG6Hknv9)X0n z>-BB4vqxSOoRy39=T`fqf5t-Sr53i3Ll>W&$l_`|BtBN{BWydt2dKOwSqFR3?|8Bo z->)l*E3}jpTRoR6I0lYqV~Re$b{X_t+NFITZ96*^|L&cYr+tu~^)Fof?6d4UufD-Q z`d@!@wSkWmCI0vS-GBXz$ie*36Dg^zPtXl=*UF~W$2j|AucLvYjIy4>#BOXrhV_7>+fYz*wjLw zVIh(}SPxPQu3ov$c#!v6ZE}5l&Uxy_)3`vfzlWdC39}_MwUBvqQ5d%FbZIGxQI zrYUM58D|_Yl0?zA9izb><5aMib_|O#!~F~|1ZU?<@Q&}V z#nI`EL`vGWAs3pwNb#;?zF4rR8uoX`T-h7*)`N#6DkmFAv`Fx_!C8+LlJQtaXq=6N z8flv1vkZh_I-f^akLj2@$I&Cjcr>JQo~Ef84TdPKn9fV`Jf&LB>FR*91|>Pa~wuNp*j){$0WBh|Golv+S(YKSVldvpJ4-eg^4zVP zy#4T)&R7zyND{q1SEV#Nqal04r&sd6AErb=<9PFA&R{TPI-8(^=STN%;Jx9ScRtHE zU->S7uMLd8@H{8&Px9_8VK-AuI!{|Ixq7(ACvM!}=5Eex`lrmhBhD5xzVhaRpZdZL zjB9Dzny#9$Yz>Wx$rq%gH7!y~tm$yRg)So>7tEI}zAi7UlNL?mJ)L#DXdK_}9406{ zu^L?HoJ1;)&(C@1jo10{pZPhy@@L=Ti(mW_FYLz$O6r8SmQTuHtVo~&AQ7?J?mX59 zCS^<2#U$gZV(gs9^;Kx+Vnxnahd14txG8M)Un^qLCy{|+B2Y=%v&jZVMXUf%8_UD9 zGv>>RG)a&`;zWclH^$T24kZH1&SI@a;%SYivj*!{>-2^xb5+f)nvUCrdHTaYXkB{yzf=kNaefBVDiaUFpD`mg^NTfxnf`&{+Lt(@HzJY_vwuPc9B!R*#K ziM3ZOdbreMUh&#X+UzlA1!IrGy22q=S<9Y5a~06_vJo+`Q$2g*Q#QMGOq&-r0pEq! z!jruM0QT{=T8Fr=L}{`jL+F$cV)i#lQznxMkB(2c zeJ~)+cQC=B)q>G5#W~HZufEB1w{I}q9aAhi=EqYWOi!84=e%(Bko(gGB2Bqr0;Xr) z6iG(ed9L5Q$^Ne9)jRJnsWKMjjM;372%4)`cQ9TuJW%|_w_jnndxh(RA+z&yCUr}- zoUvFo)KW27lw?>^DXF`TOeeI?abwqE+XYGPxPJ9TrqcuN+gF@VrwA!XL z{SOKgB3yE?e}Gb&a?#Q@6+%f&=fHcscjz>Y!sef)T4TCIU*=+Oy$3;%rSF1*OPGd~py9jym!BxL)a{*3gCExhQ*U?E9$EfYt zPcpO){K~)atLV}5+<5CQOP}%PX+s?>W!R;hE%=#F9HPd1yzywlXn339^%pRUj7jPF zy|2H+7e4}O2?+$Oi`U;J)80UHUM#j9CfjJzdR851@l2HOz_6LYIUc1P` z7)h9nhq5s|5`m>xNZcCIw@RE%mo<~=glW0t&fPn__3)HygMG40bF%1oIR`C))?Noe@q+B+1uN} z2Gv8Adk{;AiPR)Xu-;>wXP{$6*f`Iy5Xe?hwXt_uNx6nEZ+c$^*AMp5euNC3KlsL5 zJUFjtKfPgq-N-dOzFzjJ%n&~##U_n=^VThnAG`-ih}x*gHM&Izfn+Q2TW?Mx<;7~B zbDqw0DY64A zkwYtr<;YeXDY21}Gf0 z@!WgnJ=U~_JglmH_PIloq9`F0&j(!K+&Xot_O7*Rt#9~$|9=aQmmwi+RxaBnrlpkU z{E+q;uQLhB z-~IP!gRkcAgbYEEUy7Kwzk8Im*E@5tR_~!ia5hu}$qJ{OZ33$e+pT27{Q;^}5^V)l z;|xos@HQ;)N=xjcdJNjl1htYe40%AM8kk7)nY;=WgXKM?L@U4xq*OR*QC?yrh4+E6 zZI!!Eij=mWC+lX2sEf)vNTm=0HV1v~Ift>1nN@VE988W(``{hryyiDweUlgFh7Ub^ znJc3n*d4Uej8e@>?s@%Wj-B7;_Hd2IcDA^>KEhdt6NZ!V0Y$F3vboO5;VIkw4#jlF z$@mnV#HJ;Q%Mu;KsGBmI&$%=l@#NMnUw-F08qcCI%xn<%6)MJS z&#>D^N=diVqc$~jF=Nr_?v+j=jA1dGB5X~Rq(o7N%R3v)7X_*^bk|1|wPP|l#o&oL z8PbE2p1Llv&XXoFzIook$?>?_pwlkSS-h_?CP!*b*2!3uWq`sp+3Q$67<4HS;k-qO zpeE@e%!yvW}O4)9_qXaB^&se z!n+z;f{TU-qiRI1&r#iawU$5`j- zBxmC5<3;|!&t(&*tOkeoTPKOBt)&x3LB<3qYs;N`2dvdIZr*yE)5=p8IkDDQXZh;! zJFJaH=$jG8rxV`1ev48>j5aH(BIjh1kQv7yN_g$wlxH?C(a8fG_2%h zm%lb8^_X^&=H>Fhf5skF$t!<)nWoK)pTkwGt%sEGG zEwMY(aTn2z zw@XxymXiA<5hyx~6Q-M1l#J2ZtomVY2 z-W#-3IA?Ll<}sY_myuH7T*%yqe!nwvYA+=%x2AUvB~{oUeF*&AO2Br1QBw$ylmY73 zenw9i3j2r$(Dn^1e{P&gXPRsx5@!Sr86Z$BpU`w(Qbu#qPLGYP4Mx$B{-BHVo=&XE zCljW{j9xcoV<4%r4(r?NJhQV+Vk-{!4mg;VY!3T;NDDezGcPM{k7JHp2V*Qp$0vw7 zRN^-JYgDrZ!<`Mv5~6;Jj1!)I>Iq(d>s_vHud%+~kqkgGDTWPFMS$QE~pnZiipRyci1~QC5ls?*}cqm zKjUCL;p(L=j2GmK1vd_lDQ7u3;OYWjmw44@QP!083C;^5rRZc`tgUcfQ5ZuMDNq8H zBuL>I&lh;>QL!Y}k}OtKIXENF;Jv1#u1(0EYYQp?YHSO;iX)Wrq;ZV%9u-9tWf?RZ z9r(&LOf6@XZs*(*!exE%wv8{(h%Xvq2=4+f>+kAIJRv0xglvsmPK-wW5Yj$$#VEDMIQWqr_LY!tO_)dHQS z_(t;&rSUpz*efM#8yip;oSaTjQe!#mp9A>IpZX9SeVH5cyLcaZ;O2`QYb}Sf61TZ> zvOlYc`DVb{@mJV+bZhMs)v3FOI{h<+MuOi z)bBCc*dXmYioKvuXWaSE=m^pf;R1priYRr^=yNW}HUMZPxYAE}W;Shq}; zJjztEtLNrRwmDF4Q6GM(^+mtWh0O5hg)JMRTTxFsA6XpVgyPkE(p zvEBUJ-XLQQ-g;{1uni$f=>)a0;DU&`aUP=-N=ULaqucA!TN^ZeM1nMlm_-pz2B`v* zPmoe_ay+Fj7j(K^KK32Yb9CztxA%{EdGClT8|ys2HKZ1r?NQ1{A9viFFQ_Vube6KN zIGxWKSB`6!woyQuc*do`7)umMwnl3l93NxqlDw*^Ys2w)L7b&*^@o&|CAXga@e!M& zHJr7q54zwTM|q90HIHo$IGWDL^OA*uok;V+r>=7E-~f-JM$zkKR8sKfy+aO5!_~DR zC&y#7b95uc97kD~c<0#L+oRhbqLikSBvjs!Whvb> z{dk1Vm-7x+EpT}P9%^w$Od*B9M;&UFU|hu8*WX~jlsxwA`*|auvAwlLFH5ma1$@=4 zwan)EL+=^s0C_bg%-dC~>KY{_vqerCM>yw6W6j}s!Za@^ijw(s%o{KNA!?D+%{p{D zDeoRztPj0y*OE1!`QV5v(-UG}@;x8>AR_H8@6l^1Gljs^hRMNgl1`|@WQpcqP9X4OtWwC(}tl z0W6Wb0S(fQPg!|%CqqRE%FLT&okUBhP0gyP%90#R1#%^0>T`YGt5%w@=OmG4(Csj< z4AvT=I7UVh*3PlU(nTVp2wxhEbxW;1DeH-_PGi|#bQlJ!by-?^5N zHmo)6{@3dLGe%ed03ZNKL_t);W^v5CtVxpu9mPauj#83=7LaD_9h|a%d`f8@Tb-E0 zxnWw=6vK*y22CUss3_15uTrtDuy7+a-iGzpT8pX{%lWAZEeo}jNDpzGoCVF^tyE9X z&_6Om}QrS?#uR5wC1V)qTl;8fHKjhBM+pKSlxOHof?X3VM{lXVs<@2Ba zB2sC#HrM#%C%==etxaBi^>u#xw|jJI5zPGC_FHt-Hq@%@)|IAU-`s3Q6Q)e>x?vD(sQv`t3uW zyS9VxTJ9aqD5{D`D@Ju7Ns+*)+d*2x$;lWgBs*IhoK9!Fee)(FPFc*SJacWAK|f`G z9MQEJWWiV~%G~2k=+P?joFqxe1(bPC9BblaKonP0MM)gR_$Wr;v988g!}@5%>G*_R zx6j_;0ljX@bXsESKuVA$2}M=ny~KOdICmOjJXIA4X1q`oWldexXSDquyt4>@rh&Db zLOxV;WoW;_+E%}8#hDCKefbW{#$lt;nRFP8GnB?q;;?uet6As^$3$3*r>Wv5p`S`? zNohg=J?9+1B^Y9hn*`|xxIInG)-xHj#te(PhOs{oS3>|Ha;ay zVsxCK8`*}ougx@2)-;YnfME@0C7@I~LMzF=TQ`~R-zH5HMuSZ{SqHo$%Q`4k@ZtyF zj~Hwdzy4KhF=u=CGAip+c|j~3!sH-BFPAexB1M*=ukfjjE?d!px8HdeYYpC7yz?a8 z2oIdhOV&D`dF9y{NN@&_%<_sW>jOUX;T`@D2vRD-@v$d2kTT-`{rsCuy{E;gD6P+p z@B5#7mY%QZCL;!$dpIOTX;_qw%6O9d;twH2fPFs#KHS5zD`&as%BF17dUN<}(oOnS6b;5~61 zDz&u@?8nY|x>-t+WE|eT$09G7xESFKQ3M?=+33b$4XO=UBzS70&mYH8lMMApLf`Y( zTL!bHZeX=%^WQX#-#Ut#3TSw*0<5|rg=-$VZOC7iI=!N)>UkOFX;m%9MmBzn^A1}F zSoW+a1DkGpPLo)1%{uM8KNs-HM(XJsPpSXKwTqzd!Y=HaU=Q$ze691p&c#T)$Hlg` zH~F!@{qOTn|DRvrXaC7R!&%4U*RJqApZ-g9?$d!DfBL)F-dg7ifBXt*mh!1jeVm{Ash{SVXP;*G z@(wS*{0gtU@*0;eU1N85H|T9O!J3pB=PW`rq$qyHX>RD1LpEO`MG$CcpOKI1h?(9<-%kgx=-QyF=9mDo|pObkWoI)8K+(d#aOSB52 z!Nyq1vc$Q-09w1CS@rUpCn0lf1JQ&MlJ&uetD^yx%uOBN7w}O<5O#y)aHtauLLVy$`hNPLo6tn!t7l!GZ1fy2xBnH^S6XO1y)~ zWJ+G<e#13^VZ_;+Dv=A0wZ5tm0@vLxbY zR&hKln<`!)O_B|S_8etIn=D!xoA#U_zTJ+2XHnKvCc#^SxKD6ynOu6rLK3Qt?f=3Y z_V-CBIPV*R7*CpJSZ7%4_pmbN==KeEu3V)ma`Nc}r6ZC~=aHIeN=G1^6)z10s~r!I2tycw=hS2d9g!A4~*3m4$d zd#4MIZokFVyZcyE^UiB;6AjiWg(7QU#}dcoVa6a0y|Bs#H*#83oJ=NU!;JOKD;ysl zv3GBulUYTk757dSI2_tLY8wc6M!l4It(YxJqO};=pd@DZD!%mg9iF+o#YVrw;j9cn z0uagQY!L>^GN+IlEpt2xS*)pyE#qPDc%kxJ>NWFPhxW6qO~+&}JBBOnfN_jruc z9+N4GNzQ6l||VJso)SaERyffdp9yfgny&rq(srXdu__?44AtOob3g8CouuQYfponmq9? zkg}br(_?FcCM3;7^BfmtMOlYyCXqS^>udwR2gmWzU+df5E`q)byRdJ9J%BrZ$g!{W zHxEDOy^4K;?cE(d@rm!`fB4CN#AA}L+FsSR(vb(0rg ze1RYNki-OK?uPgfAK3sO7b85o&O+^Z}hsn`}Qq9AAJccB7Wk>e~eBi=`vP)A;@b+#BP?ih2_7P%asJ5) zffu2tF!X7K>WmTsv?fv-DI-uRaiq{`2j_!!SDi1AogQ7KiPHoDoSvQ{g2;dN=k~NWLnllPO!7FiJ=bqd;K9Qniv`|U>LRC%6OtrmGMg}+&nQbncQhomA){U7WnfDc5>qX>y1B#M z!+jRc(vgZNji?%tIEnb3iK_f7b0Qqn0qF;bth6+}s6x~FV zL@GF!4Gc~-vIs5A_EF?lXYkGt+&IS)aS$nq()xj5YbmU~vhK8j|ElB*v`MK42TK4+ z91}+oX}8P%(II%xu#<6m_Xe+gRiTzi^0BMD@ue?dJ-qPIPqN4hY+Z%m$b`y?ai*Em0xcE!_!!Z^ zT}@R|7Zp-!#wVV#XtB>vh`h1n)xE<`1pw3wJxJid3~T+2w_ksS)00#B{UOg>(M+qF+5VjMPJ)aAilUPkY+X{<3A54@ zDMhc-DULJ0b)WTTU@wU%)EAjb_tfQD}>VDA<1EgwiAb&;o&<@@1Rh>co_ z8@nPK!ScP6=MuHv+jH+xO9W$`A~H7UAXaTf@QmO_%Jc8o!+fpPWw>1zF6`lU5%gWy zg?*Fl&yy|c=8c>D^Pm4^e)MmAhSTwcU;VY;;D`SDzr%1axbIBgz)2g&Wx4PQtPRe4 zRg?_-U4#&5t;wE{U{SMy8RDP)Y}AW3;IbkeRWcXOzkit!=4mv>3BgmRuUpBGIwyvwHtU7i?bjPr^weB~W%oG{E1p1yXO$F6MCjU?mQ zl>Paf`E0^btjVH?t+gIf3Y5};0ZnOk`kS22$IPZVog^k6#&kLzq9~>;3+DX^N2kZ6 zNk&oToGuDFal*!6KqTSad;1JJ36sf$qeVft(;+VwU8G6KoSaE31 z=8d%>NEAXWB|KUaf}n4?I`_QCx(aJU6`+bFiB_Q^?mWIFfBADE5a@s7=3_WdLkuQX){{dqk6nrd#~Y9 zQqoCcq9i5O5qr08Vm*{~%@d;`#>4eD-{9_@8^J-95n2giC6|4RZO|9stqNl*QXD`0 zm%p3qM+=^L{3^fyJHNrL*T2AQHl>qgd^RsBXTdS|&SB~b*Yr6WQ=_%!xu>3CbA62> zUr^*Z^P*}-E^jmJ_c%NmbLZd)cWHxnHd3ri zd2)Bi?x<#Gy~j%*eUg`7{W5WXn<(oNMJZ4A2bkKhe|L|A(+P{|1R{+}QiL%Sg(a4f z+Dp1w2WuV1T6*1#lku3n(|H(15d~IMDFYFhbEszIUU};dmzAg6>2P#>igjU+isA&s zS%emlCIV+Ir7M;J-)e%i-TRuEfQn+OGN27gN?he0)b;V9KW#EEaL!zJap98L@0OD}mW6MAn~Y%w2u% z9vHv1B#F89 z_+@_eSALDz{I8HNN^ak}#S>3l;iZ>8#6SFp|08>M@3B5wAc*|Mt`W zC%^pj|BCPbtDoTK|Jko_e00>nFCUdf3_+c7?yT>L`xj|(*4r1F5Y7dKzH#1Ian8TO%NGs-z_Li3l6d0#S*LT>BHRWQ=%{vEt`OYyfJbr~M!;I&* z`qYzxUw`E-t{;s#juqqanD<@XWwY1e`NywvP}aDzVm3Qv*3D2}ur=(m)=P=vlxmVw z&8G-siFHJ24TUM0o-A1Ft}*CmtYtfF_4<@`MOh|vl7dB5^VZ&7QmwhMe~;72DQTyR z^8y^GBqm7{*46_7%&^;|masYOFwZNpB&Db-oUO5@B8fF+QQ@nSg$)%t6=|}pPa-`h zr-xKUjnbOY`UZRZ_n6PeWLe7VH{Rjt$DXE>br_FNan{i5_30;lrn3ooRU(w8u1ciT z)Yf9XYsfq#&Jm!zQlpeWX-`>|v|-!*Nn+>R0}e~e`0D+cGt$O{?GzNw1-3}<@!kf) z5of8a4MNY4n4=*J|LvEl_DUf{=!Yt5%aE{U4LUKCJiIH*{^;gsDFcQ6nVTNWT6PB5jXd9K~d@qR!`R>|Wud+huJy zKt(aPPbOS_@)>Tv^*YsJj@6Q?tt>i^R4vT?Yz^)Adi>%)|2b}q3%2`Rysa@LjCL>c zz85}BT+R6G=Re1^4g7o{p>oc#K3d~bFMWdjgF}W53PkIOtlOoiN|LnG5a4*S#R2oG zLRM3TTU~zOqt9~X$!D4D-(c8(jJ_N(4bAs`?0H^& z{WWe}e}}c~Dbye$ixPHrwpi;%yqFC`Z&)=Y5r#Vlr{E#(^btz)#lwm;j(Ps6r?{R^ zI37=%p6)Pa2&mC0f>`-LCqzndWoLtDhC{Y@F7xiq+r07STilrmOgcozXM{K{>5}&z zXU+$DD&WCsv%Lhebw8DV~4dbNtdT{0cwt13${z+KBa0 zpDfGh^#=UizxSW;!$17v^!o#zdiqKJi~syTqu=ZC_y7KX$v^mSf0FXeR;FRoFdODtC~1TSlis7+YNzsb#lV#!9I1p2(UyC zest!9h9DvI5Ly@DO1eLy9v`yx*kj};chO18#7K0f2U^nY_mS#^k3M;Ys~ZJ*Rd6~l z_|lELEXtA>AG^dyu3YBrgF{YdV>)<--43Tk!AYL8SS;A-50FWW)G3>t1n+B(XEUNG zL3zXY_?Xew2Ai9GirjEA4vc{(lL`Ct1qb6(7DbL!8l4UJ;4_c0(d{rVOFG$QZrr{X zgoRCw)QU`}9NQ8hJW4o*{Xr1+bsD-g`v*thEX90A8l_A*#ibppyk<1)GM?o)=SY*7 z>12*Ein=x&?B8i%oSv$#n9ODjgk;!X!;B|PCX=R@FTjeyQ?1O8L<_32q|@mH-3}qK zralMyZJA+1ugZgvp;hOm%|nE9<=L@j(gnwn2g^#!d8q<%-nVmxS_q5>rYtv{e23sY?kPQS}N*NM4v@)rI!826zo2z!uWD-^rL62hZ z?4$`cjS+c}5NMru*YrT8am>BDH_5s^=3erJmp{km_9fPoXLRWr!_7+^@89O=&J9{d z;@)sXJ~^c*iltUySr&{Zj_Y^saWtNAC>8N=laGA(BYgb%XL<9@H~5!-@cSSH{r(VR z!m}8zZ?U#@na7@dnyemU>rh#mO(sYsIhxJE)QkoLgqUKj| z2{OIM@x8m;jk_Gqa*UJQIXvZLGU0SGAulg6eCl~Njt8uDJLHQQcW+*&a5w12ibz6# zw2rH1XbIh3N^LzKc;*a^^73mp`0ZB?xcoOGp5Ey3<@pS`L}mC!8&jiPiEApiVI37* zy7CxFoijc?q*zRFBI4cCnz%Eh-_FNracnb53$gv?z%Z$qUb1W@dFrMA{J8TIY!)#jw|- zn%^BH+w&`%Y0)4T2wWT_;H zR(em{@2q0DPn| zga3-*pwFwXzRu75%+K=V)7QTC{Tk*lCBZmA1X0^nlY(;^fC7`W7d%1=eMB`4d%GDeIk#-Qj?*-oC}_@4U^WGN(7(2+!ETJBJ6127P)_ zk626Adnwk{?2YF-+nhlofSdqlAQ5a_!2Ky#3YJ$ZODAVu~75R*c6-^n9Q7 z;RZz&gwhvf&cX2k-U~WepH8pCdcVh4-g=9)lhNz-P)PdyA+vnOcwQ34g5j`3UY2;T z@O6b39*keIz9AG!8me**gx)(Z7=7NASec3qY_O>mpBrPj^0NnQLlM8K zSLFiP%Y}jWfx%ONsp^o#QHo9$#Kp_nG(nt4HTs;!x&~T)#^|`zN^9U|!6A1|VCbRI z>|B@{2M_C2(pw{%&nX_B(F zeTBWfeSYREuai-eC?6oON)z{nI1hPQ(61}DdMRJ}>}PrI+2;_>qO@lJ=!ok#Z=kj2 z`Dfq9*3}K1pL6}_9-n*TE`vPcnM>=a$noa>9gLHdr^hJa7{s1I&(P_l3A4>Zxb=)W^ST3Ht{~^n%iMNY&zjVgIFH`(6IgFa0k6)(?J$cSL}BXK8}< zXNePsLi*4v=aC?woA&TZ6Di3sTjz~8Ugx*}@bk<|!=2NTJU=9#&*-k8Ngh@;YJ+Ql zC`nk`eT-;$nYG<(RK-oQ<0HnvbWxy_^o)3F{> z$kwRG-oXJfTL(fwIH_#|mkd3}m4sAkD&y&OQu3-oR5lPrdCyn(r`$N2V@yrf>CuTK z@7wK>YC%y-atQ}>1G5R+otRN98Lzx+2oWlL53BSEh2#9XRM%A?4|_;(XN*}gbOKFl zhg=G_!+Ch8AiRUlLv0L2T@y7yUg+0vntR3hq>F4|&=);>-?&`_eHV6Ne+s+5$@o1^ zeCXjo7P%r2ywY=apAY#+bIyidI`NIa&axp{5kk`M4a4TM0zi`j(ljH>I?H28l5#qo z@PGebKgXp@mw5N>ukvqv&!|wV!XG*H2gY+m18%N0cNhz>)MO0S=(e}`|y^7}- z@POd;c?(%Y#Zrj2{aaZsUHb-=hcJ*T4c<{mNmK!CbI=q zzTo(D$_wf$YinHw!!GZ;vdiJ&5pUf);9xSQ+v$eN2{HtprbY*vDcBwMsb)F*$7ANRISLDLM&Ug&Nk|23Y;1CPdO|mj5#BM&b4J}1Vnw7i z%GVqokMUA6naz+YqLaipTQi;u{)1xj8iy|k_3s90sdc5>FV-e0aJzavTu91ZI zMoN#;F~Zh(dnVTqs?fgC$hu!k?ToBIHe%r(foz?rMwSyRlBTXlbnJ&sez|sgaS0)5>LI^5jmhiQx)%9vCInDT4?*NI7hwQsP zZ`?a!T2%C#B~1X`ED5q2<~)%UfKZ}=@V0U?=e5GhGJt!95S&iN9G#A-iX0&XH}32) z7z{ZXj}ZvQdwaygEgpOB1MJ^=m(!yIj8g3Hz02DA26a*3z2ok?ZxSUbTF0C=t}J{2 z03ZNKL_t)H$5>-2>zcxZ0HMm~RQZCe)8&KDKhN8DkDxB;WL>tdKEc+`F0U4A1_xF|2e62q5+vCx-Mp zDVdJhpHKMLuim7;{Yjp@^cK6DBi{e)Gms<@okyHF^I*X6*Rmac>$9)${hxk`pH5P) zUAx90Zlrk#!F(jcTHnnQB!Z&0M4Meqs=3_RWK()xEv7tiX^YeO4Hn}=W<_p?z%2%{8iHTygaj4obFDs^b$+b?8(h2S?BweKu z(vn7&h?*MD0?EQR?}KkLfT%}T`W|A=IjXWGYW#w>N`9uFryHWSAV{++2LCX#-k;ll zJ)VUrLoyg}@{DzNdCu2jfxED;X%|7?gs5 zYMvK<3Bx?=ulV{l1R)yLKH<)N&!y~|VHR`dlR5QZNEGP@GSA)^j!qYJvV^$X$JT+3&sj$y3|BAh zAZ*Dyd;8Spm@mC^hijWdhP@6V>9DoC%Z3$4^`O+q8mDKj4Kz^ z3U)Wvd2{a`v&xVvNvt*Ny*`!q+&kE3Iy>gt+K}O37#uTWFbx4mzR0PZXRY6(o5V=1 z86+8XZOO}=B#J4Df{o2B#PJbT6~x{b#e#wC2VFcNh~o^Tpe%BviU5H|;Jj=|Ai%m% z!BIj4gr}ES*NTJtC5+7bvmozm?)rv!qdm|DiD&tpGXjxDBNC5wj%iu4u#UbL*heIT zzHO~3p;p#0uPjQ3YI>|Roy61c#K0M`P9?+xiJ!J|vlNSNj(Z;}Dt8VJIi5^8nM|oy z#JTf^B@u#i*}{5PEgWj#Y|cKw=<0#Gu9?ghEQ)gZjvG_+%IjYyO*{1a0|uRxer=f5 z72eeNIHR^Tk<Yh54=n z!PImz;pp%td7g7{bc4LCsLJ9DA}J+RS@MUkU*~`Q#{+JZ$0W}4Z{Cw=EBIGOjxte! zf!RBhh#2|~P8nW!a)T#En|$GJ&C#*uKmHFl=yVe998Wk!kO~P6)HG6_+6tU&^cCAg zp7RVxBmTx;{XS0TWvF;93Uu0GcfC(p)s$5rJGr*I&DLnhA}_deaLlNiaH(zUHn3t7{3j(F0>#Yr1eNrQ+k;`4u;cWxSvkuvk)W`%WgWgrX zSTsFC%sD^0Ht)sW4jO1k4r-ihO}}DV4G0#R7vaCxSO>nNldPl4y^1Pti?a=S_Smbk1P?QTy)xiA_ zIA?I)p+q~U1F3;?p)bt0$%AI<`3AoyM38J~D}HUj*H-!iNd)-TnYqt92^B#}WonLQ zGrCE}D3WAR*05Aun1Cf_wla>pg_np~^v(rUohjz=rT zZH>igi?JqDUYmPPm!i^NpW~>b7^QW912iQ{Ydy6wn1&hhOpH3f)J$D5nM{~27Rdd5 z*0y%J`qcXeNen4bZ9k&(SYe5^bA>`$$*fWJ1*K;R)-k zl9Jk3Y_Z_!7eB_W+)-PHNIQJ$;~z(gKjQAadrTIdL4U}1zw{E3R@~aV%i-zdEC7%a zUoKEl%wRYo$@;`eLK@}lY^-y%9)dotbpsTLLSR*wD~ZgokEpgqfKZh;b+xP}3jw`u z7oGJ2l(8pVa`9& z{C$;Kf7TD*Pa|p-f_P`$IoLWY1T0F1e&*Fa$d;MbBhX6X^jRn(q(C?kG&k?dgRj1t zs|?`Fs^UYw!nbvw^h4|-=)15B`xddQS1VHLl5KS9OhRsL~+>&RQ$tL}N{Xs~uH3V^NfJnjR3@^i4SDQCiVS6G{^*2nLOh zx~LSCC~#hIQdtJ6M0Q8G$pIe0v@Dq}7TkE_4L<(j3q1eqvwZeTUkXl3Rgtcz^m;ua z5636_EM^6ZnP8MDimJrb6?LRhN~3i|ufIkwOEA`Q_xOZ6<0(a5Qxh8RpV}M|Pa=*_ zPtZ!^kR-7t)|%n42Nu@S2(dn3l2_=4np@$SN<%2 zP!$eqJhil>NkUQNq*+Fmc5&8VYlqfqSq-jD&2%!O-|Zo#VqVUJ9-entCze@wtzy9A ztZn+dbkN6Zdb04LcLdG{5`ib6tuQ~RAN5}IGftB5y`;h-u)bzqS9GfgAL|B6eG&Y% zW}RnIR?8>SKIz&6&Jx8jS_PsVCsv%k``-OlXS4mC_nxw@!s_M%9J8(N(FE*m5ZE$c zF0XZ8rKl_l%Bo&Eq9TrnwWi<6P(?{zou2z#-h1l0#(9W@z*x(CGR9cP#`-p6;i!r# z5NOmTC>4a|rKW7E9zu#GSx|r!hs0bC)Xe;0DSdtX0t2s)KkOLp2wd5qV*^wAIFtR;># zrZ&rWPAY}1N{*)m-9+Kz4pG*@T2Ed%oJtw=!dl$l*n-n_%u4Urpw}agKJ>HZo z^0JXin9+9y)9ESGSwT(Ab5B3Z<=r`*EMxo9C9VUPcQ0`?&qmh@Z@e^*jr%~>hV`%qOZYZIQal2o+`1Y)KRL1;rVcCP<;NmSI#1u5{> zlM2szKc!A1Ce7z+6bl~=>K?H3edxYjBb)c3n?|@YoMWs37nq5|z8>~AB^){mbIdh8 zxVrh?T1t%dM2+N4A|oo!g1bj0U{({j@;!VY@BMrJ{@c+mg1!s8ux~M2UGVJx$KHD^ zOOmB`e!rt!guFUMsje0S)ldX59FB2F*evD=_#Auy8XtjXT+q1XmK(kSH>4S9G|(`Z z0R{~;&|TeKBGnb4T_3^4agT_bnN?ZUXb__AXI2$ALwI<&A3uJ+@B0g9dp@2+2t^3J zO#CmgO_2KqLEzwt_4fDv-D0RvMwr&o^RGcezDew>>R-m6JbGVzLJkF9go;7|^!LC3 z1mSVGJ`t9ERX=iiTw`hK8hhl~?gWEr-zNTE5Ko`H){xy5Q3$9j;lf-gbo=y!E}dre{# zEe@p&fsTC}uIKc{bF5J^AVj}NlQJgxJ|IJc6%C7=VaR+HL`)FJi3{}nK^eW|`WH=- za5fr|X+x?FsWP~}gE$l`BA@Wvy-4wW^*;Jx3I^8Ga4uvBq((=-(B8y4E}rGyFgXgvJgB{YDV+R{eGsS9O+@RT11GQvppJPNMvKuFrLpbJe>tw zlwrHI6#?n)9;AknmX>a$r9m2mA*4GTq#FV0PU-FjX$GXE8DQw4JNCTae)snmp5wTl z`--*J`R}y8`##PgRfBxtvq*YdIC>7;Yz)9E04HYu*}HF-*&W>jqiD5Lm?8QoUciEZ z7Hmgf=9df^Mfx*Y2NDC>Xq3zD@0Gxk4}`uLe;kzKt}xc9XG2lkO^}`N6YnSM_o%mG zOBC18Y@v&1tsCyZ4gYfYxa9i#c+o*XWmqfGVH}}_G#|XOmbEdQZ;zZu%{DTgS>go; zg~s+-uix#7{Qh_+tCD#@0B1L`5iI-lt7pVva<)Xx?mUy4yG&Q<;?iNVI`BqJU^%3s zuy+ZY9>0{~Eq!2TDdw{_&}hNTSI3NzF_vcXpYM&%e1QR_m_#aNv95=y3PoaT?{%)D z5ueeANEv(Kg)g9Pe}IMOvF$Aez+9tu2>^-|m!z}IO#)&>)N?f1myV^Muxz?kGy2Kd z&u9}0haIvTC>P=vb`^GbPhtYS?{6a^8e1jFAp`Ml9Oj}!h3LKZAv4i7ggixm=u<+Q zfQpl!&a9N%S-EGmXb!^rhOJW;=RrM+ls_NI;Q}TtV>et-= z-Y9ub5FMK@!=$)Kg0kitJulkb&Zw+Okd)hKoz74w*1^mVtUpUz9!xaWs4=QC=fAC$ zW8_V5ip-iGP$v@k*R&cJHz^Vgv1PE*$uDH0RVy9MmtVPpk;Ww=gi&i?{j2GBuy-=- zREZ*Rzp47;i@^e%96=?jFu`z6Y63w<=#>K6;-uyrwraX-BBRPIR+crx$SV|LJZ-9) z&XM`lN&FX;0@qj^I%J;8d(L%yCswpf&+W4Pye#uL6|&gBHD74_#I2_AXOsY{7;}=p zM`(K&ZUBq@UF#I`P79jnPq=H)@4PXOki>RfHjmbl6;N$3W~p_)ZcJyrCN0=sR1Cgx-T9(*&-selD?0yfF zkQOHL+<8vq@?xP}CW9ypjXyhHG~&6jYta^7uru*BUtH7J)4Qf2>6sfm+%vY=>I3j< z*vR9zKhWE`-`C4eb=28HM8l{4<*1QacB42Ui3ryxUsdQpa9IOi|}n#L8^Yq z9!91-q*#Vx_?hp$rSoFI>9-sS9L%BqV>>;Sk(R1N3IG75t*OtGBQMIYXIui73;Op! zqmGt59zP|?+5M)7dtZHOOgeHWm1qULbNiHCB%1YU zUwF(WC$QH}IA29`XaYT>Zgc;E_PqJM$yV0(U6k(Qyl3E_oTrKPjT-h;%J?7f%xuD; zl1H2FE8oW}=8sGoKohp&Ss)_p-YC=gSIqR3qn^`3IA;;DhX1)6kJKJpWO=~>`=s0- zNCAVH@lQVE;kGIl5Fy*7_~+_+QR%PLAju)%FO3Fcgvkb_y_j)=7eQ@8z{^XO(h)h% zTbahSZ^((y+U5%kqck+Ioq4DM?xU*{{O!fX$HD{_TuSjJ+ z_G2#cN%?QDIdoWIwKQ(De-?79znj6T?0vk0mvvY=B)%m^b$zz5~@BDK7qKjS2 zSAU#2#B!q}0tj4flX6w1T9-9a7?m=fo+?0wGkiNI_r%L{7-!G8P@%@yiYV9Tjhle5 z;{y_Ws#&}3w@gE)PjJ6Kv@g0ZKX$@sFJOw1*`^8e!~rl383nPY#;=K%fw6(l=M>9L z3LM^U-ipf}4Kp%-YJQy})@BQW_T-(oYT^cGTRX~ay6^86KDKPVJXcI{_3_NtRBBxZ%c_=N_?D4i#=QyEdtpQ7BkMhFFD z8=JWz(Y4tPu%!jLDS(VtHf12X*KM5f#b?+`a;`Ua2wXQ(0iy_Q>FCLt!^M3Isr(cz zvyNPlIv2Io->ee5Wv|IbCbu*Hh>*XwQtA(myXw*8ETmbDg--(hry83S*T#q;t z!%^_AS-bEs*+;%JPIh`J+|z6Uo`d_}_7=}cJWSpU-Ozg8Uv|pldBMB52QFG>Z+yrv z(pH`;w1+h{CBuJyI;5)jeBpI@GS$guVw#C--Tq1rgZ-J~wCY2v#1Qg7yi*(p-C1|) ztg8*9%kzui0TC)1Oxv>BPDoqx8#Rr13Pp21DWC(I!m^}cO`G96_rAw%2G&{v;>BVp z7=&*-mdJ!WG|OEtih{h)&Lq#I$fYkg|2H$Hn4Of;T5%MYCNM1Z_=h`$E8FdHSDyB4 zBnsA`7zsx*q!h#tdE8D8JbPN-xIQ5!i_oNCvj~VzcI< zWmy@Lh-v|)tP1UvxF0)wD59DPq zhc{vS=j9y=X|{D=OOvTAqa=+aGpo_M*6Y<7bKVRKBh$(TSoVzN^?Rf#cU;^&-k-B? z-tVR>F5yYZu@Mikh_?v8gQTl!X#UmsyWgVhCpat?t%GXf)Wp#GdQ)$06|zXV`)k*F zVMUoPO#Nwv)^Bla7ZY#~hL@6E8NdX=wibZU{A+|o33|q|rhM1hX@PX1}nImZ=t=Nu>iIEyOYL0l)t@$VjLu}et>-5;kW>Ce{90rA*}sKP+>~t!fme4N zI?J_oO3%$N(Dix{mAj4IKY`fySzEvR|DjeAo_tnLIh+%l^v`Cpi~y+eXT=;$&uYjO z4~PRbct|o)wI|enb7^voY5kk5uQ<$(J2<01NBKF;2O69BHo`Jydyp1Xj&~w-+`3oY z^CxJO=;u{*vcx{$WT-!P-`imIx9@JAt^kC{@z&l%(VOagiZMv$CH3V>Qm2q%O_{2e z1a&_At@q(04`1}Rr6edrNkbw=_Gh-8A1jN_0Of2q)Cbjs5BA4~KT>Xm&*Cq|2BaU~ z47bfaz7KrBe}V3(b*m#L*KR#Ejh%XQ> zy#kjQ78jR^Uan;chs z4cvj-@@#C};zU<1qrL4WK2VsFZOe&LRbM)g8!FNI!_jv_I2qYVNThp3_ah;Q%qqC= zLGKqyeB7`bXa%>3k2H>U7GVH)nPiy)s4aO9;|4snFLvLf-wjOTd0pND;mV4w6$z3X zJ__aB=xRBlRy^cdPN~{f%24)EU?eE?*er3#)?t|^$Ee1j$}SVS|0`I&!+ zt1U?3=N8}8v=R>_a$dO0l?OWK5ytgnY$>^X1%QE|>P8*v`ng<04lqQ^zbl;I_Wnof zE4k2sUj-XdKsJVw#mN!rM1d$((lzFZ71Oy!wK^^YZ4_hj%J=~df30Hp>@EintzhrW zInT?JKRNdi`bE`fw7Ql8X)ESv;;wqrjBkTT-d?rf{Zf9op?>6rdYwzjeBY`2lC`^S zCxt`?Md#@Y59_7$v5#=lx;1gRjv4iH8pyU&Otk9wkdco_gr+_`PTBF0UuwWYUUhcE zDHtsoK!r@1X?&%DpUs;`GjY3+^WhVG$N*UNelt9R1&KB2u7&MZY>?Xk3E$yY^xy_U zeZiMth{I4c*2Nv0%*P+=2JDzmWI{^+wyvoawICdh`gUoSKr0}+PVh0#LB=p^k#DS< z^FxG42TB)(pG4^g*m2YG;)#dG&-uo`IuppJlMp@*<7OFVps=DsBDl&+B!QO;Y zKilI|Z&kmFpy7GB`;YxA1Zzzp%vZf7p2`R(@Irzm`RsnQAJ5>sw1YGmZTA=_inW(L zbS0&f3P!-?^oxmV-61^{*%wiYTB1nV9mJT1p7XfvWT!2k9uq{;#(CZY`o1S}ZGZ|^ z_dUad#PyD|thI~oY*~(q&PEeRA5s;eiLD<`&xPoX+qV4)yL>I6Q!hV2RiW)kBN|*> zwV$5&lgE~kdu>z3ve<_f(kd666im3I*Cs*UCn>DNw9&o@N@1E&Mh%@>IJsE+_lE=P zOYA^NGLg1>GihDze11d3?%_i#JBGO(arMQ#aL5f`mW2ZE+fRie=Vvi!L3&;;qwpvp zrzmKr|LK)4FHK@sm{)k9xl2b7nU`0fGk#2F-uGpOyok2R8+`@rpF%D;x zqt578$!L1a&d+;`Xyk*Ur~$$8UP#DMJy)x_AE0y{E|*jqEM`-y2u*M06@zAOWutl4+yI4=aCh3%!Q7IN{SijZA zQLywu7V{syp!oAw0m0 zc1b<1FjwFVTMrKV=C9mo?CgKpWzi)-wc%0Q@+F?9*C6V`12?PvMT`LKTNw`VY+X*z zhe3;22Yr>5wu8qJ42sQz^GH0T=Jz@Es{AKU zv^A^5+6(VbX%YwG4V7!z(o}|wjE(cGU7zcami5*|K5iD=tV7Q)!L)ubi|5@t{*3Zo zh2DZF*8?9J9uY~F{(Xnhk+-zDg@q^VP(=2raz}i4mn(DVUiE`Xgu&33)Pg(y$>RD# zWE-@^)aW9L)N@=8sLM=fU6348S=N5C_Vk|aRC0QWC0VnPLrz)1ig?gduyp*#2Kw4@ zSx;Yccy2t;Ki*qBJdgwogCp-cr2X>@3Yju`8>|LZ!WNALEM+TwDzHXrN-3Y;oP(ct z!$HJ!?ozo-8WS3={u$0rO^}yF!vlhRV^JSC)GlzAUwEvUSJ&jSWsb;e8Cr(OeGez> zh)BHEsWyBr8~*Ri8)*`|`H~73Qn9FI`<9vJuZ5rbU`W&C9Ud&w{6VstRC*Y~8$P9g zkJj7mN_)K=WtmdI5r9&%PyN^p+vV49 zd8Va9Ume$JzVMMJF#ceeB3BN*k(pvarWN&TQzODQ9jF@jG#_P3JNFPm!jG_rRt%aaxWJxxU8zU$scB570t!qbKjY|+W z>RHQJ4z?z%UQ$oY_yl{6b^Yk@x`zz&;ki3T+U}Rv?W?g>AfcQo7Z8!D-aIjrGQO&^ z@qz-7%P!Zby5MiK=3kI67F;aGU2tK>ebwZZRL0(GGL8@Qe1ET_fM7ZdX6&(u zUL}UuUlw{7>Uk%|lGiIebo47u3QspLqzSZ44+sUvov36`GqtQn%{=S5xl9wB=2yS~ z5a2}^@;D`u>AD>eizNJD&-$GA3S3x9;rL-k?7hsoCwoVnF{caiRA>+-fj590K>&12 zvrf0m<{IG(jf&cZjWHgoiCbv#_q<+;FV;8ByxlT6J4BQM^S-CzU!2?Ir}%^JdoS98 z?tdBGLL^w`Z{>kkOvtmiR2IST);W_~|6cUH1R9N340y2wT_3ugH)L_8?WZaU(u?gD zj~L?K@7iY7;e2qLNn%jMb&9RGTYi85iymNa^hp5O!FnCFlQSN%Rn4G3b$mV##FYP< z-YMpzjIEjvdS%Mvvzfm8Nk1k}G#HDAAmQZ~BGT}a6WQn4I*xCzX#G;RAL$M1lle() zNdM|cF1v~id&@2Gb$C5z%JP3FkJcRQelGa9adeby-oA5r6us0`NJ8;*@J@wIk^g~a zc?5mrF}chyMEl1VLmDxkXp%QdMs&Wcq3+p*&DjydeCKnihuDaoB!2~fuXniBfwwW7 zV|ruUApzlYBEn68puRP4ZgencNPH-?!IM=?mM8>f#XEY zf~m^#oj17Y=4jSH-Hk(*(3@dNu;heYxR!y5f?A|fp9rPIMk91G%9c(QEi*Gy<_r@i zQAR^TImj%LV5!xnBbiwA`!FOSx+~&hDYNdYdhf+I$sYXo@Bu zRA#}&l}krWk4Ja0pjmYy4cVy&32_=NaL=7XWCE`>6yLSr*Mtpwvxr*|!$d2GGE)J! zUR1h@?Y~%%j>Zx_D-x<2S_lLdysdqY z&ObfQOvX8ASk&En{&S`)WR$w&Uo{uwdP_BG152H$6q6r@>L@t-%QDA z$1(a0S&v>;`ol_5S-?p6I2S6XVcRmSsFyIhH3F%{_t#rn`ZN>bZ+`RAyx>+%-frHk zStP==RmY;1iKkLLRXSmJ74{y7NcQbd*V5tf%Wgaoer*NH{ygNoC*P7e%+0L{3U$4l zpv$NVd#{@P?B_i%JG?NdDsz;+hx>JEe3h_k;6}>|I^JuGU12#9SoWvKe_Xzje-kFu z+-)&~ z3Ppk^uqbhRcRsR;2Dm0qv??01`8;z(iHY_%+<#olW2 zVX>*$!QkKI(3uEP`Va24{-O7~uIGN>6WC3$9ZN-e+cBLCckN`TE#{_J(_yJ_m}km{ z!3-Z@|CsPMc3Spx0t441GH7R{p-&->IDMl{kNb?;%vSTQX9vrxGUuv(Axd}wOEi(M z$urj^aj-UW$5GzARD4;us2$=N7Mm6B?$(v(le;vJ_MoPryuTy8CW;08K*ryCd};En zeVLyqG^NH>T$dId_!#1)&-8fn0{4u|#((@k`w>R(o6(~Do25MOo;92U`8C*dzQ%up z;I_Innwn&^pWY+nu)&jbL zO@E~OXcq-b_{;n)WRLW^@%rh?eWeENXw#P7ORAGmaQJf*_*7gmJ3pC;`$yFw{rEUp zTisAIwa!nC1#BKH&;HOMx0y7F_zdm7CRdFR{&MTGI?F?=a(`J9lH#IOvEcPgV?xhn|Vn-QAOatp( zzUb;Si2tK6w3Et?A)6p5hN3G0XEbnF%?wkF#M{iZh!vci#peTN2WC+wteXLfjCM}r ztCyJHNX>bu(Q^6Z6xvJ}WAQ2}BsSs&DJw<5M4dwIam3#|xwL5}_j3h26&`;ik>e>{ z&pj^=x=6}=EDOl=LONJ$LHEyJZN>;GDi=xSgec`Aa@8~5bjv?!uHFaY;NO3$Ji>qT zHAey+Nt4FQ!53%ytid=Wc6ae;J@-<++Ro52MN@~-SQpxsQqIDaBwQ|)!l*AU#XxRT z{~v}KC_T;AXV_A>2y|duolkUU3rsXJ$JzaZUKWH()vWv=h0DPY}OIQ9p0bNvb)KE&wHK$3h(h$%B?cjqlbLuQ_}vt zPA%__W{hzqBZjy3u?Qb3x*yFjM*x%w8H8r(&Q78$hBW5*Nymtz=F;2e5?!C2fH#G-kS z(dMB>Uux^ZD@*XF1>;hSt4$*Mg{4?et9z4v0$K!2aOFq7X74Q0l`h8?+vUDrvjvo9 zF|vem10zV>RHT190Ud)!6fm~#lOv)^8EPUVRJ)p z%-}gJ@DBD6h}&gcKah`VAKYB(q69gU6F5(~Wlgg}d%j}6#*_6k!jsXN66d%CNc|hA zF^2fAP?J9)BE?3i%bmj?9$tpwmA3^Ia{-Xu5PAG~LD6SQwZEEw-q|wV(3tI=Qv1A0 ze27>(#3!7yNZJKhQAMZpQ$|%(6Al}tK%pK>e*sUly>08{ z1gDIM`9es?*!AX-cxzi!;2q)MxUB$4+KhT#EaPHiOoW0~p-x%5-23jUYChvUDtja^ z{%vPTs$bi`86LBS<+^gpgQ`IVMaWR7`Bwua;F17`xWYIKEGi?_4^{v_gLXfwDg0E| z;fJ1uYZdJ|NZWHXfzAvNvnr1KDH5eC_-rf zK8cq=!zVp``gm=ekLMHDh}YvCLV#vbGr#%!xHjjESL3sqi%}51RPKXik53#(d2!J4 zJp{Fq_*e9^Q@B%(?8BxHf!L!!4g58h`ik@EW=EI|8J&sJ5GNZqkk9~;txyIj#0a?l z)*Bai7M*Wr*B*<<0KKq#spL|^f*WVmR@fv@*!*pjMpphrWK{e7D%g-cHS>cAlYb%? zxtcA7^uEK9~L@(P1I^1$QG8lf=u>$o6CF9wBI_jfgyQ0n zN3xD=c2V>NA3clDqKE$H@u1GwYhAJ!i|AEhGwXL1{h?+RYh zO<%GE=wCPNzE4EXncTk&&;Jcs=J>u~IJ3;gpg8l+09kc*V-^`;sq#arXHU>@2A^ytrd-m@WcpCvW1NaX@cFtRqaV{5L-(CoYQ?T?@r zK+r7Mf9*qXCFcHf*mVtVi<-V_qV*Y)hHXa8xEkv<8Rs_8Q1VPGX4X4U59bOXLnwja z{roH;vRCD0;BXli862}^%$TbdR4`trq<;sA2+Xc+Z^<_6drGI3mMuqQ3(iI(^5l&L z->viBCM}~6Zf2(w$$wv0XBPaH(ToC{we4;6yQYW_%(R{NPTI1gH!BE?-Ot+S> zT;8I|N|u7^tLU1OXFAgZxZhV^UmL8EYfyRn?AfhMPYjx+mu-VbmqNll8qJ%HFV1&= zOjr+Au?uG9aOWUMtLyjt;#%meI-F^>L9&5cl=){v#0p`Q>E&*>L>$y}Hs_bN4Ng_!Cu-OUgL}sx1Dr zwbc2Ce9pHdu~B>?C_ zi;IuU#WE;MGirJ(J0;2JDuG%U0nkcVW|g_beYzu^vP%!hcq%;pUBsfyGz+vTaZNNi zt7B(t^N?K@aPUM?Qbe4q5%>yCr4s=ZkpAf{gm{wi51#oslp>K&mXa{fUqA9Dc<>uy z$Kwyi4ZiYGU2a1B1I$%ISQh^3_vO57SWKq8OKDFxo;yRdgsuZlF96*hH(@J(o=y3S z%()RpP+~$cy6y$HU3&hukd;oozLTKPFIOL*QS-)K$jN)I)v409o{QDx;jA}5+H1@v zF5W*UkETLiX2}RF$YA8{`QfW`;v*KA?@hWkF1M4D?mEJNv}FdYp}4_(iYHmVc|?jS zP>}p`(PPEOw(u^$`MM>?)~ov}&3Z5vgoXV>vr1ibeG)KfA4qEwd%9L{ z;v}SKW2B1R9L1ku+}f>GnI%PVs}EDejMF~rXxYZ_Sni+r=Lzd!BcRz zH$uE+IW%kfKvO|FwfLE1DkUKf@jveAh4OHts$AxP%2xv%{&gvrJ|Lfh3^Mz#w$|qW zA0zV~ZtzTL{(tjqO!7>&bmk2?i&|R<vqiEtg zNV_x-#K^0hQ-SRL7PzLcRl`NoX74grt8y2AIyfL&z+QLkoo~uXacMc(bsdHq$y+ub zpD7)koZ<~~5stGGQK63(!Zh{E^L~#R>ji5ZWHBh7&4$?=o-4oS(H72Ab5MAS@Qz~* z?e&h8RD{iSPo|aRetu*u3o=sNqZl%%RU}BCnQ5X{QYtSuYh6`u;2334sPuX%+x>pC z21C^A?fU@ujra8yqY@MQ=fSQv@t#n5yCK_T9vW4khoRAo^xw9WD&QHi!-AyUJ9e&! zC^Z;WuP2`>>F|j^J)6M5S&i;@!Y_3VuwN}tTmEtvkl28sP z!aCWsI54dl&e%G z;G_`eu7wH5{DY%Gij|v4jsNg?n1ntNf;7@h9Vw0PsMizABaIe3>67Uy8tCqiJ9Aex zZQaAJ{dTf$`T@mPpTuc(K7Q{jum&(-sfHEd_UcXhaU|`_Hlc|q@8jk%_N$T-=53p} zyR9YH2F>kyt)9(H<}t+Es;XBtpT+?AVDUK$#Q>PB=?0<{`KrXz>4POE9?2{!-rmMiM$4hY2p@O61O}uvzk&X_;okNMdNhdThBx#?hh#|{W z8&Pk+3!Wt65oc81ukSDax>GAA)~A6?C4M)$@75amHrztC_w3(GDLN31oRMaFz(nXYBCpC!F<7e5sGjG-{EpJb zaRgS7H-2jyqkePGS6RJ(O2Xj{YX#~FaTh~4IPne7<8MjLxRD(?hdha!Fct@S=>Xa& zKH_r9!KoTcjv^V~Tz$hh*)vk7(k4FXR|*t@CZr{A+Q$=QF%;@k5Tgg@mzTc13rxywcugB>{rio*nXWc6 zElWH4j;#_i!)mjGszo$_uHZebgj0Rqy|+W5u0xrMn7zq`aq zj}=amqtQAT_8?A>#*x+4a6xAMtky1+yFO~TBQddkkf62n8uQM~F>~^h7kppqUv7*3 z(^5d7zY|h7r?MYfhaRXE=mFkXN+GVia8m5qIubf<>ME<-$W>x5iXACw)}#Fs9h3`{ z&aJ;mz#I1%~6aOOIR>59b=1jf$~9fdMuJSSL%5Wg`X}yE)fFgmF~Wr=JAnk zq87=v;F(%V4h94Je^rAlBa+?nc-m6V99ciBUAxz|EsV_^r2x)u45|neLG@C)36lXylvWX+h3MW>o@fX!J$Gu4UrI?PI3=yRoQQ zTo$Gufp;RQMw}Jd7gQ~`e*IS-vNhC~H7Y70ZX_P|1CTa4RRSg`cNE_3k&%tZ&Wy1kOtlvqWUrHd%|=zIJ4k80#vaGjcxo z0}T>d6*JT`>8nYNCiTuL%o)caM<(if*xE*|XZ4rq_f3V#zO%c41<&U$g&*n!b_BmK zav7BD3qP|^hl>3GYqAlu{S!auB9;Nr0pe$Fx=gZ-!=1a#%%FpNy5d#O+`~7|gHDKm zQ{N|{mLGI~pN&xNp>Rm|AC9-_O`YbU&a-K=<&@0+E);e3JJz`Li02()UzJ{o$fT2-?}cg|2qXU?gU9^keZL1_UF-bAQxYGPEq{y$@ z<)6P(#=6i5=&@$WX4v1U@;aIr^CFJ#^5j_;$yTz)9}t=b%}59f?A-0sTti`mBwNgU z`f=6v>{L!dF%Uc@_-iU{(%R^gb+XQ&TQW40kyfoNBZZDQE=M$Cq%OH3YtU0;XXaXN zPjz`yO5)cZTPjS*K+&k0M*qq+8{*?+W%gi?h{%JayJR0nRU6wVb_{yd7FeO7EgzNq z2lO~J-Gqn&nHEx8xPn&MtJmH9ljiya&AhzyQ<}}zBx3X{=KTDBS^z6!jX8)`orA%_ zkS0G!ySNjRjFG$qbCMaj3B^ng+-X@A8WIo2!=!f!IPx*cc|&`;pB5jZKIa(dn?!$l z>oB_i=d{ByVKhh)D6CwHnP{>ncN=T=MHw*S98VTa)ZiXLba_zu-Fc-qjXIy8I?&%h zTuXOU+Pfz= zEH8kqn(Dj-te7*^;``u07KRN*$&L&o78#?4h|@Wv5L47pdT{h~6P}ZS)a_c0afrK@ zhWuu#z_dF@l15}ows#7bD*sqpe|xOF8m)Zb#|6Yt@=F}BkRFA*Ub4zraA32bIT!p* z*VVtBF2mAc*|y8Cd9~JHm5za|Xo+Om*6Y~f-o6l-R-qB~wSip-?n>2XM7P{RGigci zod;d1k?sx19zD;h)&Ni#Y{F&RV05wlgzM$)IdOFyGBSs8 zGfatuVUlFoMsq1_j_`yYZr#RrHj{t%8gj0ZG2(k0xHn3r8-cqp_jJLctzv1FPVcK_ zUbHO^*=r;5fxKlzZh{-2z~ROkn&AQCKABEs2Ftj8L0a7my;nK5DxB-gs)XssAz!uL zcuaF6A!Fwz??Ud#p=ZcIWF6R|LlgUg{lV9T3l>H>bOMV%_0da~eYO-k@C(tnQ)B!) zIY|XKaVfMvbyVEhxos&UmQv{q8bk$$w*MS)6R7fCX+YF@)j>~Z7dxAcoYQ;VHhxua z!{_O+q@oBEW}G_8BJflYXa0k9-g^4(E-O!-5I-fJ42$w_P9J8Qs3Q`+@J~(Z3@W`I zx6zp<&W!uLM9ukV3V!CZC}lQ|!*~ld*&R)|DdJotgOoIc`)G1L`u=duo)Gx=>Cbb$ ztiG8_pM!wKfqz4kLSok#c~bSC<|yT*IiG$+V^U%RE0ojH=2z8A+7W`mU@F5xYx0Z> z^K!AZ5RKd-kYRf9zCR*(h~<^LF`PDOfQu|DY@=6Xw26_QB8y2OA=jst7Kk#0CN-@h z2Q+r-epf`df0vAOYOI>_iA%UoHfEeoP)c{x_aR<%tn`2)V%VKumRfxloy%;Yflo&< zg*70+xMp!)4_4|^o3A=oS1GdX5W3p<2!A<4iOZn;#50Jv#p>KpL&j0HyO3qQ|6)V?{~=R2E&I;180Z*dA!v3PD{R^ zlM=gHQXY)Wjfp$=$l>)W{ma$=Y6O#_O^w6$4SV z0@ZxnFU0USP336r5e@{2>_uz~d?uROI={zF#PX)oo67#Egp4v|9jh;)2Q(K|3Gg)* zve*l>Q@|DS3r0rV=1ym`6aLWxukZ-X2 zYsM`yzQFdtxXC!^owH1qJhD14l%g@hjy4QxsT;Fm2tZ-UVxr+zPxj9D_sQA&ye07Q zlM??AGUm2D7fC6FhR?&r{IO^SkzAM985_0hb?Tc{@M(j!wm0WyL-U1IX{t)sPd#(i z{T}NxhEeVjtkwJ1%5$zD-1gmd_kHZ*{sRu2 z`Ww*`j=%heOWUI1&|(?<355Ro+%lZO$=Cu;(LtsD6_5z6^Q5UH)@1ESZupfKmpG4B zzWzXTa|_7AY&gVT6FL`Fj!pHS;=*E2QT|s9pi5J+2dkHMSo36Ubbe{?8fx9pg&h5;(cz{s3hX+wXbFMpU0&FN?jkqh)vNAd)BW&(~TyFhr5(IH90RLrLhQ;!iIb zNH}YTPgRO#u9DZGZI{|xpZiE)Jf(gc;~4ecOoDE@jcUJuYQl6z{Nv+<7EyLA1tLp< zhL}Tod4ovCm})AtjzzORGB^1(2VGmXBCsVg79Lo)0U3!bR@-x1#?3T^3mP`B92xi& z)Nd6It8dVdeDL*ZyKhdwFx5#X#=GNZ&ZNDyrHlSHr~8w`t^j3Xa^2i6 z^aes)g}&n)oY(z{Z>+eZ3Fo(Wy|GrP_r>4=^(_TZU)tjblni`ap%2|&V(j5@9|Bwo z0T?d231o4OzEpu@A0?nB+EsrmyuWGO zy;mZSqusasLp%0?bDsstHrN?xKk17n4ltKQTiKAa2nw!h%*>=npWu|$z#_@{iT1@D z*(60KucvXvp((AS=k4v%*n$zCDh~iZB&+e#c? z3SzT3{5J{xFiKhF^Ov6hx*<499`=lI{-I<=VMPLg{oBGIGcr5?om9RK%_TymbxHsQ zah(e;jh$l^5lJb76_SKG_-f&2oY90h!2*iyVR-0Q+Q3UOjgB8+B$|M3Pm4=*TX&Hq zLaAK$8DR`Sk5E1yT2{#V+hvPp-SykdCz$%Ez@mHqG{0$kW}_3uA+5zT>d&|T#1^XLpzSD#E8WxUdNf#ucM-YVs?BQEnP+CZ^ z$r{h8F!NXJ(C>rSGha_PaO8;khiE6RI!^fs6fMC|#iMbuBD>(NUJoQNvDy7(_NjQy z+GR(CEysqVpR2}*ljS@!L?MOzJ|Oo{*d!+lD!nl=mdaA}>o0$ja9({7S6*VhrS#mj+VobEb*@Ik&orV;(#67R|rTUw0!H~!Gr}U$+6G>M_gM3|FKvy?9OuOAG0SP!Kxgb7P=Sni*$MO@>(5`lbX(VFj!7l|?9+ zMLeNOOWgI~Z9abOAFFT91bwjd?rmGO08L%oCCl2iOgn2KwtiuW34OLSD)!>~lY-qH zK|Xy~(-!f_m1QMb;isp)cKLMJ=n&q2(+$4=%!0_4__2{#>Y|gDGRHFP8?>={H5M4f zmEuG4?8&P<{qAl)LWgq#a#l$LGo~|^s6;OH;DqwyF@?lfXFp6+)`o`Y4OZ6gTNrggfg#tn!(l_EE0u&KLK5+y{ zWyHNgTcx-}nu#{v>9=v--&fNN8F8@lw8<=nj_OQF#LuTHOrDP3Ck2eQ7kH1YMsnA! z@gJdj`Z+XM*Xo@u%`&TyR>&~Vmc5RLiW5u4k2I)n{0LT^Wp=ymq#9JPTxi&j{28Xw zNc189XP>WHw$*k$ap)-5FEfEG#3mT)P!wO1+cBdtmVHW&S(p1j|Lt$paOf)hvq6bS zU-;+iehJUHI)kb|QAhw}Z|lXACS=ax@QgRV#5mH%Vanj-i2TLkA))tQ3xB(aZ)6T$ ziE&fZNjr~}J z=_sC2m!{@@bsyc-Eb7xB6ZOYWU}tL!114HOhzvlkZ!<6BpKzQCeC~e1Iur{$cY5(n zxj@(XHGB02rn_KdVsv^}GL>c-d5ouSz^wItrjvWjF=0Y_@nPZ$TR+7tS)zO!33J*5 z=bl4OCd3oVKTd$6eYJSJvv>2im8PznacaIt2LfSO_w^GQyAm7K7c@)Rwi_?LUo{Fn zmZ+vFGW2qDSdJ1_vIgUtk<;*fANhp!PltyybVXlGb_Vv}6?+eL-}Sooy_>M^hWV}5 zh}bW7gdId|B+bJojW&|xna2+!SHDV@xEs837nAJ4BT)T`##}*IKK_O1l5+-&!#@Y_ zRItrGAnCrXbt13qp*ZIv_yX44agx#X&{iGDbX8TROqJ0V2fJO(4{5$E=zkEuU!s~h zcy^gPyE&c67-P}wuN|8;@KEMu>#($wi43^hg1hX|L2tPeW#v=sqt6uBl~uAYzHo2u z$8{s5TuMwy!9FK<)#d5UPSuOo)M~VuTnQKnJK(%wg%P}*V(UebuNEu~0NzMJ&5vp@ zWzP8pWaN^lw~KiQv&D=%2UE^Qd(aeb$c5q*Gg6b?%)n<-OIPUyzBHBdPTFUh1sj-x z7|f@cu`MMJ^T^XsCA%PIs6E(Gfo-exGoSRoVeewi&;Q6KoO9I;?C~wwzb7O84^3|s z)>hj^i+%-4f#Obam*VcUxI4w6CAhl>C{Wzp-HH`=DFuRiAwY_|26s8@-}~&FT;|D> zWWDd0V-B(4)g)C;!waUo{*F`3yC9U3m|?^qH^e2Ve;Bjm?kgHFS^CudjTmKhZb-hA z&sO@fDFq2$FO{DlJ8Kpa8B8746uY`rIs|dHf9g@O2+ylX_IXcBKlvkpH<8wOJd8K; z<3-H8*$AUrUY%ZUQoKs2CJP^rBpEHye_ab-!9`F7u32xbxjo49zf)4~Ql?9E5%d(F zxTGXU;>A-^6aRC!fSXH>yQlE?mh-qm)$uCybg^H{@H*b8LpR%|=bFfM^b<&9v2bd( zxi#st|eH71=JZCo6t?b1SjbP^xN9oRkc6Y9DvZ8brm(O z6^v8mncJ7_>poMtFy#W{q%pLp*-Sk5V&8qxU%Z?W*`&)qRwHT;%l6A9Gh;kOqcHgQ z+Uv2^L7VjVNpv0KMxz>a#Tk1&gfUGAmuRX zxRq85OuFi2^|m+uL?o<_;qmp`*KW9e&6g$>sWmTuxJZ8Ka0m|lm3W=exX;TI$uE;` z)!3S5CgzL@2LHXrJ^$+qzSq6pN35r#*YiL#lT<*+MLIau>pv3s6#ojenu+giZ-AN{ z>N^T5D!0!68*Z>K@$st^F6^BY;?5)lH!T2c_cCTw4i$d8YODv;!@9=1OK1ukg)xM#@x8W#Jp*M&wi;ZwvmV8>Y!LP}d^{n{l z0`~UOnPj=|RVhB88RO^@Y$OlszY=6$6Pqw0ZGQSC{v?7a)y~Db_kI_P(nTZP-3NkQ zNJ_@VZQ>O5&)QN)JdZK$**71(B}WQk{`aD)$Hz}Gp!4kqzrO+r4kGJ>BbBBoDs)OGtO(KnVkv63u?3w-O^wW!~_$n0R z6fWpYMlJ6MChP6dJ1n47n3W}GzKDk?}r=TWM0T?Q8uh?d-x?eDOyUquVQ81fafRUR*3{(2DS72}OM2O{#7!cmswlatopZNXF6GpsT9%jv zX4^dzzZeL z^#F2wi2HrJWxd01qbQ&DmNj<~+|4v3U z{^GGuu9kpBaw+2H?_qB2Ug`cf_f{hVbPPmlBx2lx4Xk_?lRoUfprV2bOdND#`Q7BT z)>$|5LUCL)kVh8z7*&u?>slidfki5weLjN8vvU{aBTDB2p2IuI=>hb_sf@k#7!r^9 zq;V$SY+_XU@K|E_u_Shg#D{sZ)8?aV3;_Ws6P6_z53Fytg*jto2w#WVqJsj@75r|g z)2${laqePlxwq4Cf zI9{^wtb9H#ouQDZvn2SU`0%0#E%pUIJx5NaqJKhC>{dH?TFllYZj1OS1Vt!m4JEGs zgpHH#uNZ{S%~AU4XVZsVrG!k$Cgx<8WP}~E4%UdvoXF@!8jzUx4NMG7nb5x5=L)ek z!KtQfCPo19w*PH<4&tZpoJbpFEjG$kKlN=W+SD+qYfL)=M|8uq2tgkD?USzu3GM3bZ3%^K*~lA^{u&9b&q$j}I+76OM9th&$$z zP*1@({NCoXh?F&m%J{Y*eYzF@>i#G)Zyc+DYpQIPVdR(DC)- zF5ZwP^^H4E$140_PP#0n2|3)?;mil;SX9cnE=xBjX2Nvo(5sR>$~llx>U;}2eD~ZfTiUX5`@{Fkg&Km z?a_SlwFxl4$78Pq%_6@;XQ!U)P{y}sPF`R`icbrc!=kv@-Y`VT13SeJ6(4vWFO$aH z9%qLs7{56HYbr8ahxTE5!)FYVNkS-X9Kxs0R_ENs3>2>Vc!&Ka+1a@ub-)?7ItTk5 z(7N^XgrR9mcuD_}S)GB@W;BUgiGdy9u6~r{O`b`i^?reH-rC%!NQk)vr8|KjUfHx6szbW)kJ|OT5p+6|)M=*(IPa-Lz-@&~n0ElMye2W^W`!>*bI! zaCB#fU@Bh%B%Vp5`?2Tsp7EbOsY#E&i~#z!g-oG5{;^2w?~dB|4pERYw@;?Z`KlWj ztb&4CC^6m8t@d6_n4!|gq(nYSa8{;UEfL%G@6S}7Og-e{T@{)*+!$=6{%O#om|b^~ z1Hm6zX=@xwbpl+XA?!=ygxTq9FZ9bDbXM^(V&dr@SIO~m$=Fp_ zpj=nyqCe-j(2FsOQA|a~KauCcjcB;#-48)_DQWJ2>sjeA1K-mk;0#Orx;?Ofc_8#>L{)2h>JA4tjY^iF{L4U<2b z1#*h2yEgTZLu~P5G39UrdaLOgGVd*+gQ5|f#6EtA3i*8>9u4cE5MyJE0?vV7{AIA{ z+;-_Vk9td(CyG!($@ay(VObTPc%bvB;~FPb21!2&n~r`#$!67MwtKQ;vf+N0y!I>& z&tC?)!X+Uk^x*LF@~&mx{TaO-`$kLPUwe(jh{n!7N#be9%{(Kcz=JL=$t_3C*SlJZ zDh|PxJ_Y|_aqL2qRJM^|GmUWk2${Y{32I9-KF!RNwg$FB@Rj6XF${-Zvq9we5^qc1luOdbix#QjOfFWnT{_f9hOpZT^X zRV9nfVr3|As*XaF~= zf0&zr)(Nr9v$Ktm0Ziw#%jowxSi}GI(Wj64|E0<*Dq&Gd37T-s&-z|V4N>w0^oH`P zYPzC`!rjJn@W->yed+#7G8%Tcj11&As?7w1wGIHYPU1ZKR%etgT(2 z@>g;-6Ui~q(1!T6bq!P_^*1SbJ}MlG;%ua`+}_JC;AQZ=Dy8W9oLWApX?NxzYOrbd z8m7d)+GAm`W1U>65@4k6F-cA;EbxKz2X=-sBYOWQD7|scLe@9pq@&6N`_8ML9oNVc zzo*aH$A}O=>9e7?DD+8Vzh;n+SdAqTNc$b_@qZ}BT0xsb|1;2)FSQnl_Xw_SUF;c} z?k*NHb-J}*;XT0hVsqGX_WX2p`URH;Nz!V#N9qP+=0*hqbfQ4=QF`r!jTi* z{|sbHzA)cC8@74hL?8AXbpCpLVEYo2NHy>-jEu7DG^k||Yr+WAA+Bq#iJo!l7P-#b zN+`O&H?dPS$EL2!;dJ&9E97uv&2&0+AR?lG=8hOQciYbN1GX2hv*a*s@VOmUSFy;x zTt4n5N*KkLH&`(LMS2?MoQ8Mtgkp!rKu`} zP#H}*uUyCGPht;A*HBc7gvT&TiXv{#(Dh5jbGy=%D8-P*E8ZJ|qz#p;VsNiKMij}< zfVEw;a2-GDnfc%hqKLx5@Diuy$ujWqQ-u4+ICa7$-+hi*a_;JQPJ|F z$L+X}^dI)!0?FRwUdpMQ9JM;N|2qbJmF+A4j3{k75_luPYehW${c~FW@=GA&{Dz4r zQMFywpc&NqeP20CUKxFCDQ(*mxBsUFkhU3|@STWG|25nYGw9DBplVLcDgf2ZU@fj? zw0W#nK;|;g+!4}G+n_baU4ZJMB!eFDRZVpX7#AJOU^+F#GIrY zOO^Q$RX|tQ=-01`!>k5*)mqE0xOwZttsD_e;Y(2olf8b<=@#-9yM*Ow!$46hydQjQ zvYD*KUL$0qgH8J-k?sLhp`j_ZrOwo`@CAd5aWasb84@LVtxfXA#Pn89xs6B{d}i}T z{1|!4x3TXLzi>k&+5-pfhb&@i`a@DMozS}!;b)QvH|GQ{jZlr9kysxQNe`1v$)8O* zqQh-#|Mt}e7<^bcY}7U4paWxuHneDXtfV{kye%Oqem|fLokl=fv)IeIF`#ueXJ^G@ zClTCtQuH$@wB+cKf6s9R9{WFHi~IVV>MDe1oo(_OR?HLg6)NyJP`}hM*ew2gC7{Z~ z-n^9Rg^>FqAq$qW6)!rm_>XJVI_Zg8#TPOUxs_UeOS^e>bQ*TSV(!c z)q|_SIq_cm3zdgw$+e<9W?r#Amd|$_ZjwuRg_r^raaESN!6oiN%Mbojoc2C8Zcm#c z)p-hsa{1~NSvJE=6fX&1Ri}@cN70z+I?D=KT|=o3_E(EjesIOR)TLEc#s&6v^lS`{ z>9X$eP)PU(UB0t>&5MqHe<8XzbAZZ@X214r6#WR%bN^=*OQ9k=+ZP9ZL*ZTqT?7Yi z|1*5+GN7mtf3enr82P4?ZSVlG3#h(0|`TD2l_}Na@Gv?nuuH+HZHZG-fF%dt+ zwVYrFd9d4@m!U>>bal}|hUe!ACPlcD>sLi{s2(~OPmNPQE2Fpt z>I@2FEL&FqiE*ZK*>CS zu2jk8cO&9$`PW+wvzp=+fk<*rnjzP}S)L(-p&<}HHc6p90MYQn*e3N;X3aCd5g!yP zqS7qwOCK$r+~DVyVxht5LpXnq;3cVRM>t}M+q~Bj)mUx`IZzg~L}!qGEa(%BDIB}U zXG4N9Xsl7QC6Ik+R7gB`7Aj|QmBep4p~&X2{$c(F8ajd$Pwdh95+ZYq7<+ZMMoS9` z$C{BMLZ}UdW?sQe{_mLVq)t6hZ2g?x#cAR9ykKiTXl8)ZTLXi@9q+C>KNLEpO?Xw% zCG?OhCp!y2ZI!7$Wr-D}Y}iRHlzJ1@cTH?|Rwa{pyblM9i@#P0b5UUHG2?!Ci&JPW z9FGZ;8uegF4PyO~Yj_mt5Y6r?iGF1a+%!dhN4mZxX$}3wFVzc@7g)q$OBGz3^B{EK zoxt|#w5eMFamV5(d04shfY=G*)7?A`CsI`m#0q(53>`g1if)g?-_-Eb-rD(6Mil4z zKs}XXzO67w|9SXNgUS!3@0;-_CIAf)ui(7>?g;+D%1}`xskS>{8H>XlkDjlN;PGkH z4DFRIDyC@o0EW3ICZG6aM{Y$qZSPKl($zV@eKl$CUZYdX(x=FvJ#@CoVal?zyyAUR zir~@oJFyJ;b4C_v_cUyD3*{)Ph{Q6u%PQqmv9yB9WWJ(BYl;cActh9aRHwMnY*d>i zh9lJ#Td=&bT81)1la%dKj3N$(CX3|N0Ah{4e3-_&v^}5M#g4oc$9R5{;NfYH#s3_k zf98HyB}*tc#bUz!wxyP4XvY}40C38HP%8ul0k9qh0am{JEzf6*e-a5dhJ;W5t%jOKm{Ob#j z&{gJUo7t|-gZZ;s92Vg>i{GfGcw$E4Fa~!9EoMLNkvw~!xPzwx*6P+AQMqQ1vq4|G zX0n!e`TSNzS~|sy*ODzr_K4D{=yejMoV6_?bqP-H-KUCC;-B;yZ7>rybq0HQ zNSCML{<>!5jo9F>`9z{@d*3hKG;?lFjB!icaab#puNLKlcYFkCI zOS;(L_nX&te<~6iC+mCK&9ioRrG~COLp-*mWui*nqf!uwO8zckcL^!ANkV2EkPg?< zT<=yAIK!6Pb_}me>!@^DVs`A5eRxwy;8Mb%z>Fep=3bQ7(U(N@_6NR#5~4Ko&@jqp z{ZwgqqT;vkH#kXeWNkFlmcIuuhZe9XRaROOdzb^a%C%Ux3WNlbF{4a9i#6g2PTMhu z05mkpca5n3+WLx9kZDoau|ZkVO0%}UPXw0?uOsP~f;F68waEdkWgv`m?57-MQX~1; zlo4*Tv*S1v<>g)jnI}Pot5v*^a-f^fr)2`L*n(l-z_T=&*ROxpX&sb71q|F{9+I@O zNPR6B622E1^Jd>Wv{QI9SZP1HQi%J@$Nc=pGD=g}thquG=J>N`!NA1-I_c;-4DSCz z0_=?jur(=i8gueZ>w4NZ%snK_2xk0{5KMKNz~SmyHac|~Dw)cb-d0O+FDg11jz%Lm zUL$F4j?uh0aoVh2eoKCB`WA6jh(b*8^MC!E&h-u%J6YM+X|-@4XWu_o(LxfnjI8M+ z|8{nwiST7c%!LPqhSr`=)?7RTMs9m(eQut3ZeSRU<%|G~>G*!oWF`zSbmP9G+(Vdg zVAw`($T}e5kcZy73xnFVrICyQdw^d{CpnC~I!#ISOV->R-j^x@&;9oH9tPQr2}{6! z-2(_**=%5G*!bl>pP@iO?&Q*vVUI5u(i{X2cp^+N zB)zxYaGGYtA)YCphhx0j+S?%W6xrKn=*@I?{RerFySMyKeK zm~4S8BADZ~C3$k9Ag(vGbIrHn*l+KVH7W5^gN1r_@Tx!sQOcI7IYMs_houNaHOA@y z1K<|kr3Ly&4zn=Q8Gfz#pVWV|u-`T+#7?-qZ0;DE_3rfo*^WFwt>)`HlcyD;>a`~} z&${YtvHx_DMlE02bu>|$r(UUf+asGUUrH2wt^}iav>+v$K9~Al_Ky#=a41bu1By8%U73QP}Me&eEZ^=EO$+x zyk|@P>GdHUEs-~k5~)6L84_2|d z*g!^BmWY5cB|_*h=iPP@vffcfhM}~&OO@9@0rEb5Z80mCB#jgyMWrq`B*ykt!AUQ+ zR6=D}l|nj3pD?;vk?0eIDTkZB=J7sL8@YVO|rPii9emlD5 zCh*eNOBvj!%ko?yvhX0KUpeK;pS@IUBuJ+yBfp8*Zu!Y01y?{utf3wI7*_o$GD#Bk zp|+G18O;J)cm^za7E1gxni-&SkAIgr3^&QfE;&k~L^Tu*bJsdLV*mLTDs3BbB$Cxv z#G_KCf&i5%8Gg@AU++mZ_u(7*XhtF~y-R-(5N{`wb#^t0TTSi}6h#AVD~QnJY=z2x zD)@01OGNeZGnEKIQX0)i`f2vgV~>w|dbyHu0#3C88`_GFD#jY2H?T_QR8uWiHts0$Q&43s@a@~uHy?Ptd1UO8cI}Es(waQ>U4noE? z84Cn^>AK^VTBD+JA7#Gbb8F?WTK;Q7Y@ex4i#A6#L(c}@R^S&PoZ&F_AMqFd%sf%T z0<5X(^~Qbtx-I5}0-jc;GT3&ze6G=c;#Y;o1Inzli3z?Lk%c1+wSVA4mzS z^W??uypmby`~iTU8`nyWQp>oEL7ox>^e{*uXcQ$=P`1LL=IVULiVBoOk@2VMZ04SVF?-F=rHGLVN&wkN7snbP`ba^Uo5;DK=d|1P8rYe8ZEh?D0fNzT zdj-)XyPQChYBDd~Zk&&aMBo}*^oa7Xt6y+wbP|LeFAfgtG=lWM7*#uEu&D~_=ec@< zKbPSX!8{B|iCM`<2we6bwmWvoT8?JQq*+v2DTzBs8^*uJ^kE`~03ldzk$;Flzsnsz z^4Waa6Ky{nfe>;@Gew~%?`&H@ruhNCfo=XDrrZQ+xCVogYR)GUeA=XpD_?+BibP+< z3&VVi-iP<_Db2#%0F*0dftH+FK}{mG`0clWl7B#*1r^a z>cq(n^RU)DVpfBnDSh%$p?CcA#xfUdh8t-gJe^5Ii6g8Nt&O1qq=&Yy(k9qR4Yl#7 zQBj6+GRY(WfgPHVV&Ng3^d?4i;abrqs%RjF#4hs-F(^aD%D#n^CQ&U<;$2d06|3Sy zZFv@(Q~$(Czuh_87((o)r)Ulh+%Hx1{a=OjYXrep z5v*fNMy4}O7_f=3hDw|4U2hJSs2E`O_V)9HT<^S%o=0t5$~p1uk3Raidr5Nbo@T`P z7?%u~AESjjSMsi}C#=zV2HSt%@6dOi5b}lP!>33A+m*aNA?TCf_^um6BgwM)(^Hke zm4;dH&cj0%>y*oa3m81;FZ_Q*WMZ;P)zPe0TB}{)vEJ9)*EorXt-!%8Mzy6A&w~ec zr2u%i#OA~j$?H`93(45M=-bNpSbYr!JhmE-*>{vNh4PH8DGF|fQon_i=32C6KBF+d zAbzb&d+q_WV@#=&g&^qx_u$Y_G%RvqjtaSuoJiauyDjxqKQrUT6*$fWhOwid-$iJo z0t~23ncvoOT!@P9eN}t9Kc(-`=V7*Mu*&EBm5u@;?s*h@iC&0cRQp8oLmj9ED9sB(ujF4)yPF)4c+p4t<>rfzxi#Wh8RUh-*0k$Eh7*+_=0b?Z3%U5 zwiC=<7(YTIqDGRS31s%0w+>0W=zl!^CK~xp;PkN!?g;KYWx4e*ece9fSAXsoM)G~X zx#5wB)7)8g(CI)iP$BftMsb!s|IPJAX^jNr+WVB+yV0s1ZmAs5{kZ3Jbq*KsGu85n zH{wpcLw+>;Z#JIhF248WD?curLXNs^$sTo;$IH)LU+aTR8NE*`k3M@`7*C&mRyX5a ziUYrU8l%PJ&rpH%^c>Va(XoAGwIhU5Tx;t3mc96ZJKC3B_)8qi*bEmS3{E}>#intB z>JC>hxyryi7W?qbqFX77q)}ogX{6)NP?^=am12``rER7eG+cLxH_2#$NAX%r)2>Nz zUwI7+rIh4>u$V7dKbLz3Is7tmK{uTsP9BaNffG`Y;DgrBAFNE8qo#85RqF*RI)>9X zxBR`Rx}qUxVd*ccuUy0%N*dU68a9GMe$3a4BI4^%P)_LWRew+&=Fi#Sm=V{MStd5B zl7F9en7aY9{c-eVR*vR}?HW)s<^$tsqExlS6WBsT#9my)w{NT}Qj6^L@lq5Hn9tKu& zMjT@)zdpM&@3ZIM+I`6Cfsb4UH?VTbx&;KpVJDSO<%kOI9B$^Yj&T}X>x7O~D-6%wT#~nKX z<>{xhAFp>P;fHx)?Awnqr$o`L% znuoiqlorafjfHP8>j``Q&#|}BWi}bw1VF%mp^|f)C=W{Mv?gw;rE}bBcvD-r%z*i+ zWzE*%!v{FIAESo&*>3MWybXpXm~eAgazImn`*n8;cnPb za~{+H?7zxsW){99(dv~mqjt>~S<{o6}U?TWni z*YyP0R@O&~NMhg7EHjlG^+!dL-qybVj_suY4kf7bH98d z`98wk1%S68GDiAUM$+?DhKDpNlI9H2^4X}OVVhi$Svg~gX)Li7PGFtQuh~!={)UYJ zM{_mrO9a(HI0YYt2xuJVyYjm!EoLcwwf&162Uy8J!5S`xeiSKuoI+X^Y$Q_61eY^! zju``*9|1)3uY@UFK4_i`Ni<`CCBT>4f-9RVcf$JCuc1MhU*kU$n=Ir0uVONJv03>u zg9{m3G)hoG*?;-y9&~OVlTx>A zzgR5*q!caqt#uiDh-l3{~U;vTiLa%1C}CvG(_*+T#7*3ilO@rDe< z3S1JGw?s3+-x_1r!ef+T0D_WN!Tb4!{LP{xKb?(Yo&}IL0_8AmT(ze{IxJT!;*(9I zem??CNsToukYXGgtHi#!%*8-uj)n$3_DmI7ve@@1T2cS7`B6}y z#=nZOY_h~2{>tpI(W_}~D%)Ca@&igQ&^G--aAClS;WNTNtWzj_il>Rwx< zck_kZdY2Y$T0>=EC596xvPfhJ-c%}y+?5>V?F`*LbRXqHi>Q#)Dyb52YX}s}1?{jk zA-~}v=l%CVJme9A;&4|K%AJ>S0vqK4;jx$FhrX7b-*7XJW8GXC=(ev3`1 z{EV|fb^dZf(HB7Znd%{ud81#o0Q}z_PGI{QAL2rK+{@CrapEGpQ-m2aZ2uABa!#IK zFEPjcPdn>$>7&n0WK93h)W>QElsnj;7s_Lk|3>Aqe=KfdPb+>$8E*(hm$NXtr}e*!h*hgfY$O^{M9 zqsQkt%q6i=ua@Z|rhy3sd)}%Rbqi2!TB(pO-(Ex)M)s ztQd#fbNiHJ%8)#aFDC>{8;wjZua=0E%NvRRc~6a;=eX#*dieAK{9v&{(^^Dnu1g<_ z($;P3)apIBxBgcC=e{!0v%;ge>=CO^+E8V1&p6CL2}%-06K}EKjaG+^K(ZN-GTXhV zVPA{VmuvwLQr~jvx$HV>ql>gHP*IZBHEg~@0Vz=dH@OnCNc`Auc-pzw3DK-sykJlX zHq8qKdsKlC(c3TfyWDcKEwK z2@rH6UhknD3-S{5-4q3kdQ!7D>=2tRmelz@58Df42`D{#*$qz;d~&gqqIRZPk1akz z$aUGL>l0Kedz(3~&?FYodHkGO;#DHg%B4QeEu&F}Z~V2fc9D;ev?y?iPDjsbk&nEp z(sFKrDfMfJ)%~RAF1-$^l5>;xzI0=?EpDMu`)yDuRz)oX&WyB}o|3x8dGnD$4xh$` z`VeXl7R4Sy&w;hdT71TSn{ZxbKoSK;9QX{IYJywRsgvscpkr&P4QWcSnRYOzx77?c zGDtsbtE-eNu5$I8u{IRl+j5TivRH^kVRL&{I5fWM&_G7xhxAcJ$5P9=;rqO2Vb?rM z`597%Tg@#4+m07N9yy63mJ_ zm(&m(I`@Zo^KME2ywW>3xx~W-i7^oYhExzlua*c3ST~X96Wxt?kS(rBp9L7((FA{%$K$6tUM7;q+Mz8jIXRwMLM1@Ut~g>381D6I zpu~m=49s*<6Q1GqGeeDbh6~wjM<@L3>UVl!VkByJ#|L}(#aupla~^C%jEwc@;@*VO z^uT%2H{uV8hB`gY&Z0*grdi~j-_1zH9C!!15DlIoqNP7U>NH+7q$>5p`{cn z;Y$njeVORnrJi|kj+JL#vgdz#vXa=^zKVIggk^~`1nA!7>gDvT zhcUiB&be>WA5i4{yLvI&I1B@u3(=^4cL6;il2TWHYj zNo>LUpcHEtQ1jpG%0D()i40lV3Px!(gBjPBN?Oo6><9E5nydFLg;Sgz*nUe+?`r%~ zf>W;Qf)MJ7rv9<$to0iEf4w(Xxvk;q^Zo7BP2}=Q)=G%;Nx?;35CdDM@^Rhkmr3Q{ zUNgg=5;#)R0-R@Z>(c_zEzIwFbSxqPew-J5x7;t3mOGpOv5OP_uISA7*rRfkqwj85 zpI~=-OXf8yu~lO5Bl`kBI9fkFK;m)hwhzy6E-?N(AJR+viYM1+U(o0WN0NrqMC_WX+GC%aL-giD3mLpW+$x zNI4yTqbboMA%A~Q!GV-Lk4A|*YbM#*K_jog!k7CO_#q8g<@d_(2I(zZBda@+J!nn#@ z&JXD`V@C@ajE_`)tyS{Lq370Q;7#|GFoCzI zdAF)T_*OqnV_Qg*CC96ppA^{si?Rr1sREWvID_}vEjHR`9X>!}6SQ@&ArVQ0QmE(P z8BQ_RFpR3nl^k5Wl+sb_VcFhh$(^Z=FQfu<6_ub-!S(Fs z>ON8WDv-@LX|Q{_WsPhy=yp*&MLF|9NuN;LBjf;m;lp8Bzron!V3u!Vk&TF_%6sbd zuDydTx$$_j{H+;##knT`%L~3Y#Idp^kqm-k-E(+ZWjho1A|_n{6C*R|HGIh<#5MSH z2cctCZ-Mji+84+wO5PHZNt-3Xn-4Y~YN>29aTvxOOjiS9AmeC0%O!JFO@skx;2tJm zWNLD@-8dzbMLE-4FZ}|XDESkM_N`sLZQDRH!#gXh|Ga!3gLO$NR^&C!ybk>>IJ^#W z^DQL+RO5|Qa+$n+qveZDyZnj?(Th3Y?0y@o$~(ErCWG}jYxDDL5;E&h3VL0a3l=GW zM{d{Y`7G+5-AUz}yymDB?B8xCtFt31AQ&j58$EMLcRLge2M*7*SCkaOtAzA z!AK-@>}^uv`5s_s#JifeiUV8Nk^>eq|H)()m&AarKYx=vj(u+vb-{($_8O2qjcfEZ zODcd@(h{|ExpDcgwMLMepeeWN`zGAu>O%D_moy$>;Z2plT!YB+9;1_j%8j&Y_<&2J z&KHEeKy6N?7v2mkDRQIx+C7jTxPMx6la)f`E=gMW+$SZ2YHjLs4??UzXO<-}Gom@&Y;+JTH{e8StPAX7A!dc{o{?Idllp7>^G3(9C}?;0H6wFv&Dw| zkxl&3?MopE#7o!p0sXf|&fA)Ur%E_V=eVVM3nD1rUAW4)FsJ}!B>>5Di=_(8_kVG$ zSRIP38WydVJglmG0lIMovnDnXBHKF8fP_Hj-h!5?hLgiA)(LrBAYhhN_ehFlSA49S zviF9NymD&wi;fD+LzxjY%PYd#=8?PM`#&9M0~0a9+w7A#R+Zt)r)k~PvIA;TeH{Wt zMcJPvn8B@1Ye^-{@stwg+;jXKQNAH?c;LZGJyZtpZ*KQ;Z`@x37+K`_b-%xsZGH@?@n+VQF0#L0~L z$ox<7yYlKZyMXhmUzmG0*1}=f1d^0u5%hT+Cf%PEM~LzR33a?;si#1msvieEbv4XU zebVZUobCf(ZzPh-Q8ghA5Mv!s>_!#QuSLfeQ9+T7a8q&G5!chH-NlyO=N%xJc49AI zLdIs|@;V8~;7yjp5)#jwlByuAfsd+4eU(Q#F_ z*>-@XZ!*P1!-vt>0S>W%K$cS2kst`msg{nE^RzaT|s~^HU6-jdR$r+OXp~vCp zA$jylSn>gvTr%^25D!7CnSo=k9i&Xf3ZoE3Zf?=Ocjg?b>$$XJ?;A7}bOp9$K1Xdc zjn#X1tfHHcgy10973JrosPEGH+g2PQ!#_vp5t5(0ne7*#EX8D9V+_^KFub)moXrydUKv5TCp+xj_hoi|K`!?N5KNZ!&Ry8iZUTre8z&D=r>O~S5E z;0i?H#3X~Bh4Y4=iJd|#tBg*ue&zJ;fa4W)?r+X&=aen9p%#6Prm!JwBxDkOtWlgwEjn;Y7$Hk1KnEI93tB zrRlm)hgEwijY3xC<=&JTDhy=#IZ2kxy70;SAydG|DH3pZ%me6#hM(?;OYz4BgB|Sd zjwoM(=4}U$TH)R=7Z}vuz-T`iPCPBv=2d4u(il^GfW-WEXvi&zmsUZO5cEK-JaI?YwD-4uzs9d=mLKc$*>0!#x%DueF2CEjbMJ9~&w-Fe zo{rk$6X8`;lk?4M(!79|n4w#EeT{8SEXl62(!f>3(R|hk@I3(v1TW7~$4(I=0mV_K z)as1z2qVa?VCvM$o_E{cAC2vJRvN)x-$lHJJcj>uuS zGE;n{I)iAh{4hFuNm^RH1&z1s1*B$al zmi?}utF}GUroe_HR|UBDH4_MZO5<{98n+fklsXd6al|5)sMjjk-bW{`A<{qSkiHXP z2UE38TitJl3MA(h^eqW4r2u8f6~TWz;#lXmkwU-hb2Coxjndzl$LTw?FX}YaoWv&J zY36Fkf9GR^8|*KW%gI2jtlUEJwLfdjrzLm2pl$sl!SjY55Wt`ZZV3Oc^ zdi*#byWG+`9J_6rPgDHseN-d}poG%0WgQeFd|`RsNNhTUS}%fCyx1iR{8?0kL{$cN z<1wJ)Wy5>?XfSH@gTdS^zsk5C(?o;A$ORc2!T;CZcYed!w%w-?B}I+VTXdp#LUf}> zFHsVm(HV>yBHTJdNR-h_^cW==L@;`1v@p?v=)Lp1a{r#So_DSHUwDsC^WpsByk;)P zvG=j}j`-g61srZ;5v%&%zVyv#Esvyv+)MRd&xTwGjbz2=A|3OQ5APl$a6Yh<_S$0d zy$N-{x;)OJC=gUKm9v*NCE$V>*FVk@SONDr3;)3T@RdH-XZqK=sXBzVe&>}bv(3v0 zV}H=rDddABp}L|?h$Zx7dHLhl)S|X*D12n%aoMU!LUtg!_X>aQBzcCW&Iqzy;i&=E z(YS>7zV>~ahzdR?Eo3>IA!(;_w)W86Z83cRDxGCehk3-+$D8dpWi-z!%1iq8*eF)n z`erG{zCi64l)At3y^?W<5f-&_-24zBxx+fDk+Xy!PQECr4V@de|9fXKxC zy?wHq$lLw4HCY}^xe-L>a*ITlfVWCi+@7RNq(tEY{Td6X~W_& zFBCR#Af~Zj*I5?lS4ChtuZ+GnoH(5#2JKW*3#I)jCxwQwh{Kx$c#$LQL5*a`R2XK zRpjj-SzU)-CnKg0owXvvmME*1_-h2l6f*}r+w%IS26JOn>VJ*5v~PR5nj5Zu&FG~h z*2=uOTsybX2V{AZsiSM!fE;(TOg$6IQK35K@XRXiMM_Y_ovg-vz~4G;ot_ED#r|7U zlZ`O7*H6n+B^~}hL0k%eEvZt1TE)yxX-6yZ661hXMd7jF1OMyb{NWsF9xiWlMzB@e zf{b-*yT+BcL*N`J>fWJuUXUNcH$v5XmYT$#st@~>H?bz*OT!_EMV8AGTN2*}5ca9k zl~DY)7nf9zBskFub`qB&6IW!ef6)!Z*M3Q21Pg6xygg=Wt_78Bpwd_3qgl{UfU?5mWN zz{v16gE~{8*p^&HC+9g z5*5YeIV8bFQb8LQbz@KjNt9f57VRdZByp&)))5=s=}AYRgrJZ8PYfEBqRUTQ``b9rA^{~87`ddgsuMH4O z|7ri*%JHTcCB(#qJH5Vez^i_a4)DIuL}cYAJ6N^NIQJaH0te>~y>1S*ZxA7u%WctC z{AwOPKFTIGDA*x=AzA!ym8-2{fPnLC7uQ}o#n|gA?4{WDCfp8UU+ww;YeaN|RNCwX zQOug;E4Z?k8?pyWlfY)pHx)fG6f9t9V}l^yv(rJ?s?OpNFubs63+i`jvDHsy#A&;@ zz@3V5Xc&P;oI9(fvpD-aRb#h-p_5fk&wsB%GV6dB7QM{3VQ0}omyC6Ldj|e&&9jTL zDj-`$eC=~~W+FR5e%`NznHTlVF4>%aL5d3Ue9|J%_Y&SiOK}0oK+<=FuEEPf-e|3z=x@z&jOWsIzcR_CVbsYn%1HE-l0tQYuzX$^G-4gM{hm4U`vEzYl%Tq<}h>X+SfFG^%Aq_^& zSM@#W->%VabY$&8zRG{3LeerA*L?WdWJtw>DLVtF&q4VAxckapp3Yw-bi+VcU+!Nq zA~$U>q|PrNZ3S>r+|txTEr`Qjg1ebEh=0gc=wDH|=rJi`v7n4T-0yPbcrp}2F*gi+ z7sI51n_Cfzc^_E{(q~Dpf1emDs3yOD9lk)=P8y5L+M@zHVoZjk@to~FF#Ot!i#G94 zM1Seqaiw$j<@=^wsLgMsrWltvG8**qGpFTmvz6|cw-vca(aP^?k+l!V>I+QTXRf

?gZU#|_EvP?zJI1U zMyV({T(eN^3;{em-@kkSYPpin@$i7ZE#KajwEF(3hht|=J0uKIJxCz*Fw7T5$MLLk!3jlOUusW87gL~4Eio{1)NzekIFveS-xor5O(SJ zr@0><#6MHJEk~a3N)4>4b^%lJPzh~rZ2@JPFSmF&vu1J!C6`jw(nEKS{iCZ6zP6>! zDYx_egy7EPVMG_$&kdbr`*)zVLe3wY+b4Y1WFIKj+0ebCB(b!gA8Gbn42Y-6DPZZ> zMcei|=)+T42kmvHC0--m|DlaMb3}b>M)kj#(f%20R9cz>aR>_-d6*d z=%vMc6>0w~{9U@3&8I*HHE2{%n``iWBoUxPkNKhQh;9iu{USN+iL?zCwg&X(Fpplo zko^RpW%!%8A>$4WiW7*~TEANuOP`;yojU0cv=YA$E~p{Thg;9!t4>`2Zf{&QRSVTj zD#ZUHp03G{hTBmAOn;f!nl#KWd$eR_{!)F2kN3_eEi|BpH!uKl*hDOw9!@BVV`G*rTEmvWZD<%402{ncwua|*GL8z?n;StpCOtLnt=XJpvb_Ye z^8=*l?3HMx)`bk(k-m37wp*jv#TjB?7nm4hAfS+5X3f1w!LQrTN6|1|oL4O@N|X*0 z@Mr!>uCNjnZajV>sgh}HkR^Kg(_q4d#!73|z9GyD|4b@6Yq45<`F^=!497*DUZzmw zT=%SlzMD&=E?EDuecfs29Kwz|SA9n%LvJ-t z;UV#9<~G$4P#2}QNt8U=Qde{iS$PuLMTTpBvhjn6tX?gOMpm|wPlcpq%w}EFZqk=4 z-~`c$qlm@xNahq4Wh`XZX|Sn{vK6U_E-csLP9bt&V&d>lN{U}1tf-^6f8hdgbW$Iu zRNO0o1+V8z(t*g?s$E`dfQYY;e_swv^$B~c{;AA$X`WUn1iC|A76LzQW|k4V?-oL?~I8Fa8b0$3EuW0O*n@50>cnF{yD znYAVwYE^Y@beI+1q^CKsQWXl^`H2x(pfZFGHf^dxe}FBjooINNqgyFE@*s}#Iqh#W z)Kga-)_UPAgpJ-27;5q`9x#UjTD*gnBOoSB}#rUAY7L=x}Ch1nH|_CO*Ag zYVkoR$|@ESh5ve`%Y<8=`Vi6P z*VQjpbsjUEy}x(!-dK=2Z(X?*;K$C>L(XQj&l`YEz1!baA&RNC1&-7ehE4JPWdN1_ z+@&Rdz>EH;Y)l>fQK=nZsz=oyNO7rOm||wh z=NAHt!w)@-Q7<5th8?|AAtJMA38cdovw(IN4z-8d>_h=)xOS>Eqld9|nX~{#KIkx) z3SjgL<*GJY&rz*S@l$1qI$wQ}ao2V*xw+a}AZus!X$7EHjnhhE>SR96(Z`kdNYvS|<`z1Wk4>n%J6JTAkQJ&NwRXb1Gp-H7n4~paE^eTg zH-`XNzu(0fN%BlABTkW!hWEW>fdy}4snuq0N$8zBq*Y)=Ieq`S-xq(PR}azyU_&jJ zM8x=|Me(7Ie6i!ABETxps?%UW4~O#dv~~#;yG_!KvmCIDSB=A!Vw-D*vI%K__9?FA_Zrmt{dkLVa_wqKI+h?<+41%kPIo?>T^OJN~8NLnUmkPh`)A;KR*)+d*gYZA8aO+ZR8l8tUgmEU;)0Qjx*= zd@la{L@HSC?vn4vnjO50*KR0iV#6u#*-%jf);loL@1)XcO$Cd+Z5);i-fAETqxqVK zZE9%VO@4Q=XWwuk8j|c9pC}LrzsIs@_+g-1j8sq!Gp-#$z*KoK_s)+T<5#08G})F-oZq|FgNir#*cOeQl!rl)$%gs z346T-ty4I{6Is*%tEpKPsd$XHpB0U5v$>UR*x*AS*SJmgm!z8MZ0j{EHa;3~r%KV$ zG_I#%8me65W3b2(LVF&~DfpU1Mco9tQzLej?1t>+A<$8y{UDcW(@iC5c6d)#b}IVQ1&ZR?4PO_hr;S14=7A7-Hydk z8KLy{qfm?9KOZ?^SJ-!xw&2+f%4 z%?3VNMPf)YSv9R>NBP?P;}!O>*A2IilQ(!~B>}P%mqo(Oq7BF~yOl5@4R~nTB9U;K zpwALz*D|v=SSIG|`}9;h0T9bvF`e#ajn+TL;Fgvy9UO!)^Qn?`IZM7ZGPt_PNE?+J z3#g=c7@nwsW*tUIGvNR+36$d=Eye&vo@cBFuQsd9!|m`N^l7lxPYv8Z*Xv_efD8iJ z1&y=rFsSBOnjnIH*uo!28&ZyS|3wN*+t6fir)FPkoqq2-W~icGfVtpWjH zsQG-gN%zh0D-qhb#~3B033j3#>)2dZDPDxE3Ss-kzJMCe(9XPlO0k)l1t8z&=p!+x zTV3r3?H~U30ya4?Fpa6D7Ir?B0vO2)+_H${%B6)Fs^yvj$GK09$S2lK4y#Mf!a}EF z59I@1)9=3U^DQykzavw<{hof4w&K{MJVl4qyJe=+qs&tHUtfh=98QoGZ|AY2(cn#;qBX3B5an!LE?;I-HPkf{w`M7pX_r}SEy!s#0VY7Lu76*(c1+*A zf6_^AQ+lq6?VooZcQ<*|)SBw#$*l=Fa~JsOH7PaOSWJFgkEU!o#e9v(D*2r9OtCFG zi$lph)ZAcnI=;_900~&+!}*6u1sRY6?6%-dS~VquSNQtJ@qOO!583f zUUw51jiU3*U(&W0ReD&^2F}+*KMZ?@CH}Qx8S$a0%4hu1G;5WW-;>WDB}VrVIRPLY zYSnbmIrs+oW^}xa zxAS;Wmt)8h!FA4qOrt8Su3h>#GC*+|8l?|k_;9B%4-bKnw0k1|9YzwvhRTKBy!j`q zkiZB8%Ct0F%3S7a^+SK-`Nv`ffq|j$ozr&Do3n2K7kD7v%KCZ#B;lGBaWWf0k=SnP zJ+v)2m95q#22cX2ULXqJX_0Da#XCxMl!>`Xgxn$o`95i}(`s5JOz&UeGqJ?1h8^3f zls*OMHgNjh&NXh#Dm;HCHb0+=*Y8hh^m6N|z5$PR;L#b0c6RX4`8iDg(MLiuoK17h zsn*5gIp0_n-0T;@;lyD)>Hu#iv|vb{Pip^&&$6wHvSVh0Kys8;vsOiNo!6&}9o zIhb*H`q!r8e3V4?Y>7%!&&xkBRv-ezXSsj#hnkp{ZqO=crrsMfYC>~))yIqgv3}NI zi|vPZIekDP@J@$&F7CM3&3s+hnKqDW8cJMSbO%=Jr7#(f>-Hy_MKF@3O>BdncFQEX z->p1)T^sUw)DhBws?Puq_lRGQcyTCVxba7*I=lr}C+8aOWYRN?#;JH?ba(6Zn;#GE z6TnDt{R1s;9I{ln2K@wcAx;bCX4PgDV6wQc9)Wuc$a(Wc7dOuZ`}_?{sq|P*w7}(O zE?5cweglPS%c&V}YOOiY)sqKlzg?5Jv~t|VAKSi#^4irY-Fg8KlW;!A(pjpV3D5FT z5D+AF9^^0}h)lZsWjE}`&ugrv3KVedugH;><*B?Crm-f~R)0zDl{cO^KN znKj#Xf6>(wv7Mg!A+QEa*RB8Z{2`34_TOCtdgd*vZb!!6%F1Egb{ho_e3)q2g8gMF zu@~bX*85M)OA(jEDnI+|aL@ON{}4AFF`U1x_hi`G?l+X(nmY{Va2~0uIo1peS6~Br|^dIK;8J^^=RB_D(&TabCu7M zbPR-s^#nPSbQ8J3dBSlT4(aI?Gc`Y5H_CGl5KS%u?TfKWB|t=GuMXm=M#67$Bv`+f ze8g{f(p7yD8e8)S^sC9GBr7V|LDy53JCSL``27*0h_9I@4*ioR?n~1X&^=CsIX*tF z!5sQ?Pj&B@b?>kgIDT-xqX>em&fkrgxxMV7zZWpnE$@7xR ziz@$lIW;MaCM(~(G`Dip8OP|G)_5T555?{Axh$&8=koM3Q|9YAr?s2b`!id$W1};A zXX5U80#+5geAaCS2rz0BUzY(*0&%glpy>ZHiDzvFXW=n~0Ffcml0Qy=k7#k~t0r0P zzEHNTp<-P*4Wo?0SIvU=iTgMb#2iO7fIzW*B-ygTlt@4H5~^TZl9Y8rqZ@mBmFDat zQmit5(>e*T+q&>3N5}NPV|#Dl5cX=4=(DCho`ipeh$$ra`dI90`4EIN7pmrWa>i7A z_-rQ&kyuEOCGdCWrV87#2GD-^R1o68JI|GN>8Ra#_sT5)M%@MPt%71zp2>tOSr z7&kCD++4%DdjL6MKyV8`)yFMS2x*qPX!h|!0q==C!g3E}8658$KkuqC=VQ$DnSv_h zIKNF%wSBUSwaPMEh(6Ak*T}iysijOnpZ)C)Cm`D61cI;k&XXg9ICGXVKz!$aHbE;& z?cX@|Fg5e8`wMyoBKTH@HGIoI4IiWZD-B;8#z0V~sAkl!t)rCQt$%81$^OJ+9vleI z=HW1ulty%UQoYA1>_yV1`SGSg3<2{VJ?j>GBZGJ0o2;9qaSutk3~3z*hD4-;-DTs` zvTTDrmEkM}QAG=`C8dNkEbSAhpiAm3JL#Z|mQBjMUHK}tTjti>q4y7zm-NP%27aecv?F}K{E;VOYWMH#13;0#-S`= z`6O1HUHpK{2Zej8VaQ0O*GxDTY)9$>_+LhtM|>jcw|n&BUQrj4t8J9f(%_Mlk4F-| z3+?%Ldf>Lo*14#ZshYbIy+zGACAD#j(S>4%s(XW1_lQ##4rZb}20}syIxB|HDzWO{ zbDj0TQ65ji{~?k%xl0u^yvT6!>At}Cn6GPY=TEx2k=2pe$D!Em?A^Bexp2xi5p?no z1sRPo6AE2=KXWq6xPq{t5h&|Q4<;o+LZP{T4%hn z&~KR96owIJSCpnUg?3G))?42ta^wDOvUevCg2|f-xa$~O{-m%}2#9@t~iBQ=Crwm9A&wOkz?ZD1ElpxxrvS zB|%S+mV3FkFlT@)({Tp+SQ4xXPpOCC%Ov8+sgNI)G|=fPv9>`clQ^RuAu#e|mP8nd zFfY7>#@&aZFp2zUDof(QmoYF7Ty3pM(%>Y@Ub2c6+<-MwTkR^`~sQLc>2-i09Bc%h5 z?j-&GDlez>v0af%`O8u)Eym8SOa2^??+6#ts~7}`-N9n@TJg$g=%mu|@Rr__tp{e7LP*)Jx%AYfN-*j`5- z`;cNY+A(QwrT1BSk*zTGB5JKG2E;=lfIOSA*UnoX@c$s7L*Uu6P;J^RI93D!R6?oA z_MFi1UoG~6T)jx+g~qbw7Tm?et~tYFxlDWHJfoui(xlwspTXDBo47G9U7n3^WT^7J zqjaNdcz|`utp%#)(%)TWX*~5J`O@r3Ib*biB18UTtqjk@fgV9~yk4XGnO1!KC(UBu z?{)G0THaukU9OPh@|t>6U@$(9V+mTJmsuwbNEBMnmehgQVa?9{S)RDIVlXM2t(>_5 zDkvcS5CLopC$SJ@i1HfaGknrUfA#%V8R}qOL-b8Qdw;>c4QiMu@t#;ozl)p{v}0I} zvn0Q`aNnzWUT5;DtpszwS5tl5@1yIyGF?Jzja#?ViUaL6X_q-5vPKn|l1G_)QSLnO zXh#pW|Ctq+VPmjmvA!qvP7Sl^?+NIKd4mF2Z4hx#6pF|rDQFifLHwT)0pMwGvTr6L z`GMNAAHKb6yKxKoqd*DNSpn4n;{N}rZ2rHy2LG)+f`6Nf|DO221ONXn{`WoP-%j8G z7hJs5c4K)nq4Mu-uMJ~wwy`%~h~B;d>Q@7!JpQ6~5b=Mn6ZjND8*wlc`aeqh6Vp1) n->37RIt%~zLH_rI_NQBhrPPmjt(N>kcYp#bX(?99zX|()y?^<( literal 0 HcmV?d00001 diff --git a/docs/modelarts/umn/en-us_image_0000001340265309.png b/docs/modelarts/umn/en-us_image_0000001340265309.png new file mode 100644 index 0000000000000000000000000000000000000000..8094320e8e784cb59d62d6c677ccaf49731352ba GIT binary patch literal 154806 zcmd43WmHw|*Eb4+gh(g|(qI6JbazNdm(nFjw{)YVq=?cbjR;6L2ndJ>NNl>hyYtNb zf1c+Z}N7?}Mn-*UQhXp-t z-ObFz;F@-xh4zo~5r~(6D*xZ!o&`#{;uTQ@7cn>Gn2dIh49eYeyw`t1uv3y5k+(vZ z5qSec@XA;JDZ1oPn(w$eM+7F!)43WINyrg5?ma*WGdP*XxVF8}k=qaLas2xIGo`1C z!_#w4_nXV8A(0^&8oEk!kqQrZS@U8X83G74pTj-o|9B=;F6Sh*Z!NGW;S`HTuZ9{g za_)mNrG3n1!AJi%eV-=6zby8OIAwbsclw8;!P}TsI9?|sZ*Pxxr~G}P$b~oCRap2GuX{Qx zEB^XoS~7$}KuDIaSf@PdJ}r51d=xQa?P5C3n>nyIgU!tKB4(y&zoF)W#WHN-_t(}0 zUhWU^zs*s47MDwmdzta5g#N;5B>xfaelP9Q_x1GjILNMK)cS9uiD__=G*+Lpp1P<; zxjik)>oAhKl-%0d#HwkZq$g@QtZx^7t7os<^D2*>ToOvE*z;1LXcnqZ`CfZ_9IP~N zjFydmjuhZdiijZRDfPMZct(sPCMGuFeLNXX#`R6FpJcf5J^7T)z=e&)D2|o`PY&tj zlg4chOYSsltDyi)zGR2+VPUVZ$4x#TwxZeMJ_mKHAAg-08`Hk3-V-`&JDiK`tg1BQ zF2=?&aPKp*clmm7e9Y>wQv9!L;p+#Djg5sp z>#oywSwrewHUrretFR^?HSDQk;yy@kdn@BEBpW@*OyF2ogmpF={S);YmT8#0(ZX3S z1z~BjUrBv(5Z2^)A%XFg_Ui=SKKcd3U7^Xhz|s3V=U zP*c7di2N8IXJpxIH|J#cJlasN+Nx;Zo$ruQP+0t;q2Juw8};qmQ!YOu_>$7JoQvO$``q)(L1tsR6ep~$rOQ|M z?`Boc1GcZu_(J4mgPP11HCEk$rK%N4*#(A0jj-BO4aZ_r+PlmG$B(PlQuvd)+k<`2 zuO}St8qGMnfRcLFSZx?Hp9`p+GTo0R9Xn*@wFz#FzK#=j+=GF>P!Iy6eNjtJH zM5(FPH-~0KAd3ZbVz_)MT=8AnU8FLj8&XPFz_{-7KRJ4yCq30Fw zUhhWve*E<5eV_S4Ne5M?MzKCaXbTkgXtl$`(2S4q@xLiuC;A>;Za1Ds{`w`+(%P!k z=*8FIc}yoHltS%$5q)yvM&-V6$IHu$fr-fk@$FSIzw@mI{itR`!;O)W372sj+C=|s;kfhdrnKtn>PiYahcEpY?(Z~(_mN&^TY@?;rbE=zq;frsaU$QgmYISPb zaSUZrt*qbPDw2TnbBt+ShDD_dS8{aHl5XbiQSY}d8>!f@3=(EA)a18p%dD%&?+(~gT)nnlz`Xrx` zzq5l6Ry*#quwmnV#oEiKfYQ3}N zrgPyB1Onl8vfANwdZ=ww=e(S~d=f3P+5G#W!$+*;-|QPW^gh@1 zn;pM*UKhto?x%s%?CA#AT@%m6L#$pUH9pUMscv+&M6QTW+LEwdoNGBR^^Mzh;qT$F zqsMQ8ztdj+s#>Gp%Ig~%+)rkMwMyS$7dKt;3wa#O1XKGc`?5TJ`hblsnrF(*f84Gq zg{AT2J?vV=RKdDDBF1*wWwIxk5Z6bckyc2rykaES6Nc(Njf`oY4PQ< zCCJCi)0MteRGaK6WCJ=;5$XsBlSlijUjw z`d!+T4@qm#?ZxJr-|v^7N=s{9tp_+FjjP7BDlUwXLZ{nOO*9u5 zo7QMm^^Ay!;OcB$lY{+-=^Ny$gC`6$@P0Xo&WLtKpZU(--Zwdof^6UGtA)kI*G(gV z98aHyPBnOqu3o@Qlb{j0Jwc>n(W&e2hhQkpOY-`lLVLxnWql${-WsO8<` zV|o^rsJS_#o)iITRn@!Ng~WyG>qiz&Zf?9d#Lv=ma)OrAd{dFttND!huPH}IB;|bV zdQq!6C!E8-^+Tqm0X?74(9rwBUc6)xK>)^285y5EduELgzNb;V!xW`9Ud) zVl)MJ7$#2PkAKMMn!?+q9@1vaPv{^Ou_p3pNd5J$btK6nugx{0`<1CGdnB95#-nhI zVl7@iBz7$_MSPQ`aCv-zi$uYOB>bmj{2z~-l6N2a%U0e1;W753gW zcwIy($q4@-DM0*5V!Uc0?}`O#CBIm*?sQ}xPfji~3;FPNaxWKQndGgpwe)0QrB8@B zLaNoT@sd>|fi0;!D~pQO;xBKxHj~Po==-=*J;?$%qF%a6Ia92p#*7|9*tE5ka$9o7 zS4TQtw)HBPRrwMA{;+cD#rlO9d=(WH5>isN))DMv5$NdXD9YN}e@cuyco zX`Z`Ri03cuey6)_J^dV7M0%U|)vzWjX~PdqD_)J{VnzHtoEdS+b?Mr_Jpx^4DM_aX z)7LL^al^Me{;ucTrMkQ1QlniL@#iUrzErX4fGd|lx&aEwpRT;3lJaUVjc9j%9KR`{ zc~o9E&cMF|T95F^`)t=)HC{z0IEUuyyWuqyZQb46-i~UW6)hfiZ2h%9;%P5pEEOr9 z*;hX9QOq?orj1h3Sa99rIX|sk;3?1ZwkOr$ROgj5SEFzt^>2)HP?I(Ff0+=`6o+#-0}1rh#?kmCzj`goeskTRIU7(y{av?A^+%)V(abx{^e-gwdUY-;y1IY6 zSenB1iTWdcwqHv!BE3pHqovdl%L( zoRs6_Xxz?$m*W2IAUYBZ*Hi1rKJ!P99tF^+x`}(oDeICaH&gg!mTTj#Wt64n9^@A% zp?hkG|5O{EtyXS3?X$~xIY>~ls#4yE%0MD5Hh4dA>ILspE9IxpJ~Lln><6DGUf5n4 zvWWRs<4&Iw?xymEe0vql?*HA|!Zz#DD3uFBXxLmJQwzzG)b3cf*JIPP{uI_|-PaG?k+qnk)r)`J6zT!whaYk6A z6bYpQyfR{8u^R}9?@_McG zh&6Xb(sLUdCeyMSggX{Keup{dr^Lk242Fe;g^yppEWVxU?^nHrgM;XO_~y-gKi+6w@eI6SIO&REjaP3+8qz);<|i4`=UPzbo(~u*Es5JGLj9r0bCR=I zX4gFP64A6-W5Q5bBy_{tXqpz~!m5{;4v(rWK;doJ-vH$nu9*8AAZA$4mF?}q3#3u*&z5uF>!jDIcp@w;{5M4)?jEE1&SIPayZJvF zID%eJ(w8r7Yls4bARak)c#Y#STyg@RBVB=d;oNF&np6RT{dEiaR{g$W+$n{SNBQ{0 z{RpY8nq{h37K5H3TyhW+8PZWyf_uLoZk}OdI5^v5*!eUi4!$}2qI|fhaETIi@ul0|vK_ND&Q4P^Sk(J`5YVe3w^mZO*V>>>Xnaonp3 zKJTsD!gO{m{?Ezfqawn;W^LRUbvGmBGi1;?+FTeqJ!S>M&IxabgktJjKb#P8#FG2ZvjGu|^@avq3}c2YavB*Gs!(Rtg`$7Gk} zQN~}o7}-#FI^1r@2g zT1jxxWt6$~+6YIJo{}T){KpA|M7AQ&ww#znqD&Y?T2zf+x2!l6Uj1;3TVU}o#d4@v z-$T`tl)O1LH3i7nlP1y&#_4y%@symkwY8a`{*{N+$ zH6-DZuQg(3A}nj>BL!AKOw%vR>RKhECg$#V8WI75+>N{e_9kDmwydmdF&g4S+LCR8 zLr+H>pFL6fG0xf|S?;>+T*Qq%K0vKl?YzZ{tg+AWpq#6u|eUr}TE zm=Ou(9m(SARC$Ud3423j`ejut+A-<4?YB>gyi>_uevotJa;~w4a7adU_{QqnRrjJ# z%yAhhGI4UWMIL&Mfsh>Ye?qrK2Er>ZY_9SCCnpXmc#rNhmzW!dEjloofDDWUBh-sh z#ocaTul(Zbbvh>~SGh__!A0iCj^>;*+3Op+ShU(=7?OAXSLn?2|HCZ$|3u8~ZDuAK z63K0;WwtB|w{Pl7T7df!4HM64<5mE{+*3%YMn<*FIj~nG;{NsV@QY*3zI=t#bb%4M z;Oe3{mPS9df<-G~<-X?6xcc@9B+caysnMP_+zgU*aZ!Z-b-MXq0sQ8MEV{F|_}b7z z2-d7({uA=Ap!B5MT2~z@o$spb%v!rSm!dn1WA!2aE6H~vphl}_%k=iI?T~I3YF+D( zJn(-peI%y?tGgWiUk4DC(p&ehJ)t>SydPpun8oMM7el2O`;&2;rl8z=pAnX`?8SfD z0!}}M(4BY1cKUJcr|nNBBNY5=+nXo4#%U_jjrW1<=GN%{(^YV|Sxz))@}*@})t-U% zY0hh9`sQS)7uTXce#`;_Nyz+nZEa1RMNM73{nKMP2Bt5K^iU#2icB}1n=Vm@($MSX zMZ_ivGatu8zhZXE#9Oo8d{!K#g_y-fKnZNUUDGAET8{kdh2PPIvct3X=?t<-dl?M3 z@Pg4$z&SxArxZmc6fflIlGN|eW?QvC<+<$uAEs+kNMH}FgQ&1{7|Jxm${xd6sW4cm zM#1-b6W{qdtuL23Pc9szQ0hG&{JaRFYqOPM{EefdKXP?I!+%qS$nEU5U1XGU2X=oE zDxrV^Yokx|YI%PX2kX`?%eaF;SS%(R?6m{!|MVrQhjaWCOSJRH%CUW;jf4k!6;&l|vwPvTktwCtocW1ti}8gr6u7vx;2p;;X8vvXVsz9*r0u ztRZkDRYvOFImmfzB~*y+5)vYTC^74L(sepIKK=!`@~$|RT-WWXtlZrF`6!XQ)YQEX zqXgeaMd4>>XMfkO2vy5f{5rhOcK>!br9kYsZCzoT%@6Iui$xaS*9R+o^6{*we%(nt zj&r}Tw;IoF)xT?s0q)5tDz=-JIe_ZmiZ9;fUO9}rXMh9E6`DfawX=YqbCh@gLhs4G z&<0h{w24g5CF*W|x;cU|HoLDZ|Fb{;uDU7$1M1%J;$n1=&t8GBVPx5DAxh*cya8 z55E>p&V%PS-x0-O^b5<0pYquVh6%pDv$vQ2@#DZo@AJDvL;+P*JlgrMs4r~zl zSvKNgpgA}=fF7-NUJrr=H|tGpSg?VDZ1lZ8pZ=~>Rn*!q;&H$raz5JYcCp*h+0(Q2 z<1nN^TBlKsUP79Yk#RmdPJit{Y^f;L3u_je+5NNJe%17x<9x))gxkXsyy zR|V=N_Os%!9KWZpj(8ySff%u~9;Az(9}0@BoE*|;)J{ZF3jDKVvEhK9yVOw{GuDRG z5>in`78kQSt(}+KRo}`x}R3#Q-)eL9tcD`99AV6LSx_*qSp}d@v%KJbC zYU0PJ^>>4&)DCLj-}X zA3_Q<0%txR9=LH|p9g=6jSZ5@d=2Jr#fd*|-;y)Yv`P%C?-2HUSZ%1f+(D#>n>Dk#!yVuo;vK7)YEo~BCy?Pb*MQYxo z)Faw!3f`qD&Ibe2n7A%IUnpy&<#a*+WWLieXuwTD^sWWX4Hu?Hi z+D%8Q7nqEdTQD#4LAioe{?*=|!4TSI^HEGCAdg(;XxsN%V7@(K9;zT68j6x(#44Ja zzcH!3{L~8;5gG>B&?VZKZc1~Vat}^SB!jRsO6cf5FosxWA2QPly~jAxZF8sm$4Sa=Qg^c*b3^Ot?9dzSozK#Fm(LlP&e8 z^_r!5w?nAP=X1d;BO`+%*8+{=rBSoA z#~d7lej_8Aa7_kYyEoT|3R^8J*v)zld3%);hn`&@e91)P#6n$n)q;Uxp;0 zoAU+szcv|Vdj9@cCY}47<^%vS8ryuZ{}OoY04V3FNkx1FG5k@3Tblv>v>H6PV5gH9 zctxvZN~vgRb^iIn=)RO-#+o2+yWaojk1UwOqA7BrgMZH{M0}Dd1zhruwUh*R1J2eW z)BuII>-X_Um~~NpzF|joK7OCXZ8JBZBmz|I*u_-E3a5K@+Ur~DFfRpnJtQj1Y_=uv z{%s7T;{-GWw)>kVk_N2`BFWi*U!Pw<3oxs)GFHFAV>c%4B1=YSb$z`nk<01?QS1|_ zm^%-jsXtvxitjdsbMN5j6@32U#S4Ec1T2HRmKG_joY7#OipEj;t>$cWr~kVh?`A&w ze|O_AUcaVA@$vB)930$9n!dpB`~CX`mCwm*;F9M-x4)*R`?ZCUWDcytCYW#@)(fZL z`$Q$=@xy6VHPvgcQ{5&MU>Jbgq%HIwaxdH_<6QZa${!dUOnCo(Naq{Nf0GbaAV1DY zh6}A_LAad|1Jq6iA+4JGq_}Pv>=GGSS);{2UpB@n!btSp-XZgLBx4rpRP6!O>(+dV)A#rQ zbp!<8e6%^9AnfhoFGzIn-tJc2HoNbY*V*xQRJZB#0>s;uf3(h2)YQ~)MKaFzxyY#B z>-zFoK8ZU5Nc$bKr4roydWi!M(Cp)jVpvypq|1mQo1MBwfM?5@n&`48#`BLKLOQJO2E53w; zgbOG!I%Z}(|K@b&`6A{NL3egWjiM~bGA{PhuF=;##2-I?tfH#g0tGX_*X#*vjVxOw z;nPPz*}3KApC8M`qS)^($UyS_Gs&*;L+2YKh_opLT*B+NYQxIRdiMs>u0BIg1Euj~ z4hOgz#ER1I&d%4Pr6#{RI@I9Uuy0Q1sju^d<%d_D6xKS=CV7-n1@oB+A%7hz)Nu#K zJKes%vP<96=7K;6x!GY$oU9cT-)DalLKbq%A=}9<<|BxMUZLaUuIc2PEvrL1cCX^x zk4DY(m^1s#L1{n9g6BQNrKU*$YQ8Xwh#&KIAap{uN=lI#ftLY~U+o+#U{)NkC5C z1>6QKjLV$(_3Kykn>QhJ4$ojn7Ign4uwAdVslQTxr}Ou38t5?DPS(N!WaQ-+pwYPq z*|Vj6+1rZ~2Z*8&Vv)KP={!^u7XQ-4tf%Tp3)RC~XJ??BIkzlw%?HkZO>O&ZV5~*Y zXU#S3dlNo1^CxpmxRU@NXF8(fe|5<*;qz==Z=`T_4R0CAUs;%^`h$q&tEcmI80Eb2Pm zEy&CJIX+I_psxg&62>E5^v1+Q1^6gk%%7hjP8^GVIiy^9SjY_E0=D9;Juk24(KDWD zFAiWTNP^?gsP2)FlvxbW0V*IYT7&WJ_Lyxz{=s67hBdYs&6f2FBgsZ|vip}GDjMtk z+sI(i>!42>iB8WnfB1pLs7)Ty-rAeDrRK~9O#Soc55{B)gu^u;>FS#1A26q&J7);F zySvXqmJByO2D2OFTT=H>7S5x=huSqOOZ#mu$CH=VCO~K)IS-&0VZk%$ zq!x1ufhe+*BhzBN(83iPpQcSMCN5j-6$`kRBE;RrPp!uKA@r8Z?l9XsZtH0Rm z65U6mIz2mk3q)RK*nESt`}W+z`g+);t+WbJ)&Ay4i9e{EefY)n;dMUy85Gc8sB79? zsqRapraj4>NQpR7VyvfV#iVg+0ry2Zci=Va{k;5qmE44G)0UQ&v5B_~{IPIf8laVY zcCh+fR`%9I=}2qkVwlXsNb1gMTsBB+Gx`d1AToXnMasK>hOE)6sh>$qD!g(D$zcEd=shy<70Gmb7u@a!f6eYTM-T8HVqR~1h_&} z=QKup{hwiBVee|2Q$W7(SI6voehSUV+vlGycIGN%(p~I0eL0B`_Nyh&g8e|pV-95u z!&spTN!+h1v?05<9cIudKx&R$l`?Dv! z_Pd$ujDaja!yJT>u`#*4t43T95F_9sIEO)xUemI#ede$TU!l@Wu8_nIQ%_)qT4(L; zfs;TVq1+ep;Dp02U7YTV>Rm3&mq9`A?Y&JevLy$JL(XljEaLx4P0hb`mIj6|z&Pq( zQoI9JgR(e$Bns_pFqvx}YHDh`hldSWyTlN9{U-pMO}KI6Mm5lp`_<_(2zPmCh{KIT zarXZrcy%)WOus%}qZ`#92<2*GbFr#Dn`>2rLHP;q)3Ki)FiX^|Nh>n)_T`-l$3(ELzLso*j zcUz#X3s4sj6ts6Rpaeo?#4tT)X{{gP*mxw_n{>rOfZ%?%nw!Au@BqmYNO666c4!Pz z6Bpyt`amu|2ygnPlQx#Y=DFnUB3$TRXqB1nu4G0TagxD622$ODFJbjQa{@wo3@TKs z-29H?(qANT0fV31>Vkm%T&=7`b02C#%@Wv;ad)aq5Lr!3`mVApU9cto%Z<7 zoTX7k9sU}BZK{C%jNj)p<{rMith(r@fjdsA+P%)F%yBUHqb4aCIF`*IB9b;Sp}R`= zP%zc0i6|w%wDfM-=mg}m;+zlkm;(uhvtS|jW`n6S5SD$bP8WM|O<6iTH8{knO0@=| zJ;;zqO-X3+EdejB#4>U;*FhnX%{Zu@kgiAzNQ;q4SU9!9YS_hvJAi=s1>~?0xuMA% zBmbuPPoOqi63%}T!zqD7D-lQcU+*;6JEf$c!6#B={`UcRVK3a z7ZwJU|9bA%2aueF2$GyLi;+#2(y}s8wr0@RhU_Ut#5dJ>eJDC5Wpc~HfJv8<(X}Y& z(xGN~WwCm2pF~uJC~lxooe==R5Kh>pM@c90wMG?0O5Ao+8+%nm}f-+Ubm60@*#tM z{ov40C0|4(77mW_$dA`BhT$^8&!1Fayr`%#LS5*Pzij6SuEQS_zcGt*{OxVY4ht^| zGU@@awZh|9IVs!t1s}19E%9 zZ9Zbf>6-LyDpP-K3`y-`*LM4x|Fw&{8v$KO%g&O_X?2l>8oitiu zMFJro3Xpeu1i3VLdGNL{AqsXOKq^zLcz7P{qyo=mz^FAH>U}| zU0C~tIh^B$#1VS7eTL-FvLCIoqb3f!DX9X{u@P{)Lgc%B-beKG5Q{?+ z^cCVeFz)$Hb>q{`&CM_oplnlf3mcmjz8?0hBT(RmivVOt0Hg#|R9kPUNr;GG-J+wT zk?Xd%w}-K8r>jc=3*wY84cFv{u}t+7^ytaidNJHB2?+^S==&ac2$7PKN=Qo+5)(_w z$Y8^SF&HcmN-{5A&~S4PEtY5yWp}dOzJ1%elv?unVONHF0oeEgrpv8AMj+ExS6AUk zNWz-5HJkbQ4&u!us0!p=hZ?xLx`MS*5z2uAhw%hu9i1*9@>O7n(a}+G!y!oe{NcQ= zej5V~w5BB$(i_FL)z=Fi2E9@tili3lfe$0GeRDi+wl%n}u$UY(8U%k+8Nrx@wgE>cz|AG}ZJr>i{j2|ymQCDB86n2Ou&s|y@hHS*J zZ1l(YCJ`|)uo5!VfyH`R8#U}2Agh5@HWZ|%60>$1A>*510{Fu>F)%*sH3wN^8; zzFwf;QRlqARGG}z+YALrB-z~66*e+5LTB>TZn_b7@8z|w)t0WgHNXvXvHRAabCCFn zSt`DD5|r??MRg0n7mQ^ZU7YN~P6t!c3fcw%tF%`=8-(n*8;w`myfHPsOG+vO!ygdc z|BYw!hcQn)YGI_OKyJbDDm$~Eey9qz3(v*HA*j9|j8ARapCto3#R0NM&_{#xb!}~J z5|WZ!SW9aImKDe-j!CDIfSP)H&pQi(mi}S`5hwr<`xLzP$W-a@L1J872Y8yD{e5G| zmH-OMPxRH`6ux}k59)Kj0g-0Vt3zrQ(s??>~HL};=|l`ClTTq}XKTY}3yDQe5R zySolGO_=yp@;*Mogmm&W4<6j8n~a2L3BLGXV>IS+4IWlqwIcY*?X=@o^oYCMcc+e&?q!#0#O#r_|Ib!@5lu z0r{zqTv}R+%*`OyLgqp+$Vs=11Zh|%@RSbo5j=C7n+mX>mwVH4Aq{=0u5QGglr+I` z1feC2bD|u2XC)_7K<5bts-Ufv1O~tXw$9xAJdmw}mO-5d>zfkI6+{6E&xAc%etv#3 zufzM56?4EXKKH#xfY6UPIX}Z%mbq^015D4B3i6Rgw8K_&GL&HZot-LWN1Xr(~d)%>x4xP;xaoyKY zRRIA3)f=*LNl9|b$^`Jx2&BA(L`1(%_m@$@6W^xfC(7&#X=)M)%~00T8eKY+mX-z+ z@EoE-peP`@!=+VKcy!`W<~97jM}xk1;;nxL8Y~b;z&ss1hw)7*Tu0Us&<Zf3uPJ)SR#tpfB~T9<2ur-tM8YM$^%pNSyq@=jL{Y1ZXH!YyGVryRA1h=Z>n~}4fXZE zdwQ0j?}7XPk4BI29TJjY{bCp@>jzT_N()h>w8gs&%1Rny9|8hV35fu0Qxs^R!OlK8 zJG)CoRYaB9Zh;Ks`^-Z^Lo>i}fi{lo&@WHfzb|l{T`)cIMH03t7}IcIijEGbpKrrckWA$-d(3m{eklVOgCOJkR1Ad9aboszx+yqCoUYwg z#d}ltWN<(mX(}j0bQRh5Of&0N{~W8ZLRdiVwOM=h@1HUNCNk#HTO2#dx7IN+HH8fw zi8Pg;@rnc1F`0UbgrJI0EUO+b#|Ud||AVT4upV|53|If!^hnlwL6)ncrx&4^t7bwW4KhDyaV95Ujr(xDZs2*W@Y8j=IO(+c0Ing zS3`-VZZ(kO@j1+)Ly8QF2oAbZNo8Ajlvhk{fDif<=EDmRYo&q}>W(Iv*>4fNlp2eZ zmKC*nIW_Z>iC&f5_$jZJtG}^O;02F+RC0}vBv&pCzhUQc|1qq~XSJ1@)xhQ2O&BL# z^b&S5Sw#EhiYKU9SUEP^iE502l+5@F%*;MypBuIgv^wEz#xD-48bROm4-Am98lpfC z^=obI*(Yg4EfKVdVYV5`KCoz@Q~(c#e#;Z?SPB0f=zNzv*AzB zHF{pKSTF~zP`g~d0J#GE3Vby$d=E^|nt&^@k!?K{R=mEt94#@%gp^s~MlGPAF?#|s zhlL4u*zDCdHwVo*yCW%=T z<>ut%blx1Eo~DLRDW>peR|*de4fPl5@BphK$a|pG@VabNd!q$$qRAIk2jh_{L|iGA z&`*FTWt$D{&(TL_?ejW1Mi1NXnXMOcd<3+y^a16LP-oOQALkPptO(#sBM_+SJC$f) z1M&tl!RLL(sa^o{v+ZzQFi>Ez+s*;~_?S)!Q`4L=8+s9u=|w(0Xf{C0ps%k_p>oua z)gDNh>l*YtoMv3d)GPO{?(S@mHly_}CYvh6PQ!Q2<52CCCm9z=tInh+OdC;^N}EvAJi7D6*$~a_OYbvZ z87?*`d&rj{PX`Q!d;$>=1?*2on;^tRprR4v!CiNNHmYnVd7V}W0QA7Z==G7rgV9rX z`<5ND`!)!fK-Gen(JqWdYM zk&11oNUxxggAHs8(92)wk@3NoVEDg-P9fZ=L>mPgukS(l)QZshg6H*K^#)e@?61JU z2l9!(JOSdONZ&MJZ{&$V0<1E9=x~Vg6UM1QnZR{xHh5IjS~vMz!j&UqO!!JQP0f)v zzDeDtm9~?iZAL26$~d=gzd=IdxiZ{ousU_k{3#tqz`0(RhsdWMoqxWIg~(1SimDsr zwPu+qE}$2LoUDdFQPQ{Qmf&`+r9TJIQIN9aAd5gYnuV0gwF~+GMb;S`T;Ioohq=Iy zgLdl%HLd;EQ6p<$l{xEk+fxzHgaX^ZwDgdj9qD}`shn_O3j-|(SaiN`Aa&Thfklbl@_~406Ts|wS|5#3bgAVg0qJWp*~5dhr=0q zK!0brTBXbIIx{6DtN@a4GiY3Z8R8~g-O-LsPKE&5;xJEdZmhz6-!vmLQ-{I^p{D!6 zV+$4wdWtR^viu)MtxW+#kwzbyfBGedOA8Ctlw2wykqZ=hFN@p@m-7#kooa|OaQGdU zZmSm@CQ`dJeETNrJz5Jg42OjIH}038mKEFrdH&VhFj?_+U^V2!ht-YmSbeiZlF`jl zwFEGZCqeQlDX9yV69ow4H`le;n`NG>hg}s;K;U$$H3C^rl%(V zGYc>{`0{YvCXRn>X3^fIlW0yBX^A0-=;t4ngqY8vJLW$rV+x!i^fvzcWdGN!8VRKYc(SIk||XaX1a+9dRVm`<#;uOabna>s8@R3Nj{so7QD|{*?9QC$%1{ zyhOO>6$<^0M0j;*Km0TNT2aYx$t0V%?9&Hzs-!aW&vkO^;z@AHB8WM)ImwvDJjTBw z)&`q4Bd@>9sSr_wihDydX>l%*sP1si3y&xm?dHw8=VD=K!5Qhg7rnE}YkNA=ZGuY+N=XYyXa0S&6Cq7_}L(+#K1!+o<5)z!Sg= ze&1?}dDHr+HtVp~W3=D$-&)0eCk#;8+WvSAieZY|69&%bN-7)n4kDVA*Q4BR7@@82 zCRM89HeBi)%`#N1QAUVx@tAEhjqFMjLi;b=Rjce+uEuB0_v&)`8Vi+*5nmBQdh8q8 z56zthN1ly5yM?ENOD2s=7L4}rRz)n?)4q&h(%row16_SJ*JoVL=YQ@+gt4n!dkxfR zDHasCMw+it5-kgV6WL@pS7tT;9BWpOUbdY)OuMQdLH8K1x%Ulv1V-i*is+XLVy0W# zxbapLyc88Lb-wH7eQQ>=YE2)wOHL(zO{Jn~!_E9zL&d7hp{V-vpEWXZG_-#=e5RzF z;NcD{ToY3qDgVg$e<(LDy{<((?T)NvTvCoNk3J4`a!i!8Z>>qWqCFBt33%~Q;vX^Q z&3o*d%Dfx&!L4}LBtE+Z)SJpDeLf`*F48fTN*B|gm*0QTnjxVK-I6zGLarLl5yw-G z+oW$NVyInG9k&a=){$BML$6hrc6yZ}NDo}x;JRh!J*^!Uy#M%Ev4^$&Ei2^nZ61vh8cmX z)5UGC;Nt6xDhowhwJn2XSKsC9g(H8$$R$P>X>1KM(z<~=X0JLL24X$;LmO>YQf9t1 z+%9;lZM+)(5b^1O9A9dh7}mz7{EIJhZm*2bQm5LECBw4xH0s~lL|rdzU*4MbxTKTr zkiM8|@`<7L@(Ln)_e?>?#BfqezsWhJ>H7G}TmST^csJMgO5>vVdc}94>B{a};Ht`Z z$6$ADma6F_?J~~mptzL)Q)sav@rP_ys{VHIAHx&OW=t+%%|dZrqWg6KyPxg{aNYDi86FSVxjhr;1% zpNOg;NyL63kL}LPn@;NbQ?_mD!qd8Gw*=>b-8}@_jrVZY@?US}zUoa{vi4-s5U_uj zPIHj@4_$)xg^OzaVYRca>vIxOf93ZD{L6jc=!llrRab5lIp&o-nC!w5CKw93OaO!p zuYVDr&vSA%?DN5QPvDwod-5v8s!okgQYs<$vwP4`Xo`cPnV{WdW3PV*qIRAgyC3@{XYda_ttFTYI@=G>K5e@Xkw>cwYK zoMtPQreE++KoH3E6o{NF*fw~CrFn01xtJO+H9R_Cir?zLXgcq!yC@L3cz(6k-IP(J zj+OmU^=pu;S)y(b-4Oq`!}pcz4E+I}Z{*eGe-lQKgmzlD7slz&byrO-`JQ%MpD&k6 zT|OGa5j}3vA2DO4MW1!>+vU^2Ld9>byllZ38n0%y3PL zDeNp1QkoM2v6TB{^ljNCjwRGQKONCFys$z#8XZsRCicCL;w?~*N{Uo&Oa%7cFnlk1 zr!q32x7Fv&?5|1J8x(Ze$8o|e{X`_`SsiEL&r}Y2Bxwq{T&cQO6?zL9G0mT_-k$fm zKP)TrS#hVw@Vn$Ku?P3GACKccr9G;0zVp|Hf12;P)w9)^m-eXbl07|cza9}XWNTFg zeY04R5}JB(_E2(|^yz%OylP_Q(&ag>!TFj-o_!!cngUQeacH`2rsN;8>(M}EYeRpZaZoPYT)4L)htt6pf*Cnx? zm5D3cyxRNNSdl~y5C7P0b!^IIl8uV2_(i~SulHog&g`qu=MOk3LaTFUn>t9Sy{aF_ z)^}fh>bfmtsQt)mzKf-d$NiVU@$vZxFBiY@nhD@v>?D_|EUMq3v68}Jhm8x#H9^vTw5>y*gSk9p$rPvy#r`UnFF-sj+# zM~KNbKk$qbIpyf=>w9iXxxt?D`H_tH16n!4y_dUy3i!gcQLFs_5+C#?M0QZlD(TxE?hs!|`8&z92R6?!C<@D6xQlW8o zbKBlAS9Ffga%RElrZ_=;$Ai_i(5R^^90Jc7ivapuhW_6T)P9pL{x1P1yr>CX!w`+<9j$?75do-7K6|rpvj0!(p0!ht~E2O$k%}wP9NTFQ)tlqWG|l zi64ZHvloI&6{vrO&h6`!v1qW}n9(+tQOA|gUcEzof!pDM-h-nnQtwuXN2 zeS9F8=rL2x%NO*!vNNwyT;4H_E~1IuV2(^ces==39z?1saoMc9V1u230f| z=Eu^g-t|k5C&r+CbUwi5J?(hy6gc#xc%oX?s^@n<5h9fXMO#Wz%#O>DZOHKZhi4Zv z2UxW1@AMhMIozn{&gO+$f6)=JRxY3XwSR`5AJEgzRhIC4Sx{Rx{3X^l-#R6IL1@0> z$yr58qpfng3q^-#{S05_y+2O-T~XX_}I$K-GIPF{aQG z-_+cz*GQ=lIEqvw2((Sje0|0!&*-KlZD)9Nvcwq2+AG%U`;0Ti^}QS+BuNsleY|L} z0j%?^o1S8nl1a<_!9D7>$G0o?wdZfWdW%9j{^FfG9M6~WnYnYvjUeFqo_st)DM?*> zdgoc|6?70fM_DMu(oore9j3IeJP1&!AoGG|;jrE!GJzUv?9z|~!6Z|3&eQmSE;O=` z^eZrDhR_7kv7qY%%Q_&FfDaO_JVt0ndCIIvaX#QeU{yD~e{apBlT!qNEJ@heopIyF zO|D(L$#T8sbg^W$Ua_ibg6+6A+2!@U1Y-<8doAbI;Q=SD=4_dh?p@>2qQkT;N=kP7 zF~NJrnZ|ho!84ieU`&s9X6p=)0`f$WCMinA70P?d4?eul$+7|hS)$QWvbVR-FMs>n z{GH$W_qcNFHBg$9#zR|yRQM2C_r=4HdH;uh#hY)v!^z1pUEkxqC&>$7qw-Nm0oF#4 z-1i;UTC`G3CSxY$7}Ix5XL}f9x%H*jSyvT(m@_JJdTUs%*63%bp1vsd0?@s%7xtVM zrHFzM{3n9C%g4mPj(swqliTwHQb(B`8;~>bbsMG)K&mKv;e2B#L*UXjQVjnKVK{$0 zzF`1APrJA$y96Mf@qHmhWN8GEof$UJged#88(=6t8H79+OCkh(iWKp)nvxP9JX#9W zc4r>@*9C_bacT7-5FB_ZP((IW4xh_`{Rv@%q#M+Aot0goiZW2|A}Kp;tkL?#{_AOv1Y zAToBF&Gm*x3I zA&x8^&p26DNT~?kgHl)@m=!sVX^?~C#Q8vPEItIZ(u5Eh93LFU zMVWJ=1Pm;+_nzRPY7Jf2qog1aP&-em6~Q=q@9Bj=Yek|Yl`$k*pk(0wSw#pQYkG`x z80S#sE>-V&_~2vqr^n3puW)*L!raF(_wmsj)AwX)f(?>@WM{HqI!XybVS^%gOW#|p zv%LS{jIOsV)+r{HIPYn#=c9Ya2qDl~qIE)%=XmewO;6JV@DQE+!BHy$}ec z7)^Fro*q-?8A28W*OR6RMXCt41K+W1dJ>&b)oWz1*q}J8EXHkhweIBr{f>o}Y!cBds$Cltk$b=5O1v)JCwaI);U_wj8G_omo- z&8@44983#lWx>t;DUXiFv~9z~(-o~Blq*6oN)zw~T#p2D8K|0`cG&YLS~9aBmBcuY z>EioGDoOBxqt3V~7U6;<0sK&2G>yEDG=<=41%>t)JR@Gt)LTikzek3>r* zS;ARw`Nqu&zwynN=tauczxgv9?oYUPcgY90@9_5h72b4Aii}Y?X0=*#P)c6Ay2HcM zGY%&i9fJc_N(3J7J~_h3ngz(1fUvV=UK@ z@59#92@$Ni5H89121<^ng$p_#0TKhN5FcmkkIi?(0O~f-ZUf#v4a9|~z5E_yvo?S^ zgt#dVKH!2Lejh=I-2kx=eGY!`KX?ag+mZ>%I1vP;A~+jjIK7Y{Hz0HXb(b1A+}$vQ zBfz@=df9N$!-G z_~wP<;_|UCiUpw|Sb~=X8DmYeM4^>JDTTLD0v9)cnnXLaQj|r`?rh3@-LhNel#?lK z{~^YC>fR!i!dQb=If+)JTGRHHs_D^E4UQBMuOk&s2J$qe$WjO%B^*A0@NNqNjkP$y z4s*A49wh|HW-dF=PM#wBKyN%XovZVh>=sQQ2(B}~g5f>1OjkS7JpRYlzxOmCT#CA*UmnGz&g;cSJ*k)|2r zQ9;wRBp9Z%DXT^!r6EmIQU!a5J6t)~M+;ai&p;;3uD;B29%HZa=`OCT5P)(m-Qox- z1#K7Sf2t+kCpe)fi;Sj`l*-c8OJ=#I?mdUIoNN2L6nREgj%n*PCyN>>9)X>ZYRBu> zXB_NIF-t9Vvy^Wx#Hz(d#qbSZ!R!& zLf{85Z$dz6#r4Bo?jGMqXi3nL)3w71PoDsW_}cYh>rmTp?m|ZJ53uV5QaDaL7!^o0 zI16p%KnwIplc?yUQH5e}qUe?mKd{Iu>Lys!Eyv4>WmRFFM~dj+Dn}(>|I$_doo~Lz zSAXf(P~Dur_|rc|1<#HBT?o*2EkFO-6>eU;N*5&M{tfz_J$!Y{m98U`usbPuc(g!w zb8Zx#o3HP%zcU3}QD&NssNNAmMrJYq!J(1_B?7@4g3w4IC{oR&$S8`EU;35r@Qc6n ztK9n1>#?I{R0>sRa~{o4Il22G?|$$h$HymheZ1!+O|gByUB8{P(K%>iEW4DFJWUww z>@XXb)NRjnJSNLB%F>f&Io29$*v0>&eY>+pb_J0HBHqR)2_i<$S8SFDCQNz$K=iFAJ7=p(+8yR0?ks^XpMcm{{8ECqm zg-j^5L`Z?q5`jRrJhee!~F%W^!e?J5O4Z4lflxNK){jcxUwiIQ7XnBDJi!S$wWsl z{ATS`Qcx5nnX-7(&>4d<7VA81-(zh?nI=df$-7@W zM;w-laZ=D*2Rm5=urkKCtvb&uH?DHHe?Vs)HpG~wG8I&n;K95h(Vj#^H$rCyhD9i} zmh8_;R?^4*N(6N$NVH%&N-2|=*rjc1io5{t$g`YGI1YAqh_~&qs=)Rwi^USUo;1w~ zG9dtqrlYqO7bKm7rs+79uv=OlEG#~F>ZYb@a#nRi^YDx!l`Pg3b=xAPAWzd63uzs7 z-?Of3WLi+Qh7a%0$%_$*6r3y@?#)+*PfB83< z?(FkN|HnV1X&T1x65-H$O3k9jy>aci%{hHLCMA8`Ff z%9Wjhu@(fOQKlgTKPWtWlw5kZ1#=tIgStl`=xtovzW%kZ^Gm<_YkcLaKSy4SAyay2DwAcNXU?ec#iyHNA;4>MTtt$0J6glAY;{*>u9F z$oawhw<#wxlu~FZkXlE{zmOEA+=4(U6~^ zYLgJ-vWcfMYR{4gdIrhl26Su^pKSiz0KrWgwt(xCcjXt^g{K@Hgb1>N3)`29bMYL{ z%D}j>n9K7iEqLtWkH@?Pe@-WDvQ+v{0LH=l%MVgKoxR;=P5>HJ@k$L$-Q|7^&Trco zHq_>0NmO`DZWAS(7le~hJvBhG;U`;ZYJ9NrDTy#;h!W@MsgD^>pXF3}B6f0MRt13| z6e76cG0HC_D|gY1kDd@0qEjT`@LrH;jkyp9rKF6*ba+OI7BMz7MCs$K&=hHku`aU6 z&a$W_dDig8mtLc78=Cv~(TT?TD411JV24;Zr4*g980U~uV4Y)GH%!YiX5S5gG))nJ z>3jAj1yyHRwmqT{__ard=+l|zn$9}LMaFOb=685-?-9nvWFOXf#Au53o>d*Q!B$mE zpCwoqH`Kw$wKLB%=2ee#aqVu5qiH*=2jd)}ZI~~Ytm>A@xM0!2C{GbmBa57CHzpKW ziU@{z6EI!P$Hz6rXoS|9#k%68?|3v{vbQs%$Wv4>jGC7Dy5-uz9=2~;)eTuuQZ*fq zj*q#vGbWJ-Ca5_aE`nl^F>= zWh&^5qw@(}P@JAN%*rB4V4I4gvzDXNig9Zgk5Zbpqyd+NAnZ3jt<|=vNbC+ z4hqTN|Mn~V!f(Dtvhy+@{=q-ud*A(+6h+R#D5c0#MtRCg2<{!v`FPQ>ZY=kYR|pxn zaqTMaoq2LyTC~=T@|EP0x6Xvog0C2E>z|qMG1W%qMNF}Me7}w^UPQnkY~C&Vdov!~dl=)bl}0E{kV#DH(bou_ z5oAMWBU_@CMkSKo7*^|s#t8DX&gV2$`g98s;Y_jEjsqjdAjmA|(v*q`PIo zEc5)_je_R;-{<$={(b)9&3CwS++tNhK++kH^`JxG?pcRU1X-T&!#j^SJ6rR&S94Mc zvs+gkY_muhllLUX5?9pb(X5G zKq#&p9560(%IPjD&9K&Dt;ae?-*srM$#snR^da!*(L>ZAzn_jr;2n=1+~uQ>KH%Q{ zdq0ll>LFqCS;8;49{oj(5z_Ac}Sk;uzM{01}5#Tc-sTyOjP z^Xw?m#E?v63-q@w{M5TVKZQUIHT8rLIO8rcg4V@MkG->mAn{#fd9AR~JL?!S-o<;a z5Ug6mPMV_=feHa(16~D;Q=~(G1skHo=`lcf36%NGco&yI-4h^J%9us-;_pLvD(mY) zK!@{KJvaPjGk)D5X+wqq)D7gO+qC6^lp40+?J|GO6Q6z2O>Al!ZWvf02Pv)?vPoKm z5TsfSlGoVZW{9uz9-M!xE- z_e^GO2_t+#+(2EAOJ~3K~$roM?lK7EIB?~M#d=wgw>={voyfgy=RpkG%p-n26$N-`_QWOl{Xv8EF{_z*aA9n(THp3DHKnwF!}1tA1_ zqv_3>ho==f=_!;$=>(&vw9SHfW$5~zx_2y=70!bi60x*Sa+W7p>&Uf6Bn85CoGu%d zt!GjsbRmIsjL<2kb&CxiW8ypH+SO|u%}-gaj##&zAQB2KS*$Fp-g8?dsUSV&Vqe29nS&|TJkMsUqRYE`}8mVK! zn2~G95{b2*WdoJpFmi!Jq*!k_+f8EvX%DFmOm-3$a|aVm-&wkK0BaGc#wbv^LL`E+ zRLEXoji;hP2}$c5RpUu>$=Zr1Kfhdz6YmE7OvZjLcpDPS%E#`I@~_@YZp~sL(um>o!S}(D(hNIcHT)>l3(mLFda8<++bZdtd=W|jvjI{Kc;P)=ab$4l-mnH_rhM--vHZ4 z!Tccc`a~O8yC-UfJxwzPPpC2lpaVGZiR93yHR2PGZdj8|J(*`W>T_ZTKSz67b7SVk z^Mc@h8}~MB>K}WNSZ_&&p#5hOWFa7A1VG+LptL#PMs^ll*s^mMOCp4)0KiS-gF6Q& zo0|##Li;un(@Uq=WF8SHEw^B9dvti=Z`jFADF$!57^mXxkSQ$!GNC;qt=UOa{CveR zf!!=4)frh5P*NgXz$-^@)xgxbC~b8S0DBSQMt&nj_TpK@7{U2VMajMQL-xqn;tgA@ zEXC0?&$Fhm_&txvqAto(|Qrl-d)QfGpB)Qbv;_yTMTEyp` zYxXC{qLkPYWXw3}0qYG~Nz8T*coHS)ts`)r)mLl9>iCq8=QX3r1f9f0Dg=+#iag6` zZJ_BoPUa_6Y0CBeU9^%hn`M$vWEs|bPR~wx`~CMQ%Lz_stTU8(j-P__E+#?g9lh~* z7buh1hStPX?^&YJeiK_6+uiwCS8FqUtdDgk69n_+ntpkLN(-9KaJFtp)`9o#ueg6& zGf5n;UYk)FL)&)zlmGY6nPiGt5$AB-HhAyJ(u5+@%*HugZ!u>Ty$dK2@G7UR*3_M2 zf2U+RO6ht}*R5%n4fDC+=Ji8#2vn_Ov1lk#MLC_)cRh7ep$7)U+a4(`7R9*C(F!K1 zL@Ny<)?gc@mV-herAK6%qti7LrFr$0DMen;_p$JKf}vZk$c5#rFJC9g)2;K!drzK7 z5MW)w*;&nNH&bSllE3_TMbmaMJWwv=2 zitT(00q)J$glWJ5OA9iQgdDtZSYg0J{ZWOUXyjPpGRaCRy!N=0K~m^r0;ff9k%=T# zf^sCWHjs<~3#(I$$|R*w2pvc>fsl}C#m;Px^>V=|iv`7TBf12ro6W*_4-FdMQemg`n?y&K8;gFU-6JQrkW3v%%A6@oPni;WJ{gp4OHNSKWRt6jH^P<0d}l z^R$7$`RsX6#Ce?BWqUIEZ3D2j&*2%*+q;2(73aV&_z+n+C1Sum(YyaR0N&K`d%V7# zP%-Z8)9!5}XN%1f+YdMA+@5b30Wn#G-o|PohNJ@_ZcJSWSRZ1V@5a(?Nyl{UP z$D;G`+lhcg1`^?8VPGlHsl-c(^&pgHJkj*FW4&(3REE}ypyYFt-xmSf2F%6JKmYOm zlV)|>k!KkyY(?V;hB(Lz$xk58WvyanN`#0EvG>%yqw6F#=$P>?1QI2YNlKArG`*#B zktJ)rWNjR6+hGFqRsf3ivg6MDBa$Q`OB39X9K{4pK+y|@laLCJ!%=sJk%@_bloE)< zBaMpVYgh{s9p_Iv1eNMErS}f0$8>VUTX&c2ObtyF%OT`;MO&@8F;Qr#Su87F*-2S+ zIrFnM!3360(_2TLCO8+JAX%#M#CSx7<#2z8o!Jge(+~(;PZH8JLn%q-M$E<~i&cY4 z5^h|*g6Ufvisf{IR)T|r1Df8We2X)uEExhx>x%l+k-VzN z6^K622Z2cyBA3{eAt@7r3y54I(*Y!;f=~y%^`wbnsssx=I3D_dF@f2RVt+T~z4sba z0NQ|>O439iOdtsgr8NhKhdg|6XPY!cYmIe|L`gK=){!kfj;3 z*^H}K4teF)mw5dvKSP-+e(#_DQ%>h|7Ihm#Jc6g|4OLaKS}h|h=FE_&C#Brmr7TOF zNI5*%qitLAJZJylfZ5J2-o=cTtK}LgCA-sDc-~lxH$A3r>5ZlC4MkB53FQLSdc~th zk2pFy;`rndy)i>^bAc2RA!1F~M1cKbFRL#A-3xnRKdL<~8A@DahD0QvVruxUyO?J0 zQ#Ol%5Sa(gLC5CB@QDSag@|!n!9_XQhWQvw5(rX08}eN!q#RJ^v3(a>cJY_zu`id; zr=-Fm1Erfux0jVWQM46kjJx$khKT(VmNg)fA z^>||KD=A_hZo=859Dv%uB0n_|PZ)aMhsz-ODYgMtp6CV7WgwNYPZv#K)j*L2{KArC z2}O~TrHVuaX0x36X-#wLNOng6X#K@5Ui*32(XwWgIVPh(o+c=H{_K4K?;R#Yx#gyu zhw~2SJbTkIZ+>*2fAQTvrpPj!>Ci*+6sa_AQ{mi@wb45|>v;eEF9Ov7?Lca{1YiST~|0C$bD-F0HYgD@dnzGVS?Kzh0H$+>Rcjb&i!mKeTD%Xmy+vu5 z78#G$ry!Q3SIgQ9UjTj5`lc zah?dSf}kwNy!7%dc6ax<@zTq9XE}Ow#QnCypy*w|Swr14tk-L*^@`pYv~De%7~pwJB1Av`h5@@6p>r-{JZHn6 za-nC~uuo=F+w4cNb5InXxZ0;C^tjlL&)Yr)2tEO*F0@8$!PR9jHyl3i<1m1XP3^gG zStA%?jtyLhsqi;oZgc&3jhH3uae#6OMEdYto#aD^jh1Q~b0LR4>n5yEh?wn9=oBH7 z2m`2rD!jH=@NJey5fY_r(*?S_Snxq!mqcluYQ_fR<3Wm~KAN5~n08QAj0OS3PDrlqygH70yd_$YIj_=rt$5c;-B;nvT{O zrbW&uODVD>CO+BrV{~>{7oWTL0W&10NF?08dzW#MGcNKWNkW`Ud6qHul2M3i24gKJ zwWIme+U=L;>c+%IvIjKA5;}t%VwjwbNhL<91|*$}jCJDzO>fAXhV^1WZv%2U=iNIG zS)Q#ps5(}y#aqj1gw7sP5TdXw*6cWqsDB+E3 zV@8SM?pedpvSX*vSSL|3GMNXHoZu~;^%y7E9Vh(o{)*i~V{=JaWRyipkqD+`hCpy- zZ^WyIBR+f-`+J;6$z%lelBTm{Nen-j6p}oVjCT(RO0oOWmwEHg|CG)KIu{U1$8svB zBgpgZ>_(<;jlp(Nt~lE%sq2o~w1fweMhGTKGagA&FIm{0&RbOP*_VRG1WsCyTMuJ0 zY}OQp7-JDjnkb3sER$4HH-Rw;dL-$M$2!aD(y|^azWUOP?|pOzfKF0;Uy~(@ozaNX zb&C&?t%HsH{ydwploIb8LwYSjfU0wV0U7I|o}8YdoWVLno+MZx%~B@Q8Fdf0@6UNS zKi%do3kYu9yveIydJSU?eb>_WF$U3E%Y1Rl$>}LgT~Rl&+(MQpM&%B>(o4D=$a8vQBKQlQuC7?GE84cDH$7e7$K)}PPTns|&Z8<01!bkAtQMM(Ig*@iuQ#tnTko&<)n0DoPQv~52Pn}7?kY5pnqu=(E8&hs?e z#C7=q2yqUgKA}l|kf4SjE~mE-&jxZEY3AkZkQ=ZlxAp8U+C~zr5{Xm_p)#b{B%p}T z!3B>uV2z_2lA(CvX`Q99fv!_H(G!9r9X1^RUd6k#&Y^HfXK3m*r`4FfJfqZ#jzH=_ z$_SzY0gp!piP$(I!uhClTRetLSY+2x|gssD(RLj$~dxI5(-H(xAX!; z7QJ7=oG-e)amXNGC2k!nrbM-@w)9s-ZR3yF1}bQrGQSb`D=9p5u`T~Q{QQI?=(TqB#d zW3gB=nM&qoOXg?mZ9h5hsn=`j^$P0)U0rdySa5Q3!u<4%s;=0q9=X3;T@r|EDMZ9Oyg_yUrK2 zOR-U6DCYefP2QBvmcvUVGCQBu>vbMiwgDNJjy^{~+mFr-ewuBhtrz<06Ts&s_2UpD z^DH~BFn8W|6CklMSw{#%ok56_L}Fp`&AB(s<)-kl_c5L*_{aM28BK`ih_cyB?Q}!y z+#vlF=kiV|VrdI8ct54LI4lN%QhF%bYPQK!;;;}TD#Im$&U83aQ};b-k+OBO4;yC} zhL}KaP{HAiXI)pEjY@X%3~5|s0u7)e`=zAB5fCAQFd;l)h#T1;&GYtOo{LW;irHY_ z!LiYHj)%*Zdnap_O;0WZMWQHm%DBwPq{sEKzin@Z?4z3TEMs)6(22Tq?ico>Tdct) z$x_X%%sH!C=Ie&0>)D%(*c}&9uInSyvWb0kK9DAgJWFY9^voA&iq;zMJaPlJ1jN-W zW&_T&V6uM$YkEeLU6fMHcBV9KPnPE_&dx})45bw1s36OWsEoFjwyC)D-dmVz0U{=4 z(MoR9y@vq3^JwMJB4Dj!u~^Y}ElpSP;d?*i_~eB5?#?Mi$1mT=DEDUU&&Kq|#(Hq2 z;>u1=)q85A>0HmrS~5xl&N+;WYiib8Dl&R!q8u(rd{B&wj7gD@rx|M_I9V(irIOah zb)oAl#w*fHMTut*(@~CdJ@d8U=&V5?&?2x78rON8m#n)$-L5f1=GsKXXV6+tqGHJc z9|U*WdUuKX~spuk9|0W-EoR> zp58;I0I3i%peG*VEmdct3{hKxmM9lE=^e6Aam`FYB++@~ThHF4WL-5>Hn2a**c%P& zr-a4IU~FK~T3)+4~LkMR<510ze`&k*KVylwwc91vgxf|9}hrPYR`b z(@l}4kU~mTcU5)eoJatH2*lu{O*+1bi>GGp9s`ibp_$e2D-sg!=4PsDs`{S3bMLw5 zNGdH*((ihb&6uE23YyOHU@e)+ zIKF?M2PY4CaD1P`qkH`97r($e$HB?_9NoXq)#)pYF?g9J`h8pT==eTLY5IOZn4G=Y z46PLF^_sfqQz-z$wby}kk5Q5_tO)@5G zM@W{&Z{OL-Y(xlT5UCe!G&Bn?5Tv{*Bnc$!%K`J)0d4S z7`$x~p6-K$5>wSmjDC1NCbiS;BErQ1FEt==ZrGgzX+T;*t`*8lln97eKxGJ6dKP#) z-~g>;s$KzdRF74K)pdp zPM}vJ)OT08^4o-x$+t?Y(bglTd%GDN?Zt{!({pb=Wm;xvDPZi|b)ztd))~YOr6gr; zNQtC?_t#Z$k4n`V5%B@;pFE_fCLA5zqwhKfXE}K^V~Uj%keBvLOmuYBH5p4_GcxtDyQxoA3ZqY$>|as zC9{da+C(iKyku4yCZ(n-4MrOT(DjzfRYwSp*(5`E9)ur7nBcE|aLm8{_%T^I;qzAu z{P33Xx1IkTA@_x|MT8s2hY`}N%LhCh*sl`ppG7` zpz8uM2JW|zXNs$B-?2+zT9{-BGoMUw&eIN-JU5gIUM?EeLmK-Z9nbmn%f+JRpv+j3=58M%UDqHc8C9-vaTw{_DMF06n1G5K8%eE?7$uNJh7zN=lhWj4 zvg@`oaUe!VA)xnx)6Kxss|A>V7kx3|yx z?>}Z<<@6r@)8GF+hkJY6pA}?z#(Y{LO+h3n(zctlt^e*PKS53R=)e4eCIr6t@=HGZ z>{G6m3)-$vJ@n&v-zG%lsC2qrgKVj-H=oX1+NBTzALA_mr=)lz!S`K^J3x0AckxFG zArdX(TYd@-UF6{E=HviE0=CaDy?M40$M9`anOa#jy4Mtpl>g#C!epI*&Z=r@IK?Zgx z)kjcqYhU4hcbmB|R(VH6WC<7mlQPUQ2*Qr)miA@afe;)5F$w_za-h%|OeKhYn#}^? z{VmGp?rx2VnnsBDzGJniS!6j=CDAgY#DX!YGUk&IgOHJ=-1GZK^nu2Do?k6_da-7? zu9=s6CzQphLTH;%U z5CW|vnM}zKCB*0}m`>!P?%4E}!)d|Zq(m!?Lz4mhXc>^{BzL~rY$)=afWRn54?EbA z(uKq*&(p8I;&47^dG-ojOjxeYu~8vxLs885>^DoYe8Tyr<}g#}yu$Sjt9nD%59B7J zAFh=Lw`R&W{jVa>v^^g@y2s-_;Al~wizfgg$?i@{_ayA98G!JWLzu< z4$H`cSt8fHx1W=jIg8el`!t|j`mnL9U-=SaL>@T8XG0Ym1dG_veK|vHQ2sq z-8k;;Pr!9_K_=mlDCqhD?-fEo+gsYfp+sPRHf2)f7{B3@lNtXm%aNhS+JM%QuD3{? zaelc(k$P?iks;|QG7TuYK4ANvgK5F>JZDbimGvBEnxm6POfp5xr|gG__eTstu->d$ zwU#`~5fKR2Dz`pD;{=70{P=?heD?B!QckeWV*3tpkc0R5AZa>JW4HZofhraDZ1i!HiFPRp!FmYn&8GOp zf>~k6B&<5eVhB`P@cHSQAx749%isR|bE3{ToKE=k%QKFTk65`Sm#Za(h@352eAhCF zBwWL&xNDT6?QOb;Zy+0^1i6+3lau)#B_fF&A#pxZ=GPCHq(m+G{Icisi<)KA@74fQ zAuyRii^!_$INU$v2Os~4M~@#ffA|5SUQkw3?mzxM0K@VEDJ5B!aoI&q_Ca_@y*|TE z=Tv1ur3JJ5k0|b)@OS_5Kk$$LuDO^b;HuDpmlp^xNAHy*nOr^fR;?$C!s)l?efSW&!n zqQ3wDAOJ~3K~#uQ;yfr3lUXh#Q0L=LsMOg=@Dd|DL8Z*GkjWOqjS6SM_Ba=jTH{2- zJH7jMQwha%it~Qg$extQJe_sun7rx;9&At3Y*-d0^P(iT0aQj|EnXOu-Y$$|0-OF? z6h(+Io(BaoE%;KxB4w{*7>&bYx+Y2pF3vCc<%YCvte4ciXK)*wPc{q_HDG2r!6tANLxT{J%xE6%1&;Qo)Y5T3H+=H=9uo!i z@|@T;L_MKv)|;5$%4PUm-F2DAdV z3Pd6@OF~W222`&I<(m2YeKxBlF$D5lalBWt-wxcH8ZK7D=nL0yJjp=jWLoiL(a@Qe zhaXLlLK32-9z37ET%wiYU~l#Yt=q>@ENWDV6fvf2=Rjr|g@#ONLX0DmAc@5+Ysa&- z<;%;Ox*Nu1DYv%mti`JY0Um$w0iS&G7nD`S2giH-(=T3;mlZJvLSM7qY%m8WOlAea zH;BHWy8n=@tyx@NaejWrs~68xwQs#*=#r31KUlO@qcGL=0r=N^Tp!5m!o(5W4``1(rV z*Ty?&mEm3X1X2=z51_X_)GZb2T{X7R&20o0?=m~z(ks9DI6UmCWdV}%;WvC{t7xet z5ED5!0Z2gzk>JIRvKDv|M#ak;_s4FBH$sl5I|`zR_#n8xni0PxvTW~(nCNM2P0O)} zXz7JOXvbEM1Od$uux-nx8`!wO#N;>v!+8FT%8)Vj($_9C^w%Jk(MCWBgqElfQ8CcFJUE14A|2YY+iIjQi?uGPH});frX~b%&XX%iR!lfKd4xl9e0-d) zg9w;h@bcA)r{|Z{n+q-$3#Mg(^`5+#&8(!Brd30;~-PDboGuAQwt*Y6@9_ zh&U;DvFvH0!1-hw@x*m}Ua^&@~whUDA*#2a=AyI#ZXA#C46BzTV$1}_uc zW|Wdh04|IaTp#h#j|YU@E>A0wy@FA<`%^li<$u2>W~T8N}5*I5@h;`sxBnq-i&7+JT}f_`&-R`KK?gn3#dStvQ+PGoNQvvl5Y2 zxYp3M4PF;WaYbVX4z%Lo{Ue^AUGVt$h*BJJzF4D`L<;Bwbaue#98(s!zQGuWphe&* zvXa?!g3|@{Isu5)@(L{ig@%`lj!$1K_~LYpREiHCr0Z~~({;VPxZ>&gnuiA^5BD=N zC2@nLcZzPXOv{Y-@6QQBak=S7+5RL{<&bPVL}{4hhM{x#s90^2enKE5%rnWu!zn8( zI9;r$*K34QtcJk*#_%t`H)H5}o}R9FvT|52ILZ{0QgJdfoVQ7QYU{D4ltv)1&W{xH z+s_E4Qh%w|8Y2b0^*CVJ^eorOU_Dcwb_iT-dcIl=JiThUY%GJjPDmyn*EB~8B4QjP zK15uINqhu|!Bgc1gQ8nsAZ<%id#=_sm*;26*DgdZ77IT8)i3zs$rIv`s&LNv)Zb#*7=IT+{0CysJG%SL{m-}VEy3AcEE{ao+DL z+CD~5_|ck1zsN2_~g9 zy;ERiO&hKIB^|3{+qUgw1s&VAZQCpC*tTukwrzKCvd{ka-Mn6NR@GZ$jq!{^9goZZ zR3u;iDhIycCwt%9+dcYAFA-r7k9hq{^A_Il7Qa}?B2C`H60H6slW6HqM7v=BE+l z$q04057CO=ckX#F-CNq+wV${%D58N^9D&*}cV4w=+sR8k?0rT?1iXFm|32p8hl!27 zE;^UnQ8Z51NPYWa8zi@%X7Ge~hL6b-`!&pNcUd^YWD7 z4GFtABJlGpxSc={Y?L{WDuD>;^L{e^1bpHtbYl78R?x}YoKt6UmDAoX-=kH}(Jv)2 z73F*tCB)d?*6g&qXk()8zBozhQ74zSeh!z9%60iuaOHO}RMr!!$?(KddJR2}Zlz~6 z_Qb8`yf95fQi1(4e3LPk=3w~UTDX7p#exIXP;rB`DouOe{r*Sf!pY1kErZ2+Y;~0d z2Q0hnh&MZ628E!OKetfAPAIvxm^_TGzyLQ#Tt_aNifi}}A-H5@^}*^6Xd~ignkX!f zLQ=~bTIo*P-;bWD|1I7{ci7Q=@quewG|IJQ$L-}=FBGV*B9sfd8|?4;e@`zJi#jWx zx-;0rHAbx6ZCN-`_znjzZsU=i;33#^z(cJC-HFXe`^{O`2o1nFRztozDyP8y?1F(ri3Vhk!wS!)Gv zV+n)REu7vlJ@wJ8;0aTk76ne;o~cY}Y|XBZl1`+hlIc=gFCYdErDamjjK-Lr@;ImD znRyNNBL)v~r6z$l3AHcCP}tRqxAGgR6GhH01BDsc_0+gfn6~D&8t2i^02G=qZU1eO$d5n@hAY$6&~SgD~02%XLYHz&=4 zg35X?d@!CL)#ojF>U|KC3lF?KB2yuMz5Dkvnv}mr>l&rUMh{G$$1IAUH)=H1wfKWT z@pJB5ry~XD{n!b!_3on>X43wUXX47NW~9hC8+WPiTSHQl%n-l4oV@0N><(G#FK~U` zfBL+47mTBdWd|{f=KXkL&?mstu`U);-PasyxJ`91r!lPfJ>BDe#^7hRM49{$114t@ zj6bo~!HtADH+Tx!H3?F)@MEM~m@yQ>DG3w3cG+2T2zcT+#DR%Hfu*?wv|(W{rRC7F z8~HH?J>#=;c`BiCfwDJiIjAym)Q<2vIoC0+n3OWiygSanqyW*&wCp%3E)sT1Mdg!h zadLm$f3}!bd=?ln08yHpRCJ2+Qt(o-xa8EXfnZ(~Gy(3($BTexPABaeD|}VZSgnDo zsk4!b#wgge>kNRZ>^{l9)KNsheOWdx4p`g*x4b7(*b>X|hsXM7TGmp%^2f7y+c$ssRmUYMocy%d(Im`#Vz^!e zO7qhCMuOt;>+)0~aiA5RnhssaYpNT~GdDWXM&tj$c->Fczfx#6`exbm z<&f(NVAu%|XepqsdiI7KT9GT-Tt89oUA#W9(Q?{Sf zLoskh2`-I$lVhH*%z5@E<|C$RjvJXHL-6AmwrMb^(~Ma{HD5<3BO90%l5uXMPMmb? z^ZEzQ4|Beq&peQS_8Rpf_u6VPHW69jGF6W?T8wRgT9lGTT=IfU0NzIt@6gC$)Sjwj z`Fds6Gjx%uvcQ~`o|S*%%q*0VDeDFea|rk6*r{CmVc)RNe}BLy2sz=@Gk6|TN&i0Z z6r%GAi*c4M3A_03V2J42MMsL6T+Z#j_Yk>O?dyC#MJO+1C~!bSd?EX;E77Mo#GnA) zs{u>kZQDnCXV3em+%NyddHKF?E|Z_K5V6r#-)`T-TmDW?j==&_!j4W;%919{U+!!MdSn9F*O=&rxogwR_{t2~#OWx#jcA_pXK%IBoQ02I_E6zil$Fp>Ma;AH`pKDZk+BB@1DWN71&dT=97QZh zqox(yp3zIaw4{Z?15Xz`ILHr%CNMs3zMpsYvrO7r^X;rMH>8AZMvT_8fn%R^D%cCF z6EQ-TKl`GQrz(J`CUMQq!*#ov7Biy9sKY^c2KKs##UC=YFl_j5|yUN@+< zw=1V@KIdHp;*u)1{tdS_TZDIfFO*4+$CU}A&5JPTe*4HcBL@-e*C&*rT=0piVN@|c=ZF1Dla_jxJzEL z^}X8rN_=?|-pMh0QeQyG@|YoH#KB~o8;KxS649__{y-t&{Gyb-;H}er0EVN1R*-Kh zp@mE;E_-N4^W=aTMgw2_oZ2n^Ds$c7$#}T79o}@i4=q^^7&99HsiKo;yAJB@${O=7 zEQ>Yk96%bj85x}HfBB#|DCQ^SQ8OD?hZ`sxpdc~86Pi~YdAA7-VR?{^HB>?F!jHXU zRyVO4YF>&M>hC-)Y{g!ACoI53SwrPDCSDygBEMX)E{+7HY%Z_oFCrp{REMy~Za)+91R+uq_EHD$`i zJ&2Vlsg2Z-xwMXvib|7llrISJQU_rO>ueS9s0||cYuOqq%s)LP0)Bn__kn5{Y0)BvnCrq+4$fySp5>TvolThLrrUInbLfWzD{i7p@8@>hNf*a3F zC)yNgdbOW(ISPmGR|u*M^FSQIv<4ga)2_MZ#9MJm!4Ric+^rL zgQ{9-9r9oQjLxH@c07S;p5TW;&gHof(MZ-^r#j_@9roE$kXe6`Y(UU#KW6Z?DuxLO;&pJXksL!LS;Mu z<_t6*Dl{T@S7Drem&olE+$p~hwh~ottf()h{hnb`{GB@%Qkdj_1F=MOFn0ef-4S8> zv{ny?ybc#dAo23Q&BZtp8IV$hGDWp%!gJC&s4DK>Qg$N3b(z&!Xn4vcn+01~P$@Yp@;F)=G8pnz-EYzILaVg{^o^fT^g zBp%W3RjV0)7|D$mnil;pNCxB0C7(jE0a3)BLjW`)7IV5eseP>|m)POo=E()3#!IYl zF(uV}JwESnI;cr1ecp;i4@7KP@Vo*72m0ItZs-IM`3kSO3Yvk7!KV?ny_ymlCPr+5 zN$}{PWvo%{A}RKMex0NThWps#-@d3PHMLJG|B%d{7< zu9)&t20rIw358x%5J03hNwf;0%GDg@$#UQom|3CzEeMCE^&vS`qovU*1L;h*&(J?F zvdLLygNDKc^90j@F{Nki z*>8o0DbmVGaH@4>knKw-%p_@e-Z{Jz^4uWDH$ERkdR^XQM@!&d3tW0x*#RlMlQhdZ zaS7}oEz`&)Tp`oYJ_A7mT>n)Rmu%6lrQ(I1#KIP>N)Fsj+WRi$`Rp#4?n>LeeI>?z z58>Vm5YW`({d)kaWyJnQc%kz1h)^4%R6NkoZsO&tu&_QKPZpkiPAR?_lK~|m3S3SH zXpkmrQEAZ9heMrj2c0rW**9OOC3Eo$s;HV5*|W>LxmFe;D60TAo@6FK3A0E*D%$rd zD#$2oU`dZ8aELWwHF3;4vUz7Lv=6D$7mvj|eOxmps1#1%O97T(J@&Y5INA{6YB|4jkrKnk)Rk1oLjoa~qGY90#2evG z^;>1qCP-^l{Onaw*JlOl%Hp`2qgLT-B2(m7+tB}=R)dT#zum74{FLhMZO)VJ=AVZn zE<#U*-@n@}No>>T?1VOK7*aD5-&}H5HUzg?^a5>P@Z25?>_F=`U^ZO8IcS_W;B-Mt z9jrm)HtTs?6->NNp6$FdGIpO`A`dULR?DrnK0ZyGZ=YVg6xnA((#oBTz;LRWy5dLY z;{Kev;b2+g9JD}U9Iz{S7}Yow;!+1R7#6ON-enHzBqP8R!)QPL#J4*;5aktrbSF-3 zMi~=vGsqO#IN^fEVzPV)>6AV|reQids&^yDh;#@tP>-RA@0iN2&Y3bo!7qo{8Fi?5Z8-mNRBn^@xgq+TXFy zT$>rS))3<|@ng|HkvtdrC%?$UK5P>8A2H#PbcNs)QjUYOTJVIdl<%HA<{(tjoRKB8#c>RqeHU!mU3 z6tR%S^(2n8#S~v!6=ma)$`c*zAe|ODVWifFj8X*Vm&Z$y6o~Dpqg!ji+rs2r4xd~o z&n7Gpj%v{8)t`M=}~mGiq!xsA?84e{qy~a+D(7 zoL>4oEN#SJK1FWX^u=*x)Bs6ecub^*ZwQh3f*O0JJUV-u(oZ&oqHw7IM7Ao}elyT+ zhdVtN4+e#d#LZF1D+HJuHy}_$7hmozr%45N8!7Lt;5dLseOQ=8msa_}Q@fy=tDcN7 z?0g9bKG1HPf_E+Ky?u8MsE8fiiqd$dIRP?@JqroKE}u9zq$*^Sh4%0iP$N{=aM3Hy zG;n1&Ff0vtJ{vo#%UktC4S!SnI+FDGqaM)4Fo(~7@>_9Nh*!D>|I*qpTZXAg~p&Q1usXWh5c&X zJQG0^GI~yO;ryolB&>z_I`x%I8+O_ah%+8xIGKh(LBa{Clo+F1zK6=g-15~$wu)j2 zs-ecr+#=!qvPe*?Pf4=Kf)7{j5?!8Ps9<^htO5d0UN1s41@eNutbk6`19glR70H&= zWs1nFW#N6BJFivouLp!CaGjPBSGWX9qk6VkZfL$zQvbtFvjwnrTgfH(8#kjf)uYLc zE$uUsa445lYAmsAg<}t?)^O6J>#h-x5#NM|vAFNZ@7IQ%`^si-cke+8f+?yAV0-mj z$M=WZqD|EAh-{3u>S`~7)k?LG+b*BQuISG(>2ngLE8nNq38KJYe@ZgUrKLCE^0tv7+e*=2?$s)Nw-Q$YDFxC?9;G$Tbh9XHmmmd z&i!#Ic*pg)H^vPwSX)jfixOOu)xjpeD{;t9+|3b*_vSJI2@e-$ zsm=p}MT#}IYvo6VMix5AC?1|m_m_r2Ph&V_#HAep#Vx^zrOY8owcFOG5E52=h-t}U z=_n(lo6Ch>wb}LQsV zhhwj^YE(T8`4PU>7~msE+i_l(EM|&vUJpM}zK+9Q-Y$pGCR|xmaj#rjS~3557xh9Z z>M@QPJ@-u#Cii=WHKFsxHt>TvKf>|P!_(2fDkzijIax>NqSC|A<|?s+S+GLYHE_W& z8D?$7UOUU!0qS*Y)K;J7se1pl$mHvxC~T^nHveyWmJyb`EACl~ad(>c)M~!4;ivol zHme!@<8!g$cfEo8$+0h923%XKni+I3Xsse9A15iO62%13g(-4Suzh{#JRo7Ucwia|RgR zre}y3C(bgn0GGpH3rgWlo8;W)*|lyk+-laY1lp+z!0~o}`>#J|F2koq`0$YId7EE1gmZly60d#6-nvd#(xm#a|u^;^;hz)2*%z&bi}CB^UKk*fjFS4qvK#{q!1w{qt-eT_z; zgtrXWC;caWH>S<9HWBYrIF&+cmUBDS^PYMMPnW>}@jgYC(M-WWAxUN7HR0H0umq;7 zOKZLkK_vK21A)Y)Q+XDa-PM0PBsQ?)>p47=O`kaDCS$?b!*_EKs2&$)K71L7Oj6`5 zL&+9&kXpn3alRB9MCmW%%Bv2;sUO!B7N0b`1F47pg8hd7jLGEoYkxSs~R&Sws ziqb#7Re-_6n=`L^W6_CAz1zL%Flr)n+m5Dz=?N6p|8Rtz&Z!Yd%Z|6~LH?K2GiWY- z(|o;qXR1S!%yX}=kBC9H%Xq=HBQkvs1`o|q`fe$G5~|a!715%=kotqZF9Sy;0gJf^ zis(rk|7PZGdyMD?Z|e2x!O8dO5e|oVsM(GchW(GiODsl3?9YGUza1VMQBXDO1g1Rf zYvroZFe=zi)Nrh$y9>2dJeQ!^L%c+rs>&tNp|6_p>SMhL_{SH$GLHY5NelqR` zk}p`M9&RH=-Of|qxiF3+*4>14ZEH&+`8c#jQNY6JiUaVB8(m^NglJzz8BG(*`X<*xDJyH zMQ7Lp-DRe@iC2r3s?e0XEG;eiN5gPViMgOkGPLpf+CL>4@*Wmjc`$4%#Dnl&nTO{~ ztS_W=%ufif9>E;gE8NvpVFC*Sl3_eW9%0DU`h)?2QZO)OqfMmd0OwzQu=SJjSW}3g zk)uKi_0S+iq^sz8Y)nw;)p^LEI}qryGe`P$^olQd83uw%b!Ha%V1H3% zWMjwRCb~($(vr#AE7MG{+Q#HP0thto<NM|UC#4SFE^sB@o z;1XhBVtP5)a=QOqZ4JBO{koP3YU2;$iA+=O&@Coiu^j`aOumvik%(nk4#DGugc4?~ zlB#TQ4k~DH1vB`;r?{vSoK~5rCM$dVsYY`>ZENq`JqQ9WaWmfGnVDT|tB`VtD{wztPE z@CBrbwh)V-uD`F(Oc*P-35!asKrXiKcJulHelobU-!nC-A3Zs`_s(+$I+EMQ2b~m# zqQl_{parB-#ZFWziuO;VS@#a@%4Le@RR?M%Y-Se$7zA)1awcL9Gq)dxLJK($eL-GH zVhGhg_%0{(ot093q2h~e*OYcxIHBAaA)JAw+QYv(b8rb=Z9-$CYiGTddql!^#EIJW zu4h>3k_phhPHp?vbq0Q#3fN+Q0e=w4H+N{%m*^N$@p68wPfxKjG@bPiGWN}ewLgRUi7g6Ijg=di!q52lyTmJrTpVZOR*p7Kl!*rbVAfpde3Lm(TC+w z`I(*aq2j)qDXIYX+DXB0fkFa^%M(5|>02Vcj7A0HzkiJAI1Ocd-3QFS^O9srkPb&}KHEFdVWPMEfWHBgRc?nJqfk^t=|BSC42yr_$AP;U$2&KQ|4X?f zu7YOo=os9SAKlAJ4RB^%Z@-m9BKGs()Y#!o-XH?qqmo=VcuwBUJSt#P5H~uW{j64M zY-mCY$*~Sn>Zx2+nlM}R@81J5H9wkn>22A4+hc{zXgjhirXpQd40t56sbG}q)l@HKnb9*q zP85G{r@Dxwm~h^B_&N+1U$ko0SiA?!E=@i^k2OS>sHwdw*ou`G!ntXyurBW#PnV1hre?9d!|BPue3fv?KYPyso$CWGj!IM9jPQ4tnXou6jBUFlVz^mM5 zw#62OolLOxaYatv>a!EgI5tXY+tz>r!V|hIm()a({M|CnIu(!ICgSrb^Xby8g9EocgF*N zio7I4VvNk15UBH{B^8l^Tm_sJE)N7Pq~1yaHmStGGi|1W{{m94v@%P?xq|7~uRyc( zwc6DE64C^u7aw!2CpB- zSvA+txb{~Y;6>pCXcG=ZY>-p{1R{FRapHdq-hCE)$8~YU)LD02|MT_YT)?wBZ~HfO zkeFa>gyZoduf3yvm?zue_gn1@UU~aFJUY>sXOn+(XhO-|MDn z+3)i^eXA?9tb1yBj>zy8E@9bPyCV>Lrp^RJ&;+~DUKG213qK-7Sp^!b&VCEC`!xo| z_bP_u&oLXmL%O}VG&6yxQ6ZUv*5qm7pr%|Bb+8LfSoNoNsbG9QFY53^h8-RY4V}R4 zID0Zv^a+}#dv-^AjJ=CL4Msz+Jr!xzsDegJg*2lOyZ#)#9r%b*6gJl}o^%g|JQOhv z07Zxc8FvbAA;BP*Y)d#d{lArM0W1`S`Z+8|8x|8yR`086Xtx(Ls6CInb3}(El6*mP z;*gSR9Pa+v#6U$Dau4X>^+{X5@Tv#)1JZ%XOWx7{@^GCM)7wqix$+Xh@=HY}BzF)g zptbO(OfNAp(85?I?mF`rQ*K%la)~K%B(^Z4*^Cp8*JEt<&D>_Y4O2LHIXU;f9Hzfo z^?_<>&9SooA?cR`)6>n^k*gn3?uQLcf@|5r7GN%T?F8_hUJlvH>+@-?i0RhGNtZ*v zEThFPN=*=-CulYP+3&6Ygq2`bq}dxUC5V#$(%8dRNx*)b7b5A|4cKxn_ZsT>CY;w}Vpm^GDnd5yM?0+A-+~kT~D7!jB$)II+ECIS% zMUCtCML5x;I(v!6i^u0J&jK85Ync#DX=C3GpB$FwS)2JHb}>bz;$0Mri%giQ#xfu* zER+1}EmnE>58auM=ZF2Pgdh+266YN=XYRU>_ejlH8IY>`prNH4ZeGJZROV!rJOR0Z zZu1Kh%*Wd-88ZpRc{!$!iKt3Eu_j}cwrOf4*;|;q3h|z{;Y4w<2x+P4Ht?glKS!3r z1T9A0+L^t@su?Y=T1IHO?dtV!Oag+$6*LP}1+0O&PU6(OT;Ey1C=k;KSA|Q&A|%o< zSlKy|UNeXd%%&4PKh*?_!1Gz7GU+f8i8^JFSiA|Ag$`U%f>d$eL^Mc(Q%FLBv&ep! zh%z&?;Kuqw`B-Y44H~0olAW?B!nrlB$%0IxQ*W7B1LO2iDJM6$ zw>bdA44xBbmE<72)YiOkXmvTrp^;k`$)Xa;pDCr>R&WtFW<=&8l6D!+ONUoTZ(6+r z7pz{L<`Dm#|KY>`w+_MmWU_3z;c#wZUg5B3fSr`3l|8+X$>**!alKI-VdQ*3Dp!|^&U6#+ik#Hu>TihS9^=eYXGMy!(frl z{C8^X#K{ovefBvwiqYx4?0Bd`#1@{GVr1kp7Hr2pn}kq&*P+m2Tte3~TdvbS*_Jdd zCf1|Yt^ewi3G#p|`A(UPEdVTe1zq?Ubxe9CHS}kq)Ogl9n3UoXM zV-c#dj+LGid@9P+R5i!hu93RV zMb0vCeTSqlUqAu&gLk%l_INl1yWy*+qO8mrmj9Hfr_2>u7*a32a`$_JiiAi=bh>iYMoXVHgFBbknee&N9)^NRDSE8 zpZdnt#rhr4N*fj;<_Xj+{iWAWA%aCTR#WbX*Wt~%sHaejM8EC~bO9fA{KvaDCj|ky z8R{T2TH7kK&l%3!VwBN@egN-cloADiYBF|#{nU+~v%pOdN7AJ%v^TF3qH2tmr|$QZ zQ{d2ETR&9w8>Qi3e;wD#ZXx5%PFakQP(geFds{RxoH4y}e^wMlmY8Hm2}j#M_NSx{ z0A3TldP)`XP*RCZmBCGmACLmv5@Vk~&MP`TF4bPRs0LOIH!jkPknrf3gb6!EC}SGf z5w)NYsTVO;baM^PCh0QyiNLmlcNMP)K{`?O@w_Cs+&r1SpDFZ?$B)T>|Hek3V4=6U zL40A4;L%ApjHEIX9gBCqz<%2!TF57ohLiGq$BUsS-O5V3Z#~?dRpcN_Gp`aHn!N(g z8kK~hw{vwjWyw#&ci4q!E&_$y z42lE=0DAne2xxTu9$ZYth6oW{62&wE*!rjp#B9-!HEHmqs!01oJdZ0Af#Tqww8uk# zAQYx6yudq926C0|3wHj2MuxWR_+~I{^BnG(4|mp2csM42OXow;p=G3cIVYwWa1Wl5C3h`X6>^eONs6drBeqX4QEm@R* zImOn>6gDaJyf7y*KzxB$kWYtHF04iGgEF@j2Q-AEd)`2)SkiUQtDisbF1M+e*7?k= zrj#0AtxkIwUxaJlY&hzid} zkQMMk8R^}Sx!2T6A`CBfbPP{@>L#>RKnHpD!|T5RQ39^5~pSG#Wdc8)A% z!Os_pa8yaEh_dD=yg?&1wDjxaZJM?7l-v@`ZVL* z@N~Ofl2h%r2BC)z3Q#PyoNQVL#;nI{IS$vGWP|5PHB!OCXr1t{cEUIraf+D(p~lMx zF(+yhAwCQ1LDq94(yN}?T|LZ~E ziOcu;#4n4X>z!TTlgAC~1g#rq{q7?(T9AUsv(S4f908x~=8E z(q|=vhtTWXpli7B@ObKdePDsg(DX7w2$T1o79{QrJfXalekX;r5w}9T5ax4xbg>im zg?i1eo9b$fs+5txp68vr!?URRs%w zS}8H+)nU_L&Ok-d4R-{p3gDPFp-+D!TxDbsNo%5)X$Z-3V83RhkMwftsib@?odd<* z@H67VVp@Bz`oK-6>u5pOi1#rG$Ell=?UnNyK`N=uvF`Y85^h$sMLNUncKWzKZ+SA!SD@h1XT>z(p;reUU&1A+^|PxqV*$+XTPJzESdOSMu$@OJLAdS zBqIYgf4#};R88kN$4>hQprS7Kka~kOgm6ARgTpqV5KD!rq=K%W3qjmz_>IV2LhUye zDbUQ7cjqbk0EGeW%oqtWj7#993Hp0y#VCLHUOzi=7-UW<%NMubyRJA#%97 z|BedJ);+oG)ZkM8v#6BZ&lUOM+AzsZ1$`Pl4^vn_^B8j<3o)JS9q^f&3_a(<3Olf3 zHT)IBO^Jn=VT=M4go6!D*Un6$LV^4+_%w|#Q_(>CSa>o-+2wt2{!oZ22uoOOb90;W z<_IYn)g!cwun-%yHS~l-_X6HIy56Y>>sU#T6^b8?4W1 zE~uAE!A}1N1RVWUfmp*VR4kPw>B`>#oVv9e1Hw421&;4%%}nuzy<&|AVYlcSBL{p~ z;Sc7Zl#+Mh!@^ML_1c3YS$`UW_NFzXfWY}-%>Ga5WN`}Y4Rvdj`jakQsWsQg(WW!) zo4|Lm3Cfn%9r4a@=p_vdNfv(nS)vX83AxT2Jh21&ep(`bT}$-b5b;6U(=1Ts1|MH- zciUhnY=h+R2+~DEvKCN%B&ZV+pF()3AP)n{{U7Y)@jV1ZBGXeRDD*a0u;N7XaPO2P z(`eO~}m@NDFOn-hPx4_X7?)kg*u~WR*Cm1o!T|DJ7pE11H*ste! zrTZc5>mm7DeJ|Z>TUYwjrRLU?u#_;H-+~Dhi|~YFulI?bX{DQDcJV`a?YXQ-KxBB zjywKdq=)&sH55^1u)1qGwBP%<8ZfTStcejeH#G&Nf~&jEkMrRUy7%W<63$IzN+}1O zdn10Q3)%$ch`%?!auKBz)dH5#Ps+}b?FM|Fz@Ra1B?;=S6BJ1;gcnQ$A+wTVY=t#A z7Kf|Bt-ZCh(EI;0(NRKW+xF^CQ|o+ESBMDK`#3*{e@X@=)BnB`dBzXx)yD?6;RZ7X z2=@r=9w*ayAH=LMO&7=x3`qP=@$q%g%P20U#Hn^c+wqXS=gaYoy|$P1=^QD_LJ)X0 z2s4|;&~+-50TzDWXw|j4*{+_AtA43g(1{eriAv&u!6<8FjQ$fC| z{^%X_tw}1P#FdhVUI33K@)R`N0C}eVaW(Xmf;a+}aW z>w2eQv2?xKD#_HLWEtG~rnzIcVTYRxHA{M>)U{zpo_^{$cRN&ARBI>LVOswL*9Y@u z9f(%tGRlo3XPP!>cQxxGOUi?B-m;%g?)m8FY7g{#Q1nKTJsb;PNx91%@~KKPBA{iF zTWi>rJn)y2Lv_&FYYR*35K>!MMlnG+51Mk2$0Ab`ssh!EWG+GtmXAWNKsx@B3?D#m zYp{1*)voNO8?nH+qOe2S z$tFKHX)FqaMSmkb7e9BdeQbaPl0M7+$15ySvIJv;}a%Ty7~hGfxSqcc03+5@=Qao1*6h|ZZ10>Ia8Fy8pBvaj=<0Y@!QMs_Ku|Y&brE-=1>i>BGaQVH1>#}uI znlZK!%Ye2eBKahl56T`f|2bD~EM^R>g!&fF*^n=5#G&5&$Dr9?yHkXZj(Gmbd*$a$ zYkUHx*gPm$#f6b+qql;I)2hSMJynev_0sSkq*abf8Ec9CLhWQMIA*3lNoc)j$KO$U%-YJn;@`}~=goZXR^IU(zLdM8nA>3i6~BIsH_(ytdn z9m2GdPbjVhY8+usNjp}YI(zb#kRS{q;JRi;MtTKnWaPJiDYZ@j#j z^*rKWT2Wy~JqZygR6Ze)2}%bzWNRoS4hI+;D~ED-)ZF0zOq3ig?@1;DO)*KI2Z00Y zb7Ra9pZo)Th={DRQ_I7(Rcno#ZGHq*s2&Q{pxEq-f#Op0>H|*!`yn_w^Q@0u?ibWc zeLlowudgmccZ+vZd;lelLevQ`qt-G@EU>$)^uEs)zhAEO54OwC?@A3mKBhR(aN90& zG%Rj*ET0(yV(2u)c#T)-DR`PqE1z4#o4^03u=T{bDvo@-^!#(+Z|33w*ZDx1;WcI9 ziVAa|U;dl7Z6*KxQAaO|F@x5o>umu7;sC-x@P$TMrosBv#kxlm=F{!Hs`qu%b?{8O z=l9jkkmI%au-@T#pM5<`_tm{2>v?oD~}uQvNC|OvvcFX0o*>{BR?) zu=1sWUwQQLTvC0uAOnS{_@i<)P9kw1-G>(ASnP`$Mw@kH5Xla>10U?UjkM7VBaWA+ z$!u+{8CgQ$>>=faEh{G!jPxT`${)0avK-k5J?fS)S8F zZ`|LGa``C0Gk8*01@P}q=XCvRVI&P>5aVY|t*vdD!%>3WCapz^NrP;2l*Kr49`B#= zuZZ*0NT)~JeLM$1j$VVeBp}^S=D73C!TPn5pbD`M!96x-6Q~8xg9Rz$*zk9PHQPz2 zDB%x7if~$(VKc8QqI@w3n{VdwyhiB0%rp4I)WxBMaK);wwfy;1t6q8i7Y)YtH#Bo@ z3j|pwdSfv5Iy6TpuOgE8hWaavXJKIbsO1WOVCcn_mfPDr#g;Pmpc~I^5AzSO(uRvm zG~t~6cyaa=1q=A+UQ9%YpZJbO;itxFD?5Rz<3?5w*g@AW;|E;*TA51HX+A0`y!@Po zB?nlx&uK~70~T#?u>hn|tzdr-UB#J^U0MqUl0O|nYIC8v+n##P(@IS>A~xpsOyGS& z==^YTFXI$h)T>bn(OTj~Xx;RSUF zL}(5`1+57y3Kskr309k4WA5KivBVGzg6!G#_OD#pSLBmXp>|hO9*pdqH?3|-PXEmO zq6A)Uyx}U3<{YNC3V^A0I)?w=qaLpb>^t3BsuJKHG>m@H6f~vOGdVAS{od?sp7Q30 zf3lz(C`bV{9vxSU$zwvVV%3ybRB1I-G2+Zl)!xCX-AeX;|BL@u)!LcW?s8_V(U9A! zm3dR&%L|f@thkJF9GG|(+%@8bX{E^pB4K288=DVpuUw>}Z>~b5d6|Za(GTi;$q>(=X_$Lt5^S^~C6INVG(mHjFqTsiK%v9_l>V}RyQE4Y2t6veh@{StR|A2*drF(`QJdZj&F@OdE=WFrr`#n; zmB)ws@Se7Al-uWfVcpaAaXg1V&SjYkJ$|Ygj|gW^^-q@l#DfwYsaaJ>W(be&kDp-f zq{Lu_o)g(`@7w3h!t8bMWd&Lh-3FsFqBe@6X5!t4i|(k_b^1?Zq9&o1y5w{fCB}co z(#r(GpEr%IunG^!#(lzH`ef9&%-(V=bLWd?k_Ad(e-}5m1oA53W_C6WoeCzLRqOdA zEnvpB21xK_ht-jsT&zZ?m)WHN?^4X2E?HKe2qDmj1&|{YLK_k=g$|r6@s_ecQt$nu z6Zd~}M$PTe!~YPnc}cV=U;e~n5jl3^Jr&>QOoYOhI7pC zbZ8=tyVI=i&hZrLt)@CeGF2=Vs3v!!j@j6nE3Yk9@?~mct3vGE%hWABOI%>T8j0a2C)yL9Fd()w8s%^k*A9u*-xvU&+bg_IwsPu%@-7(ca*`I zp20D{JH3O4OwO}frwbB7RH?}bp7i$TQ)L61FnJ`gH9?86FAXj7A!5Y;p*JyXCUkXrm9JK&vH`4MGA_@R$G>3So@MJQ#7U_ z0~4%ZMKTNheq_31f~rf4Ak3H+l`ol-j?ql!WD!M#4AP+r!&Nc-%iP#TW!Q2ya9foF z`etF}Ot>*}m~thHE{HX$!i<;2kM+?2UBR>X+r|E=CGr%-F3L#@8 zwAW>Sw*G~nSMsp;sPSa!!!-Dd)Da(PFQL;CW(90H;a$IO{|I4~4ttYTR#SWqOcrRu;+3@8f3?E!2Vv^s$r{X zrd5ez($o&<#7&m;snT=&tUuWLfk>TQ94Fgz(8He9cGh`P(I1UzmO`V24|m> zETvW;-JWx7?d(yK17eehn;9{wf2X|rBbtAkMjZ{v@b4bVMk+9Zk)&d(3-n+Wt+_3*^p5JJrg4ejt) z$bkF*xZG^n`Yq9NwWP)wvm?Q+inZy{F=na|i8e|$-igRW2|Exgp0nG-639Oz{0QFG zQ(ML9{<;7lHU70Du}k6lDIMyv;%o~JSGy%t1?|gAw@#p*)tdr_jfe*#CB<6 zYEkFMenTfA$)<7K%FX71TF;$-#fDFkVt_*3RJR7kp*636L&-FsBCyS6F-8a3Q&f@h z(sZ{^bEK0}ve9_;Y;LahuT;{!zmY0y0B?!ONY+y9wSMo_7G#o20D51<8kV)A#s3G~ zKqJ3%hSZh}K_crwLWi!bXy(9=FGky50QLbdaVYf8MSu}z6@>;Af(2zQiYts_x_ z##lTp(q`cTwN9aE@2^{%1BY7i-`u*)YkA2(e00Q{#c4ogPbvQ5wH>5{|NYSs$Ft%c z`_X$fB_)ZLG;Rf})-5R!1Ze~P1Pk~D7v{n)9J&j;us;x6tm4n}3ctO>(-Y2|U-KZo zy+>lz_C*=DXknt~pHo=$E>aZ^zh1L;)-h|H!}UV+<+g3~{Yf1p4V$K?&%`ijMZoYKeCS=K}Bv zp?EN!(s;>@oef^Tc8h)}QN$pbF3Vi4vR$ zwS@bJ2aIRqwqD88G!5v8rmh0Lce~(KEhNS_mc8Q_MZsWegYCh94`07dQC94ZMikZuG77B_KHQhGDA?$A+1lta zD=IDzJ2)Z^=X0zLnqK>dhs0XZ7(-Jw^m;wAD559>%Vv~nnx?_n8hnJb9%>8Tp}j%G z8XLJFSlyC1$Ry?Q!7;O{q-+B9cdYcP_nb~LUD$gtiEAIhb248-v^86P34L55mjx|S zK(zfa{21q!&}nNeX>Wr;ugk2szjPjhww+jt!UwIoHL-0OM9hVfVOuGK%nav@MRAP& zp+83RFMo;p7rp?_VP1a|b^QkF+7--bg#P5mN&k~SMLgJ~dE+hefAW7}cOQ`b=^sZV zF+97C*q<=}2mcuJ8(%@x4*zHV!{s5Gs^H|_SGauZqcFchemKU~HBBz@)}ma+M_;_j zjjaKZmVB4%^4oVteCz&EXnXs-dhuis%@gXh-t|5yi6mrXmdKkSbFIuv+{&#Su>l@*ry_nG-cf|&Ml4iTpc7F zyNb;XMLBV(64FRg^kL9MMA8MFJX*?p5t8^y2wuIj&q;20_4T)zlnu{rXsWux3&GBy z!|teBN?lmz0*sH4xz%kA;dDAhQ=^q&R#rqx5=ELUP3UI{2-VIHmTsCXvq!>IRP%~! zmsKk(U~o>5mj%XHtd|@f9WkHIky6kf_PKRwi(xNiFzC_LHB!L5XhUx~v@iIU-YTS%BF;X|;^hYb93dm)zNu5m3% z(g)Zs9J&j;uzyWkU)(Ps5vN}J5K~*sg!U|4&pSK;#saH)g>Sc=r~h`U8K>Zu@9gL; z7go#fI*gMotjU8=?RkgSb}iz$*tb0cM+#B+_B-c5E<>$e(5886hEtrq7pI{^tM;y+ zOcSE|Yx`Wp(L=k%@*cZZG<@mHo!z%zuy87r48ph4JB0MC>(4#a7N;!*!n(k)=Os9c zP>QlP)OpSBqa*G-I6!zunk2k@^C}yigp=8fXLdFzt>a{#(@irZo?-Xv%7f7n zR|W$%dp+vfF!2qYEa7NUGMUcU>}G8CI-C@9VjWS|4S8OHl;m}VREmvWhb+>RjU&^N zJI52cX^e{`C$kwz(qUc~T;3ki$x^=gV2{g#0e!7_{oyWAq|hoQHx_ zXKo-pyz$@x%)JufXoR&J3U0G3&O zDdlqUD5NC0dY$T9Z!QBnemr6NxBnh~K1XyihJXDpQU3au$$#MsbpFgwkp77uA^%7J zC((yqCj0k(n91M$do*vn{nTqDYs1Fjl$V_Afe`v@)?;<8nI;67oRkc^n!ba&OnB|iKCfJf z0KsHBVldcXa8saELTo4Gt|2l7jdjfC1wCaDBIC~PA+%kS;BB{ z_W^N|AcP=|VyBnHfTIO8TfLtNYP1SlvJ&=63`=!4O#10%YYV;V%aw9970NrwbW%xz5~c8 zX<_|ve`kY#as9_}-qU>7iu`PS!CwINVxNnEMYp&7d}=Hw*76;_gkPB`Mr#3pAKMx- zH%|^+AIcJbY}etT)9-J@HGkoix6W)M*0z0r zwoYpD1T2X2;fH6la{)c}bVn{^%|gw|S+v8lzS_G`Ja0pxc&J^mEtF%q_40fF)knUy zZ1ehxHU!Sy>2|ly(z-LZ?kEaYlO5uGN6z`v`!r7Cy;Bc(U+8LiYdNkOes^AT!yB#~ zk9qWX!mMyy?{6W~m>XBN8I9)1DC1}v>J*jKbo(7LWsxf8&4+ie-g9gF5?dRaYz#NJ zckdCMt^!l@VE-}4(_^|(L@!FXG}z$U&xwTR@aO<9Jt|H)9?v)#jk$U43fmhSWa1v>bR3vLz2{^) z;rit(sIH=^q1W4lT(Ey~fH#I%CrC6=6cME{k<1VX5>+C-1rZZx2}v9T!)MciZ$!I*d{1^x5aW-yY2isDiIi#?;8{Ep&N)!(iP1DTfh*|Z zVlnr`zNO&xJ9o){>ouOgah1<}-)A98`NlWDfs~TEm{HbEo57L@t=j1SO89tv95vQb zTYD<-g9r?p-kc_~a)(FD&rg2xr*NZDc$h!@0{Z54M4JK0-}(y0Kl>RtXRKNt!h7V2 z#NC$^c~0XZgbUC%SI(%7WIUNcCh0_7?mV9H;9yD?%TQk`f=;QHnpA9)Y`czEtgr+=)5VdwV>507*D1MQ*izA2Csf|k9$Wsm$tUJJV?ol5*Kwy$}xJ> zFdCiEi8UvALF8*XSrVY+z|6E559fG%Tu@5Mk#p<}W4h0#Jet=;$`UbFw7mjkkB%81^U| zht?rudv&8nKTGg&+Fl<|$lpeQXV6Lb<1ZMxQOe63DfbU&)W&iDxZt%%Gp3CtS&*8v zkNl#?X&nL=u?}^OMHNVYvN$5ulG+GL?};KEP!dw&rN&1c^1LCb4BZx`ad?!|Ji9}l zcAtLpIEC~-*K+51YhH6xrM!^Fe5r+1ZfMD81{ptpGT}4*&T{)#vV_0&;x*nZ8jfqr zk9IcrOuxhbw0pqoMa2*FdweuY8JC{dT+T)+_)D8RDBzbzv-ZB&Wp;jEH6$^_I*bKM zYi!*td#Z6E!@7XB2k7Mo8zyn#&|TPt{XUui5_tydvR0h7e0kOhd!MPhb=JpI$6+A@ z@D#F?bK2$G*CNO!6Q%*VzDX;V?NP$xok1&&b7om<6r8fvd8cwEFuE9y&q@odG}_$h z+O;(kE$qg5cdp2KsK;?nym!|5wk|Bh-+gI){->TW38|+z_REFhnIkOklU%)R9XMmy zuN+@|H0CId*~wu4u%f9w!%oK4%i9!nft}AW&NHtZc{Ah6pvzWAQCD?I>FTyJ6l^QEy?pKlku3k zwp_csLzGHvrEqnQ%?qO81}aL3IvH2GUGmw4-NOTJZ1l)|16Ff$`w~@MVVtEu7*fpU z%*v9&JD%U#L?tmrUQr$$F|Qlub&c_wnf4g#>7+5v7$PmXy0gXIy~oV+oJ?y*c?Hf9 zCn>^t%CaI$GRBi5yz|6KOd7?|)@A8P2Vw}4WxrW0;)Mw8npvVSUNE1|nN=mm8m3K6 z-B_YnBRvRfK?ms(?>zoYE!g^g*WWwUH$H%jD4vj^TgpkUJ)_n+8eH76Mq-+S4Q?0>+aO`QBqrih^HKieU{OedQHC{*jm2+}z@=+jscd*I(t~<2~Xy z4t1hcgXc*YqEd>;Ihu9n%eQSlJLYsUsA|xffUtBT4x?jFtg zJO9f+Ag}*Ed1W{rjnG;KG(daNq?Akah4Ky==~ewQVSJQY;jKkz4Rvs?zy7V;{LR~U z&{9w~feE-V2CX%{ejn#xw*NN!RYh5r#Bo9=j>vRGH_;3_9h{2kXi4e~a{+~vZoF~aySb!>v*{m$jgH5ZbCoy#9AO_AOuM?G{-f&;~M83X&RBmsuh`U$7lgg$WYT~ zE#5nlwr=s#ut&e&XJ#YnlPU8e)GQv1=X~X@hkW{DFETxxP#GH_^+E-5A8)z+c*M^U zvz>XWWJ;1GA?qd-NvF$nGCMV`3$bE)Eo)1h_+@No&`B73PnskoNrJ3RJ5Gqe087h8 z+PG!yn|Ho#3*f#znX21$#fLla;ZDNW<^_$1&u-IRovGdlS zyP#fXGTRK>V&wPu_O)MrzV19O81NSRb8F{#cEP@YW-gm41q#248n)+ow&v=L-fM_5 z8_92uXB_wYeCnlVc!dr}$B!tiqu(D;RF?f|O`$Z8iwPT951~_xS6u3+B#|dcB{#1P zdGYZ8nPxn5^D-y%lCR$0qn`u@z<$<6I7dF8b9A^*QP(ui(uot0g3)ZsbUr6Z64ErG z--`jkbe2;)LvB6K-MGdxS1xlfnuKhww;WC940}ByskpnhhZd5JG^1!5vMeJ@5^Py= zG#!&BDfb>e;1H9^t%4tB52{ zh;$st9?F^|N+{}_RA)4GLnjUqYo%47|BfSslBB6dYei4SL{cz_JUwaf#^JSMKAlpO z72Z204sr>J4?@IhktFgVqvF@@qi7e_o>iZ8XFBmBL;XP`MsZ#|6^OiYa6|oV+-5Xt ztBBhAt)NL6vS&&KsN5p5U;7YTqR830c9mjOaeMcWhm(7J>8-bU=@Z|}^Ea;X1E0Ih z&;RVdVm56FOq>lcujNn>#sMXR*tGD?^11cSQasv4zw|uKmwq?g!;KAkfB%1^_*?%g zve%>kAOBg7fATM5zV#;AU;c~eo$dE^jq1j*e>fqMA*-{>O+Y_Q$TH1wB`KYt>;3XN z)Yc(IfK(n%YK(&shJ)Ge^7~pz6rL#7OXw*C(hcD~D$>LPtnk$BXV!x^mMGF88@^`7 zI?4uS#uB%1)}kmF_WR^@L*?4q$GAr)5xA&_FeTo%@IvX4u|{`dv^7}gNJV>0;AqyM z0Tn6kY0EQpg{?g{>M))+RP~%;H|4lCJUlMxrZN3af@#S_ViD@y4#owQF&JZ)ET7#p z1u80I*&Gh(54YLfKV+ld#S4XVmPAV)9*y`X|Ie>cR1LQ-ZHC~2H-JM*$;Y3$#83UZ z*BDLjGF4q3@7ElZlG?{e83Kh2aP|NIAOJ~3K~$-QadfmkHHUZ>b#dWb%f?Yw4OP`p zH4S*T)Jdc*8GCj+Wjm8h^NN$HrPo;Oo}sg~1&nF#Z_|5TvJ+|cYeT*k@OVCn$(`eI za}H6@8M{~2OpW6sos6Am#>iOy+|CyN`0$w78A6?ZaX9AJ$8%mvW3&{!kj4ln*i#VeiX$XK?trmrLw?fSx*c1%~OcCE31S$YByjVCzei#3h24yj}-C-cFz3y1E) zF6=$pd9?UX+wT2E-TkR79l-JNh&SH2y*S$Y%Fuo9_kQ>LIQF|W&A;oh#oA`xhlGW3 zVX&erFWMq%A(!9xtb3TDZPt0Lm*5(#TSPYl z2GRoFdcd}H$Uf8-ta)6Pm2Mw4f{D(Ury zNFk_AjTDlyEEx1V*rws$?gPrYAxct&tr%tz#uyG}V@y#}T8pil7Cs|Lm8R4Yo&Eq< z*W}X)c59PCugBI#9|T<6?orO?eB2+*y?EzvU6 zi3PcYcB87s{>xt`{oMCsA3Vg(ro^B8IJ&H`_jXa=`LVFrzj}%2#%1DAhODq_Xg$S9k z1tCXc9a<@za~Ma+S_)6Mcf`HpP5$pM-RGMRMm#*3GbZ0?Sg(8pc>T=J{5=2spZ}wC8o$aiE#Q{#Z3+2gmhQYQ*p42Tbp1)nD2?1Z0RJ1>iOVgwSUFCv_0ruopUa< zhY88p@Sa^caZ7my>u8K)($sV>Z}GY3o~K_L{(tt~E7-E@yz~35wRSl9o}0RF=K!FQ z0cMGkMwCc}8BwNW*;bh&kIPos<+9738JB0u4-Vy$pCmuHN_LG(8oSE2?9!B^2_BAu z$q`A(6d^_cL?X}tI(OfEPuh6}KdgPi?Y;m3rICsLP>p-<348Ce&)#dT@B9AYD_gf{ z1cG*}gOrNAn$Q%EAW)oI=^$(cE<$0jP7}9kLVqMVdHy)Vy%9n9}#+AUhm zm~OXCUu$etkrp{y`@1+}7^hi%bJ1 zQV@v>Wi3Pj+8B1n6A1hQHciItjS^NGO@bgmMj)(LX2S{JZ(q0;!Pg`li$$O#?&r7h zp8ReM>%{C48e{)taucef~CehGF*xKdMHegh6R?>#5`1Bx&I3DljFX(D4qgjSNO)F2|7 zfnriwG9%DNqjibVzKbV>=e-3=V2r;zDqEvI=9i!EEZ67O;idSSamb0aWkMxrHXE$0 zZJ_iRha@d3?(7WMI~Xv@OH5Vyt`T!<17rPlZG<4GJ4()0NELvPIH3r_fG7;flL10X z*1HXilsF?$QW1v%oyssSYwU`sZ6>Vq>YB!24fK~+xNzj3Yo+DY>2tuW9W&VJ#^+5d}1*AhV8-FL%kA zb5_qcBA#kC354LqEaUa8xCeB6Im@|KRD878=J9reKip5bl9#M>=B1E8z~4Hyf^+bT zx3;)mR6N#f@C#>7-gBM5T~rQ&>HJGQFD?kq-5t}St+mbzk^MRO;r;(L?k9Bj<9@uu zIMks#qVhh=;ly1Jip4+O{I!SPK3_CA0<04}_V^?G{LlYQ4}&=>r<#gzESQ;p<};st zi=&^u<}H%Bw@Xx<#nqFGb3U<}E>QhgYrWzWVkUwjxO;JS`1k2xJ*~nma{lb}F+2B- zJ8E0b-McVe?8@EE&+{sk0_@aC+$VF^9u_=sPSnYx|4yuRv*e3ecW%zM5!M}o*Bv%# z2SlyN1z=rmi8b#&OkZ#7&s)@stdB3Pptn=Le&tQ7ETiA-QR#{-$*HVml%z!3u-uEm zS|(}EN++frdBbEQ95QOcc!JFaCLgA>SLI(jmyF>0A93Zqe7dP54 z$xHIe5E+9qL0xdM7*VSrT99&H+*vGYh*@{=@cs5qiyYrAF_RyuoAC33i~q$~Tl-J> zLYG*u1(DPi8d4yf7dENP{Q07`mZ*vxyM2Xr+-AwM5512MJpC9e*$v*fxl2`f2!A6|9wDL)<1|M}uN2v9MudX^ zF={!?bB~jGCI!9LegwEk;L?K{lR@DrGYOUuhI^|IZfrPT?(QV%3)D?^Ve{xDJtVtJdNiN&x@0e%t}ftSVTQ$ zx#t%z#_h5ql7cgF#I>U2uXQ_o?qJN4QZ%ICp=QiV=q*@&vft+)mOG5@66&cOJU_|! z^zl9y8qut?w-~V$EQOMt%)j@+X2i$(9a_P=@0eNUoaR}P)`=Od1Lj?Ow=mc1w+BPK zC*poWcR%jOpL5uHfj?WcV?AM5#Jm-Qb)MF}1&f(*L8Mh*3gSYHn@q@#>u&?i>t>g8+hux#lgXGsNKoGY zD=SOVO4GI>+Igit99RQVTC9}d4kvAIWoa;f{x8B8@~>_vEK&sizPRhuRN8ZB8eupq z`rOSoohDyi$V1F-t0~V_YeSajY~H>BLXso}xz@yiqEU_!$pIZ9z1n6J(rUzLW9ovC zt;g?q*UwE+?{|+WWrf|mcYzDw(iC$oJ@hw#-9I?`q!@#}b!YZx++E?pkN>-bVFMxS z1+6X+Dnd#{m5*s&91$vsQi_L<4f)Z|gOqv6*4AytNrtwTH*VZwG#=7vHc9gWJB4Ig z$EC9;`N1FjYn*xL2~M6lN3SWleS3#5eg5}2cXFLxtI6)}5Y>oS>Mrrng%*!K^e_*c zdyTI=_X4lJc>^c>{cW9@ouAw(jEJy8;Hr{12nd4!X+3PT)9sPvCHq@LS`9}tG{nka z8;Yzl*lDNAfzgft$Kwy2;otbwk7BH#Qf)4rJ5Tn;mM^YV1^dGhm2pIYq%aT%ie?~L zYARZdfM%_h7$}L*b*WDXva+J6G};*QqGYMR>Mu8=i9 zrge`RCy;{5IZ7jFL@{v~;Hq+N$x&~6e>j<06Ewr={=iy|oE2f`Jo4%FUuerHH&i;V z?;Xo1E!i)8hwf%kQW(d%IHIE@2gY!x)GRd>m2=!EO4>@Y8U{SxY7j5*(iY>HNy1OB zE%EFm=k=_(=lR??K0lc7b89R7cg+okmE~qxF|vm6U1nKR1=8vKe>W!{pmI}ehGWrI z++ii2opO~8fomU(xP z`<-Lng{!kTS3^1$zxUa@ra4*tn#FRp!&h#dQ%F~LGv$2x5d;WI-|;jCTUo9QGA{29 zc;ANL=8e~JLeWwYc~xR6OGhb|dJzpA6KB{<#;i1(^p5peUTV^5hJ;bzr{p|ku#>Q} zz0Wgu_PKU@i{T`#Ef>64i*=6eg9BE29WJbGP?RY`3Hr@0{a%+a2ubpc%6eM>tt%RF zgXMOM;c%ZxlCp7pgD6z2EiWSlZ0_xmXF2UwjB^5MHQnW9PHY_K#+_{@lM%zogdlFx zQi`A~8CMnKO0%)P!cw?Ov(;v0rBBdqv$=i1jlErx%Jck=FZD@^oLp;^5ClPsZYO3n zjM+>oa$QjxJfxxf`o1U}6>W~8Y7Kz(?Z5d)~4wI{Suv3|Qhwr|HHKhG+ zPK=WNCY~Y$ruUzj7K84JLR%_bqAN{hEV8vk|7U)gNoNIzW9{h*i5F~A&arANePevF zrA8b~$6sq5CGd;LcPLZ!XN)*T-ahq~PrF*g<$J&Vd4BxszfI2=E)=Cd&LZ~thOXKcm3=!?Wh7wa=)#$eupsdhW1(;#?C{*MLri3S+pk#q(ovF zC*=$RTWL#PLp@9DcqPj@*NFI${xZM0d%*eHOnf5>`Ki@Dm$RIglZ+3vTC-)!US;?% z-@MJ$dcOm0_=W46=!H4(wXER3y?Tdhd4&P~>CTWBlZ;RZc1y$182(u>m>&jlRBsQJG|r(G77us(;TmJH+91Sz$mRXGANS)O#yd&o{_H!^;0_N= z)=ez~=Akg&3P7ks#=P$|M#B+rUcN@V*W(}ZD&vxfcseJ;OxgNGiu$m(jJz1=-t z_|^-gX~jbioac$BADbUn-6ipDrh5m|MY!bMhT`|eVt4J{^XwL#y{GR`$|qg5+3Jzq zJmz+IZguCB<>;FiW75VXSz55Q zy~mOjR7t_a7@Epa>Ktn=-ByEk9H6wJFd0QQp%=v*?>0EG)*)_0KB=e*s!7V$^&NJT zif`WB=J{K9Fj^y2fQTA|Q8e#z#0`ed@r|1|(OJsrW2GdUgaYSKYn0a2IEGwEV?=R(qq{>Tn2Lncv3GGnf z;+RpEB7%UTtjKbW={TfuIOWhr5(Ykr$C~-IIJL)e4vEA7!Jn>C|6LW@I+UsX{AA6h zn<;(@f}l=34sqH1`=pSE5qOL|Buap)yJf~`q9{g5 zfi{M?PUsMzwPS5%nMcl@;MHrM1L@~J|L^kFh+q2smk5O6jXN1psECyJxoft2oLFCB zr5%&x1(V#=1s{KuN=mG?Bw0bf+azu_={QSKwJ1ju7UxD)clD$YObSh545Or+y<&SP zAfA8rSF@a-Snkme6<6|dme9p@LF8IdvJr)}7JM^H=XfERLw^Bom4`hS$HwwXI`7sc z*7DpWKU99+U-%w3?^4w?z>VVYHs30gxEJH#`Stz`S|KoE+D3ox#{Goue%z0D5%07^ z2Xnua=@=t0I0Cfs)jngVWaqcq!W<-CX;7-JB-e1qJkR;;XFpG%B7W$He}Go2CMh40 zSOIudNuOX{y>gvj`XBxQ#%OxIB{q(&^YcIdvpoOoxB1`x&)?*!Cm-b>{nkI_!yowo z51u*A6t`)U4`{Y-n?I;D+9Lj+c$Im#Eg4qCH}3>^BIkvoj07T#Rgc+*Ze~L&~Zte zbZ}@I62AE-JbXrIfpmhRGIV2!G?MfEKF&Bw>sXCjXse0)4b~ctzzN2A#W)$WGa9kp z>u~beIwy{|=ygKxQCxbC$M(&AM!WlbdtC9tjay#cz0so4>C)+T>GnD^N$AmNf&;R~ zaf~r+-MYpwO*pr4f>xu&crf7j+A(fzZ_;kJ34(}0u5r5L%*pe<U&6ojOiV|#ZG zAtc?VPv&-Z_k4GBI3SE8Dy=~(Y*jIyOb}tn(&{lpkyXOdTX&C5 zq4fJ+6c`O^8f3Moe*~M*hFt$S*v|?uVCvc`<8iZyzB9-A2x?TtaFXMWEn~tijkrOx-Sr$?Q(^N7t>q22 z!wpKc&!o)Q1@@O4=uUxlF=4O+KsyXL`RHj{?IrFs&+%%}XVA554pREM%~r9>LHALz z$XiMj@e*=lm6ntg-8M2vxF!OWQRFM@taQ8ROpp%;1X9oq6*7pR(gbBm7%A+=Lv$}W zWE|t7n5c6=u+&2u&F#G%Zd@LsIw#?T7yOv^&Q?~teDLvg8lfV5^#&n6E<)zn~D7ZG#VjA;R{h&VTf(UI4#jyBb`2^3Olvosf=ZqI5zhrtq_V* zP#H0st45XM%0!bG$1~%EKOQGcjOA)x(3X;|N^>nK`45(l@qCeTJJT5J=m#E0Fw&Mt z_%3yB9Sja&!FJ&82%^Z_X5{(3&xIjUDaxXlLt2UgYL#Lf1h2n_G6p3DrSof~cQ#Cl zn(fX^h4j4|_Y=DNaX;Qg{CRfhri0L1=X29k0lxI5ud%+?;nBw)MJh>}CVc5jf5M&H zyZpo_e{`PAy-j$svB4O_n{Qs>{P~mo{lEX8QWgdO$FKim#*+yL!!iHOfAycyY&Q7j zb1(DCE0_6+zwvQ?@Z;Z4GEVrL|IWXDXp4u3+_xS~51qH8&Y?4h#7>TS?@@0Phq!$H zMQ--K`sY+DY~i^DkMlW3f5snpci-{!`?Gg_k=ALHRrjpkrw}n=HAUsQkAZUNvLqCi zpcQd;sf|^Vy@M336O`%@MFF-b>32KCfuPs!a`9A~#_}>&Yh0F*B_+GtLv|+xZ%#BX zZ6BbeB8VESZESGv%qg0Uh*4FNB^gEqbbD=flAP@er&v3+#--O*x%JBH>>M0$e0hba z)gYOqgd(8dY7w_OB%=vS$CkOXy#*K=ag%nZgK&;XT8kwJ$z(kCe7i=2yhs2DY9A&Q z$5aknRe2+NA+b0*%{BxIrvx$#X_-X1U$u z_RgKUJFF2x5;%!cfqyR$MIoWHOfrkkD*`EKN1CcIZ0+uIYio;9sR`mHxwYiOAysJ+ z%G9C~3WUV^+!W~?CNP9}#YHtOQ;Ws#{oUG`eD{VG(+-?JMN`P7L7=q(<$b7ZU9brx zGwu@jb6FZo;Jn3%u7xe8#hD`juZJR(@NYd8v6nhNd%42Z5U!tkEM&DM`L(b5&Vv;2 zkuw1oS0takT(O({RZQvTk(DLQD5BYHkk=}W%}8uLaPxXOIMGd zhE8bEdJkbM`fi_-Ykihm4c3;H7!D_lc2kPUgw3r@9)55x>?54U1uZU~J2Im$I{$uxko&xxzyC2+G`2Flts3PYBF2lNl zgO#T+ANnsaz?9jzn1bx#XRN!m7b9S~{(T!b`PBTm-ooqNPO~usJ~RDey*~ajq4>jJ zrqa*h8a+Q|EKmuob#+l!Vna2b`-JEENhyh=m>{SSVxA~Xle*Nz~P5Q!y!2MgPl2V^Bemm-`F1k@Ee=k{07_e4-o`jwYM&;mf&d{NdiTIs3*=T z+C?y>G6+II1R=&aoU>?ZV7~(;0|HO9r*aM#H;I1rm$7fYLRMD3$cIP51kt@|oUsli z#Sv?S`QHpye4PLQAOJ~3K~zRt&((ep$Nhxve%z0D5$~o$cVzr2-@CTf=QIEO_ehe2 zM;^Y!*S_`*u3x>)kN^0`IrA>1fy<)e&W%0JpFM*B;wa?PpZ+PrD5TSUjjwpMe+POvi?k<{|tCr+K<4jz+9lZLKgaQc`1ZLSm|tW}}HxA*C)sG|01*G!5ymw%HjTkd&~| zUn0vg_O@=bKb|nEG^uP*7)@m=vZ|axXH8|u%8F8JEP)qBu%L{<1#|HX=iJP)0I_&g z&99;NY#tthOajYzV$~L$jYb*4ly}yA*P|ulVg3O7=5HClb8>q@o*tCs~`n8qtd5dma;U zi`9)Y?B2S@`4z>vb7!y)n*9|-m67X=wbcvQhrSQn?s4(MKf**9;OPw}r%YAl-6D)#KTdb*2NRW!WD9Fo-Nl{W6%YmM}eRlLN z!&yg`CVx4Uh<6#Ka}=gVfk@@&Hs_NYG38*_+;3e&6syWmR=PIP7N{wwbb&bLgq$fo z211~%!#YDzfRu*jc)&_JVlyecvhZREyqyI8y$)YwbQc~VIM_uGhW>R*;gmxC(dV(R zyiAx)5I7R;kLUPGpIB66lZ1_xKCZLI?K_)TU9e{%DKi>Ko_zc%mY0{gbL%Fz%aZIr z{yY9Nyo>W_kr-oU9ewMjiqdl=(@{Ly_fp(X=~|?ny7?pZ+G7?o z5f9+WCmupc$?yK|=Xm9%*D+f2qd)e;oWF49T^~UZ2CS_2c>Rqxsj3oP8UFB#f5H<_ zUg8&j=^yal|JQ$yAA0or_~tjC<12siJU9>aoI*&a3(`MrEzCN66>q_kU@@6n3^x-< zakxWJ(y9Mkea@_{l`iXNm15f6^c^w3u)xlh*gR*#c$G3Mj}&3C&YO$dIuSF@dFYn_ zTRWW!{H~`g=E0X+hcCO`XT{aH} zSZg@7w#J?PebQ2Mc4LDioiG>GOQCd@!CAyZa@3vdyRYheT!se95Cyzo%dd(IFsLBc2X->2{1i~;%QUuV5 z8b}9G7!rj6)(W4{OF>#xbhTf^l#@Lsl5Dyoc%qn>Bj&*rD;U z{=bwGh4bI9^97rr<{0?`+`ad;vGDC}!=uMSPWL7Inc?Ae1quq|`2O<&pLwOC9l=^# z^5(u_nEPXl13254bYscZ*swKmGx_*q9YHS^+!|TVEJ;Eo*qT^&6URzRaH1N@s=--8?N6Kr0s z&F2LaCCmLT?>}{xDjoC2&Q)%_c7w4N2o=zZ8;B&qCMmMnAkNPcW&%0hCN2Xq=0(!G z<8BBxPiO-vS`l2k2hP!Z;3R6P%lnp2BBZ37t?=glOSIm+!sO9Qv>!al*+8*^BN9-? zmJ3flK=Xxfb8R%{L!CAkPgtaL>`umnam-S0iP7zw3`ZlRf*=YBjbypoAqYZ}G$pA@ zs!|~YwBiO*DWp=^5`?vAtx;fkhFpzfTc)qtRr}KQ}d6$>95P(-CbU|^&Bb)kU|p1 z5v{twFqVq2(7f2}xrR2o1V{}_hO5@0@ zlFUK3*`QSvOmszb^*YsaU-k3u>2tV;9|yeeo`YI|&9TyoCgJca5c<= zTH$ap+DK+R0l{B2-`s7>aNyp+D{!y;>5 z3^UI+Sm4x6Raj@gyV>9Vr+>)ei$v40E(!x|SuyS2NnvN(Qc}+D=V9(RNF5s;$x=gb zW=ZnumN#?$#HD~v1nX@{Cm*4vUN zHWdHw)r#kCR$N?_eDaZqiz}Y1_myiEKXxhLXP=Du`ZdFT=J?R*fbTmO@cXY-yt!xj z@EOI=y+7i{z+i!M%ZjbB<%?G;k^(+@HsB{8iTK)8!yEgS>o=}*`Gr6568%CDblSv| z0|tBBNELAM{6jo+@gnVR4=EMH(Ga2cQOzdSSXx4|wX?(4mDgeIAtLQ?<%H4j0HcQt zcW=$$f|!apYy;q&K}ttamINyBP}w@EPD@K!Zw z1I0iaKC#?oDx4^RN1F|Pa=A}C5PWVh;+b*6Cs&pTgy3QjQ%TF`wzj!dYQ{-{ZTRm= zTiBdf9zUU2Z3&KdC2<7DyIytjcw2Dw!0`BnGJN%#;nVMn7!{6_%K~Qvzx{m4UgU4QJ~PqpeyYka}HO#TWVv#d0#yx;BI@5v|8-w36B)e zmUCxLa{kl_uHL%C{_udb&^YVpbvrC~J51+dpB%Ya0mV6wCI|!HQ4gfxV3d%KM+}pK zwD8HCH$|^#X+x-#w~Gh@;y9Z1;HTl|^TQ|wA;c`{T}kq736;K0KVKfpr| zov-ow(x)@R%hf;q^kd)g_}*zaJTFcx9Nmo5Hs!9}H8?x#W}Q0UZS%>blx_xHl}8+* zn@vh}^6BPN%hA)s^cd@eZn1(m+t$7N@Wk!I2N0AMvdqz=0nQktvsgRl0Sv5Ql$LB~ zIk8gojiD+F_R@l+$mf+=W7*F%2a}we`vXjoP#QyO{C-VU`O143`A(vgG?!L5d+7ql z)|Q9`Op=VRZSNqZWV}71(`^$z{2;C!@q}<}NN7eetDO!9R}XmcnQ!sPN{{zlIDwfY zZ0_$<3Q0F^0~NZ`tQ|kWcrs!y-zHFs%2*nm4ry7Sf)K4O`v(U&psFggwYB1-S1r{> z1zK0MdR>mM_GrZow)YR%-5)Xp#Z7|H8;r}hxT5ZUY$}j)bUEa`5!M={@Lr8x7(&z?w`uY^Jr;6lU9r*> z3{uDL#IfEMeE3wr7cOhQb=wzBULM=hUatrxN;6uK;Fv|7m45+*qS6<~uVR5|N!T%i?8no3}u zpb`fG0Azb6_XgtKiM1biIp}V>9+VSA-T0bL?55utCJjz5q#uKKv}^zt{FajI-nZ~ zHWN!L5`52@kQ)bvD+iXJeJbMC(6ZK+oasw`{i`KEa6aG%&WHSqZ-{#jBM+vG;?|Y@xk?R{T46~TfIVyAChwgsdkNfdgA?iW;FKvN0t!DyKC?9ENYu~m{M&hT@MfCy-W5$oL^ zTQ|3P<_pj8P(S8;D?$pmou=5RfvXIS5GrEgMu*+OZv7nvNEs3-&kkE%Ugp;3CS_5g zjq#GzW$B^1wX&PF7Go{_UXM|lkYyR6RE$l+dT)u9c9-$~0k0kGArSQYeb(34Sr*>( zH_i(BWyS7rAF7IRmJv8dtJ%T=yW_ngtb~{DI`t6RbC{;Y2(>(N@ATuIR{Fj5d^{)CPWRt zlCClfn7dEgiTyAVs@ed5jH8af>&SsCe3C{tWprr2qeyGO08){5rt7? zMMV$B?m@xmXORk-pC6+vE1cc9C0xY3F{>PdS3EIQE(}b zdDvNc0i5f#IauH=406Zz*z)v=fH(IIukIK=^;AeJRDAuKp%V)_O}PBJnBL2A zKcTxH_v2lLkVjX)-r1Xvc*xCReO~$k zjiD$jlFHBuJ#?l5@^L~^R*bZzm?n@y5U7yK6zFV%)fLo+xBVyp=jgW@JiOdupbf9@ zSW2y#q+^mGz*&oOhGkpP4+Jl!B`IJpuJiGJi;lINj1*DOA=d>CvhQkp`3gSkCEX!%P znp9hj0h262SW6IwC?dAUDd$&uEXN@-5NsX{+1%Wu8t)Sal1vL)%?1-I5n2+e z2%LwK8Er7ep_D)x?e7Jp(7K{1b9Ak`HtRAQi&YZk7X2CLp@oP1DJKxZ)`s`Y$u;kh zsGLBGfIx~l?}=&Gd=XYDW?Dzu2ny|HaNVij-(rW~SWC5Ve+N={A37hQR2EbOI@ zKnWf>E_h?tFe&{0zjfX4k+T7}MviCiRAklsJ3dOd3XX=zyCWW7{Jh|Y-`{zq;-9}# zQs{+_gY|+e-zl2?F6TK#RJAaZZMLtyO0UzwhB2}l6DZ#|Zg)CJC0T8ax$x0{>Cb8L zj~Nv?2uZN@EWumbjJ<7y!1sSp@<(4b+}u?ZS%JMCU2uh;6psDe z@)IWlzI@&C{2jx;_u-iS3CHifS|MeP@u3Mfj zdNcGXm0=2x9F`T^dycC^K^a!F$EW)}*=r)Ksd>DLMkI-YfJspj2|**6&zphrMNJh1 z5JtZ6E1)V1fA-FxP*dL9NQcx5Ho10stY07AD0G;ErbL*8k zzw`BFofny!-s4BV=DnTF!Tp5pe%z10lJUV0KE<#9`v3HfUVAUZ@Xv-Lw+PhRKoS3< zV}AJV&}8bLyZ5-J3%)78PKbMKzH9#Cv>RvV+i|WYdwcMkwTHU`)>$l|kJ#_sf(_FN z>LAS+(AFn^nXX8R3MC~`vw^UJ zFfbHFL1i6Q1_&W&hF*B1s!}>}i<7PYkG(hh)iulSyFP1prgzxmnNL-9RlA4kL7w9Q zjEyWii6tNkLXbojT!4{4+#s$X65PRGzy&u*pb!h;#5g9H$Z>2OtJ~deyT=;3PMsRh zw8wXP-e+197wdV4eTM2PcRLBL{gckwd%wdou4k>^@cn(iDbEQ=oa4^4qA?!ZR6KnB zE;p`?8DG=#&O3`w z2H2)%*;;DT(HcWhWTa_<6oUPo9j0Z43JFP;k?MrK@rcE;VZNBLpAH!fhAgU@;b=_L z)O4of=CzwlXXni3C3|~2NF{M)i4+2341f0a`z%aH@SgV`JV2P1@y;#>JG%^%6z43{ z#gx`sZtU!Ge>!2=M3&7wXbuaF5D7`5Nz#O)vpFY=1!YyySxeW(-alXLgVzLlr@>h| z)8U@gD`_3eWyO)4fyxMBhio8FGQwk}6lfuO$%;N$2?60(m5+aU&)2^SRwQ!m9Ba98 zpgkh?2N9nuB_+D=T@+G~r%7a{6xV1Gs4&d!mL}3=*ykzxOu$qMzp8!f<~qwB%O)k(W@XlOA|5$_o@i@b3SmdHT?B`O^fHKv)sxOOo)uaN1f$S z>v%p(8A|m8)ZM|Fx?c!EwiW+c4^;=J%fPh1H(!3o^6&j{%DWTKejzxX1wMC8G0p|4 z5FD3*mku`MJKk@Q9(L>ZZ`^l$>WbvFk`xI%n1S6|r;G>FU*@RQ)>n^(#Q9Ai=KAsL zh%A>rME2;Wi`AyinE09!eGu25cZftO&Uvb3$y6J9J{6pYMH6@4Izi{)ul&$8{?lLi z3I6^czQ+6Ka}uQqA&^NZ1{t-vh@LSv9-}0-If&rUQXrDZfSTxpa@iAh`G55Rd7M^O}9xA%*7Nn( z-e%T0zWCBD#@7${>{#Qg7G1XZhc)jlj>)XyGlP`-u$0&n7I&l0h*bO%!cYb80)T9qwUmM{8P!=@4rz^R9~(FCnqk zFdpyFG&MTWq&mT6hD->qjYqscn@6oODQVgYX)X5rocVHz2p%aURb6v19@82_Fdg1I zQk^hJQ$~p-&2oa2EE>b3>4cN{f_8e2Z_0Q-b%GEB7Udb%b%bCeMLS&3eToBYGrDZo zvM6Vyi6lW#hyi!i-{%rykRqIqV^hS6w-@2hE&K78#zLHBqLVDvJ#<=15~WbW zuU`jHLeO-vvXdqWL8>TL40!9ds-Xd`qiZ9jM59QRL&+zd$ypOPDLvQ6f_En#(?ccS zokTXo+Q7XY_E;JCy*q}#`NIiEWx~-S@T0dC9(d)*e{kXc88&MIe|q2XH-9+cufLpd z=fp8e1uq>+e)HSrdv5?}o#D(_@&NV2_mM)7Xo(2EX8}&Bt0jJKm(!)8nKy6^f=CG2 z3xwnxsm~B{BlYsVEXaLCdYcdo*kgo?!qwh;gw}ZHky`h(A8}1VX{>W(N|9#?r^fNs zc}*7re>gAsjg`b@$vy{viX%=2L8_}YETyp7+xd+I2g7+o6c=zL^x zb!tHE0v&N>Q>-dtDngKL55o=tV=OMjDzlObFC;7J9wh`y#tL3-qX5dHjL8X$LhA${ zqP)vqE_vZ_zT;*+>y3I z!4rLfj~CA2NxKZ_F747T?SieQ#itPo{6(4%7b|Cv$6Im27MpDTx{4N1JquohRYoY* zwszji9%s(FXmPRheGnOZV0(#JRSH*|6hid-T^*ojkOV$^Z9tZ#+*gi!rK71j8mTG0 zpsYJ=XOK9uzE^R@PN#r0%g_!?*RZHMa-})QQl|6B^qHgy$VgW$l^~ar!@S`3;T5j# z?V+?H85oi@)xpH_wp64j77&_0vPQi~ox6e3BPz0K~qpBS9 z)??EQB~otP$eGO-2p!>|d6Q8!Ev@Nr)-qo#Sd=w`B1ekQ_dzOJW1*@@(+q*5?P~H& zW7?Lk?GQpxRwX(~&{|{K4(}Y!2Xv8h?|gy@o-EBEc$Cy=9pQ7{S&Ae<3P=VygTWBv z96pHHe+hUYu|jb$7*e?i{xQeE$B8|Sa7!!Lkw;czmBE#GY zh#Rxhmc}%cwxe-5#(9*C5*1QLHn`0Y`nbCr7fnbR2|tBE^obGMJ7&Q(;?l6e^EeUa=+U;ysu(fv>)Axi|AXoNt^+<01?aVh`QQ ze*dlk75L$6nwPJ|@Bg2#blmOx>h7Fc2I*#A=^VUq;walqUtMeAkKVC7D&yzA_0XbJ z;Ky%ie(_TY-ou?qgz}bc;OqA-_qUGi@$auJET)**8A>V=5ole&c~HS)>Ls1f98IPS zUKjxiM!!JtH$Z(AG2Vkh6L=fJ_pacDgZ~ixEFv820$KhJ(DwSDhG7CqB`77)1Z=a6 z(9gCcEB4sm9q>-m@lF$itM4o~ZvNf6<8I&k_0Jb2|9nwi*!ORp&DOua+BAH(ZO}qe znSf|62Eq5|{=&7pbLzNr>Mp$h>wk6;XZp&~=KXgko_8m!fIjehZ^rGPECav(^^4#8 zM_U2-JCoR#Cx~NU-gfK$E}d*}q3Q`3u*Zcc+CKJ<#j=NWdlw;xtM-9DXx~5B<&!^j ziyd*s`N=6Syzo4q9KXc7Li2y8n$Nv(jhpg}I}^kI_01zrE04-@y0!%&IO|~GBAaTW z)Ov>U9)yrvYj8}A4M)3(Vu!Y=P)eaah}7epT36P+b~x|*e(``*nxYsGff zuCaao2$Xjhdi&!%>!Z+mzX+?xG@e!r;^O=0hWX22mjT_SUE04a`vGQ@o`79`AKMCm z{y_f6zgY9$6M90FDAL_e?EYEQ#ccYOf+rxKAbaus`S^aj@x_>hB&w;} zc7fl&Yp|YmKvx_1pTF9!;hnXCFTL&f=6%ObCTVP7Qit{N|JJMBh3enq-(Oido6xom zsRX3P`?l|K7Pzh@(V9tV>6Vs|rG&vRAs+pYh~^EP{c)VQf)w}g@+xxlQ-oj$?Hy!$ zhmd}Ppl`(cQ~e191JmDv^b?4|&w(5P?Hc|%DpK;iK$Z&<0f~&gibLfXCU8#RpjkX~ zWSmdqdugH>4GT&a$fV$ww-+kg)^wE9DYLR0K(sZ8A1ynX% zKZ~y22$zR$NpOG82Er}d&o4>{2yX~s@nof+I12Svvcb7XmNV8VLP~TmC~n%0rnAJJ zb6ok;1FM1_t_1k(Q5Xl=j#_9SAX-F>d!dV$ggCS^riPVvUj8H?|G+qO+gi3@>L zYj#E>ocElcpV71pa%cbIbr531rnzNdg+C>(+mUqT8OVk}Pnz`cj_57}x=XvX4{JXV zHo8cPAn948egKWU+l|(1dn(TSDee1D0iFvWNV%aaUhP9}IQe@8oqij=CfrB7J}Y^G zC*KI`Pkh*@a?`N>IMk8|pY*ZE&J)%Z7CeX$AzvYg^`a0~BFDjR{aXcu@w;w|s<{s# z0f`|x`t7oY0I3h$9w;W`jBh@i;Yw*(-=u< zEX!$NQO2rOo++xjWmz?xPA8OY%iit|#)9;QLTJWC%4m=?oitQUjrR%Huk2DchAb-> zkH(aBgE1Xhet?iEd1pw{C>Rd0(Lc!wy4KK{jvxibI*^gZTPjJO52%|O0BfSRZIP5s zgRz#$Opt2D(gz-#o{}aBD$(R2>q%QOk|gD1K1T^SE=w9?5K1CM!1mY12TxU&aYXy4 zKyhK6SP(+49)=i%ct=}zoGliN6U9L9pmdJ%F`F(u2o=eLl%Qw03f6rGV`N7XrSOO`p951iuMpG-G5g1WM@;@6_=ccvg0H~M zn3WA8ID)fGYe#UV52yvmC~4AqMdt*jX=!?*kf+)(HV-3t|trmD0FiPOlfbbCXW}l+(%?#B!EX93dUyiV?wCB!Q;uIP4!L ztu?`0OizNQ278<}j>WRXRwZ56Vy#<~lsqn@6hdS&-rw70Z!F_N!{CrH*dPg@NAy!V?V+oi4TGN8M(OZ$-aG~w>2f9~nuSy2sdzaIE#lhTI` znb!N=o)p)w#s6Y9E363{)+W`~cl&8n#J!7k7X9(~?fWk|D8we%`W_a1f9&#U0b!(k z7E*5lGl-Ck{v*E6ZhIg54!mCxr-b!+ho=PF8yLuAulN5t>uAb~6u5VO&NuEKar@de zKE0c9J`vnq24td1(lk0>!0pk1Y1`omB-kiSon$;JD{`&Kw8STxLI`$>0sF%t)2>CQ z8EIbd-q|Up9dN$X92v)RSFf;~FVVvhN-4Zwpc2KhZW&}bqn$A)=QF0$j=kNSvqi;p zHYYgG!Ct}H*@RRi42DB0XYn3hx_zC=d`?$4WJ2)V&OU8bQ#B23XQ}HNA3S@-Ud$?Y zC7I5cO()#A`T|qu@EDvmBuP%&Im&WQnh!Xi%*fIqwulg=(P+f5C`fg}V!4P_8n>#L zdW7?|Hj*lILNM=I(2CYtG?1hT)_dm0&@~OY)ZCj-na^i*#-Iu0X~AGnfKn{VDZY){ zTLq3Qi@vD5@Hbo>{N8ogddaNHsEeFIl9DQka~{<{OY7mXdZzkk@8fL2y9m1!QA=x^ zvGT&a$;yp;`W&hnTy6z`^;9|H3H#yQHQohU9}WM`x$_uH{po{1kh)U*c-A ztR+=z5XBzV-}_tK+Ru6Wyyf=Zkbpva%gNcC7hie~DHIRRW^|+HC_ej3*x4hhhp*F! zNW8O#?TbSShKR{A(m7HE-LfGa7IdAZT{d{FIXzK~@`SzJl(BGhO)t-ojcA47WSUWT z4k-o0!2l^2eb2O9>$)i!wO?b-P(&R}>oG2(N8DC^GeQ|9GT-66P3i94FS*!n=Ikc0 z`xj^*`f3Od;e(yx7@(~f^Lkk3Hso#nW3@2258R&c0r8BYp&<6O5qj1=C3$OB^ZReU z&(6a$e&+TSKC?gOxU85|OUi{Ic+cK&M390T!{WmB9yXQ%ojdjoUYG@!;r)&KRcEIbH~S&=iB5^La^-l3`IW+8OcS=!9yyB*{{As#q*b z%0*=8OEZZ|G@We_LSRkoy;C}g_tJZ^03-=zyQH(0#A@Jb|ul zX{|+RNoNf{1WadQsP- zeUJM7E6ARRsH|J6_F^f_Whtp^aLpXIoZ!#jMjoGXyVES>oHMT&4o6gVh47Z)^E;Sx zLH+1;@T*?ZIBs;3kllWX;!2KWf?Z5#7u7Sb&7!L)OrXm&1D#!X&wlj**|5MgLxww= z(f%F?L7Ap79AZ~awM-)9&s#7pAR3Vi#5p zc<=}}N33pw$l=O=49l+|y6?tqC`l(2LPzNqp>smlvaX#9iIS1@M%oyBLyJXV*M8U$D$A>_S$YJAq}5@BPHbGHiS z9)ntL2XoF_be5x)##=+Mh9oNxqGB)}tN!Lbv~Fxu3Dvxmiqn2zVDFlif3pRZ59-0WB2WQyc$Pzp6Zgs$Pn|pYAE|0UpKm z7e9$}OTy%w=Y}bJdwa}C`2Bll?Boj;t-;zr5T1~a81IaEd8c4$91j~qk);Ud@uotj zIV0tmTEk(QF&giXra7H;T;JV6ozEB*BZP)`o0_Y|m}XhC2ylCEz=P9SWH`HijjC-K zjRsV8L%D2eT0|LXk0w*5%Ox)zj`{X?-(WD@r!h4rvjs(- zu(!9%{&>#WY|64V*idm$Bvfs~!R{WzEa!(l@ge~Cj*dXWa#=DM?^9Jv%4Uf#YX*ZM zuB!k~o)7WP(bOGn({Olj1H7lJn>8Vha~7#ILMwt)oL5zJE}Tc{6z3d?OvqA+>jIOm zAy`l7g>SpILmG)vvH5e4e)rd&dQaTJucCa3TKtp8-Vu1a;_DX# zp$ST&WWt_^0`h|_W!g1p5}w<8k=-niog8sxAb5Lf>DrEial*WE_&#|y(}UOMFEom~dSAhK9?5iYI? zgq7}_5Lo9)grGB?L9S6svAa8>vyRqSDq_on5+(+c{iM}80Y$+Ny>ONH&(C=6t$Ps} za8u))qb`^Du3aZK0YsV;x+X#|g+lb2rmJ2yfYulil`$%EM4E8i8E8#>#_YwlbM^c^ zZh?St4n!1855dPooY+Kg0`Y`zvAwO0i2BB~isv6fcmmP$v|h!JlU)XMmv(7?f%d(F zJjt^Hh$lY!k@vgGVm%{kB}9+@iWJxM%#hpLkRm!j9zFmZD)=z;Qc7Ob8x|0}-Alh(Ul<@`XI~6!zQ!rjYp;1 zBWF->vpj3R8lPr$cYbs$T1ZkeLvww&?J|a~$P1ppE);c;HzbE_o@o`g1GaT>sBu+4y z&5=Ukn}(CO-k``2xc1{e#b^JGpQXHi#)F3saRi1t2fa)Hm?WxxCO!L#-o3KkzEzT@ zNYF$!(1T@*2rwSyIIZ~X&Mq%ryUx3d3Fht%%*inTgM+JtwvLH8?=hyMu4sOcb5=aa5L!X{(bT?a_1MJeg$s) zUA&vX{C9!71OifPI^&V10&l+Dw4JpWV>mxQ=f!KgWJN<=PuJILVOKq18AlbIPS2UlmyA-7o#WB6 zBb|4=aO)bUHsSEvbNFI+{r&BV(CTh)-}RJ0hd?VZE)tZitP6cVXHVXf5YQq9nyy-=ZlQJWf^>6bYHNfs+}wvDX_-o|xM>I;aRvat;Eob<-?Oo|}{ zBB2Uf&*|gjL?Ph&z%hhC-FB3VIXlAvD$&@sp>Apx(|ErGVx54My&tf;6jI8K=--x! z3;POB(|1Th-lCF zC3~`mZNdRo-xRXXqHQXO@i;bt+~b)v@BNzAe?_&7P_P}KZB-|vlxsFisT9uINNL=& zN3NMJU5Gtq>dEV{UKG~7f}6!9{)?}>bpZX8ew;1G(z?+1*7fi8|J+u;Paxhs>jjxi zQTaZzbjjBro%3cQ@%dGL`3o;HE>ds}QW|zbz#ld29*k+N;n93fXQNQ;YTwTt>{FH{ zSc?pvlcwchG-jNo$TVRv%oq=L$Wz7by#c3XOCb}oL?a0dMgxSgESw=z8F?Ef}SmY1zhmNkBM9*R<5jiYte^?2mVtOiS{j zpy_+)>b6FeimdAxjz;Y7?J}8984U(Z=W{~vcmrB!RGu-L&!Q8aXCyLVZ+8!^B+eRA zy@T_P!C*jVJ#E*ptn0{f=uLDe(gXp$eiM;xRDiJ^K1A{rfxwvxZ!EGG zcbH_7GQ^%~A3&^;DFWv_wht2GcohgAu*ODs-^GVG{xBx8jE0yfJ1c9Dny$}gS|6|? zR+b&H?^?@%kl?+>T8ogHRJ*7rxSjBQj3^{B>;|sv?eH_7yTy2%^V(Yv`JJN^`039f zf=6IMBm|WrR1(`Mwzm?qRIH&2L}7GMjJdvl1ye66x8ame3FLqa!CdeG7d&m((WME3 zKH=spi`j7uctzCD+g^vGAMo@9wuj z4su%4lBJ4#G$d~=N7KFsuazJ(YPaA(D-xY^UQd};fucxo-Z5KNoSvRg8OwM)Aj@(V zWrg>U<|)=h=-KhqFxFs;i-9ZV-VmV`fyi1y9bX|kB2J2-Hi3&LD2riD1V7ne6>&aFcaUwEGcAY^8hqsYd z+XYAmBM1g31m0NiZk5kik2$3UQpH4s=<^jTcyI9XspGgW5fOZ#Yf9!-!8FSlD%leq zcwA6OA8^D)H*7Ose$0Zom8n&QH2C$jvr5QBa){Vp7epk82qE-AT@*XkD$)x15U@Sl z-pZegeqZPlQ~@PG3Acfi1y7o!SZi383t~O~;!L%j92ySCxHP%oP_KF|6D7&g*aL+K z=&YbxmPn-`&8eCeZx;k>u-4#%q{d-`LZ%ssjQxL7srbJ0h$MyJ(ay%+P9Kl26*0MT zmy|SBBuR?U3ElK8YP}`#7_GHEbS4KMT?2%~2cOT4YUkbcFx4e4)gy0?!q~Hr* z_yl!#i^-S&2Ua3+tB%daUN+cTB)sGusH6Fb#yK;$`rzROc3kp?J9X` zt;N^~`!>cBguvU5*|LUgL{&HI_KKCZn&{^Z>n&*#E1SmFSQpo6@4`BPNQiUgW_*J6 zo=?5_JgIPe>$~6ObUNpYw}*W4^FJFu)5BhoXVdp(tkV!R)MRqAC$GMorO?mWbW~6S5Z`mIC~S z;cNT+SJ(Xe=|}Q?C)|gz6+wntox4&DMgx*mb7g;*=Wbp{O2Lo(*cZ6Eze}bpxDMaW zFs3HiYgx?Z=we9jI~)OxjCxHzkmd=4@gAx046SEo0-aX021Wup3#rmHo#E>KK7(wI zR0;Pd3zkjItsuF1kg%wh=u{zvrfV%K&AEEzfVy4MwU&c}edhBiJENUgX|xt01bJ4l zHyo4Yic$tzsZcKVi3AZ)I-wZuFdhuhbxqmSn9?GJq;8@^(sd1{X}NjhfZ3u1G~U-N z%96n_XOtH#fFNRpq;A^Sf8zqX<2{yDg%mIv?qKqQ*}NjnQcT<7d?YmKnl?i5lq5+K zj1PF>kt+7VS!+lKBfN8@Jr%xG3M)H;F|mhh%Sn+v)0|rCAH|=#Pb938*%z!&Y)Bo) zv@ZGwXEj|WmmVHF6MC_MY!Ed8O zyh{9RaFVirezojpEGg-%ThBQwTX@H{ETyb!>ZXo@^!;3{lzO5Hv8ZcW7cNYY>)=y> zR+@Y;;M&dSxpC_Sx@tk!8AbYzH7z?@kZIZXW(o3C_XHntMX)XgUOGw8(&4-yRg%`qbucH0 zaup}DsvpCS1i`d2%v#T03MZ2V^D2%t+p|MDYq8FvrHq75i6&8k5G0+kw8pG^${t6+ ztc11)MMgO}&k&-Zy7zt3R<3=WC-G00b{WuJ+NFIT_Mzj4 z3xUr^b>g1>Y4MDS?b(MAOAe6&zOM{!^$KjZ_CZtQ)AqK3UT%MNo1kUogh?Sc!rlmk zh%W7#m_44(uKFd_2TEj?`%*rKI0F?-Kl60N*)IlJ# z0xK-ubQFoA(2n3W@134AOonvMapUSCRlQ(O570uMKv=-Vg;Ys)#$$xkWLbd}lB#JL49BE8 zVK$%9SsNV{?~y9O5rf>>yrt@Ojc7kJUtb-^J|T>NI0-{ z-=8)XB&-=$*W(tXOnUn9NOsT#kLfGEQLjmW^ARrGKNBCiy6Jc_z z*U<&B)--cHTut`9o7;CO&NUG}8(&yKuKNmuwLSb>V4UM<5~=m0ShGaPi_Cqi`J$|A zY9uq2}Zr=AW;T^&|EV9=t+e~swN~*SFT9(+RqW#9#Y3+iL?D0SSc8Og$zHkR_ z6ayZ<`)SO>B(g5%8B)UTFk>{>p{wSU(<822c@IIy-Pd+GvmYmW?6F6q7}TN>Wk;?>w4uTebZykGGsJ3jGdtzA>4X$;LNK)vU9np0 z?%%)9&;IRyn`~!~&^5gBmE%ZU;yhA#7EQArQ>&k$PZ_q>^~slbAC2?OtA?vuZWtW} zQP7)MO^?+%OR5!G#3X`ucD)d2rPlk>N|R(cm<}ZcZR^OhjAEGKq~cG%{R-cH^Zh6* z(Ff3L4L1?x0j!OrBTZ)`TW%lFt)7Eb^2qm9v*ohpPrvm#<@B7-y?Bj(>-n1~v*b4) zp2N$l2kqjRcw+qY$RWI4uWzHBJ@TU9tlWsdul7m*jD^t44QwHPEKJOjp?`Oi_)G(i6ArdH94^j)RUb)VAkoQ_-a(#TxdFsZ~xInSLho8?0vn6M( z;nBP#RhpZ7y9DProy{4hDQX}YW(i#gESJkDd@EX#MA5b#qro2IRIr$K42v6kmm(IdrpG^BH$rl}bXhA6F=&P(z< zrCQGE>VUHbB?XDr5DbHC%(AYSlnaV%$ki)XX`7m5T_Z&7UvfV7-?dE>doP`jPJal| zIqFJglQTx+oHR)=&QRAi!TY`n9LI?XQ0P^r002S%zIf}tKP9ofeq4l%C5{jwm7!}% z^?*cYq)F@pTZuf*O zD`+O7hce0-e9PWx2H2H~(?!i}F^@1#Vo>RW?2}CN`|P*r177FYT@9xJEh5x1q7DRE zmLaubFy6sCOEwtt+^w6u^YEC?SQ4#B61_fGr8GOEA$!B8SKxjSrbIyFcvz2RGTKFhaW`7VFE4UE3{JSXi>@ZKz8H&aYHPg^aydbr0YZrtGJZq97_ z$IQDU&K5Ji`qqM<{NfFaYiZk>u9~rI4ULJ36{Ms!EmBIX>2SVV&tK zPf9_o2$cdL5wS|{Jk|#$WlPn?1mddF>zv2+)nn&kCCyleH{F^9DQxvSD`G+?k%3_% zP)XXet_DU$tlUl;%fquX=F5sSNsvO~M1&JJ#?#skB?8ONVy#8uX^p3|2J2Sq^oCJ# zRfVmZj#uA)h-+*1hbe>OQ`&mL85uYKySs%AsI6??fGmjN)`6we(k^93S6Gt=c*TN z<;boeBkP%XU9sB=RJYDatUX@QLzo`(c-J1#W{)wO7kdKMl?k!RHug-Dt6;5{PKZI9 z>RAq-ve~U;+Pts{<}SPzp2X$@*vHp_AP7N0Um+A)BaQ6|SbEqepou|d9pb(crOAp6 zp;JPLS=%H{nM@`;IzHj{!GJX1!32j^3r52f=QOXq_7>0GzQJ&JOtI*gA5VEOJ!Lwd z^TO3b?oStpH06c~n4Y0hBpGGrxqkB|`@5Rg?!3#S%2<>$X0s(CXs%w}!Fb8=K=Eha zd4=Kb6|N74%+Al5)GgI=#$wq}OT}bSl3_`ur0zO0ozObRja`Rr7bLmk`qdYiPAAOH zW>L6TrIcO6mAyTbSJ)63cu8eyvMgtBUvWO2BBa1|9kX)IaF~-Mio=5gy4H{;ibY+M zrHaGDL(b1<2%!kUp|y_pKPXIy5XQm&0ZM7gMN8XM2qiI{1Ml(Pq0=;qf(uES=U8h| z$|7Zg&JUMLl1B80s#N3rU|713#StF+LT<3}I-`7PY$itSEI$h9KT2|}v% ztfqrtwpjK(zF|G@i9SGl+}e7HvX0?yfjFxXLa;j;08mvmX_g_C=Fz)vb9{Wn=;wct zPu{xD&BJ}(dT@m6Jgte&@i0%RTSL`046=k}v(djB7dd5LWg6uk*vgml^mK^-plhF$U-DGaI=)v9jm9r}bd! z2B9R@7~Z{qOdHPG+uOkg)x(E-@JWbC(j-Z+-ea6+pkpQ1IM1*U$W~Fcu~%40xrP*P zdQAn_5BAZ1gbbeF`{vs`IIn0wy_`w+r1j_VaLoX2$ zC6VFCmIPUmEX$e)l4^?0W_Poj-F?S%@0s^l(;D)ys`lCE4$VP?ps3FWT;SX~b*lER zwQ8+z_<#TZ<;(2fxe1XE+MmcJDn$r^WF@FuUQQzA`6`{Yj;gLmlWs!*BrxrM?0g^| zYa#10Bzw)uW!t2)l=7S((mn&0J$x4i7XjUcUD&t99(2e=i~Twm%&oFiK9aFg1b9&= zOWE1k`nvb_57dvu*}wbm(FRt{-w7E49={aeZh!YEYp-|aJgweCiQq`61|}6wIa}aY zo1a_Rg8PF{sU+G8s>ackO5tr-yp@*NM`8NgjRLij=?nQgr5fl)^O?K~nf~QHr9>;h z3#3#yX;EHcBZc>Y*=&{LPKuPaeAy6{SL)+#>w~;oy4pSVs=Mcy!3$&^75P4d3kS-;aZnV>swsf*k+JU@kxfb ze2Kk#dt6#yCzh5wN2kn7!&mRzBhTmbybgUhk!Deq%*zrT!>F4wo6os49P#AVE?;@) zIvUTSFwATarWGp2YtOLTM@mVz)1x*ub1`Gl=*pE&B8*`%n<8vYl%zyahs!$~%ohc! zGIZBR6t!bAImO_KIvLV~lAgLQvCfkwF}``;!GZC(+Mv@e&RM*#F(yZ9P1ea+lx2XY zHCg9aJs40a5#hW=iJ+ZSyW-*VP6S89>#&{XRfVYy7>m;KY-S75P}Mi^q%c{(q5jvu z$Yz}b=N{^q2_+jyn8LdnTC#|SRHAAGn$OY48%t4ExVB%aK>~PhY138vrlPJJ8h&>s zCf`=@maLgAl+0L6oio_#B1F+%HV`@}NkZ zMD*tAf;aCK?CkdVqnE!*BqjOu7!PPEiIoUFXmyEm8riIsbyeWX5;r#}8aQl|$nzfN zR#0h8BuxWv4Cst$Ij3(8KlsuMy!hgGu+i(X*6%XxWo(RwB=-v5zBi>*_mOtAET?S} zb7g0189wkHN`Jsp!5hOE&w@Y@(-QF70lOv7~?%=jITl8-ecsK7!%V_CV1FG};a+@f& z#!}Um2#1c&mDq3ygttpHj+T=9gSsc_EXkj-N_-=89-z`2ld5B+s#1kGMbS`JF@97?LN`S}pk zxs_n-{$ivM9w`I-uKkRjFckI?51_^AFMnpZiwO=2q!_YOFim23|Cd_)U6S~D*z zZjWP*TnA$;N5?0KI#kd$`fF6P1;d>U$`Yb}ii{JUe(DKcf9qYYZm+Sv-s6=of0fxH z=cNxl$<2df{`lQ@Q5J@2%J$YSIuRIa(HmRzJDO3J(CZJmbuvX-M=yOEQ!c0%HHwJG zws+V&IVFlyp4q+3c0c1_JmKo4EsPiBiv>3hk11z4IpFF7Uzd2*XHnLa^$E@kBBkhL zU97EeUQrlB6e&;wl_W^v8P6AZ>rt^J){-n%R5>^!&mgy^q^?cKdTR?R0cvaug^DAT z@}zN$^BxsN6lEDS2_5*#G^{6Q6>8_)5>{n>@T83|&qyR1k_GPrJXwmMrPTh;T7j)< z=J||qnvxO`Nl631LxIM0bLzxDaAB9x#q5kjX# zv7oZxr6!7E=G6k7ruatF4yExrYgis7Ya1I-7o40sJH#%b)%T9DRiw^SgK- zdd}vH9BVCyvl6$tae|rlX2fx{gj3AtbL!gA z>vwT3P-1JPSE>vGya^f2M!36Wwroww)hb^Dxe`7&$nuOitQ86tQsS*$3F<=7=Dk1& zw}fZ`&9QIbDX;Viwi|ofi({<8TTksAwjmcOouD=rToB+k&SR892}zb_bbDQTYlEh5 zN025FvnayJK-X{b2~tWOuW)sNH3nxrQ5>U`#QUa>qcp}@WK*Rr>zc$l#*-P& z8+4>8%SGeMI21N$9j;8!Wp8Qov-cinTgUAzT_XU*N|MA_P*595UX&DNL7hZ+EpbBQ zjKOQkIo6f;sTnVb)R}HnOQ3*n;8iUo+y+OwB|EX!P&dJt^A6F#@!DW8Wcl0bT5y(i z1La)Jdih1ZGL|dr$0KyCoOiftfy)!{P>VB?2Pp(T>QJi$<09U^{s#M{Y;y)}h-;dH2|2edr~-maOs2heuqQo)G(z@BPGw5ovFEk6ueHDFmiAOb%|7 zbVB_kOEjl*!)#HI^+r7L;ZNb-c$HonbFz1Wv5tA3Q#+`e{;R`Dfs`RRv)0kf2#Ahe zCt`PP$fBq?nNC9Weu<_HD2{e~%F3fV87fLpX5J*?Bw9jkYF33smPA`B5EU6ypX;k$ zwbFz=Cy6wJZijhgu+|X8F*1s~VXbNRzgGV=i(}?xO`0U=C?+yckw*4b!5g7*-@SC?KIgMS-4om9lh&ts1HEHmtwaT2!@I&QDEfBdDcB zdWhrXEFku7rRsTx+7VKmb#VZ-u|zUd-1YJu2xBXiYRIG-l9`pUQ&NOhgZsv-C6hV# z4g=~TNQ?KAT?BL&c42=?J4gJn+H1CAK_g8Dl%IH5&>_z`9QRvQ8ouy_FL81@=KH_@ zdpJ5e;WvKcx4C-t5>Gw#WDweIOsSU5v7tp@bwWi5dM48;zw^6)#GRYBS>G6O>((CI zTLIqr#V@|f7ryW%q|$6{uJNf)eJ5L6o4oq!>-^4d|2}JL>vVbvX`1k{kA8&R-QDH& zTkwf&0$R~DwYM8bsLBOhnpSJ71+xg@mRNf4J>CW}Qd*d%_f1d-U}zaK&o@eYmSO@c zPWjn^M%>*=RLK7>H-^xF|DZ$Xod?&zHNC{^2(6&g>CuT3(kw%Ey2vPIkr(8h3MCa) zQE+rH;p)}P^f$N3WX-!bj~O42dGTY=9sw>Rl) z!OfE?U%k7}qA1Dj3HG+-bTa1haE;0v2AzyM$0tNGL3q!tyT=^O7TBsFK0QSWN#tRF zIwh|v?2+aA?L(fswuA3l?j6l2s)|S}Ms*+@k-(_iL0ZGf$rvdlJ6ju^PG`J*^Clur zS7~MT4L(Z;GZQ4MOEUx#Cy{?ZW?1eRTaouyigQnO`vj~5t@wA*mK2%|4Xm`QdR-bFdQ4CXk`3}p*8KcorbQp{?l*Ulvuy`D++2i-O z(*Yo?#nV(@lhD7UwWKs5xN**50ObKbdv()m@jG_-X z`DX>&bAeu6)f9Iv-lq%_$GO2X1+7fy$aF|5S?5?xPM6wp zcBPUWG!#8rD)62-4i(ng29{vwJl!lMNiq)a-eZv$Ok9j`hA4uLmTYw6um;tJED}7m z(dSR%s7Z8sB%#mv8!ca1Q#Vl8v-xitW^5frP31DYR{`qW5Ue#1-8MujOC4I#RO7r1 z^R%j#VUxMrPp-k%HRWFvRZvkOV`-l-3@wDO&}(v#yN`+4Z(uZ}-sfbV z2Zv7v=QWYw$`Y-D;H@#1vMg~fFlE**XfnP0=1Isn+d$f&gk*g%;_7I?sNW;$#u#Hb zJUJ#-lJ!AGBrPYW$1G+G%DQ6z=!gV`LQ~d7VEKy^lw1U1Tkiw4xYHz2geI^8LU2y| zy)H6W)QzaJloIb?GMSQ>IeA@Co>8s7NQ7kKBjS5Z2mySBj#ANwSujSXhgF_G5vM(b3yWjdQST4wbF*Qtw~J8y<- z)KyI+Lm*LE%fbZ+^4{r!quXzB_3l2_)V%ZB+eCwPN}weoqJ9JREkpd>}y@K_G~el&N*iH5wSSJ9{TpE7c$b zQie8&r5wnrttLji3j}9p%IetKpa}^v(LBdRSy9&Enn|S2K`q;$?!j?<^w)a7*+oEi zVHfsI+59d+@4kJDFGOF) ziin^5i65ua$+&v;GH<>0&VAocEa8+&s>U&U@9!+7#M%1@S?0ON*pP4+QIptO;zU$WT!`0Y2q|N z0H>#?NFi7st6|)JL{Y@^PhMfY*X7o|1HQI@$jy1l+lNO~3ZA~Sjgpe0D4CWukrV7}Y+|T`zFdEZ z%I1hTC5uyxEtu5>c0+HmdC8m6*1=P9jWYqZAE$(Y6d z7@X&g>o+i^p__Eci#f(y;#kukbScZ2#bSYXmb%C(am)ZOB9y zc^Ozmg~U_~u5Rvd_i&$uvvi~)N+YU9AUIUeEZ$41s-fj>#$C{m6S&axk)#P)E0psA zDn;w)uyis+=r{#fLIGQIZiUs-1JX4PQlpC`<%$Cd;BeMa)i#j$Sc{k7T+qM^ggX9x zU91a`$WVETl%ktxl1K%|v4N(kMjD`nzCMab>I`xj0vG34A_XEPQCdF`NG(O7SJs_2 zs9%*mfi~gv;9vY_qQ&3INKfW8}3EDjOH3De_azVwH`MYg^RN;6;N93LOC zyVhm&-5-G@p`Qp;^M2eJSN!WYj`{5OJ;Pc*H`et*a_S2WYAX0|_Py^|oL zfPUyC23wcZb;7LlL`u=?^tfp~uiZXne^Rih9mbtKX=6>$b!`6s_SfEG9QBB#=kVwN zr6je7(?xX-&f4mkN-2q>80YdclALzzRw_isI<%C`@|>cq$W(_#S>jiCki7R)wPTX! z>>nL7UW5dhb(TvT13v!rB_f-n@a&x^zH(=bHI7I@Cyl9H80%HIU?B7GK^wM#5nJmh z>Y7$YVPlZ-*k+$3j=6sK0N?caB^pwdun>ZND%t2{Y^^2SKFD!{5PadA%GG-XbIS)3 z2vX$fWchyh6_TJ<=Ebf^C$N0)D^I!NCe(Z02mecWsU;FjnE{U{SMxBw4P)Y~5 zWJ5Oa;InA^1Dc+e=IA*I1Vod03;IAbeQZ*O4OuuscD~tAy!YT*vGDgqsp7@`Bot>o z3L;3IEkDQV{hW%Bz77J`(0GO#AtQ{jRC&&Pkux(joxvJulG5wMSm8N2Jiv)i6}o(R zhqd8=qAD@Q5JfSnzd>zkit!=4mv>3BgmRuUpBGIQyUS+=U7i?bjPr^we)Sz}oG{E1 zp1yXO$F6MCjU?mQl>Paf`E0^btjVH?t+gIf3Y5};DNJd0`kS22$IPZVog^k6#&kLz zq9~>;3+DX^N2kZ6Nk&oToGuDFal*!6KqTSad;1JJ36sf$qeVft(;+VwPs=cME3Zh7mrYVt%NFvLiKLU5sI5=~Pq9p6| zsDx!lNiOg1Qtj02D$jWh`0~f}r2HI`_QCx(aJU_P&ZFiB_SK?L59EQu#An zKmNR#RG}}mG>+6m8f&^~!mOy4NvPI=Q&NOJ$!1Js({!|CwrhUT3W$pa+8!X0o=6E+ z+Ip7usGh3U-fMW2lys7qC`pNR#NMr&SPx}g^TcR~@o@dkH@JJ}MsPl5gjRxB$z`8l z8_)&Fslu3w6vvPJJF3}P@A1;dpXB9Nze3#KCd#@*QOcA30j4(W-`(Tjbi!gf zfk>m06k!ZSVTq-r_L6Sa!CHs0mR>jGWISf?bRLFLM1eh2%0Q~+9I6?)SKhkAW##F1 zIvgFJVqMsyqBsF@7JUVTh`?D(>564Ax0?KH_r7K(prV+n3@Cw;5?8qgwQYRp2b;_b zoHH03_Ny{4c=g@8lzC1kQk2G$c#jg2%iHUW^9obfAwkhJ-AW-a4KhOOfNDviAXDIx zIA>Uxn(=H#vYmi8EGkD7Tb#GVI;N;Chm!&-gY3rSwcvPCVSDbIbM99HKH0FaT5Az@ zCFoj(t@`uKva8SC!vnpjN=jm_DC)poDhY%$lgR||RK_5@KgTlrutfRlyB_}kXX}Ig zdAEyz?!qqY{|9#a_8orefBMJ#=)dlGgn#%C|3~)j-eY~V#%r&= zPSTC})Th6b|Lte~Pk!YW{uMv)SHGKI_-DV$@zGHOiF{P{F9dAHxwAea?q7h#Sub4x z{htf?eB+q0;&`0}hVT8%lspr14}b1gp2JE+C^g;&h@I2Esno3MTnW60$@3seojZ&6 zu(>{hv`4qsq0Hw@4)0Qw6`d@lP7`Xg&al@b(u#SbE#;*G1;#1T^&NI&O}Q9z^UeWZ zxpT}5k6+=+Fyr~HKJ}#FH(q&*>qldbW5swp<^xxE+3a)zGVJ!KC2S5m%=3yYNhzudXKSpfNMcP{RQRf7 zVMAq2MVc(@lSt3W=^<57qqJtUzQNx9J?8T1;w?l?bJ&s}drT+EZ2~ZAf;1a@RTcfHTrEqk4blXteQO zI|YSvfmP9aytje8!&xe8gGlis=4i;CzyC7KUMYkK{YOP@84}j4K__OChj(S!kKFt$ zWuTEibHRgIYtJflO+a2YfxDK$;n&8U%kIN!^oqnIpK z)cKr^-7B1QyQ~cds3_+4$%Ly5@@(e7nF@WMxls~MmB!snT`L7fkTPR=>jM{9igrSE3{;E-X1-q1QC>vk!s zk|gakWHO#?alpK)kkypoR+k_A__JJj@>wSPHyHLGV{cq=@9>1*{lm}k?!8mC*ZTaY ze=}lxL-YNgc%D~Ze~la0-(f9#3N?txqJ*8DE!MgbFJ{Bgi&afYgyGJ?DR_uGeT33{ z>98V=W1fHNDX!-ej>prchdPWI0%|mhAhbQu36WA<+1cQk;gIc}%e;H@HgCN77I$U> zlMd1G8M#bLVC22Wne##1$_Ovt-&iZO)&xdX$0+HOFXqf#xl97m1T|6!v{v}qqGhNG zS59zmTwqL%peB}rQ3Q96C)}A-I2qF&riiAmy_cpe$`T=F zIkiyV3XAdj`jEf(cYmB0KlB0sTiaXw@DKkG)7g|HjnDm^SaF0N7WF8lc>V*=@yoyX ztNh>({upa(Bi2WKvMi(58}N7k-haZ6{K!wx?+Hdg%e8|>gkCC6;MJFi}Bhj56Xi2x^>=GZla+$Xe4mq8T>EIc5JDe5;Cwb0dv0$e^KqfI# zr)+i-ystT)&4{7|L0uFzcO6ChX4_9E?v{6gg69bT;6_&pgIP zx5K13C=ar<5n#WgijD>9vOY)gdjDB&3P2SJq9Y3SDM9~^e_Iyf2V;~daAl&GMh0Fl3{-hGoCP+Oq$-g0QCh=wK6{v zEvU+pPNx&JD}=sFmNfT@uBl19_-#X^5={ZW{n0t3`l68B`z2u88f1b_lOROo+=+ZTY zo0mA=zs=E|8??;5z2S&_a!OGYOD(*zEErE5*YDipXguLiD&pZLAN%OX_~i4?^5&aw z@Gt-H4?qa|{UOGLXE9vgVr}a(k3IP`Sv|(qp&~S!Opr=)G@FB|84U&qF~wTPAAjXd z4yK+L?w(>?#oeP5WO|R|dw01TcR8Hp7$>=Nc*@CS!s%o}US4AO)bnf{4_NDV$QLv2 z-n>rXZqSVtk%aze9aqoL61u&V+Il|p%o&R0<=1ZTJFgsY`EN!%z0u<<^BHo9rtppS zqei(B*Hm1?Ix4ty%o`s+=#$n?OKQ9{S|A<&J#z9VXsFwOHevylqnW@f%Beu?a1?jvaEP>Z_1^$4o_WPr_=-H*KbqjGxEHk zpDOC6SKK4WQc1EZLf?L86~leP>fw>wK%62PWP^9qMS<4(9NC&wYN=KCu%#3Ak&np`D~t+;N&df`oMKy$##A2c$4J42<6V(ML>697xrzpZ^2RfI)BpZ4tViH z!{x%}y=62SF&d4&>04QA*xlXczx(h1Yvzj?QcBV^TYiKe{m~!duYdLj$n!aCYwJj< zD9eH`eC~_Z`Bwvp@TDJo)ssuYbRWVM|Fc4iG_*b=9QcoJO9Ive5_& zbS8q(N9TMHq7_Ol;g|1aVi~7NgqCw+T0=>S)$!vjj8Qs3+xvQJt zp|FBKdHXIq9YtroPuB__ySl~oy1mMqZ*kNFN&e z=aXY@-;QB@Ef5o=9lD)_R4JZ(>KeDM-{$)MJ||^GUDqffA(mXb@+5D6?KScmw3e8n z#*`J~@ew`WXMMOqQ3a9jMVWJOe1P|YPS&T>>#*MM@zuB9BJE`KdOZ}9et*a;pD~`7 zM6qBv?2wlw-Ya}v;e`j|m+WW=g_4GffcN12oVgu!CD!k{$ zSg!o+0o%}|uj(TbsAie$jj0bO-Q+n!(*hS(QIzz09V#>B=J*7u6h6!=dYz>S)}JLs917_}Z<5-guqg`J*o|FAaB2OY;1Xd_JSQ zf&zJ1wP#gCK1veSb{``eUS@6g8dY(V?D&W=FkKYrBt0YXXiQ2;mLy0iFkvaD70@=u zEY}kUluocY8nQL&v3GEQ%+`Sr>`iLhKo~=hZzUm>n#y>3os_()5S0z2OWyOf{V6w& z<``3xb$WCn$p?0Oq*_pvl3c>U+`w$Yb|+>OOU5hj8bXB1+`}q#Lg6^SF4c7vh`1gS zxEW)XOqM{?ZXuUq<8U5wDG2YN^H3W@QP)IG02lh>n`T;ZKA|ETX!1pm+&69)0o{dN z*f(hRH|)O0i4Q#-NEKJaZC84{?(+{HY0lZuyC%N%S64QKB|=F0yP` zS!a1HNm5Ry6aMf2>*u+2=@Re0{WboL@BIu(oIUIo=S)epbbcOXVN(d%Iw=L#t_Z@4 zAUGbHpjYwp0v-@+K5rrWr&x-pwtp+jrQ6;h77%mSkdeE5w50I z9p1iu!eq9f$`>4;PI*CHWo@m?VA$mYS9Uo(JmRf;2OLbsbUWQp@j!;4($omy`0&+h zm?&mCnbOG=I?}9<24qziiNspHvXDq_m zriV&U)isjv-bm?DI!4$UZ_gwPT48zDH2kkx3%LwZ8=~zDMYfL8a;}9rD?FhW(*^ja zkP3u+U-!#BU^evH;DZKO?H~r?NYhDUri(I=nKYupl?lO`5Q55>C8R8Bb*tJ6Of!Dg zJ3wONA^Wb+8~09_78U(wNfSUfOM$6Jrf^UGwEPUgMdy4{~(i`N_WnKT++V{zU~7!?S;F3@gZe z0mNS8#E^a`CDSqc^9ld@)tmIUKgE-m-ePxi#0Q^!29gA#^GFeA9t;NlTDHS)f9@52 z;4?4rGfB#|Yu6aWjXVt@P>*C->$_QkL{QY0XtRq+HJ3Y^Y)a3o#gr#5ZE-rk!D4*K ztf|`CiUpEL1MQ9JP`pkwS@( zEcG5Ng3?&ZVotg6Oy`y)IctMB_xYO7UDqWpipgtBRas&sF;R>*4pkfGWyLfvxz_1X zI>DWVq^mSSTGGf8QB&htAX)h4eeg|o5B2B@-9yYdM^%zVB1)>}-GIck}P{@q=R| zL%I{wGF~-RI4}GX8hO^w@Qp3@LNuy$!kzn`OTk)im!C_EQ28{bqP7-is?eJdM?^X$ zic+j~l=C^YwV_w4D!>}70AJQvCr~QFEauE7bLzp6DAEsPjJ+`&oi6BP330cNtpjVF zvyMU-u3p+f*phel_NmP=Uw-Ef*EWX?dmTj5VQY7n8;3_rjiofp*chx4OGzOD%iqB) zXJZ(XMS_FF6HfJlwO&dVDV#AJO{Tc2AWby=EMrzI&{A@FG{6|c-Kpo^c#QELs~5^{=H5MKl_6D#9+0RoM{ zdD##ufOVnrqJ#*jOE0mm6}I+EXqWeA0o>W#^$m$dd!P*f&vG|sWE_nKA|C4;)3Rh? z9eppbk4W^qZ>=ePR@O1EEJ}wec&s#?#MAG@z!^zRCBy?sownk#6gqB>dmkzxcMc9Y zo=iEJOsQ9drSpbO5Q5{_LTy(q9BLqD&OX3=>Vdkhnama}igNjm8&mVj>t7*FJM{Yl z2A!0CZJ5;+-qiRwqqa4X)Cj2|O0cFzN1DO<7T$Yi;}hy?&SEyDn9l+mqX?_DHHN$Q z_OaH5`K|+5({wW7=<=YRdj18$VZB+m11-jir6 z_*X}cGEsr4*gKSn82S!Q8D4mDgC|CteDQA0(Xr+~{tq|kbQ10yPdG)83JDGTGg6+~ z3Y=^71KT8>^9)BL{>ESZeop6Qs8lTqblPEey-!)ylvN-OxwgB_)@aBgFSv7X%&41i zsuc^HGS?|DeBv|g-n&KC>#~13=g#qrt<4R}L9+a=U(f2gCeIgifH;cDg#Z%>L5ypt z_<1l;F-_0r5_TvcMiDo?bY6HWC#aq0bY7u^ptg!m)3?1`MjE)G^P%U>)`q+Q6=xK+ z#Tr9i8H`&+8w4baRuMH3)1qcRnK7OO;;AH!aOc4<-?FttI{=ZAp+DK$<$MZXU7Xuj z?Dy+kfjvzUh0(27cDA#P?Hqx5px_dG5Jqc;WdEeAD+>_NlekYd2ic z!eibe+!KKTQ#6c*L8r_$8UBYa>Z<{i5Edyswz;pjHfYgFji5#@cR+`?4NT5DWJ~xW z6Qm4!PWfWd^w=rty2ez&0SHcuLPrXnXsUd{Y&ykyMC z^@8cqU5-u{9GvWPX?K^dQ#|qb(xSOS9 ziAH-toNDG}#agFJKd#Bkg88JvXpIvvTP!H@1=gih#*!P$M(pXQF}vH_B)5(kY01H2 zL0Ob|XE+?6;+$n?ZH;c$p_59KzsqDgC0BxNHKL<69U)%y}VN8J2 z#YsfB-(!1wmtMEScrqc@F|)j&s4J9K2r0<2E^#!$xeDtYv5v^|f^N4%zt>}t&p|;^ zE-+OC!9(Dj!Fh)g?VJt-0nUX!EZ-&unyKd-Oq~!xwxF%lwZU9lp$~)$;9E!KKF=dm z@+6h1IhxJrCK;nhl0{j=_;6tYmN?eRIPQ*f4re9SSz15&bmnl*V4VvxD4c0dt*%R} z!CX5Yt&p@e7N;%7nou!q?m1lwB7b8}p^jpd)}f5wloqY^)W%>M2E{WWd zVY*l#_xD-b+U4p~AE5l&8x*HAyfsY6nohUJ^z;~G4CCW{Ha54Z$|58vq(rs-f6ilt zCDxK#clQ_!))1W@mmhlyts~z2%9p6hMPN^?>ZV$>;%ws)QgH8d%2+Z@TOGH|6Yws?H zr;{`1UP^qqKt(Zw;fN&b6DJ92l(Vz3&e3`Z=(N^NfjjijWU1F;WDIks_in3f%?OP$36*Ywb?`#N6oF`U-(^-!5&>swm8Z?R4 z8si+bX~nr6##rJwrZg7qbBr|s?%AwghSmoNy{D-Fts}gT5h^4MX)2qEl4MFyH)BpK z-{ZoZf28^QDysgh|Gb|@)G7q%&bo6@a#qMplnnjDt9^$pgR4iNmBi_@utP|Ja3W|* z-k0NEeKl8^vX|AshkS+ibzd?U0o{dN*q=+gdi63t_G3Rx%TU!as(m{xAjZPx_MQv< zpVC|-)GUQW3KLklggPT_5#RXTm&|mH?$^T{vQ;L`2Yz=>pj_aD!g&)i@!l-6e~3Vc zB4va?P**u+QRA8{bXC{5hAtdJP^`5gPBhjOxY|*rGZsZjr|G$nO`n2u9;FqXG@&#h z^Ka1jsEbNLi2~;ZCzWN8N@RD0n;hT~Ov{q#V!@3!-r$ojzQFU(KFjC6{N><~R2AuZ zO0U-=@^E~z&tg`vmUa;B>a=R9fF zrJH3a0af9!##2j6nj{oOPMT$8X%}Y=wsvT(mQ~!^)J!Kc`rRH- z%@=`PYu0%dWwm@F?USxO;4D!bqg5cSabm@xyYJm^btK#0dG9IfDy(iUKqT9$98ECZ z26!z~J>gU24jvS|M7od$E+T2hK>O;!cm@<_=k}&*}ahiuo9wcCl4XRh12q zkh^bQhhXD{XCCLnFMb4*7mR-MkCBs-zw(*yLaul3>;DY_rP6B2;uD3zadc<)YdR)E7 zo03IdHgW+o`mSI)J!Lv8sEK**>1VmTJExOnY+t&>b>Q;uC64AfNs>?*OEsI|oe6u8 zQmpkl35hc{c6u%f_s{(;w;RJv9${V4owkm_r z1Tz7`9klK8vL@1!o(sa|N-C^5E1`KGs)1!~!ZTKqsy2a)%hVyjYzVu~_4_VKf10@< z1>Sm6;aTse)M><|`CN@6--7|$19rae+qY|E^FDOb2v>%4j5Xi_gKyZ^!``NZLq}nb zxu)k;|9|Yg*Rm|xdFS_AE=}3(k7+C$9zdZ-P=qcNy49_2 z^ai?-BBW4I!Ql)!1PKyo0Bw%jrhNKkbg?q4YM*nu4;lm^lD)r(?%JnJR#xW9mFxSy zzv+LsmJ;g%rnh}brJ?1JxP4i1<+wO2zhVs(LiozpzX|b>&^^RM{3DLz$;>(1^WPjo zC_?C^(tn9;cDaXvq1b9c+%e)<`I`2Fwk z(fjZ5>9gl_ZNp-DN1mkQ(-~Lik9cQN;@gcA(3R-J1 zQji&gNfKn+_9nEJYPTaZio333JUS)GGHhp&LUT4f0~g_Ii`6oY^&n&AU4qil&*}R1 zn#3qt97-7i9sBlK&!LOwSfgY>h<*VkWlZ3GK!ynQ85S_ZkohXumLQH37wGwVGJ2); zFPbFbY&0U%hEy9;WpI7xZzx1W{@yox0pk0rd-Qu03@oGJT*weejgEepy#aGv7|Xpu z^XqX~h{mKUyzLkVdJi$ly@_DhFX!(~h`XwysGGgF<<-@kd2+_>V$th42!s-QA1!|` zQS=yND0f?a@>l7}G<6Y^Ip0HxLsT~2D?KWM)6+PA88p3hSCrq|KKube2Lz28n#=@Lee?(S}oE@{b;j-iL{c;;F6@B0F-f3CImxzBxM@Fc(MY(fER zpue03Auj5y4ET;k@h5shrX|dGc_L3MdS;qyf_2WVsx2@owpYD9&q@a~@ef0E+MRzG zci6Gj*k@n>eu1+Tt+s`o4-Ib5`CUC+7goDu|0}K-xb!Rb9p6CI;dsI`Nc27ImT7JjtOTlFK=+tQbw^__RO}hs9A3nba#I#W9|0Z z&%Oy=M=_+HLI#-^fFsl8twHjb3T~^Z@4KcuaIvP@HONVRIWbiQs)DH z4Y>?jX|W-8B#fMQuj*M2d;kqrEl!^jf%+lAv6*mFqILZtGi`8;T|vU1x(op|R)csy z0&E`%<9ZsAFgj6qo^xufkx67kG9dQ`gAz^o{-A;or*z9r2>`kAqwWCXGfRGwefz|{ zL7R5nWCUaW^&AVwQ9FIjexFYK4r4`Hte=bQmG{<|3BoDI&s<63FVAef;%J|4F1}b4+9FN*K@k^q0oQXP(~2i>JA4% z#~MW}jYhg}OH%w}`$qf2K&RFZL5A0wCQDK^0qB~vA(F$A%$kJw)&}+q>M8sy8a#00 zPrEVm1*J^xI173W$qEb#0g65s86@EI%QK?<`~p>0^hmmW{eJHlMFPkIS|Qma6S5Ld z)*r(`GchA`b1d0Y+pMTqY#t-K`xg2FuYOKjdl+Ba@T@Pgt3d&Tu)QKQqKsoq@J$7;a;TB6LrbM2LhWQeAjxTsn1=*hb)0J)Z_t^lI%W_#i5J zHTMmH7Ax(U@r)prv+R0TYc{$!MN@&lQ{`KJjv8c2!dM%9$3< zV9!R0FNpA;v^#d|$)Cs&Qo;-u0+3QA@-l8+DF7BHz;oCtX>N&(%ClIRR=|<}CJAx1 zC~G<+|E5gfy(s6q#o^GR@l@WQ{KHdY1!Q<_m+o&Qo5iz`#{I1!rS@mHoE(`}qEo?~ zle!(Iks7_BTa>O{BbRqt0H3eCTipubjeXomY}aA)Y%TuhP@B~`Lv!9WAgi_Eu1hgA z@;L(cnAAUl_-(!z5gd zAiAGS3bHV{tW6uS-Se}IV=L*|reC<;SLZ}qOBol@yp9f9CO^O*(kJt|y@OV?J;pto zEjyQ-1N_FE0&he!0xzMB8r!z2t^y!WY-`Q%VL^KV7$D$S#JEjJB&{fy<$dDRmvj-P zIa`SOYchRRwaJpQC{QIpMN!iMW8(ZdiQscORrF3GqASRQ^o@K9oMoJ>7-K?MPERh6 z-%@kSaJv_q=OIh-SS;6`)nGDw{wi}fH-Y@8SY_aHxYOy zb3JYEjP7=txZ;_#5gg!yXP3RHfTvGFsulOx{1SU?E-(LWV$aL2LOYp59*HY<@J!R3 zg~CEem6SiAQ;x8&ZOvXruGT!Q?|c-dBn}BTn54&M>_Ad)+;)Zhsz zSi_BY94<@Y7G!*>WwAJ1*+;F@U_&E@AhONB;pPo_VePVLS4Xdx5wPv8Vshd3tguh= zxw}Z=A$xYfGkWoU1dk$_AhZ2wSTQ1Kc&*br(kl_pWqbx(dqxR*ZVrm>t218h-flH@b?LPf07c=~B$jQ!ijJSvXW%UwVI7&-vI#h`1VUfzR8jdU|5g@%UbAq2a zG>2cz^aU8Iqv=B$HIC%QI{zCz1hF`!;nL7gQA~z^+YTHc&8eqqZv*m&>7^y6qcNZj z`h|)Va85)U)2^pCW0htM`A&6`EFOlp9f+I*=@eAv-a9#dTC-zoj<$hoaPKOQ0*W~5ul4%0}m;hOQ17-DfZZqZaOiwNs$`liygrAqj z?CfFmjoZ^Zj1I7AB?8k9t6r1G@wN2|%gc{qcQymjDO~%;o!cfsh^Uu`YKcC;m~LNN zll@_s!?^Kx7XvK~GHt*N-`3e2QL|Sw`+}X&LD%>HRN%OB311sYhT2HP)i^-x9P>Ru zwMD|Y)j?(^`C3|jtc`w~0w^@IuqhaAuT3&-P>_WzcBJ^2I290+eB__=zTK~W{BX;d zQamW*0~rda#~rHSl~v@hCA^R}HE_d-3C_omI&M#p&b7>ua-UDUhF$i^GukYA6(fxi z59Rd4f{W!VAoeWanzO;L|2x79q#pqSm9Lk*x&*skTBM5+;*j5dH;tw#LsV@GWAQ^W zXC{SF&_qm7g}Odmw=r!f-z(_E=Ks6|A|>xD_EIJat98_I#GDiyuCLyKZg`I_w22qQlM#wwgcQN2nQ# z6{~>#NfwuXN8)ZX7e7WLo|&4xs(&j$XIN5eSDKX7R9i_Y9w+K`E6()mIXoB5#`oO^ zg)+Z&Z<&b|HO0xGk;CW!87jr>^LFHOv{17OY_OtgksEbp*6FIymW9A>_@rYAJE4b) zo(>_{tbs|5ceKo+S>OB|iv&ng>{fw49l`<#4c)Plr)MhwbDN&L#ZyeKFI7rI)(|yY zY>6=$aQ0iMDpF!{U`=?@;+S_{1mmJ9)xlMdeaM}XM0FBB7+T9u{b7me?X*+Jklka4 z)vc#0;k{>2F@R`2sQ#;sIuhi{3_&K%7-^dp^iz5<7v`O?m(H`Gdcfo~aOrT2grpic z_o4wtlZrD>jWoSqggAdG7m4}Y|4JW7Ja%oBn5VmC`Niu~&u}xd8!n$>jj~)Hew@f(;ltKZzS${V>TV*uHD* zDinGEe_ro17BL1)GrP%uFIo`kqLX2J!%El~UROaHFMnn{6aTQ{8J#pe@tSV2N(aC? z#cA2TR0DHNRUYJ13E^hlSI{69^-Sl_pZp%b691jaDFId_3F7BzXYRQ!if z=@DIJ@~IEfbiE2a>}s~6lT}rI&&5O0!;_kz*#-{5F~)rM{8Dq*Qcw#V2+te(kKxsO zO|5xeFlynf5XtrT*EDD#D&^sug>YqD$+R5}|L@XLKz_^!s?oCd7(MdYa(Jyva-eCM z#ZFqL-4KWPwH;5#hJ|gKcw< zTtPQ@FIzkDLAf_Hebiyw4(n3`3pbDTzfkvh)z@r4J9T`-WJB`3UIb!QsidmY7n!-8B^J^W({9Sv#wida>696a$0MaxrVb4rv09*HaQy@XTwkWKpmT)p9HrJIzl&^ zl>=l_FyOhznv0QZXmK<-e6W?%(*fG{vZ$R+Y-KR=hZW6-PJvVSX9}ioDpQQ(S(=8; z!6H2vYNs5%a^J^tiTx`Me35x>nl2l!^eMIGK^BV@y3J4ARcK+#TXc0 zVwi`KjiIld_Ku0V6gT`V=LO#%bVhm?F)AjXt=dPaBAK{~W`krHmE|lCvOsfMIOgjw zd{r*$20k>L)e!59Lgg5l3RH>PaRlxR zfm}TqFV7!%HSQB>pdB6irTI1sT{O%Ks3P0kVp(XKSO_fyYN0oYmS|Cz`?q7EvvX7{QY zH(70t0F*lRXv0nF@9Nsx8JhRzPL<}ZhCI}2Q~S`is=IA#iJkb5V)6=#t+3if;y

IULdc?J2_n*%xT3TC? zozmw~Bxvj%PVfWcAVB61IlAg=RjBzsswLnJ8B^*vgjtGin!+6^^n*$Jq@Nu_3fDJb zd{;17eQw2=5E&dl!$cmAfY-5b7HW4)7#sptXWirIzEcWCoD3HWc%GPR46t@{{q5k0 zySeUZKKZ|sD^E2P9TLp!<=yd)_R$v>OK^Lrxqi5KXQGW&5Xi`QO=ENR{tII6RgTPW zk>CsYA-Q3e81D7E@BIX{x@c91%?`3Rc+vUrh`d*RRTh)wUn`k^pzcr_0}L1FJMu4y zp;Cv?eb@>+;s3M%_HtHc9%A=T1%JWUrxXe^p%cCZ5&xR5$m@Vo3L}gt9Q^wEyo{d#BvkT+INU4#ONcZ3O|!mPr9y(W}h8-j+3-wa`ELl-fiBKBSgyDiC`*zLboGp3_^;eXfudLr?s363o z1~(zlH>6I0k5M8l=rRiKEBqCQi4_yWvSE(&zzKjPT0x2sHz`|?8@+2R+C7Ja*jz0^ zI!mxe@*EOcHn2y8ForIYSVCxM7pI4pG2-9WjsekhVbjUzDYF?%`I@`%cwMED8;NIP zdak>>a(-A{ku6NzUX4}ONL&1DAo~!AHxVvfRZ@&k6gQZ5%>^LT1;u*)!Hww5Xji|C~nT~c*Sue?Dgb0pG zXZu4cl^T}9DV69(D(2$iA1Uu?6~S%PjwhW%m*{QNEN^mkodgm?I<6ago`G2`|y11CR*gY_BCw~XZ$W@4J=xU1Eui>CZ zM-yCPp59NL_3jFxs=lQ4vSYcoxYe4my;MdQefrXhJvq|AuwF7Jc?@ib!YNbRhEAvB zUGe`AE40=j3n)YP0pSW9$yZz!d8x7Kjs!20BIj3xB$% z_u00IHtC`B9YUcdUJFrqID`W^pQWq6EN5O7x!POQeh~2==>VSG!1N*VVa7rm z{HZmaWFFazBRTXWY`TzrmFW~(YVz>=BP8Bvn!80+xK@OzId%uTp)9zCay!? zw3uOh=|p%;VXt2vGd|=UP3NR0>WQ>>QxIG zC;F@hO^k|E`5Iw2S%V@OcKdCPTtzu5mo3pv1b9`p58v@YxUvypg}q;j@rM3z|Mf0m zznLKQ<21&_g9sK}J-kE@4d*;P2=WK?8(Xn0AqIbhPf(e6oKNkSMOyCw z`q{-!;uUjtZ7KQsHwA^+3NA5!m+Bf151)agOBnF$8~@Qjg)5VQj`(SbqwzJ$Xq4KE zJg6@L@%|gt>fbkpR$Px=ztAoLu`ES+0Qr%B=_W#_3E{boa2^bmSw4Hu#32&5235Uq z3Xy%q#DZ|jp{0i7r+Mhi(?u|Qbj_F~`yX9vTT%H10}=~8^3KC3YzNp6;B?rwW38Fk zL_RIF`iIhw9fl|~c|fq(Ck=2S5JO3k_>gGT^KtN|OT{%1?^tm(ExP{gL0Qgm>(4E* zd^>Vuv7m{(v>q(E(oK_>Ga6S%wS6|OTdjs&>*wIb@qMrq60pZ8np%#IoS|6p} zOV@D3Q;W+*qqK4)hlB4<`mWHfe&G zA;UK4uWCQ(3vlR<1b}z}F+e5S8R*22%eLQe!CMsOF9qnm($=r^-^bvKWm0IBGrUpP z#r??SCcfY*@ws4?Ir_#+B^P=kF3Q2lGP;mEkPGjncb3XAY`B^Fx*ibrbQ?V&^qoRZ z_XQ>cm|_ldVmqe{TC3r>N!~t|+BrvJPjp?Mu{O!OF%@6D2HJ~@O&n?BNEkYrGQ`64 z2EX?|;r#s|^>$BGzk3mQ&m_3NVtdx!_Wjk{(<=_Y&0h8YlwN1NQFLPGFhv-%qpst0 zNTC8~=v=_`{!T~aVg1KrqTV0-53CV2;cg#2-@T4gJq>NjHCU_N+l07_WqyJ7q)Nj4 zW^6fF+LH4kJ7VoKG`)y3b@2&Qd^6|Sqad^OCG)$od=gA(s;a7a388K|TJN-DK{z=I zb*w98v?Jhoj~)d%lizpts(Sn->prcVK8>rVXGx!S1wJHZX(Z!?0;?&CY(ca0_MQfT z_x?jowk%O40`LG@ns`>wy1&4{Fj@UsMCX_~Y`R)vYrVW_FB z#2)YG1%w$5S$xu<5{v&MWII@9KsDE^iX=}1ZLW_lGd?mytd0|K=H8xFBaxxFz;zz3 zYINy&Q0>O}Zq@4B6!`{D<~u2&2(4RPTWb8?T)PENep$w8rU9GQYMXbErwxvuXF(?$ z(puvLpkP3_57VvmCLPf$dN*7U2Hg>ftBWanyj3>JSF-q*H#*jj>!Jwm5x26Uoq^*h z>oL)y)r-eD{&tVx#+ASaIvG-XkyM%I<>|tYsWD_;!lT68`#cykj!yMQWsIL3_vz!Do=P!|QLOr9gI)rSQD|a*-5LAH1{ODp~&Pzc^;iw6@h* z6exax4y$Je{DLgP{zW`CR;h`(p3J(Q%yRB@#M-20r7BfTj?{cu&b$Dek@@GBC&w21 z5_1EpR2tLKXxdI&Fj>L-4->phTNrRS94b2!Z(=y>AquhgIp2%g_wHZLIICyhc|4E0 z@FzR|3VQbzIvq4)^L@tGi7KaHPxNOlf{*Q^|D>iMb{(wFA(u7Tu$sc&KvRIj!FubK zDoUY`6p{OnRx6n5&7{XL>HmV4_jr+it5+Z6$_)SZQ1p5FzN1c@QTI|*|gAKJ5;JwtB z{hd}5k9)>*rPHs-BQU5sjwNc3pOk}5OyKf@>&TW!8f%26|L(=%#_T0FT@f)P4Z6X7 zcV-utV%Zz?@)G#4N7fT`apuNZ*NjovuarYxRs_+#Cy~6h93Omh)b$b-HE)V1#;zXT z@=k-|nmgzI{mS84adpw-{*}cG!i15X-hX7RY>kc8H>Gq{StN>$a+Nt}@^v`y>e(rl zCN@yFd#bCuOOArrTU&)La<(Kgsv`K~i zw&BY=qz|oOrHqkN9x@-LK9B(bZJTcV$ah$Ox@{|c^U{*Vqmx$b9e0hrp;`t0kt1j2 zpDpc~BeF`I)mPMR>d@`j?3-1*eDw({zFIZk%pfVR(X^Hg4_Ro1b5s_E^y7WwQyH$u z%h&657n5R@6&aOcv>!_xlh?5?L*2)o7)opo*?l}ZQU`B)$cxGDy=0P#T93sSD=wDC7O0F&WOE~zwrcW zk}DK~xoUY2_(3&V9^o^L`WWsB0W+@avn3{FW6zUxpf%H<0mUZANeohwqpzIoCxykH z-T%b+XDqfOx=#@P)|b;8Wf3*iA+Bg*JhP2%K8Xb- zLKFYkfbl9t3cxpnOgDzjQkR3Bepx~%9aLU49noeSMhgSnhp%RQI}Xj5yY1*Z7VZSq z35Wi{7mOnYemw8-J4F8hmz5P9el=VG6QNsy<*)kdfMfCzz%6@EP$IBFg&;`*@k_b5 zL4}G#4oXiGs}7ziQpqMnraic@`pY}!!;}ZxkfUi<&3tVvWH^p^G zWv#_a<5sSK+!nSFpfx#gHPdBIsH|<*tktYm->q4n$MGvP6n2GUZInUXVO?v?&}0Bk zau%WD&ieC9r5Fuu-c!hqH9fA*(YcC&bX=EZ)Ky+X=I2IRxIox{J-!)l**_E4Uh$!l z!_Q_ef3mvh$}6Bh^AFeXpe+m@Zm{5u*HcoUdBCRmurI%paiBHUqZNK&Vw{Flm2PDm zh&ORj>mn5H51URZgHWX*^IUelxwT?qIC%<)_ospN4{ra*v{v&^)Fr=cm^=?WuZ3{U zvDcPJ8NYCp{RF#r{IKm{?83Y!g#a#KOOAdCy6vaQ38lNDSUfNybQ52m~rR#J&#y&$A8xrUS|7W4e zOcz(`^oU3^E57um+?qJ2?D4zp9je%Oz$M1_a~=MTOw;&wuZRay0gunKpFSEn*rtA1 zcF>!{_E<10D6rY0|G>{t2(?cqrU6<(F9yaxA(?lt+v&AO<^rXXva~br{!*}eQ*I+M ztjW7_B#Do_+bp8ad)J=$z~`7eTif=P|#=doDI4EWhAkXl$H#i5c2{Z1tOZtJB)W>l5+&{@aWZcjDXxaj0+o zQjLBkS-iZVp<#k0nYK1JS^Uin0M5RuP%wkU=yBV12nTQqu_F~1k36B;vRQ>O7kudVE#P0Otdls&+(EGJVuSdhE08mb2(cRWd8d8)e;Cb4(%aw$d zJ+XfokykMd;|NL7n=!xwD$X!ypsCER&!W*;D7%Ax@9D}=v=zh>KrMrD=_fjsl7Ou6 z;$r{BE|PsL&%W%Z(_%u8SrTn?FRoBY#`jI_*x*%})m``A{Sl1(X@EI<>b}NX$+`cZ z&vgrKi*`P5qVXM+hHpgAxEbj-8RaxkQ}9eFX4KnL4dn=+O{Zf-^zpMmWUmhLLL+Eh zWG0MZ(Vkb{XrZ{Bt-c+^qHx=`Jw=<$kmODW4I7TA7M%6Q?LRl>e7DYfoivOtYwmIX z+W!aehT#X7cL@9Ls^DZKIHfoI(;)Km7E>w#T`<+1(Q3H=ZHc)y> zvFfZvwz4;ZB|7Q@rdB!8+S7HjVnLpReu2Ub6*fQlbk*mhCOhNRe>F=Li$BqeFL-y6 z`HT;+H8}6j9PmKBj^N*kI-amA$2U!kDyp~h`%oGV&yJQ>Q<3Z(f&@C+pBDA{^7Rt$%1$uF#RvjH(J)|Q8TWlZwu49Z^duC0k zfK2&i3!!|B|9LF%WGUmW%AsIZ+B#fnfpy#F^BVfEpWUK{UwIf$Ee+RSbeJg}agB$+-mMztQGmUI;(r9qG5xGE-OD@LOy6Psq9yMm{U4@s%f6K>=^k zD>v>%Z;18?7&*YO6jB5NpgAFOQTzcKIntQaCdNE8YG&3z3KyPqo=lYvdNvr1cHeJB z?}=J-q#q|Awu>eoBDabS|KC2*9&vf>J}z!!z%e1wRG!qV^Nk$ZZ_;{PfFnDve(-T@Z+U?(Ae=YHux2u^8`p*aTl0}4hD=v1}I}RAAXvspW5I*@tOJhGpL`c}7iRbwv%VdRusDPsFuIuTXE5Bx742^> zcB)Yhgb7h!m)W8!Ctr8qG@JZ-M=v!`?Qs=`ras*ACAu#*M~nS)ho!%s^%8OWtq(on zY%4VNUCc!_Dfk7euQltPHEj52@FE3X|H(I=_2tz9;lbavWVYi&YSb=)fl_(-y7Jl! zkK4@Ir*{{Vw#&&AAwM)aSOEhg?lJUVe8H=Z?R0C#v&+9%%p#9Fk7j@a!b1c8f651Z z*U-o6Vvu7E;eLmUTszBbPn)D&SZc7NXEm&l_r6}dTO1sTz4frK5iMish8+CHBlhhk z?pgpnuea-a)@{u+*?Q^fi5rP=5#pC+g`+B!_4trKPa@-p=Ch}%TWFYa@czbB?p z#)ZXQM07dE*c^EhaNFw7*IG%o@8GB^2Xkg+#Uh<}l#*zU^qs`AGVl9Kj8#9TdSM46 zH{ZaC*6Dv@_X(3$Iwfr}#@Rjj51SEhgVpf-nS70#8Nr=Qz4vVDTcd8v72oML)^wY|?XY=dVvX$#9fTQm0 z2Q7GV*}3!(N>OzMRh|tF(kq@qa9tvFWDCwej@GV;+r4Qr8i|kQJU)MP`-&%x&(0l* zM5~OXUHG?T)6T-z0^{F(bbfOB@L1ikwph>hJXOyvor8#?AI!ZKsoN9ygbNm zc{+MD488G>f-mn~Qx!ePtfjd$v|yIbLIcZiS_#s=%Qw!&)xZG9JhjBqHjdq!sP)^{$64mRB91J3YX4KekMfHNzN~& zxHvweXKN4WIHW_{y9vd?1_qC&s>?|EqqL(6b$uE^K47vFy&nfw;2PPBJN%7Jbyhu(5* zKRSL))mCF~{1I8FYT3A)&_vbfVV`2RTx6Q`R>{eSFU<|9zBd!uymUv)>tDu;Zs))p zJ1!UHv@(||A5ksVny)U2D zS}4_(Q$UD;a!CcO%*EC=g{82xvgPy3&BW~Ij?D|5?2F53zlzVIXw;RcN&aE;uaQ=V z8vqGX`{eQ0LDmcmiKgfJ5ssgmIf0&5CaWmrEWorTZvIRV_7s=np6Lg_6<<@u%&)Wv-dV`jm}~<+mKDGEBM3HRcG*B8~YjE zsafMr(IOti+Te4;L^oq=R6YE6(tR7Ie~522RD>yc)hvLP7Zp;36+E#jPf1CFouUC*BHfwCq`q=68(O>_NO6k4Cur3qj9 zNtAvZW>nkP`pb~3nQ_da=Uvb>vU=`UEboMTT9V?#)`c5&EIXH~M9weL=N*KxYiosX zoXh&=1@;1rsK;|knvol0MsKkaPyP^s{48dO9kgD5$jIx^LRnbRU=iSl4ay5$WX6;G zgcNd4eb({_VVdyMcwsBU9%Dc$2~4^6-&ue7sg(V<6@~FE?$F|!g?z)HV7vLv|G5MR zA|I~*b3%rOt1WkNCLfNyAMzuF8wlKRY9v={{&<$%yh2cDaqc@cbStCD7%IG{z3dV) zzAn9)`hf?yxj%cF3q6cAVnF*Wd5SC&`@)znHWWpi{(kw(2opIS|HX8Kh>^J!Ljm`% zGtyFQI)FWlCoS%*=K^V1BD4Zop(!TvP!$94bUD1-kCTBai=J`m4ntSxhto`+{BJ&LmGL2& zyH+DNpX-S3hY_{Al>Yt1IblxN4-^o2gwCn87kU!{o9lRj-oqHR$C@9J644)RE-pV3KWuA657Z?@JoJ`#&Q}|Am{U1a= zo|?`=#hRm?$+ue8L{QB(K2-O9k%-8#cC(j*A={=F)Bsth^wn`C{*Abke+r?J(Z6Se zFQjInvE-6IUTQ_)U!e^9s-7x*J<_pvQ(qDLy|)!MsX0DdiUj#$ZgzDHSC73v|18+T zxN}$KNlSR|%l6awk6C^0p$HPq|LvKPnL%4ayR zySzp9WGP8xmfGA@J3`hYLz)Lgj0yu=%xY%z?2Hc9XGU5f4V{r*vz{%u>P+`VJ3eur z@fXV3KR;zyH>|>h7oXD_=gk_dGOQnZzb@$EJJlK;$&jJ;3OukftKi3QH#i`zTkaFH z9#1WYDu!~q|I-4HLytPQabsqMn8n4jXBvtLWfk>wv^*}A+0&ETYPD1-hzt>^a*Bid zkPAhn+gr%4iLj$rT&)^yznOdnO?#F+{P!Dw5he=;twz||Y5_p(W4fydk65)q0nZn3 z)CB$kq<|zAugjct5v;&N2}H8P4o39ooum7-y`CS4!b@}yO|p~Y@mseV`%)?y!h{)~ zpRmhgI{zQnDiIkmmKkzG*@1!Z3+Qcs=EfG$&XN?*lSB3VUhhb zDfa1CQJUq-hS5-daeRJI=ApuDsB(6K-S|K_V%9N5gR0)H^gyz;ytT_HL4~Kufe)t( zcA{J9h>IUHV;}9nXLrrr$Pb3RE6@-5uad9ljiU~mYO$u_J8qsxsZ_K={e}v5}#<+syu#U`n%L5QT<+xk0}L#jHVy=lxi$R5w$@@fxK2%y4ZV+ zIrD}PLdx9i1q_l8`8^sBeWLO6(IUCG^#<*EO!;E*PkplF6)GbhvuJ@LG<0YddU2#` z>PBN0Jaj6FEFY#l_>huh4xDM>_@=lio#yB5yYo|h5}Qkj4_5Q9WRQUrQXmZA{=({C z89-6AsFlZfs@@NiwK`EUsgI1QLy;%j8wT+`gEPj5|&6Ft4dIypx)^m2SqhUqqAYkhgLP``~W9hKFXrp zRxy2ICZ(dwkh8+2%q>6d{4SSeiM%%Zx7yP3a7y;ke%09Nre|~?xM_8s zP8NM#nx}GY>OIkazInNY42@$mr*0xS*fKwC44^q&qX9b)sT1a>KmEtPK&V0>nm&=J zB$Ze!x#)){LQ^Rh6PM*N;lVeLhssj9{#(t^ zkxam6-O95JdWGj2#L#-9hg0rPx2bY}n9$tK&WZBmE-ubJhUP7Lz{u-vjU1zv7#&%p=DOduT^GDXv?^K5;zv6BeK`W z&>k1Ra;}yxmVu{)9xI@a$@{389k=;Vv>~~f)3+m=j=z-6Dk(8Q!ntv&W~KRZ-fXDG zJDAwiB1%Rr#J`p z*bDUCE0_?GX2vx)CYK*l#Bj^|TJW~rezsI}#1c%$?(vJNlR3<&Iz;*;mO>tESn_swX_*6i| zbCn@q+jgmZ^tg`%{v_9LXC1x1HR;?`6%%?u1S~`M=n+Hmzsz}8n#X- zV?;Umy^aOU5S0T(;h=5HQe{~hW3=GSu9t?)Ob z2->i+mS=kCwcyOgl{y#T`<5BP5TPc9V-Ojo(UHU3-0dxMrp<) zC5)ESE!aC?VB3fCv)wN{aLnd76O)1Hne)|N{;S=guoLruinS8@t^zmNcdf5N*suR() z1l{WT)b{;9eOkj&@C9|Kn!;yXRKXSv3yG=50x3jtC>a|Y(>A;YmJcNQ zXiG&H9(&p57gNNmZ&#%KohpnF)se3=p3fmVq4aKlIjlNM0$xw^{L7ce zG{7Zt1haw%t^c-@wn0i?-d)`0r#m~??D!ku2dr20Xfg^~Lm|d8aY=e+9oSb?%&vM_ z$Oo)VB(805(RZe6@zw7xJakh`wmZ1=xj)~dCo?b4B}}#VA#NOFUjD0eWOfAs8dR4- zbZA2Igs8ciHg{kAj;;?Vvg=<6m6l$&JCe5T+NG3TtMuLZj#VAkPEb{=;P*wITN~{qXPlo<+^s?zo*X z!odA61BZ}p2{nh>-R1Nv2p>klEGA8l@4Hocm8>+POf2yN+)L?rQVJO9E2Q?S%u1wG zkAp-p<>4OdTam-v_8CJ#;*qD{ZawRGVEB!xd!5#0z~n%>pb#Qewb;TZ3>vlK`6&x7 zWJeYjyyGhM15PMY_h;(nmC?f4!4V}5YPCs+Bt{0bG*8ei;G8}Z+(8n*W?S4ahAsp@K(M0?8LpcW#zkd-OkOY!I=3~za?xIhv-jkE1pB8#ZPuX?IZ zllepCWa*wc#~jvIkQW_RCQI3J{&^n(6?xcNtKxtp)HYXf(edwbG5n`OiMu-vuQ9i9 z=~*n@xn~9k zhVLG?0%ZR0EUQdQ5Q1+S5*FG|=3uide*8f)<|yqo3ZdTMi2cOQIKTe;u=SEdd>TfD zo~cHx1<|g@@a&5%weHE2=al1cuN8!S4&EF+8GCXrw4&Wz0L}F@pa_ zty8-E-NIlUsTWf@7!vDr{9}n9Zl(4I@E$vnw(z7z$r{PpKUx~d$LWSB=*L=I{5xW0U{b^cZ<=~e!#d8UHX6EC*PSt?T* zvIodocAsS?7>3tR{ z_7a_Y=6^OFUdm;@?#{6tzHSpAu>NWfyz7238m(JDI7ln99Z~-3T2)o0SD6fzjdL5> zs3~Ks>yW`s>hg?H{$OsNX-(z;0~4dup&57is;+ruRD2fm_0DdxP1R!5w?l>pJwVF^ zvQ?1fULF3(d`SapG2J{_3*^(y#pSzO){+?n7oulW`!~(187}d(78HDy6?9Sk;+K4Z zp`AK=^$M=DU}$W3x>StRPevW%s_Qd{u+MaIk2&IvNUPtCUt#Mdn?fbawv#cZJaO(h zJnRXJ`w*81MA&zbf(>?7rX7?0q$E)eZMwsS&kX z>Wdc*wZFlyziRI!KtKX-Ad-?;c5Pyll|LD`=#gqNH%SR4V_35P;$ z`hiHau5~=u^r0yGB1?L=x#FXMJI$f_6GKHaHh6QfCP1k8gm}EY5RGc}re|_DVemR! zduXjs(1|i%VZ4Fd>9Z6d6s(B*toogB>*0j?G11}fK7DARLOxEg==S;ym*M?b3xF1= zgb53D-cuZRdJ6p7|AcZJ8I7|vRi776XD|andlq{9g=va zUe$Y_iuQ2z@4`fz-)Ysb_E?m9P1Ka0h<*%1I{rr@-AC2fDzCqT>Ne6=gf0c`kykcGaHmNs1@n-XVl$0l z^u+siL}-v7QjRJpbN+ukorPCa4b=4&5JBl~7&@i9q#It+SM9^Xm zOR1=-iQ+h1;!bp5J(GRxBlo|!4mJ9rh=;W3#ZpVn*bP^IGV??1)w+l(K3Cxe7zaW!?rq8r&LPJls|uBH!=i>6Msru-|ftv^&cJzi>%T{ z9plSB`bfOe{Vu4WH_vUrl?E0<{9HuB>IuQX`7{^8m!q{QpDD#Y(J^M-2w}=9}L07f2=$R~yXyk>v=||YommqNk6}f7bN&%*B(GBOW zy+NqLXRWim3E4*(-oS_E-(6PXxvs6S{r4KWzBOJv0rRP2B^^#)0La{_dF%X!amhcv zXW;c4!nd}pTTBr(pStY{h|ahz(#bI>qg87pN60?}^mU@7Ea;h(;wXxb^y^QhJ$JtJ zB-^V+@f#)CtI*(LKY?frs30wA)*!NZgX%HI9rz19pNf=|CrxWHlvTsiRY7*Qcoy!T z%LYE~q5qn{2rvfGyCGv3Rh}&I5m+cvr#4Dh2<~>U3H}pX7t58jS5(&n+R7rT_TDRb|gM>a~ST&5P)COTB`) zn9lW=L1gn>NG7nqq~{)Erf)v*hIjU5%B?w(r?_g#xG`^f~{ zWWr!>4~dWWL0dWxR|8VF`v9+_6r$<#QDC{_D(3W5qw920kZx2UG~-N3dD}@R0h;&4 z4olu{8SJ0y#A?t8Mj4nKV+p5%-fcfeRfZ zW71e?K0V<_BPyIWoISYYR#!6G!a$Thc`HkU8j$%_bFPR8ArRtM*cALHE@q)j3u6|H zS!QLG8~Vi5FWFA;do{6yH`4FGH=uWHOn5kqa$wZ6uF>%Y=_P-kuxeJ#K)Be7+~%Q) zz4<2eu43-6rt&DHCa}dboIpjeK@7Z;_`z-4qWESEQuA0o^FQ0HY?QA-!TGsIWr&8> z@Iev#F5*zcuURWsO6GCRS5)$lti1gek5Eb%Yk6$w)-+9#V7XAb%lV6||6Xp%tA~hG zbD!4(M_80^IFoMBeS{EVhERZ)6m9-(R0u=8Qv-H4nFCJx5mkOXhgQ@aPEr4wfG|T! zHkc&oRq^Zc#u$i3QkHxphZ{#KkdSNRjyi=z$1X5hNPAx(qyy-aTK_G(_J1}btfK#c zk7gm%I@5R`EW*-0mfpu*Xp;Np(6<4)E@uX;wa9>2ue?6r`S^w zZ}%e{5MxIvwJrq+Lk{_^upiZfbsDQo}Y`j>%Mov72ZRJH-zNRmN1LNHO0cC18?LY!C*+Ay2m zSA|t3MbYn#Jx{|lq8eTi^B4~o@x1IQEMD#Dyp?Ku#0<4rL2Mk`w$Zj1jp+ibH^O#eD7_K@a`{6Nu<3y^5c8?fb|R<_3fg&C+n@3Sl&o zXC?=}=^hE-^5jNQ-j+p&7D{G4;ZIPW-j7+l22$6X|J8BROBXm(chF`|aC+V*#Kvlb zda-BiU$)n5o%Jw5e(Z~gS}OVA5Urp*g>I!y=ixuFdAKWy%5hFxEt(N7zP=J7vSgyn z3h7}92m9xg8RQvvJBfj#8v}`Ue2X8_+ z4+4yT!};Oer;L39z&NxU5Tle-_^PUE_%|Mfur-lOno>9eMFbrJWrQSktRxU&l#e9{ ztw?GS!~8N3oEA}o;{Q`}g-c&IZc)8%o~!HZceBh0gGkZS`u}}B$|WzX)wZ%qsP&Q` zTFf}%>MKdmQHYX=fM1{0wU7=dsc^9Pm=T7uP0ZPBX9K4pYU#*V(pq4{dP&CQQ> zn4fpzI{jTGH3c@7tyXR{LkwXDyfHQljzdL?U^pc3(aYZ?n`tEZ)5a}@U9%7YSnK~}r8#zcg2z_l6!HyB=zInzkPD)j67b()J z_gTrEE>*k%s+wwJ&!GriE{8Itu*QUG5sO-eC&_4VejV3rIHzNKy>duoxs_H0S#P5a zfKjqF^FI@O`BZ+&D3(7$N+Y3Bo~;*$mKm((5M_%Uh98dAIy6}P`kmfXoE|oavogha z)~X8QOEih7a4Z=x+FA_L~qG z$;i|uf$I>WcfF^)`Ms#;FBH}(EIJC^5L#~#AU*muL5kv&C?UWs%?E9>b$+=IKVKs+@Wfa(vet^q6`zW z;Wy^bnu$Kym!3BOoQ(ga%r`f~YaHL6#ZcsMZ>PX)(p}tOGG=T>x&jL!iw>ASow^5% zBg^xO>?*&;$*UWJh@JdeAKT1||7sm?5F zQiwZe)E3Af8nLv1@d79tsIkX>+E|mM4)>%(VwgJD2P78wX&^h74;S0)rk)-e|LI@b z58hd-)>Oa>A5RL1+y@x*iG$mPLn*iU{=d2$>)E(Oe6)ok0*iQWd%KeP&ZNWKR5>Ui zvddmdaf&XL)Q^AS5MGD7krAS%viz-*i945TE>t`zTs2P`<5a7`w%5&HKJ7aNIPA#G z4cy>z^{W!BOS{Ehnt8n*TEFdQk0euGwi`^@*pdu;+X2JrTU!}49T*lOSQc^Gy^nl) zdW7P&<1<%Od_(Au6TN=?gBLeSGFuR|N$7JG|9lp%bQ-3E?SjFNU+ER;D6>zLJTu)* zQXVUpA|aappv#~v>Ce$0z)HfgzuT)W&+@YqMvebLylsvine(|2&-V!j2IC7$OT|b< z0e!;N(C_qU_a9weCCx`9u~pimq-u}^M$wlcpH|??Dx&UZ3cu4>j9&mjyqmlp;|8hH z#rF_u_8&j3sJ5R&`Q7na+P<4C7uKgcA>LTOC?pB<(i8|IC*oOa%!A`36!VLfV{tyK zM|tslom!ZyUvwS{*{)sr*g4QQXuXtbL%FO}!QCq{GS){8pb1Rh{k{<4UcTKlr&La3 z#S{g91@k!`5f%e{%p%H5glfTbp2(E$AMM5>yVI3aC#Ra(EuzgU4R zANZn6)m*AnkC_f{9*ZUT=3td2jaoqIdG?h)%Z#da--;NBRW-egt|9x~@fwFmhFQAY zL*&3>TP55W?#Uw;HzP#ALf&jRAAaFjAiK6A1Un5upG!63G=l zR3Gp{zqay4D&;>9IWv8sc3P0KfW;&c^ip&01Y#?~_0@DT0g4p$3#Mpa}-})i)yX0S$ZC zL4Dejq_u$iW!0&PN$oYV`&xWlT*l`nZMUf3lCtztn_~}K)WE=`wic*5Zct4qweCom zwN2W9NKSE|O39jS_wr3P(Kgv9{h53pBxOTfEyN>~#a-w%WINQ?_`N9FJ&++h{LAO! zFQ9nTobkyhO`(N320d+!U8>LM$wXV|qZ}ZNtb-4k78q|y!c8%I?!-X@2 zLkCBSJmQaAR8tbm{7Y~cMHRusUS9$N+ovE~2f`c4!aWC}Wx?UWNM|pN(}Rce|75M6 zxk!Z$d_3Rgs~J>+IbZ0zm3`s?+RN>BT)cvSrT~=ap7wWP-e|Z--!Y|%AZOLU-TsRAmheN8Vlk=Y~2r`vua+cuV`EPw*z+g@ld zS;44L!rmK9jJAyyT7=#>?%L_Na}W$qp&9!|f3pxnqu9A8@pl0OkqO~y zj}@%grOC}m^0Cb+;-aZud8XDKu1WmXQot$-z>M>OPJCzQsa>-�= zXHW7cT_VbEj`&SKJkO@&Cx~XxjgR^)fr&zNk{r7u>E61nuVcrts5)Q9a(aLHg3l$O!j?zz9?18_w`rtc zRZlGiqfIQisp?LJ(-#)Ys~>sv^k^Rg0k7UEIj_;UjZ{Yv{x<%_g)})i`KHf!*fg;T z$wF6e8s*&c{mHimdh%f=#!-J9n%o1mYB2s-UafLWY9eDA*y~J3_;B{U-p5S)0m9P< z(|Thb@3SMBoU!x+&jELV9`T9ua~AoZ*IU9L0?$eF0hYL`aV@tXiIQCV3b6tM?~9u* zi+V%wtZ?t-vDo%+mMp^FFCLpp)Vrr)FjE`Z#S?mtX@t~B_5(|eH|;nbltDZzoH<6i!Qw!9i- zmS%-TV>nC=!S3NYA*}pnlk&ENMKFSKBuddd%3UVuWf%>|wTya|N8>GP!41E1fbTxuHCbAtgSc8ZcDqz>RLcADIbYrBaE3u$8{< zT&$+1^D3PcIZ9Hrgj+Lw& zlvi_6uNy{%sLIrDa7H$jw^Kvznpn4UtR`qVqvM^Q4>QZ-1ueKotkiX<3ivy|ZXt1I zif=Y@^YVx;)LyNd7}yRn9r%Ur%Bduu%)pqYxefp1r`oDiOq5jo-N|0M-Vv^(9Tn1r ztoy8!CP4|(bluCO%Cn58-^G-KLWs8g+Qy;rC6tWLBDizDnkC8d4Ts$0UZch9Yp&x? zm8}hG)NE<2z9VB!Drb53eBV9#&jYnk=q`U-)gc}`VGrs>SOlbu5*mYH0<4Qi1FF-P zVyMJzU1&6Gq-Jr$`|N50dc(HsQd1PKfszRfQb>;YqXon#=0eP`8?MJRZt#Ghm|pyv)SaN@x*( zH8rS9!Afxbj?mgIf)dAZzTTedLMx1xJGKw_b)_i8cU%M@lx{^are+YBy`;?T8AoX) zY4S{!r^3W_<;ylt9aDYxgS?vND$yK8$}t>uYE5g7fqt(QobaSKBMo^1Ib*hngmY_! z6_4`zG1p-zbXb^Q40v*o2MMZ0HG5}>=tf7s4l+QJ2A;0eiq-SVc_#VYEKSzt@x*`$ zf0+T}k&%qr!Pw9y$DKC1;t(&VYzwcQMAl|JQcRA)87ZMT=OD^9NpZW^PCjXRBHVZgsF}6FD_RrjDDV)P)gyVJux{U2D z-&@8or^2;led-AP+tqx*@&Op zm3r}1I3f9TmxQBlCohYv5z_|FiOtfmD@1ucnr_Job%+W762Q<#CunrEuk#S+eVnq& zEzcso`W+WClUdu~o)CYU+fbD`PHytSlvdZO7{kLzS5H`J)}_^~G=RW)o2GZtEigb! z7_*<+k#G80Ckf=8u&Jpz`Vhf(5=a8N5;?#hU~6oP4!N}T(3q7H3W&T#HMjn+_)JB5 zSk)gG;70E3c--fVzWDdLqkehJzE!fUxO)g+$Qgx|9M(Ncy~M%Gpe_eY%Ql~AShg}h zo;~M07vE2H`E{!Cfs^=#vGDO*&HL^fd*_^8LKAf5;jQPM?*w*}j-C#T_PUTV3VZ4f z^Sgq+B|i1vBzTi2?;;t|WJC4#OC^|y31m$*e^IBCklL@BmTp^E)>yFU<#0iZFjc_` zu_JB$>)Ej&L_#!UoyaZ#9X)DVbM4&9W7tyjnINwn!b?*HVdxow^@MxN z5L6Ms|J1IdKmlKWTt7x~pz7QPjGIj`!inNFRvZ}PCoWyF5QW^A5Q=FGqZIv~joPfz ztHdtIfZf+tx7@RKVOsmB$jNof;8pq0DVphoVK!ch+EWj+oa+PF-!Jnhq&o3&#$k>k zw$(cbJR|@Bta>EJq_?x6LKjXpsv(ZytGPootpTRu24n6w88IoCsaBRsG|@-sJg}vPr!ky>^lQ)T*h>!J8*r%C62j%`qBNGw zCHWn+PY16$ z<5%KFcG@~RhSmBD1>65_vMB)fq^=(6wdcU+A4QJu<=;U5Ju^iBb=90rN2D9K}-}*M`xL&m&KrOu{>*hn^!_I=V^_j+xck)7Nam z>MYEynRJaGKdO#F$q12B@=ESFmRZG)@2#hO%&X$$5K5z8nAEK+B!vBpiJ>1w%Ikf` zbv|@}F8m&lLHDw!Dvu{YtJf<*JvZkopPrAjEuw!>`tESyJWc-fbu0-zVYSV?xT0oF z@3?px2?Yxw_+y%C<23suc-3LEg1*JrnE?F8=xf}hl1>)oXV4ezsrQi#>%Bbks9!z& z3jjV653fA(W?^(`?tjf$4QLD-k#U-2kX5Ae;equbS+T~7 zw2146(?0%yNrXVB%deXph6L2V-*?MoMUApYc8$YJlzhDs`5@b$ZHgX{Mc14BDJ}z=e(EIrK zj19R)&&JFnc!2u4U;n2XS)i68(No3HOO>&3W~e0* zf+>ZocpRsbE&BYuO1c4#Q3w&sSE4gh4N)ADVFq{TYWONPKbZV&12@}3h!!HrE}jZD z5>7y3rIc)Cvh+)~~A;B|@cFN!pNCM~t~h^OLE zl^CI{-CPz!$-P4or5YDG!upR+wei8PUmm~beRXHS#fuRhMRF1~Q+$jZw+FJ|Y8(HU z&P3p!ngP1$KMnaKho<+IIF?pW5fR4$zJApOUEUcF>Gq`v1q`MAK)FWrMKS9MYquAA zKPe`xH2c+x4P$VLk%qO-bwB;aoObfsz-s< zBiB8vFsKujKF^!y)@IDl@z#2=%Te64I&+S6JHueq`Ymvm=2H4ZBJdW%ZT+_qy%k=S z9&1_k2`6VMA$GoYK^T`eFKEbLLJn|Iaf(R;w7XFsw@#~R@xZ$!n^7`1vu^+BcQLc~ zk^iYq|Ndp~&?3(PiB|=mF?SLvWdSA=v+=n1?<`GfVi9$5cmClXw-k7DJZxg; z-JUBos;su%LEKMlOBKnny~ezN>O`!&eO@0dit?_sSY8$>ste~ z&a$XeYFcA0{}*~hG5f%OqXj1ZR zDcgRIv;h)V_BytE^?82KxPyod1?GvAjXms`g$qOxoZYry%V9+-xHFsyC|MsmcD>wbo$-eNO?1SHM>f4!h z48IG~RHM*um9ehE^0tsrKO3( zAp{OSSDWt2oFkw?+~+yMY>1>LCRP}_{N?mEXROrpHZiu7kW=<@ zaNfRTaW{6@UjCQ=)zWzOGLuE7s%F|nFOs!0Y)RK_PmY|v9{=!Kq2I{uXld6Pp^&-v zOqi}_ziBFu^;!tPh)QANK+6^OKjWCQdg2sf6@RpKExs{GWik&f4`lXWbV>a1a4gFYgKnAX6uG|W(HS3bcN+DF$1BxXSfRbrcq>GCh+@~k(_Hu|G z0?6rAqSR@z?@md#9D4qpgtl?>8KuARvnZ=fxewH_!dMBoA+Zq3InQpZsRy}wz)BWT zSvr+)vkynHyj_ZFhSljfVOy3z*3Wv-8@O%U0+gmj97L&yD@q)*H#*DM1+-Bs8)J)< z^BLsa`fQ`tZcPZ>iQ3xC37eT|BiGn{No^qY$omg4gwD;277h)QY@=;wWB_aS48`A0 zS!eH~RT(QIn%+&DQ}NVA97ln`<5~L~cK#mMfB>!ZZ82m^SC_d=?tk{NQ@TQyIi++M zN9{_-+RA-<&`AEnV)yD=;lRFUyL&(ei|o~%N{5W(#^d#A_eb&kQof`5UCv=k)#4x8 zz9uegt7b)R@+oAj5SS1E93cr$V&r>*X}^ei&+73Nyh|g|&#cL&s%m>US#TQaE0mMP z?$RNGx7EouCycnKga4h}ih!vkW@a|#vQ8Lj$!u9yw_HU&4(fRFP68(zqg3hZaGy>^ zir%9o)N%d~Whq~b;nfVTUw3p5w?W|z((SSL$CA7+wK|%Af6ofqTUbm#P?G4z4Uxxa z(tRr>wI*jq5cyQFdPl7pO28!A9#Lu9yW41OX3&BKx+1LCScEE5H>ROhZ16YKq+0ZY z{zq0f3?lHL#(W5*{Z;nt_^53#s%dFA8!C>!72uz(97&cb z6tZ+Lvz#{HTzOHq)|R)cyrLBH54-1iva2F@9r+=M?=Z?Mv5?@eLlXunt6jRGiX)zJ zI;CScif9sq)=e{-7n;b#(4P0JS+FGQjZWYc4Lwu; zxaAhbYZ|77Pe0G4!S;>nW$G*VSjCh#WQD@1M!5prn31Q^pA9ld6I4-{QG{XNlYCyd ziiB!XLv(PFAK!M(N{tZqO#caiR+TkC<`!qLd{v^J(la9N(+SWV^ELH=SHQWSMN#PG zS#NM1Uip-H?2s$Ka_39;ZU ztuK}ff2^y>-atl1G}2*!+be`*@*KVU>O>qE`2(ic6*kv8{i@P6^b7wT&gN)@!ZaJs z?qOSTz={!%_cP_7i=`#R>f5xhw|B!wksKGd0IikFO&>WUGKxZRSy>gVAt3iR7?aXv zZ$~aj7R%DA+Yxx4@?1w|Yi%fWYlz4~2zZ;A`?-znZHC|CquhL4l+w4Q0xu}pF11_R z#Mnjz;?n(nT(n-ZdVv{J)jWiB(4;S>9I;N-V_`^ zUQ4j`yRA-czT`-;d%+=jwzkXRD7)HQP5c9%z3321g(BCQ<}2r{sDL)7uhRHf6Q~yy zGpmoS(sbh6t>MK+4Y7|I56qqmFH2r%L;Y9JgHBgmn&E2iP_NIO^FAq8SU-{Y{xYO) zrDu;W&;vgp@~QiAX;8$&cF-?w3~Q9OZa9N_2v&$)_-<|~G#HzK(S+@j1p4=y+7{8= zjdZU4Nfvab;z{Eajp!+sp>uJbs2hrv_qBlk`@Uyk-+~LB{_+=a?_OzH9)q;sUg=}q zt>LuxIy|aY6BWS^4cDu;S03Yq4neqF1+3CCNRi^fX-~;8C_p1vis_|NpM*e~dVjlK55DANwk=)8#n`f1aak2La zG(%lW0{dtH+_bGWdbuFXVyV$Rj4jrha0ebDN6+5<(eCI_dQRpKWe;$8oje#8z4#5>gs{(=%O~T~UQ8SN z{1uNcFD{u#KL-~;=D^E-`%dIElSA^mfgS!+z()xx3&?hhs;=Fy^_!THSPkVu!uE<1 z;fPh~*uGG5UNgR&@k~wSB0FVWvb&4zpb6L3k{T}7RE}CRmb-l$SqlTsV>=VB5oHt&aS&=bU11hduC-r3NU#x4D zr5{|+FK|^(Ee@C4$#H+hf0E+7KE@CF*|$XsoFat1bQ1mK0z&jFP}))3a!W#3Al20H zaS8|CiuK&y+p?LzS?m}VZFuz1gRqo5DY!#3vTVWD`KJHkg8P@3>UUU#>D$!pXIAOI z!J~?TIokYlt~8v7kMxfmoCBQkdlOl0c81l>jU^j@8oeznCK|lLn+kKau)vI8EmKJA zrRA{X=RbOqIvJc7Cv{4PTzVHn1#IS(kl){rZ7O5XBzZGvB@0xLl(w>Vb}@6QghLrH zbSfBg9T}$#AL2IC64o*qHfqJ(b_^nP{Rc(8Xc!A0TDP| zeGVT#cw&pv***R3H|?t^HV6rVvDL1q5JUw|sd;ywTGRgx_$hnPZe4TZI-Z!|9NX}) zu%za~ce*V4Dc0F1@t%#RSL=K`uXEihKUmDa(KX)@_i51kzYm>DFJE8V3KP74fo9A> z5Mq>`v{t6c&GB12yQJ7r!WK!9#Ei`d4-jl+RE__UUgEYN$F`Wmx3R1FuEb;(*Vul} zqgNkO0@Z5FjfJL39yVxd1K*$=JBe{>|Lsd2M$Dl$6)qIL_*zGZ9~Bynd-5tT2ma8r zCP99Cqj;8ZD#P$j2$66>Q(rZBDP1H1uOEOOypc~W#YHHXU{qcScj?LzmZ_k5VlKsL&n0X z23J!@2r>IagDW0sHqO_4zsdf%#QuBxNv^T(&2ej!qj^Ep=xg7WftqKvpncjR_7%WU`EwJbDzb>;qymwhCq zEjIPCa51gM{<`~d;xV@NLzlV9CU-hlj8q}_3$#$F3Evgbvb~tT&(k((g?x$lA-mqA z=jKC}0*n&^`bb^OgXatC?tg-Wp< z>+>NI2de|hC6JDW$@l+VYB+gPIsKKjVQXLb`GN#AcoTT6**R-^oH_V_(Mo z&{Ss1o;w2Ui65kcqUVm=cw!;D?o|eg+5hJS81{bGVC)Rg<0HSfS9~tQ+F>rPu!md% z*SV9`2Whf-1Bz+gbe785h|0cDsS%2A&*u8O%u%!@g1Yy*SQQn(GLU3Dw&kjp&Pw7l zUPg1gcPFfq@ACi*O`H3hnKgpm<%4P8YwWuoZn)A%?CE5$<96N2(8&Qbd7j`ajha`y zfF@`y6ir4_|5xCZvX$N4!-bJ(Hwtx5$3dFA3w^X+5w*=_nK&`Toy;(MZ@(*{pu1Tw z)L}3}D$k(sHxb-Mju(TA5Atr{j%Pg_PstPpNb@134Hy_|M`c2FNaK@_Z?Y1$;&TVA zR=q~9U>C8L^=#Cy!B~Vrc;wIeAhG5wO$p|#V1jMdhmtL(-D=|`G&t6~l^5h>bhmkY zIk+E-6*Dn8a2b-+3&T}jOoHh`tvC84FV%I? zLF~@OUIFw4h-eog(Q;jOn0v%DN}gqnsx?Yu74P4N6?qk@0=~3S0VUlM62Ng@vmi`H z4V9f|)-$wO5T>oHu%4b{|MoG==DHvJk42Aq#Jus{jv}n;Gf{zft6O3*epwAGA^eGy zp~f3s${$E$h_Ej9n}MNre7bG4u3cNLmB);omWtepKs35BpMWC*hA{zoXK99)eNN5A zcGPL1=I=uJylvnDn8rrg8y&^`2^GtqOy3pYcw0|%oED06s;&QqON1v@t{E$ZgeoG- zv=^I>-@nYezj4m9R)y!>U$`QSKes(dhbQfH;W!(p1hH~l1Y}StJ zXDDKvM=K!V9K#PK5>bo|Es2*C>ujh@i-B?<$ungBDnoBtpn)&MhQ6;ANh$0Pj_kQ#G)F$@0ussOL|x9lE|D6^IYZh>9bCR!R8m7dXD%lp z+sFHgzuluz5R2B)Mz!&mgbwsZZ`-VO)SY(OxWf%A_YL!lEGz!oG7F!iQ#@xV+#%WK zd2H~zQY?@z*g`>TX$Aw$N}thPxkPvas`xqH5f{eT8=fxpyXr&Uzch)y=Q<*}YcVNW z2!_tWgOQ~vW{_?S1F5MgQxU^UzLXfDp-gKI4+NN6VJRc=1&5vHC>}E>2u|to9zKIR z?yf7;8(~<(yl$zu3k>QM-hWUTpu^eq>}xpW`&$fn)Drym#xDNgHW`C|x_%EWH_B>@+>?omueXt>paUvPK)?x8u5y0PV=nlc%)r~g2MSE_$Y;=I{-Wt5u zk7gqA2D|MwR^MlsC$>IuJqipcJ@JJuvVga07 z09A0~n$Rw#4yiTk_#wS*vC3smAfH}^3CS7ADtmqQw7NdcN={ESbNpj{II$8gpckCCWbYmq#S|jrlABckCwu;%dp|HQ=(gk+LX1U#Kiz68j z8bHt&%!-4n1DuSO4&yqLY9PD9UMwb__tuglmtR+J-ov3X200tSt?U1AMO?DM-r^=i z3H#$qv$}AMU&~OaE zcKW!gGulOS5>~0S+-V_t2yKF|g9tB}%l zW-B77Tgw*v72g<0mI?$5H$5`c z4Nb{nGFwcc-=$%v02Y!!fS{_^SO+U%yzeJM0zF_CEHoJJiYUCsK~PE9fL_lxE{vZ& zZTPPFR+adV=~w>am+Ij7xkAIk5|S<3@E@j)>I9(zYF|K;`%Oi5<3@RI@BVp{`{{a) zr&NXyS^lmZOYfKaRQR#S%nA>b8OOG;y68#!O3by~yty!H%Ox7jjie3#*+Ma$iAnIz zLu*=pVH^&6BXG8zaVWS?hvL!A&<#MbGM1Lw(2QBT94mCRmwCKzb?!E7nn*a<%jO;f z#%n|qw*gYYZc|3GlfL^xgtTI!OnF_2vD@QLkegI3sgUG0hM=^Zb! zB|b^ykJz;wt*J1o(x*VSxVescOT}Gpx=2wuGW1Gqa5h|^A{g?!{yey2C^m7kEgXRes0 zQpg*q%GL))a_At@3AaV-^2y`FQV>XoxG1hTqP?P@s+WAitZvDn)ywl&qD+u{%GTf9 zv3kcb#0Wsm$<~n#h?~kTMX~_pz$*BMZvIlGc2o)0Z|S6bRZT@rOifPiu3xgI@|(`r zSAK45+OnimQfuJui~KolGtu*0`|$5tPp%-4Y8$d{X3_xv87CQfpg1vyQQNIiiEm!dL_VZ zae`s5E-pDqXVt6wgvTV^!tbGj%S0-eI$33gt!ow~C{xGa93-E0DJj7~Zy8>8(#~d` z^GZ(KCH3nl(AU%BE2S2Wcrw%ywL&e=YNzbqiE599nt*M+x?1%Y_bZH?AasmK?57}7 zPPM_AqylORiYxSm45{Vu;QLoypVU?hf4>6vhe(F!XqsMayc)j59e6ZMOM?Htv9OS3 z^1KDd%VfT7Fi5(aW(a@H8RE7^oj4Hm_wM#ccrd`7g6;eSeXT8uy8nSes%Igr%{qH< z!GH`+f_Epae_qm+(!le!Iet4MrR?3Bq}-3{as`CWXO=nNly?JD)YX&0y$t>0u#O@R z;y-g!v%+AtCVg{0uML6nGLbKi(9ISX8IIfn+1FzAy43cI}$xOoSa zSxlFb$e(ef6mk!&@A@*iS>Y&SRYeEdf#tIjkID;bs5_p6%*$QV26gaP{L^sj{PPLG zpPWoH>P2f%3Wo#pHQf<h7_iMmD)DY` z%V21}PmOtGz!t7RuM zbHpN_8DtnZG)dA)`v#}Q{1;HW?C9+Hw+pigC%9U((P1A6)sgiH-<8;_SSRnk!Tr!B zWdpH$So(0ER=604+=?Nu9beRw9M_%@lWDlSsI(CS=waV9iwQcsJ;s+{f}ZU$qRz{F zbctCL3Bc!BdETSZbO15JmDlOtx1p1NbaY7Dq!$n47lhbFXVSTJ8pBL^0fiLgHLktr z=oNXYtVXE>_P<-oFGIca_^lL1@pDUJZT~E-DL$RU4vN23sXL}uZ}v)8;7=1qZqoTT zMzxf)E?9jG%h^^E=x4=MgHHy}(6hl149=F9TAC&P-22qwLtZ~yamHhe3B zF6Ed1qoEBtPJmWvl|~G=$Zxj$)~==L^}FJ-gw$RW;c*B&N2y|V!)7Bfs~B!Rg0$1{ zQ1`q9nFoSz{CL}fQvZ&Rw>onM(#?=`*Bdc-M_<2o+39NB3*#(UA(9=~&+EcIa_YSE zd{UMa&Txb#Pk=#wvu^?!wVomt5~F5INqJ3;ruU|0Jh?a<5QS1z|+6xyF*DNl0%@U8rY zgfRE&C6L0eni{@bi$H(%fa~<8Kj3I8XHybicDKfvckM4iVxdo?SqVcGf1cL>9dG3= zaZn*SuBoB$8hfB!_b+{-PphG>k%B$YY80J?S!~qAW0Y4|{ENU)R5&_mG?virNRyO1 zJ6*|!9m|%(=upJnQK3|LTBzcvm7uOC^?Q=y?2l33G;KM~rs{&}(03)OTsv_pTyj>p zBTWT@5mIVd@yb`<{~vpA6_w=|{SOj?fHWvbN{1lbrF0`DA)$14ry>eScXvs5cfE8; z^V0Rw-8ska`p^6>=H@?lGi%McaLMBEJkL41K6{_N_hFr;n#3d~(X})W?V2W)mGejM z!ya7qd+;e?l8XElxL;N@;^WrLKK=K6rb-bvsqt$~^VVv%5^o~;>Vi*}NA_=vk`{GJ z28Q3}`VhB0M)at7mbB~`BVhqb-XGRXr8_M>7TB`UIefCu=7+;Zx)v**9uS71g1D$P`o^L9*1;lwT#&b#mLSoQag9;;7YL|RQY4C}hk8+?`xADILaO4d)uq^C1z`c{ zBCFTu=(%{s=4_CxXn&sgnGJ~#x;=W+MT;Q2om4CCbj3$4QP9_W1Ue{}l(*TxE_J^l zl|JDN8UO=rQ)jze303zx?HAKa)nD;mS$;ZDVi?2`{l@m|?5wVt-*+%rqwM;k%(r9F z6YKiwY1lVoENZG$6fMZyGELaC!-MqhZAndW9WlS5ruEyx+FXvfQtRurbf2!ChzC`-3S?F|f{o`l(8H%C%S{|HS_RS1I`r(*obXIf*8)%y*_TU; zdS0LXkDrr3kLXA4c{;3jV%>(kWj51ACpQptC9o$j1%7f+3k21BMytO$PsVP|vt{VO)v^ z)m7f0>XVwyLxzo+zy2wSZ$`oJmnn`2nOsDK;?J!^)(&S}r!|jd%8pIUw7>5_O27mOq}&;_|e+?RHOOFGqoJ9!;71f_bq;!pDSCcfYjl1j;rjvr=PttE~& z)sH}n-y?tB3Fs^&!j6bOHDQP{5-~71_x96YN~v2bq~&a&NV*3}O*DM_^qpOajdF~O z)EpNDV(_Easp3k44vm6l&C`S_ZP)GbaB%F+r6#x*Cdo|F_$KVP5x(AbojOg`4sYQH z+|YP}L7Itve`PJQftl)Q>bgp2b=?cRzl11eG!2eVt6zsE8Iyj?QD9EPE#1|(uL$MP zWGfG0)A9mr9uG|5ic}TDOt8d$ubiXgI*z5`O|S2@r(R3L%HbTt#2gOfIXiQL5$b0I z1(;bAsfrGXIA798L=2ub9*3oIq@hkqAWWhox;Ok!QPT3`1OYG1&F%Hw%UoUofovvry%vw)zo=4YR6ZwX`u08zUsC8!Joh*6$p4S8a{=FuyOf zRfVeWeS);NhzS!si^e;p{;Gz*z}M8}Qg^`7%wQau9cwU|M`&4~($jo3de>9tJD3`6 zXmWg`&?(Lzo0_=a5u6AJ*;G8))8x~ zcr6q1?R?pL`A3bX%6GE-Yfs>rUoJW8s59@?LV_^XPjG7Nb*wdb_sXZow1_nnbH%iO zpnm>ayt02Nw#_d;^Cp6qU#E^|*M-pXqvC3i%B+MIJF(Jx^W(5Qzj-%^yG$0B?Yrgj z1sg&FB=u}oZFs(EDc^?3p-hbrs`4Oz=Ofgo^`<-?AGVUcCp}Y~k>`uy?!I#mT zAKaBDEQ;CpTi#v@yodo1xN{oKv{y@Z}^hab6b02C;UnNrNN>P+K8pIGfL4jGe9<*VhKbe^ZLJ{$7X3X zA@RoOdOV)Y=pRoF&fK5ViL8#(DXl$g;)xom41ta6gV_Jgd33-~9^Tf(9~6dfStKXN z9p0sJ#P3*GKB`0+Gptt{aWcvbMwwI@36gz4kv}DYf%b(JdL?X-^0`B^dHPz_Fv%g9 z6iIp-JEOyH9(*ou61n{oKof#q->$-XF3<{|%>-}1riYM8~W&kjE);+sEq547?{Qh{f`FSSW0(nENJWko#6{y|Z3=XdST^QlbU(X{0>$!hdC_bN6Cc%I zf~HXH{nzKK_@aSJ9@CVT9l8{ zj&U5wzTG}GTU&G?RnRv<{r-k4yC46<8y4Bd+8{p#tJaM@ z^+mfHor$@G4@sk-Wddw_Nd=;X(~EhFu6n$$`_A~7@CbFJmolZ5lroOmrNVTCFJ%XN zi7kyZGL|YVIOW`?J)g4F>9!ZS{ z?KIs^;6D{f;C~!v7)k8fG_>Z_s;d<%pk_4Smn!6nJ^`98dnNJ{ebZHZ@Z9C~(Y0&T z`S(nu663P=oh*W-tCUqd;Mr5nl@Xy*_$YmOey41mE}-&YDT0rGNbHZaHfCLem9xRz zA7y8mbz);jXG8X++uP=Q^q$8S;k8Jzq{K{(+2)q? z+1dSNR9#_g!CzPbN_9xen3DT#rn383B3nn8got;Q7*O&Cu`eg$_A%Z`M-SpfDr=B3 z#X~u9@v=Tbj~{9mC1?pP>F?mjuSge+u0t@ScVs4+8O#j|stH1kKaPZ{rkuk$$3DL0?TEiTk$R2piV0V~|EQTAD%MT{4*~ z=*8g6AS^|r$YkLatMpORut-fpX_Bpf?j-T@MNwV~c!W#iR$BSTeZMi2E#%rWkxDZY z)iV$mi?evfI*i5-i)gP7j6P;&j>6Z9>BCOgqS2O-G)F(Ibq;m?rLJ3b$gh%W7e!D! zj~pC`zz8Yd)=t7UV%e60^@W=?%;S1W2dn#W{bZ_=Wc%gFc*t;>oDMt52Q7C=qS`fODl`bV_H zym2OyE-!9@^S(uF0wkBFLX9H%tX;B}4zG0gK|F;#5t^x_6=p~`QcgV_-pgB^V@=Df zW3jgPcz$9sKk-2gi`(r!(x`sR#&ggB?QzkwB7{R6K3MJF;n>Tk>)`Y=9+4Xy--mb+ zib3|joS`mMy1koIT+oBK+9l;JzN^DD33zl%b7)*)X+Q4ppa!g%n^@TUJfb(8tb6~X zdexlGPKdE^uy>}4Y-t$*KqQj>?KeF%Dkx+^^}RptP+8L2LdsO5f_ZFjz|F&K$%^iF zMuaaKrwM;E zNia&NbE!G^l=+;<`N!mF4VY*|n*BT~LE7r=1?}x({H820o`K*x!=T>|^*m;b1n8Um z0j*S(3eFy5;e=&uwe~qF&S^7kc zugyUO4gx2jzsQ8B{Lw%=c#Xgq2UH1s=ZL5T=C zLktQ(z)*>x0}8|ObMR80q-5?ut$CJq@san9DAU)4LyKarkR)XP{h#t{of&&~H!V&5 zHV1M-%c-sBl+4rJdrE!D-`@GRTJe)9MJ=7L)5y4JlE~Z%3t0+8Q745=yfvcC(p5P3 zR3l;hba{z7B8jBLVDsuzt^i50<10%;xLa=%lx<|>$JZp4&UepSYtYXY^Dl)RGnWHy z+g-dzc8Q^f%VoC7!E>oswfEk=UUx6sis+A_19DeCu1g4(`1_!q1YVZWb4UEGV#T+m zjhoN)m!BF&U-@k3DVVC*;i!o?K|)KFXq$d)$@brJuYsvCWV?PJ-6?pVPnomf$-y|n3uFecdg_!>p9?#qdg90 zQFA5)Q=$6Vy8LQM%iGvTO_N0Gan})TDZ4n%pa+4r$2A7b4&SHchdTMZDl_XhGzIt( z8_ga!e){*P@kTQHo1Q1Tf_fq>TY@wjK{s~7$e@(>y&G9k)mvsSf)6ZS2d6-!oG$I4OVI;B|{ z9Uj9jb%ttjUY9$D1C(Nx!)}5G4fA&H&I*6y3DU?$wQTFW`^|^YWf^Wq--815WZfPs zP805A6?OIHe40ds-ur!6o9OR$aJmq(*MnQanNR0PCL-UUS9-l^+XhN;POn$R9&cHE z<>xgB-Ke)Y6<6-8rW_iK8nh$x7?KZO88T@doh0O$tF`X)pE8*Wk|}SzhE|IAS-MPN zf@_0J6Uc`X6zHlI2Iu*t$cJo#B@FICRH0d; zyYk~)ypBt?!jHd6OhL62(??ZN_7t%`~q^uw?lh4Ef=ScWW^sFV$;@g3)7T6`|7rJ2CRqi(A0 z;qB#>NhCm$iU#KZDhCdNfMwKY9BQ=3_(y9?Q7OOZwDSi?McI^K_aunQ{^1Fc!jp_f zm5H-+<=kwXHz@u}XFtf2fyfKt}5?e}Iaz z5FUbhOEx>HI?G;IXl|mUd6bXeflfPBU+fzFTWqQF0?QPF=gXDWx~o+wZ>Y8O-q2L1 z_UzHWD=)3pgwdbUpROuMfg=eJ=)kp-TC6qV@42P_6KNJ2jcKYgbnGo>JK%e8-SX9* zr7**Jf9J1^q%!_2_~20U(+`~5*_1c!MT^6kA0A17`fM_zr7Z=cePN-nH$dRyXDg> zktZZshOMgC5N(hFG|~e{qkt=KE69u#G6)=5Gw3S{YU9XrAyN2L>Xa0c3721>0NU~K zlG#{~HC2%}sXMJ0oW_evzZyi^8xB)@3y#{NI9G?EFF@|9RMl%KDGr_EN%fb#-bV$6 zn(WsrtB7PGk1=OwXMc(rUecmWHzk#N_YBV15{9D*yI;dmJkyGE@Dem_E}@sC86iw- z^9p7ARW*U(OMWaj?xVmKM;( zS2ec}a`+}8sz=kBdmuH5b1*SfN*X@6ho)&Utdhyy=}9;g#j2LrkE0%2ld1I0dsFm@ z{uqKXeMCDCFL55SaV@zPsLBC1XD{7`1b)+rw@j8ZSpnxdh&dOyJKA?R)X*w$LLes{ zo}(_VKkHgPxTZ_w`5v0F=MnZ68y9Y5vUe|Q-=8lzHfa8JA(cq_crN6X`oeoI+gobE z`U%>Q5c=w2ecI_?I&$|3^Khl6_Kt44F`n7ta@p!R3JzRKFb{fR^;@DWha8Z8YO3U5bQAw^n09z5uL?L z>86&E$JUY#pVOA*B}YTi15sZ2-E*IemW%us%tMx_^8%pIvA=lb%s9aUh)=7Kx=feMq7H7ClB<}hTF=Q=`QKMg}!4+P0+?ZIu;E+ zB1>ec9*f3+dkA;T+gx(?=w9G-L{5Se3Dc`8hD!RS+}29of=ag!-}seacX-y*vmTII z{uKV6spfoj><2+0v~x*H?oCW*b=nsLCB4$@?^aoc?IwWSWKcH*%sJn|eqO3)+r_!o z=%stN8Jg#!c;CmS>)`~}&2_#&ovGOimA2evX=zl@9B5=>@*+_cRI9i?$3dA>xzv_a z>zHYIQz}qm!Cjb$166)I<0%H*K^9R6L`Mzt)&-4W%5kd*I&~N|9OFlFvWF%}?9YfS zBbFx4bM3On69lr^)Mr*j-}F?yfX~dZfL`CYEA5d6Ls%K`&66s1m@6Qx?__aF-y^fr zj2BWXsT@)>L}ztZhZ(ozT4HFWMFjgtghuj@Idg@cJlnf{jM6-usOE}|vr5T75FB~! z_Zwz8Q%nDnPNy#S_6m>j_HXtPUiO;FnsYWjN#9w>;Na|Xj4AFBzu#nleYZ9=-m3@K z^kRuHo*FfH3N%9}3=cbQsi;szd#RE$>uEV2uPpicqNQG(5!^1>xq_Dk+?Fja!#{C?)ppO7tUJ6jUfpKHWvrYLlVhWuln2~)NA~nH4c`+g zlZm+xQM#6)(@89Cw@J%_-jd~{fBfk{)eXPM+Q<&*b~yIB3nbi}o(_Kor)r@Q6)x@1 z$5hUj65wG7Z4akFaUXY$iWmwKo$RdlhDfq~EGL{hqAs)vR}KaGPdRP&vNRo%;U5%(cF z-@AlGmtcHwH|O>F_(Q9p(#b^?%dkHYlopaOIaei6P0}M!%|c4GH3{a|Mqu7D@;z1V34Z_d`CgW zQdJS-`_79f=P5`J^2@hK)z4r=;M9eOg*O6+p`I(M}^((2$*R^ z8Hy_iEe*{pVw|SQ2Jk*h?JG(Ctv(^aP+a?YD4{O6)YK)!=+RRk#^?BE9#cBkAlweh zN~V-mAH;RURNOs{Hy3WYW~Uf{Z3ac2aHf_Kjih;YgyQ)EG51*7uPOjZ3-rXNJs)I3 zDHx+1bNN+!(5D*ukti2LxBiq9J;y1Y2pD>}ilxXPCcNgT7=7mbS%1pfCtYO0TD|9J zK;iik4~Jo;{)xb|lsYdXgOJm*2<7?S5%3LBAF0{ktk9R9hnz_7jIm#~zuwz^ z^{Rw=npYk+>+&U~LsM)<5Xlf5B&e*+FW8w4_$DpnBJ=1oDG00-0`_9zeZ2&cV#}9o)~Ls| zPMeKVHVB!rb5@)LKgjfNnutk?>X}(gR4#sE3l~)Ew<`{zi}zz!vdc4`;o7FAkWra0 zrMP~N`;1hXBzwt4`sK@`X?3YQYIZm67c#!WnlO|RQo+eqmPsfVV<=kKU6Ck5zd^+p zd+%X0!MviZaGfaXit68lS4pKEMgG(Z71LCyA9nq%7eCH$T^%lMf{OFYP)4*PAqd|* zKe8N;Qg`fQ9tVDqXg`q?8ik#=bBAr4*TDMoEBDhCGRk2?Z zABQLkNabJB(;KhG7~zMkv#nRYQQ(MCA%xA^)M;AIS%Ktqkaf%}AZ)^&nh`_G_%Up| zF|4ta;WVW2;2S24{V*l%^!WOZM%^S+{ENTh(Xc9dsw1Re=bmb5xASdyg$Z_|P`jnG z{W(znmwVyu=q^diAT zN*Vv8d75EFJMCY;Fz-t~S4AbQs}74eNcsddYIJ2rDWhJSaIgF5Vw0)km|>RSQSej;f~l-}a?^2&62OLJN9v^eOGx8s*B;x_#54^3+RXKFUGI z#@bNG?t@>U3Yww(Q`4QKOrB#}?ZK=yPmY$Xa>z$-ZZ4DaONM&4(I`Kx^Ufu^&*c(u zS~%8*yN#WPjDH2S$w30*VnF@3)~o#cR`!3H`OA?{$K+2ikt+&xFucz1S)QOzk1%2X z?_fL?Ui-l4o+JOy5H|RiF!(=Xh06F}aN&Q3u*&~G;oolj-}gb{sR8TDmlgdVPuFda z4dMUxaAh5K^musmxDdelZyb|R`eQ#XJZJmwr~mcOaR5h*4(31N{IB7a^Pg+i{(lXr zfq(e!f5!i#2~W#QEddg`um^CXHY@s%6nuQ;_Ywwb(yY|@stmdk(YiH1iW28)8|?iw+)Z4lH0TeV9w``f-aYQANK3j=#Mk)Qqb1fBx(PA0HwjVw*_2 z@&>rXZoQT43BL<=mwI)D7;<*&gGoh6`A+t`qWHhjEx|Ar6!>q{($@hVTDILWA#>h47OifKO41vw6r}y>s9j?UL@%!J~lpVtt|8{XBiI)Vk_|sb$kCLmS zV;2-k{G(%RFg9r69)EkIFg7l((y$j3lU!Kd+L{?oSXfw-=@qR)lDM6n9g7A%*l7`e zQURB8YbbYfOACCtc1Z{!o1f=V3qf@M^K59E3%^^J_=fx9@wfp=)s zhYi>I^yqndc6M_*$RZ;xy>i%e^qB8PG1}MnWqT|WJSMA7uEBhNt|}lek2WeQ%H!W2 zG=zX3BthM-AJPjrUdJtT+m?%0;Iaro$?WmDcdV=t>5u36*Ecs8p9&~>QZ3P(Hs5}N z!>)EP4#%hB0j(S%NnDon3=FGjo`w>t-`PCqLiu_T2mx zv$p@7)nQLz6YP9v!t3z{ruOO6+To$1ogJ%czT&|tQBqP;>FrHrRh2z2NeD5|Vd1Df z4G|I1^=?tw^(EA{ps1+p3oAAe(Li?)UaG5FRb}P%Re7e~TRe}7iTEqR#qex_Ie_|tRe zgZQnjE%F?4z+3_A>+4WUrsTDV|@6Dio0>3k3-J;89rRl@-tG1_92t?9kC}Di{s_p(% zPS5jz3pV59;}b$E5J&!W-5XD;`=_vwl9@R)fXrh%2~pW~4|MT@RgDUF8L^Mg&h~Be zN6RQE^mJvPkwa(neTN?(?%c0e1HQ-c3S3Z~fZ!>q44zM%CL}h&y@M9X7VOND2fIb_J3i zN5PK3pJ$(QfjM{x4-c)tw2#qvHkiW?*2rzP&~G!bv0}FK@IC2_@lk@c(SvQZvj(8;#1s!U9B+1MkyV+FPO~FLlKpLYukm@XK}jUnXjN36jm-_64E?>cHqnK<`!%!XH}?hJh1-j! zcRen7t*wuh*1xdGN80WO+dwmjf|3&BgZY|b?V8`lgK>QM$<8?q4L83>e7QEv6WOc)k|#@yb6SCL-eZ!Z`pbAQvw`eE87a-3`YKZ~OH4Kut$?mQDT; zI4~d!`Y@`qb8sXIxW0p9V`KZLdCFRE2WSvd!^5#wS6BDvYl4B0fG;^rQPBpR^7ShP zSPfWuh4}<+7@3ePAZ@avk*VqW&Q5>l&lh~oyZBmKTIfWafj~$d92^q(9BIMZfd*Jx zUHzb-fbl-xvIi3KS`0A5;o+gz)uJnjpj*W8u^rs*?k+tG3#|3-a2+V1{e^~T94g7` z7oU5>$?4v_$yihX$#UY-(kOI_DP{!QifMu}&dyw5uXQiwq!{16m64Ij1gh3*sW}mZ z@BS%lH8p&|Z$CQTo0~HvCME*S&-%6VnT3T#V0gHzVuOF&8xL5jE4%Zqn())Tql2Sk zBEPeRhKMMjrNqBo1HdnJsoRy864_rJzLi= zjCE6yyVvmW@XaR>;_lvF0sS}s)S9TMD6sHf_NGgKP*hpYynFRdXB(X)_2lGa_4{+O zl~LbEj)sPY5OR^UkP!4*Yv`6$)t3&0e;R1`9lAV;9nRyRwdmsk(w7#in-Hz7`5Q4)QbOr+`=7X{_jP-Z6>Dy84p^G7z(5qV7cXA3e);AYW6}N%9-n}q4%G>KPGu!K zB#bn;sYw733q1Fq2{>8rea&n425=pmZdkS6{+fAcM@Ppu?g+Ka)j7Bnz-<{KH=ya&mH05e#DPkmhDV zphlop5y0+z18vm+5&^uL-*tvg${!OkJqJfYK{2xI4y^VGkVk_W?L=TO98F}UtXr?w zULa$VTqA1GIh-}_ayh}ctvl~V1X7=lxLo#MlHW~f=%4gag|qrTdQBH=buG1cCURN6 z8gr2HnX5FuzP^r#i-Q1Sr(_F}zUH#PSIv`$w@X`>b@TMZd;Pk9JX;Egy`bxoI~|}! z*WzMoS=(b0;18fM6GVK4xwyEPbQ>pce$`jiey+1Ip075SPhkBHpamcnufqn4=hdNc z&+nHB^0fb1n53M;=|!NS=d2)B+-KmszOW|(imm;?61EHbz34b9M8a$LjYcjGj@fqO z?c28u%*>mKrfI7~i5$(hyG08=cL(S;q+eTbfu#qkkUO*8jPq}utv;M$Dt}MYQEPB= zGI2DWl8vL)`}c6*<7N+B;!&J{MgV$>)nbw!)PP&M9L)a&$#xbs4!e^KAP+v(n&qGi z7(-wtfjA#XmKT581NA=-pt7CM?f6aD^hF_N0Alj+w8>2Q;K|iher084*U{X;C3oh@ zJrM|$&|U%qWt#2{0bj{R6iRw2=jsy<7^{C|#3f7-df);ir1aViP|rS<-x&?qG-M2- zN%Lf&%1AH+&P*}`b(#*JH*T2}F1tU6mg@R9-@SvQc|oA^Fd}fxtcR|_y365n?*>bK z0Jc+BRyJjMZ@I1QOyS77ZV>{kM){o8;=421gkcj{@_&l(^O}l^m~{&d!DmU7R8(MjUy>y{HP-Hq+Frgonwpx)M=xK#BjKfbByU&_;s?Qd z-Q3*1)M$f!Hrb!8KqnVY0l13WW?}Zg61W4z(Nh+iai48EdVBLHxB=q(CsKo(?E|EiXA_wZN)lD|qHK5WuuQN!+idyd(q zV{A+VyqX#3{Jyj`J6`&?Z>i|%QL@MTfbOc$E&;On5B3AD3-2+Y0KW<7`rznT<+!^6 z0O;|_2?ap!_xD~vEd-96fvW3uU?Sbzc<-)`OgH)>&-Z3V8-9%}F1Ad3aT3l+Uv}FB zz*^(Nk&N4l@SoBJ2p8D$iF_q;z_+ul-U77I1rrycbV_O60Fk-6x&jvD-;Xbywc`Hg z&44G9k&#*77FAc{7fqq*BwV^b?hOLi)B$kf(R01Bx0IBrj*gC0w6uAHG^D`n_AW0k z(=6cy*8e`tXTid{6@GL}4rs>3JK6nW+G4qB$9sH!-uw=_n65wJG^6Wkf4U*V&d%<( zU(rhl6fn@L_cx=yhsX7$^Bx`^4#3oNxStw*D$$ns@B!YWDzWLXG58;X)li4sC@8;+ zi;b0#kU;Xi?jk=vK3*AQX)_v(W7cqfe7gyTPM1;D-26V~JWwKWEc#2UPO{0vk^qnV zc#h=~v=ayP3}|Ahn8Ley{}2r9yNxI~U+;N(JbL06Qv_yee0&@LfQZ=GpTJlEtnd%q zb6HM*VI@4iyfoRJEcnNj9Ht}3C%xo|D|>r`A+I?CfB#+sIGT!?86&vougOI6%E}jD zooCCxA|n6z4q&kEl&%7xHkjHsFkrCQ=nQxd0MOU`PW0dv)_Jyi&A&zfw(O6l15QZe ze!N_&pnDW}?TQ^QA)ulmB%}klOhA9b(tX2#N(5GNWL#89NeQU!>$PB(**aVD{{H?{ zJvlKkL|_bo9k2sC?cBryJdInh{To1)ZES2f&2RNZQjw0J;Y1d>c46jgm3*qTT^<3T z9n7HyZqQNdT^O4og^AN-!}u42NYmi*Ckj2 zSi2(PNXymI-o3Agh=|YaM7oQ1Q@_#L$h31=s*J4cu2bD`>FuV<4`d7j;K?139Th*p zWo2gvg@w8Jc%RtBkJW~kT_S^Mf(|)gB}i{?Kyh*LJLtvEL@$j?%PV^t9-WNA{{GdS z{B(#--GBNWnHbOn6hQiJ2GOOxp`iEUDASTQv<^~%;31n8W9M%?#)ot!OEo|=}HUu16pFv>p?<%>K70s9g6PIulM z8ZLYrxW8`&x3<0>6chxQ1)xXBK_|8!DSY4zwx z32=PAQabV*M#kSDR*>@Y?I+-DZSX|L%Vk{5Vyw~*A-B-iB}la{6kxVeepm%}=+V9kIQDZKW6T2&_YQ0uxH=x{|r zLD0YJywA^_kg>>yfE&Pa2Y)y-@g4Db3iup1bQ8=IRw%ykQXphEk~+ z<>a@Lyf8`F44KTo-%x7Rlh|BE`=e7E%d0a> zI_0HYJ-@y16j{yFnO+A8Z-|2^2^SKk8^XhJ({ADrHrTVymk{H zenB6=O2NA%5ks+^$=mes7f-)2r*n^?YEB_ z?tw}oi^C_{&%d;j5O&3Q;=Z2jp}_3r5?oxux*ih#={_S=Kl!nmC%VO*MvV^+{W_qV z!(Ni)hqgZIfNSe$!1Cg)7it%7RTH$ycsDe zTZTx6v-0B|>sL6PN1nf!tEv6AR)x)$L12m3(66JKaYl5DxA1yawMLp{&~u$&U)n9y z*j~u4vQl2p7ls|;5s^7_ajs`zHFR&zn!LaIX8}4p{N%@&hRAbg zu6v`7?zk~zY=`H;s?6_cw8)F89FOR?XdF`Sa&hizat!`?UkT5@2DMz*ray16eBoI_ z5}B;dgQ!bm{MwT>Ob)`*UUX3r_cZ^uZF4A1_Fe_Dx_DTRJWw@rvL&T;pPhws{KjsU z`jsGmIN{SMAnx z(QMxE2>pm^#nW^V9*R9!y_h1VFFtGU%8rCcVu`B6ow=RDwacGprL(QxOc~|C*jG&} zTIvHE{RXrVTj5~=ctXdRjjYw0jjo8vZ3W%%Zod{g_8B5c?_1+5UAs(L3!LE6{1@4h z86ufWEW)ge*Doo-1$V45{W9f68LtF(a!7xc3%uf>sXUldJk%YXZk8oMw&NW_PxX5A z3Yfe0fd9&Yz?-D?J{QBwz=Vy3;rg%N%7*Y|EE2oVR|Crg#A9Q#Q%=d&CyJ)F*3`MW zq<4Ju-`?BOSI!{T=w1%?{Np;4J3BwQw#+}%n2*ZxatgFV#1@#oko=}FRZG@yUGKjP zahlK}t5$QR`e_T~5ygo9(`JKr&JAusD$ecgFISFk`dwmR`Uun!|6(~hcW zc{De#>xf%LiWI5rDKQCYP3!+DmMDm6TUHy8I-uEq;BTUgd+dDs+R5XTGiVEw=L1*o%4O}w?c72^C{P|r6LKFbf0%i2lR%kSbX2l zI-a5I-RH~}&*W!-g71`( z30e!TY2y8e2ffv_C6v{Pp(D4|3#V`=S`WC@yQbnx?BB8-3eo3HmIh6%gf&M;s3PUt zIKLU&Y!RQiy&4JY&SnqBnm*c)-f*HEbgi^+imksR&oquG!Vwfi;ROlg{Q7UX>1|$} z;VHPmg-U#OcZCTnO48OTIy0?)?2uL=PYJX&ennmk?On}Ld#5{~;d4%18v(7*Nt#2U zWE>jm?G5WFv1ZAs5Dc(o^djUJMnTza*608DS$nX1E;w<`dUFg?V{{n2*0J&78dJ%n zwnuyr?kncr%;DBv{oDe-@hj!roTqtt4Oe|NhC6P9dW{=g<&ctsX9NT>L>C~haBOB*F zj|r9-!&<&|vNTZFUjYwcHDT5tF$zC8DD!ogQNN?3#<|!E*hoh>)i@o1sB#i^SX@x; zjK37bULRh)(latN=W4#j{~WZQNkC@CH>)iT2`($IC?`V5KifFk!nMzBBZs|wWfww! zNbLOhbH4A-%~QujDA~C+hDE3|#(OGff=eC`c(j~idMtNY?aO25$~CyH#D z;kXfy_6U1v%K=4E-zg#z-G39X%f=bgKD_)824sk(CtQ)D5A+;Y{2N zKD&|lgv*kmzu=5UFEmup^VV$@-nMwI=w>RJSZ3EjoH4L}kRbXgm?=t6bGkofFzhEy z_u7?%@K5S56@u-*<6qUuqZZGP1V6cw&dChQlnb^;7X(24Rfzb_KiB*ScgWIu{M#Jd zse(9X+uB*t-?sUa;REMJONXkh70qEH!$H|oB@vwMkO@liIdxmAK{pB~iBGtGOG~88 z=y2C*dvp$&9>$;#jKqBi{6M^OkK$ z^7uZq(YhjZEwt?Hc1ha{mmtFHL=OUKz4rKUGRL05L-5xfo6`dAOj055evFpedj8wj zLj(lN{nYO4pa$mbzOWqOplj-PxlZt8QP*# z-0M=wcSavuas#gJ;K*2a=L6W$;y0FKg4yBTuHx~Wf#USfXp+Yt>NE}$O@;3@E|!A} z7Fd(N4dEkWw_o*F@$orfiTU#5chOiYrScVlVe)1ihdCQ@i6B(k z83&^*txm+`v&0~=MfOSO?K|HW;~hKrWIz1F^qN6X?p2_#Gcvhx_sJRli$^$&x&4tk z)}8UVGV9>E15$3*c?+MqRWQl*hh5Y=>akn^*01tVO>%4(U0DMf<*#lO= z{^b<#dN2xVAvVK7aBg>fRbsS) zNb8*@ZoYplhs@GnRTy?VhqR_j@!vk34A~ zSNlh%%tWI@0dbw(KXC;S7Z;pK^5~>w1Jk;9yd>$rdx&8~{8)&y7K}TrulY!LSy$%x z%QYNx0%hM}YG}~2MEcqM6!GYU)>yNA8mAH$SrouQI<8bS>%Q%bf zTVqp2)WGaQcCrBV?oGsWGkm>jQGkpvb-UquYb6u0gBBG`5k_E zdT0EzH=|A1N+*X-S0dvq>j}@?nvAXcA!;>s7&)X8TRTa@M;#3jihi!yzm%?I)+u)F zSn~a_6Td4>>DQ-97Z@&KtSokbjHmnC?}PeZ+@G~Gu=|`bk=nLb0O1|U%R3X z^#kxie(>eifn!NL`nHoD_npwx$=ian&OCZWb*THhxGx8uWWyXp&MV-))ey$!_sy*2phg~Sn4k;$W>t<)|=AC z^##elos-{cDytg&0ynxU@5!22UhgJ4ZM|Mn=d#M-APLoFL|m))zDBNXdw6w+p7t!H zIrc-^OAUKnc)?1rC(Q#e%&qOz<&^r znIaO<_1%42cSsQ(4k0jn*zhc`PMU-87`-Vq6aM44w!V%R%3f_t&*vsDLOw$>We}Fb zKx(Ru##+Mld}?@X5b6;g(O z;IVkmb`JHVDttVlbr|rgouI}W)i9-S-s$Ppy}igqAmrL4lXM=ZG+RT6)6*umYw$7e zZNMxmxqNWVB(fcppLY^FMtG&OFc~#)GZ@@tYF;6GZD>Jbw90BO-c46j6O6SnO;6C3vUU~ z7iSF56R)Dmzer{)7`oZMy59{#Z_XxPX;Ay-wdA(T&`eyc`4Hvi)Uy#+ZYaO@$oKi? zp-mX#>4TFaoFsbw6EBh3`QAvexgrGD-rr`gWi~Vn|e%ps<4@^;yycBtRl}_Bh!|pQ8 z*Y4`bD{#!%^Cz%xeay-KH*ZO?urJ$CGz%EGv2{-1jA{$2U_zH34C?$A^Gr>Fh5 z)8AJ-;aG==*jfK&hoy^5TT|A>I=d}oUo2lUiS^>!uxqiKnc4XSie4c z{MmWwFVFaiJpM>Cq?8 zo&^=GndUa1BWRkfM^VN3cFUg+S$$&;T)L;{1RQaCGQZ_Dn-zb)xZc0{?Upt(m+tR- z{-n{zsO;$`^S=V|3=bOpJy<%wFP$J`|H3HP2sC8pvP5~E<zopr0Pi2_ AF8}}l literal 0 HcmV?d00001 diff --git a/docs/modelarts/umn/modelarts_01_0001.html b/docs/modelarts/umn/modelarts_01_0001.html index b4229559b..1c1552f5d 100644 --- a/docs/modelarts/umn/modelarts_01_0001.html +++ b/docs/modelarts/umn/modelarts_01_0001.html @@ -4,6 +4,7 @@

ModelArts is a one-stop development platform for AI developers. With distributed training, automated model building, and model deployment, ModelArts helps AI developers quickly build models and efficiently manage the AI development lifecycle.

ModelArts covers all stages of AI development, including data processing and model training and deployment. The underlying technologies of ModelArts support various heterogeneous computing resources, allowing developers to flexibly select and use resources. In addition, ModelArts supports popular open-source AI development frameworks such as TensorFlow and MXNet. Developers can also use self-developed algorithm frameworks to match their usage habits.

ModelArts aims to simplify AI development.

+

Product Architecture

ModelArts supports the entire development process, including data processing, and model training, management, and deployment.

ModelArts supports various AI application scenarios, such as image classification and object detection.

diff --git a/docs/modelarts/umn/modelarts_01_0003.html b/docs/modelarts/umn/modelarts_01_0003.html index 701134105..354a10781 100644 --- a/docs/modelarts/umn/modelarts_01_0003.html +++ b/docs/modelarts/umn/modelarts_01_0003.html @@ -9,7 +9,6 @@
  • Multi-scenario deployment

    Deploys models in various production environments, and supports real-time and batch inference.

  • Auto learning

    Enables model building without coding and supports image classification, object detection, and predictive analytics.

    • -
    diff --git a/docs/modelarts/umn/modelarts_01_0006.html b/docs/modelarts/umn/modelarts_01_0006.html index f7d9e9fac..4a33b4654 100644 --- a/docs/modelarts/umn/modelarts_01_0006.html +++ b/docs/modelarts/umn/modelarts_01_0006.html @@ -1,11 +1,13 @@

    Related Services

    -

    OBS

    ModelArts uses Object Storage Service (OBS) to store data and model backups and snapshots. OBS provides secure, reliable, low-cost storage. For more details, see Object Storage Service Console Function Overview.

    +

    IAM

    ModelArts uses Identity and Access Management (IAM) for authentication and authorization. For more information about IAM, see Identity and Access Management User Guide.

    -

    CCE

    ModelArts uses Cloud Container Engine (CCE) to deploy models as real-time services. CCE enables high concurrency and provides elastic scaling. For more information about CCE, see Cloud Container Engine User Guide.

    +

    OBS

    ModelArts uses Object Storage Service (OBS) to store data and model backups and snapshots. OBS provides secure, reliable, low-cost storage. For more details, see Object Storage Service Console Function Overview.

    -

    SWR

    To use an AI framework that is not supported by ModelArts, use SoftWare Repository for Container (SWR) to customize an image and import the image to ModelArts for training or inference. For more details, see .

    +

    CCE

    ModelArts uses Cloud Container Engine (CCE) to deploy models as real-time services. CCE enables high concurrency and provides elastic scaling. For more information about CCE, see Cloud Container Engine User Guide.

    +
    +

    SWR

    To use an AI framework that is not supported by ModelArts, use SoftWare Repository for Container (SWR) to customize an image and import the image to ModelArts for training or inference. For more details, see SoftWare Repository for Container User Guide

    diff --git a/docs/modelarts/umn/modelarts_01_0012.html b/docs/modelarts/umn/modelarts_01_0012.html index a5fe25d95..3b6725a3a 100644 --- a/docs/modelarts/umn/modelarts_01_0012.html +++ b/docs/modelarts/umn/modelarts_01_0012.html @@ -1,8 +1,9 @@

    Data Management

    -

    During AI development, massive volumes of data need to be processed, and data preparation and labeling usually take more than half of the development time. ModelArts data management provides an efficient data management and labeling framework. It supports various data types such as image, text, audio, and video in a range of labeling scenarios such as image classification, object detection, speech paragraph labeling, and text classification. ModelArts data management can be used in AI projects of computer vision, natural language processing, and audio and video analysis. In addition, it provides functions such as data filtering, data analysis, team labeling, and version management for full-process data labeling.

    +

    During AI development, massive volumes of data need to be processed, and data preparation and labeling usually take more than half of the development time. ModelArts data management provides an efficient data management and labeling framework. It supports various data types such as image, text, audio, and video in a range of labeling scenarios such as image classification, object detection, speech paragraph labeling, and text classification. ModelArts data management can be used in AI projects of computer vision, natural language processing, and audio and video analysis. In addition, it provides functions such as data filtering, data analysis, team labeling, and version management for full-process data labeling.

    Figure 1 Data labeling process
    +

    Team labeling enables multiple members to label a dataset, improving labeling efficiency. ModelArts allows project-based management for labeling by individual developers, small-scale labeling by small teams, and large-scale labeling by professional teams.

    For large-scale team labeling, ModelArts provides team management, personnel management, and data management to implement the entire process, from project creation, allocation, management, labeling, to acceptance. For small-scale labeling by individuals and small teams, ModelArts provides an easy-to-use labeling tool to minimize project management costs.

    In addition, the labeling platform ensures data security. User data is used only within the authorized scope. The labeling object allocation policy ensures user data privacy and implements data anonymization.

    diff --git a/docs/modelarts/umn/modelarts_01_0013.html b/docs/modelarts/umn/modelarts_01_0013.html index e9b0704d9..a65b6b02e 100644 --- a/docs/modelarts/umn/modelarts_01_0013.html +++ b/docs/modelarts/umn/modelarts_01_0013.html @@ -1,10 +1,10 @@

    DevEnviron

    -

    It is challenging to set up a development environment, select an AI algorithm framework and algorithm, debug code, install software, and accelerate hardware. To address these challenges, ModelArts provides DevEnviron to simplify the entire development process.

    +

    It is challenging to set up a development environment, select an AI algorithm framework and algorithm, debug code, install software, and accelerate hardware. To address these challenges, ModelArts provides DevEnviron to simplify the entire development process.

    Figure 1 Algorithm development
    • Mainstream AI algorithm frameworks supported

      In the machine learning and deep learning fields, popular open-source training and inference frameworks include TensorFlow, PyTorch, MXNet, and MindSpore. ModelArts supports all popular AI computing frameworks and provides a user-friendly development and debugging environment. It supports traditional machine learning algorithms, such as logistic regression, decision tree, and clustering, as well as multiple types of deep learning algorithms, such as the convolutional neural network (CNN), recurrent neural network (RNN), and long short-term memory (LSTM).

      -
    • Simplified algorithm development for distributed training

      Deep learning generally requires large-scale GPU clusters for distributed acceleration. For existing open-source frameworks, algorithm developers need to write a large amount of code for distributed training on different hardware, and the acceleration code varies depending on the framework. To resolve these issues, a distributed lightweight framework or SDK is required. The framework or SDK is built on deep learning engines such as TensorFlow, PyTorch, MXNet, and MindSpore to improve the distributed performance and usability of these engines. ModelArts MoXing perfectly suits the needs. The easy-to-use MoXing API/SDK enables you to develop deep learning at low costs.

      +
    • Simplified algorithm development for distributed training

      Deep learning generally requires large-scale GPU clusters for distributed acceleration. For existing open-source frameworks, algorithm developers need to write a large amount of code for distributed training on different hardware, and the acceleration code varies depending on the framework. To resolve these issues, a distributed lightweight framework or SDK is required. The framework or SDK is built on deep learning engines such as TensorFlow, PyTorch, MXNet, and MindSpore to improve the distributed performance and usability of these engines. ModelArts MoXing perfectly suits the needs. The easy-to-use MoXing API/SDK enables you to develop deep learning at low costs.

    diff --git a/docs/modelarts/umn/modelarts_01_0015.html b/docs/modelarts/umn/modelarts_01_0015.html index bf16f2150..6702250c3 100644 --- a/docs/modelarts/umn/modelarts_01_0015.html +++ b/docs/modelarts/umn/modelarts_01_0015.html @@ -1,11 +1,10 @@

    Model Deployment

    -

    -

    Generally, AI model deployment and large-scale implementation are complex.

    +

    Generally, AI model deployment and large-scale implementation are complex.

    ModelArts resolves this issue by deploying a trained model on different devices in various scenarios with only a few clicks. This secure and reliable one-stop deployment is available for individual developers, enterprises, and device manufacturers.

    Figure 1 Process of deploying a model
    -
    • The real-time inference service features high concurrency, low latency, and elastic scaling.
    • Models can be deployed as real-time inference services and batch inference tasks.
    +
    • The real-time inference service features high concurrency, low latency, and elastic scaling.
    • Models can be deployed as real-time inference services and batch inference tasks.
    diff --git a/docs/modelarts/umn/modelarts_05_0003.html b/docs/modelarts/umn/modelarts_05_0003.html index 65866c7e1..7fda164ee 100644 --- a/docs/modelarts/umn/modelarts_05_0003.html +++ b/docs/modelarts/umn/modelarts_05_0003.html @@ -1,11 +1,13 @@ -

    What Are the Relationships Between ModelArts and Other Services

    -

    OBS

    ModelArts uses Object Storage Service (OBS) to store data and model backups and snapshots. OBS provides secure, reliable, low-cost storage. For more details, see Object Storage Service Console Function Overview.

    +

    What Are The Relationships Between ModelArts And Other Services

    +

    IAM

    ModelArts uses Identity and Access Management (IAM) for authentication and authorization. For more information about IAM, see Identity and Access Management User Guide.

    -

    CCE

    ModelArts uses Cloud Container Engine (CCE) to deploy models as real-time services. CCE enables high concurrency and provides elastic scaling. For more information about CCE, see Cloud Container Engine User Guide.

    +

    OBS

    ModelArts uses Object Storage Service (OBS) to store data and model backups and snapshots. OBS provides secure, reliable, low-cost storage. For more details, see Object Storage Service Console Function Overview.

    -

    SWR

    To use an AI framework that is not supported by ModelArts, use SoftWare Repository for Container (SWR) to customize an image and import the image to ModelArts for training or inference. For more details, see .

    +

    CCE

    ModelArts uses Cloud Container Engine (CCE) to deploy models as real-time services. CCE enables high concurrency and provides elastic scaling. For more information about CCE, see Cloud Container Engine User Guide.

    +
    +

    SWR

    To use an AI framework that is not supported by ModelArts, use SoftWare Repository for Container (SWR) to customize an image and import the image to ModelArts for training or inference. For more details, see SoftWare Repository for Container User Guide.

    diff --git a/docs/modelarts/umn/modelarts_05_0012.html b/docs/modelarts/umn/modelarts_05_0012.html index 24d62e69d..d55f400a3 100644 --- a/docs/modelarts/umn/modelarts_05_0012.html +++ b/docs/modelarts/umn/modelarts_05_0012.html @@ -1,7 +1,8 @@

    What Types of Services Can Models Be Deployed as on ModelArts?

    -
    +

    Currently, models can only be deployed as real-time services and batch services.

    +
    Parent topic: Service Deployment
    diff --git a/docs/modelarts/umn/modelarts_05_0013.html b/docs/modelarts/umn/modelarts_05_0013.html index 0808c007d..1d9de8ffd 100644 --- a/docs/modelarts/umn/modelarts_05_0013.html +++ b/docs/modelarts/umn/modelarts_05_0013.html @@ -1,7 +1,7 @@

    How Do I Upload Data to OBS?

    -

    Before using ModelArts to develop AI models, data needs to be uploaded to an OBS bucket. You can log in to the OBS console to create an OBS bucket, create a folder, and upload data. For details about how to upload data, see .

    +

    Before using ModelArts to develop AI models, data needs to be uploaded to an OBS bucket. You can log in to the OBS console to create an OBS bucket, create a folder, and upload data. For details about how to upload data, see Object Storage Service Getting Started.

    diff --git a/docs/modelarts/umn/modelarts_05_0014.html b/docs/modelarts/umn/modelarts_05_0014.html index f56cec4af..35f49ce4f 100644 --- a/docs/modelarts/umn/modelarts_05_0014.html +++ b/docs/modelarts/umn/modelarts_05_0014.html @@ -6,7 +6,7 @@
    • What Is ModelArts?
      • -
    • What Are the Relationships Between ModelArts and Other Services
      +
    • What Are The Relationships Between ModelArts And Other Services
    • How Do I Obtain Access Keys?
      • @@ -16,8 +16,6 @@
    • How Do I Use ModelArts to Train Models Based on Structured Data?
      • -
    • Why Cannot I Find the OBS Bucket on ModelArts After Uploading Data to OBS?
      -
    • Does ModelArts Support Multiple Projects?
    • How Do I View All Files in an OBS Directory on ModelArts?
      diff --git a/docs/modelarts/umn/modelarts_05_0022.html b/docs/modelarts/umn/modelarts_05_0022.html new file mode 100644 index 000000000..613f033cd --- /dev/null +++ b/docs/modelarts/umn/modelarts_05_0022.html @@ -0,0 +1,30 @@ + + +

      How Do I Install External Libraries in a Notebook Instance?

      +

      Multiple environments have been integrated into ModelArts Notebook. These environments contain Jupyter Notebook and Python packages, including TensorFlow, MXNet, Caffe, PyTorch, and Spark. You can use pip install to install external libraries in Jupyter Notebook or on the Terminal page.

      +

      Installing External Libraries in Jupyter Notebook

      For example, use Jupyter Notebook to install Shapely in the TensorFlow-1.8 environment.

      +
      1. Open a notebook instance.
      2. On the Jupyter Notebook dashboard, choose New > TensorFlow-1.8.
      3. In the new notebook instance, enter the following command in the code input bar:

        !pip install Shapely

        +
      +
      +

      Installing External Libraries on the Terminal Page

      For example, use pip to install Shapely in the TensorFlow-1.8 environment on the Terminal page.

      +
      1. Open a notebook instance.
      2. On the Jupyter Notebook dashboard, choose New > Terminal.
      3. Enter the following command in the code input bar to obtain the command for activating TensorFlow-1.8 and activate it:

        cat /home/ma-user/README

        +

        source /home/ma-user/anaconda3/bin/activate TensorFlow-1.8

        +

        If you use another engine, replace TensorFlow-1.8 in the command with the name and version of the engine.

        +
        +
        Figure 1 Activating the environment
        +
      4. Type the following command in the code input bar to install Shapely:

        pip install Shapely

        +
      +

      A new independent running environment is opened when a ModelArts training job is created. The new environment is not associated with the packages installed in the notebook environment. Therefore, add os.system('pip install xxx') to the boot code before importing the installation package.

      +

      For example, if you need to use the Shapely dependency package in the training job, add the following code to the boot code after the notebook instance is installed:

      +
      import os
+os.system('pip install Shapely')
+import Shapely
      +
      +
      +
      +
      +
      +
      Parent topic: Notebook
      +
      +
      + diff --git a/docs/modelarts/umn/modelarts_05_0024.html b/docs/modelarts/umn/modelarts_05_0024.html new file mode 100644 index 000000000..89108e5a0 --- /dev/null +++ b/docs/modelarts/umn/modelarts_05_0024.html @@ -0,0 +1,56 @@ + + +

      How Do I Upload a File from a Notebook Instance to OBS or Download a File from OBS to a Notebook Instance?

      +

      In a notebook instance, you can call the ModelArts MoXing API or SDK to exchange data with OBS for uploading a file to OBS or downloading a file from OBS to the notebook instance.

      +
      Figure 1 Uploading or downloading a file
      +

      Method 1: Using MoXing to upload and download a file

      Developed by the ModelArts team, MoXing is a distributed training acceleration framework built on open-source deep learning engines such as TensorFlow and PyTorch. MoXing makes model coding easier and more efficient.

      +

      MoXing provides a set of file object APIs for reading and writing OBS files.

      +

      Sample code:

      +
      import moxing as mox
+
+# Download the OBS folder sub_dir_0 from OBS to a notebook instance.
+mox.file.copy_parallel('obs://bucket_name/sub_dir_0', '/home/ma-user/work/sub_dir_0')
+# Download the OBS file obs_file.txt from OBS to a notebook instance.
+mox.file.copy('obs://bucket_name/obs_file.txt', '/home/ma-user/work/obs_file.txt')
+
+# Upload the OBS folder sub_dir_0 from a notebook instance to OBS.
+mox.file.copy_parallel('/home/ma-user/work/sub_dir_0', 'obs://bucket_name/sub_dir_0')
+# Upload the OBS file obs_file.txt from a notebook instance to OBS.
+mox.file.copy('/home/ma-user/work/obs_file.txt', 'obs://bucket_name/obs_file.txt')
      +
      +

      Method 2: Using SDK to upload and download a file

      Call the ModelArts SDK for downloading a file from OBS.

      +

      Sample code: Download file1.txt from OBS to /home/ma-user/work/ in the notebook instance. All the bucket name, folder name, and file name are customizable.

      +
      1
+2
+3
      from modelarts.session import Session
+session = Session()
+session.obs.download_file(src_obs_file="obs://bucket-name/dir1/file1.txt", dst_local_dir="/home/ma-user/work/")
+
      +
      +

      Call the ModelArts SDK for downloading a folder from OBS.

      +

      Sample code: Download dir1 from OBS to /home/ma-user/work/ in the notebook instance. The bucket name and folder name are customizable.

      +
      from modelarts.session import Session
+session = Session()
+session.obs.download_dir(src_obs_dir="obs://bucket-name/dir1/", dst_local_dir="/home/ma-user/work/")
      +

      Call the ModelArts SDK for uploading a file to OBS.

      +

      Sample code: Upload file1.txt in the notebook instance to OBS bucket obs://bucket-name/dir1/. All the bucket name, folder name, and file name are customizable.

      +
      1
+2
+3
      from modelarts.session import Session
+session = Session()
+session.obs.upload_file(src_local_file='/home/ma-user/work/file1.txt', dst_obs_dir='obs://bucket-name/dir1/')
+
      +
      +

      Call the ModelArts SDK for uploading a folder to OBS.

      +

      Sample code: Upload /work/ in the notebook instance to obs://bucket-name/dir1/work/ of bucket-name. The bucket name and folder name are customizable.

      +
      from modelarts.session import Session
+session = Session()
+session.obs.upload_dir(src_local_dir='/home/ma-user/work/', dst_obs_dir='obs://bucket-name/dir1/')
      +
      +
      +
      +
      +
      Parent topic: Notebook
      +
      +
      + diff --git a/docs/modelarts/umn/modelarts_05_0032.html b/docs/modelarts/umn/modelarts_05_0032.html index 6c135d2ca..f4fc97471 100644 --- a/docs/modelarts/umn/modelarts_05_0032.html +++ b/docs/modelarts/umn/modelarts_05_0032.html @@ -6,7 +6,7 @@

      Solution

      1. Confirm that the OBS path in the log exists.

        Locate the incorrect OBS path in the log, for example, obs-test/ModelArts/examples/. There are two methods to check whether it exists.

        • On OBS Console, check whether the OBS path exists.

          Log in to OBS console using the current account, and check whether the OBS buckets, folders, and files exist in the OBS path displayed in the log. For example, you can confirm that a given bucket is there and then check if that bucket contains the folder you are looking for based on the configured path.

          • If the file path exists, go to 2.
          • If it does not exist, change the path configured for the training job to an OBS bucket path that is actually there.
          -
        • Create a notebook instance, and use an API to check whether the directory exists. In an existing notebook instance or after creating a new notebook instance, run the following command to check whether the directory exists:
          import moxing as mox
+
        • Create a notebook instance, and use an API to check whether the directory exists. In an existing notebook instance or after creating a new notebook instance, run the following command to check whether the directory exists:
          import moxing as mox
 mox.file.exists('obs://obs-test/ModelArts/examples/')
          • If it exists, go to 2.
          • If it does not exist, change it to an available OBS bucket path in the training job.
        diff --git a/docs/modelarts/umn/modelarts_05_0063.html b/docs/modelarts/umn/modelarts_05_0063.html index 6e82da725..d80291523 100644 --- a/docs/modelarts/umn/modelarts_05_0063.html +++ b/docs/modelarts/umn/modelarts_05_0063.html @@ -6,21 +6,21 @@

    Create a file named pip-requirements.txt in the code directory. In this file, specify the name and version of the dependency package in the format of Package name==Version.

    For example, the OBS path specified by Code Directory contains model files and the pip-requirements.txt file. The following shows the code directory structure:

    -
    |---OBS path to the model boot file
+
    |---OBS path to the model boot file
      |---model.py               #Model boot file
      |---pip-requirements.txt   #Customized configuration file, which specifies the name and version of the dependency package
    
 
    The following shows the content of the pip-requirements.txt file:
    
-
    alembic==0.8.6
+
    alembic==0.8.6
 bleach==1.4.3
 click==6.6
    
 
  • Customized WHL file
    When you use a customized .whl file, the system cannot automatically download and install the file. Place the .whl file in the code directory, create a file named pip-requirements.txt, and specify the name of the .whl file in the created file. The dependency package must be a .whl file.
    
 
    For example, the OBS path specified by Code Directory contains model files, .whl file, and pip-requirements.txt file. The following shows the code directory structure:
    
-
    |---OBS path to the model boot file
+
    |---OBS path to the model boot file
      |---model.py               #Model boot file
      |---XXX.whl                #Dependency package. If multiple dependencies are required, place all of them here.
      |---pip-requirements.txt   #Customized configuration file, which specifies the name of the dependency package
    
 
    The following shows the content of the pip-requirements.txt file:
    
-
    numpy-1.15.4-cp36-cp36m-manylinux1_x86_64.whl
+
    numpy-1.15.4-cp36-cp36m-manylinux1_x86_64.whl
 tensorflow-1.8.0-cp36-cp36m-manylinux1_x86_64.whl
    
 
    •
    diff --git a/docs/modelarts/umn/modelarts_05_0067.html b/docs/modelarts/umn/modelarts_05_0067.html index 3ad93889c..bbfe7f8a6 100644 --- a/docs/modelarts/umn/modelarts_05_0067.html +++ b/docs/modelarts/umn/modelarts_05_0067.html @@ -6,7 +6,7 @@
    • How Do I Enable the Terminal Function in DevEnviron of ModelArts?
      • -
    • How Do I Install External Libraries in a Notebook Instance?
      +
    • How Do I Install External Libraries in a Notebook Instance?
    • Is the Keras Engine Supported?
      • @@ -14,6 +14,8 @@
    • What Should I Do When the System Displays an Error Message Indicating that No Space Left After I Run the pip install Command?
      • +
    • How Do I Upload a File from a Notebook Instance to OBS or Download a File from OBS to a Notebook Instance?
      +
    • How Do I Upload Local Files to a Notebook Instance?
    • Where Will the Data Be Uploaded to?
      diff --git a/docs/modelarts/umn/modelarts_05_0081.html b/docs/modelarts/umn/modelarts_05_0081.html index f82b32ae1..e3b020826 100644 --- a/docs/modelarts/umn/modelarts_05_0081.html +++ b/docs/modelarts/umn/modelarts_05_0081.html @@ -1,7 +1,7 @@

      Where Is Data Stored After the Sync OBS Function Is Used?

      -
      1. Log in to the ModelArts management console, and choose DevEnviron > Notebooks.
      2. In the Operation column of the target notebook instance in the notebook list, click Open to go to the Jupyter page.
      3. On the Files tab page of the Jupyter page, select the target file and click Sync OBS in the upper part of the page to synchronize the file. The file is stored in the ~/work directory of the instance.
      4. On the Files tab page of the Jupyter page, click New and select Terminal. The Terminal page is displayed.
      5. Run the following command to go to the ~/work directory.
        cd work
        +
        1. Log in to the ModelArts management console, and choose DevEnviron > Notebooks.
        2. In the Operation column of the target notebook instance in the notebook list, click Open to go to the Jupyter page.
        3. On the Files tab page of the Jupyter page, select the target file and click Sync OBS in the upper part of the page to synchronize the file. The file is stored in the ~/work directory of the instance.
        4. On the Files tab page of the Jupyter page, click New and select Terminal. The Terminal page is displayed.
        5. Run the following command to go to the ~/work directory.
          cd work
        6. Run the ls command in the ~/work directory to view the files.
        diff --git a/docs/modelarts/umn/modelarts_05_0090.html b/docs/modelarts/umn/modelarts_05_0090.html index 1d9127cdf..638f71685 100644 --- a/docs/modelarts/umn/modelarts_05_0090.html +++ b/docs/modelarts/umn/modelarts_05_0090.html @@ -1,7 +1,7 @@

        What Are Sizes of the /cache Directories for Different Resource Specifications in the Training Environment?

        -

        When creating a training job, you can select CPU, GPU, or Ascend resources based on the size of the training job.

        +

        When creating a training job, you can select CPU, GPU resources based on the size of the training job.

        ModelArts mounts the disk to the /cache directory. You can use this directory to store temporary files. The /cache directory shares resources with the code directory. The directory has different capacities for different resource specifications.

        • GPU resources
          Table 1 Capacities of the cache directories for GPU resources

          GPU Specifications

          @@ -49,22 +49,6 @@
        -
        • Ascend resources -
          - - - - - - - -
          Table 3 Capacities of the cache directories for Ascend resources

          Ascend Specifications

          -

          cache Directory Capacity

          -

          Ascend 910

          -

          3T

          -
          -
          -
        diff --git a/docs/modelarts/umn/modelarts_05_0128.html b/docs/modelarts/umn/modelarts_05_0128.html index 616f0ac2e..378e96bfb 100644 --- a/docs/modelarts/umn/modelarts_05_0128.html +++ b/docs/modelarts/umn/modelarts_05_0128.html @@ -114,7 +114,7 @@

        AI Engine and Version

        -

        Supported CUDA or Ascend Version

        +

        Supported CUDA Version

        diff --git a/docs/modelarts/umn/modelarts_08_0002.html b/docs/modelarts/umn/modelarts_08_0002.html index 92805b4c5..a93c3cc0f 100644 --- a/docs/modelarts/umn/modelarts_08_0002.html +++ b/docs/modelarts/umn/modelarts_08_0002.html @@ -5,6 +5,7 @@

        Obtaining an Access Key

        1. On the ModelArts management console, hover over the username in the upper right corner and choose My Credentials from the drop-down list.
        2. On the My Credentials page, choose Access Keys > Create Access Key.
        3. In the Create Access Key dialog box that is displayed, enter the verification code received by SMS or email.
        4. Click OK and save the access key file as prompted. The access key file is saved in the default download folder of the browser. Open the credentials.csv file to view the access key (Access Key Id and Secret Access Key).

        Adding an Access Key

        1. Log in to the ModelArts management console. In the navigation pane, choose Settings. The Settings page is displayed.
        2. Click Add Authorization.
        3. In the Add Authorization dialog box that is displayed, set Authorization Method to AK/SK. The username is fixed. Enter the obtained access key pair.
          • AK: Enter the value of the Access Key Id field in the key file.
          • SK: Enter the value of the Secret Access Key field in the key file.
          +

        4. Select I have read and agree to the ModelArts Service Statement and click Agree.

          After the configuration is complete, you can view the access key configurations of an account or IAM user on the Settings page.

        diff --git a/docs/modelarts/umn/modelarts_08_0003.html b/docs/modelarts/umn/modelarts_08_0003.html index 07b2a416a..cbca9208f 100644 --- a/docs/modelarts/umn/modelarts_08_0003.html +++ b/docs/modelarts/umn/modelarts_08_0003.html @@ -2,9 +2,9 @@

        Creating an OBS Bucket

        ModelArts uses OBS to store data and model backups and snapshots, achieving secure, reliable, and low-cost storage. Therefore, before using ModelArts, create an OBS bucket and folders for storing data.

        -

        Procedure

        1. Log in to OBS Console and create an OBS bucket. For details, see Creating a Bucket. For example, create an OBS bucket named c-flowers.

          The created OBS bucket and ModelArts are in the same region.

          +

          Procedure

          1. Log in to OBS Console and create an OBS bucket. For details, see Creating a Bucket. For example, create an OBS bucket named c-flowers.

            The created OBS bucket and ModelArts are in the same region.Create a folder for storing data. For details, see Creating a Folder. For example, create a folder named flowers in the created c-flowers OBS bucket.

            -
          2. Create a folder for storing data. For details, see Creating a Folder. For example, create a folder named flowers in the created c-flowers OBS bucket.
          +
        diff --git a/docs/modelarts/umn/modelarts_08_0005.html b/docs/modelarts/umn/modelarts_08_0005.html index c7e4d6f02..b2cc14e4a 100644 --- a/docs/modelarts/umn/modelarts_08_0005.html +++ b/docs/modelarts/umn/modelarts_08_0005.html @@ -3,7 +3,7 @@

        Overview

        When you use ExeML, data management, notebook instances, training jobs, models, and services, ModelArts may need to access dependent services such as OBS and Software Repository for Container (SWR). If ModelArts is not authorized to access the services, these functions cannot be used.

        You can configure access authorization in either of the following ways:

        -
        • Using an agency (recommended)

          After agency authorization is configured, the dependent service operation permissions are delegated to ModelArts so that ModelArts can use the dependent services and perform operations on resources on your behalf.

          +
          • Using an agency (recommended)

            After agency authorization is configured, the dependent service operation permissions are delegated to ModelArts so that ModelArts can use the dependent services and perform operations on resources on your behalf.

          • Using the access key

            You can use the obtained access key pair (AK/SK) to authorize ModelArts to access dependent services and and perform operations on resources.

          Precautions

          • Agency authorization grants ModelArts permissions on dependent services, such as OBS and SWR. If the OBS permissions are not configured for an IAM user, the user still does not have the permission to operate the services.
          • For users who have used ModelArts before, access key authorization has been configured and does not need to be configured again. However, you are advised to use agency authorization again.
          • For new users, use agency authorization.
          diff --git a/docs/modelarts/umn/modelarts_08_0007.html b/docs/modelarts/umn/modelarts_08_0007.html index 612b49a6d..a85bc35cc 100644 --- a/docs/modelarts/umn/modelarts_08_0007.html +++ b/docs/modelarts/umn/modelarts_08_0007.html @@ -27,16 +27,16 @@

          Agency

          -
          • Auto Create (recommended): When you use ModelArts Settings for the first time, no agency is available. In this case, you can click Auto Create to automatically create an agency for the user selected in Username. For details about the automatically created agency, see Automatically Creating an IAM Agency.
          • Select an existing agency: If you have created agencies in IAM, you can select an available agency from the drop-down list to authorize the selected user.
          • Create on IAM: If the automatically created agency cannot meet your requirements, you can click Create on IAM to manually create an agency on the IAM management console. If you choose Create on IAM, configure at least the ModelArts CommonOperation and OBS Operate Access permissions. Otherwise, the basic functions of ModelArts will be unavailable.
          +
          • Auto Create (recommended): When you use ModelArts Settings for the first time, no agency is available. In this case, you can click Auto Create to automatically create an agency for the user selected in Username. For details about the automatically created agency, see Automatically Creating an IAM Agency.
          • Select an existing agency: If you have created agencies in IAM, you can select an available agency from the drop-down list to authorize the selected user.
          • Create on IAM: If the automatically created agency cannot meet your requirements, you can click Create on IAM to manually create an agency on the IAM management console. If you choose Create on IAM, configure at least the ModelArts CommonOperation and OBS Operate Access permissions. Otherwise, the basic functions of ModelArts will be unavailable.
          -
        • After configuring the username and agency, select I have read and agree to the ModelArts Service Statement and click Agree.

          After the configuration is complete, you can view the agency configurations of an account or IAM user on the Settings page.

          +
        • After configuring the username and agency, select I have read and agree to the ModelArts Service Statement and click Agree.

          After the configuration is complete, you can view the agency configurations of an account or IAM user on the Settings page.

      -

      Automatically Creating an IAM Agency

      • Agency Name: For a cloud account, the agency name is modelarts_agency. For an IAM user, the agency name is ma_agency_<IAM username>.
      • Agency Type: Select Cloud service.
      • Cloud Service: Select ModelArts.
      • Validity Period: Select Unlimited.
      • Permissions: The ModelArts CommonOperations, OBS OperateAccess, and Tenant Administrator (required for using other dependent services) permissions are automatically added for this agency to use all ModelArts functions.
      +

      Automatically Creating an IAM Agency

      • Agency Name: For a cloud account, the agency name is modelarts_agency. For an IAM user, the agency name is ma_agency_<IAM username>.
      • Agency Type: Select Cloud service.
      • Cloud Service: Select ModelArts.
      • Validity Period: Select Unlimited.
      • Permissions: The ModelArts CommonOperations, OBS OperateAccess, and Tenant Administrator (required for using other dependent services) permissions are automatically added for this agency to use all ModelArts functions.

      Deleting Authorizations

      To better manage your authorization, you can delete the authorization of an IAM user or delete the authorizations of all users in batches.

      • Deleting the authorization of a user

        On the Settings page, the authorizations configured for IAM users under the current account are displayed. You can click Delete in the Operation column to delete the authorization of a user. After the deletion takes effect, the user cannot use ModelArts functions.

        diff --git a/docs/modelarts/umn/modelarts_21_0003.html b/docs/modelarts/umn/modelarts_21_0003.html index 8797718a9..4597339d4 100644 --- a/docs/modelarts/umn/modelarts_21_0003.html +++ b/docs/modelarts/umn/modelarts_21_0003.html @@ -8,7 +8,7 @@

      Requirements on Datasets

      • The name of files in a dataset cannot contain Chinese characters, plus signs (+), spaces, or tabs.
      • Ensure that no damaged image exists. The supported image formats include JPG, JPEG, BMP, and PNG.
      • Do not store data of different projects in the same dataset.
      • Prepare sufficient data and balance each class of data. To achieve better results, prepare at least 100 images of each class in a training set for image classification.
      • To ensure the prediction accuracy of models, the training samples must be similar to the actual application scenarios.
      • To ensure the generalization capability of models, datasets should cover all possible scenarios.
      -

      Requirements for Files Uploaded to OBS

      • If you do not need to upload training data in advance, create an empty folder to store files generated in the future, for example, /bucketName/data-cat.
      • If you need to upload images to be labeled in advance, create an empty folder and save the images in the folder. An example of the image directory structure is /bucketName/data-cat/cat.jpg.
      • If you want to upload labeled images to the OBS bucket, upload them according to the following specifications:
        • The dataset for image classification requires storing labeled objects and their label files (in one-to-one relationship with the labeled objects) in the same directory. For example, if the name of the labeled object is 10.jpg, the name of the label file must be 10.txt.
          Example of data files:
          ├─<dataset-import-path>
+
          Requirements for Files Uploaded to OBS
          • If you do not need to upload training data in advance, create an empty folder to store files generated in the future, for example, /bucketName/data-cat.
          • If you need to upload images to be labeled in advance, create an empty folder and save the images in the folder. An example of the image directory structure is /bucketName/data-cat/cat.jpg.
          • If you want to upload labeled images to the OBS bucket, upload them according to the following specifications:
            • The dataset for image classification requires storing labeled objects and their label files (in one-to-one relationship with the labeled objects) in the same directory. For example, if the name of the labeled object is 10.jpg, the name of the label file must be 10.txt.
              Example of data files:
              ├─<dataset-import-path>
       │      10.jpg
       │      10.txt    
       │      11.jpg 
@@ -17,7 +17,7 @@
       │      12.txt
              
 
              
 
            • Images in JPG, JPEG, PNG, and BMP formats are supported. When uploading images on the ModelArts management console, ensure that the size of an image does not exceed 5 MB and the total size of images to be uploaded in one attempt does not exceed 8 MB. If the data volume is large, use OBS Browser+ to upload images.
            • A label name can contain a maximum of 32 characters, including Chinese characters, letters, digits, hyphens (-), and underscores (_).
            • Image classification label file (.txt) rule:
              Each row contains only one label.
              
-
              cat
+
              cat
 dog
 ...
              
 
            
diff --git a/docs/modelarts/umn/modelarts_21_0004.html b/docs/modelarts/umn/modelarts_21_0004.html
index 4bed9bdf8..731a3d422 100644
--- a/docs/modelarts/umn/modelarts_21_0004.html
+++ b/docs/modelarts/umn/modelarts_21_0004.html
@@ -46,7 +46,7 @@
 
 
            Label Set
            
 
-
            • Label Name: Enter a label name. The label name can contain only Chinese characters, letters, digits, underscores (_), and hyphens (-), which contains 1 to 32 characters.
            • Add Label: Click Add Label to add one or more labels.
            • Set the label color: You need to set label colors for object detection datasets, but you do not need to set label colors for image classification datasets. Select a color from the color palette on the right of a label, or enter the hexadecimal color code to set the color.
              
+
              • Label Name: Enter a label name. The label name can contain only Chinese characters, letters, digits, underscores (_), and hyphens (-), which contains 1 to 64characters.
              • Add Label: Click Add Label to add one or more labels.
              • Set the label color: You need to set label colors for object detection datasets, but you do not need to set label colors for image classification datasets. Select a color from the color palette on the right of a label, or enter the hexadecimal color code to set the color.
                
 
              
 
 
diff --git a/docs/modelarts/umn/modelarts_21_0006.html b/docs/modelarts/umn/modelarts_21_0006.html
index bc845273c..b3a9d5680 100644
--- a/docs/modelarts/umn/modelarts_21_0006.html
+++ b/docs/modelarts/umn/modelarts_21_0006.html
@@ -51,10 +51,10 @@
 
              Instance Flavor
              
 
 
              Select the resource specifications used for training. By default, the following specifications are supported:
              
-
              • Compute-intensive 1 instance (GPU): This flavor is billed on a pay-per-use basis.
              
+
              • Compute-intensive 1 instance (CPU)
              
 
              The compute flavors are for reference only. Obtain the flavors on the management console.
              
 
-
              ExeML (GPU)
              
+
              ExeML (CPU)
              
 
 
 
diff --git a/docs/modelarts/umn/modelarts_21_0009.html b/docs/modelarts/umn/modelarts_21_0009.html
index 3bc66f495..f74180b09 100644
--- a/docs/modelarts/umn/modelarts_21_0009.html
+++ b/docs/modelarts/umn/modelarts_21_0009.html
@@ -9,7 +9,7 @@
 
              Requirements on Datasets
              • The name of files in a dataset cannot contain Chinese characters, plus signs (+), spaces, or tabs.
              • Ensure that no damaged image exists. The supported image formats include JPG, JPEG, BMP, and PNG.
              • Do not store data of different projects in the same dataset.
              • To ensure the prediction accuracy of models, the training samples must be similar to the actual application scenarios.
              • To ensure the generalization capability of models, datasets should cover all possible scenarios.
              • In an object detection dataset, if the coordinates of the bounding box exceed the boundaries of an image, the image cannot be identified as a labeled image.
              
 
              
 
              Requirements for Files Uploaded to OBS
              • If you do not need to upload training data in advance, create an empty folder to store files generated in the future, for example, /bucketName/data-cat.
              • If you need to upload images to be labeled in advance, create an empty folder and save the images in the folder. An example of the image directory structure is /bucketName/data-cat/cat.jpg.
              • If you want to upload labeled images to the OBS bucket, upload them according to the following specifications:
                • The dataset for object detection requires storing labeled objects and their label files (in one-to-one relationship with the labeled objects) in the same directory. For example, if the name of the labeled object is IMG_20180919_114745.jpg, the name of the label file must be IMG_20180919_114745.xml.
                  The label files for object detection must be in PASCAL VOC format. For details about the format, see Table 1.
                  
-
                  Example of data files:
                  ├─<dataset-import-path> 
+
                  Example of data files:
                  ├─<dataset-import-path> 
       │      IMG_20180919_114732.jpg 
       │      IMG_20180919_114732.xml 
       │      IMG_20180919_114745.jpg 
@@ -90,7 +90,7 @@
 
 
 
                  
-
                  Example of the label file in KITTI format:
                  <annotation>
+
                  Example of the label file in KITTI format:
                  <annotation>
    <folder>test_data</folder>
    <filename>260730932.jpg</filename>
    <size>
diff --git a/docs/modelarts/umn/modelarts_21_0010.html b/docs/modelarts/umn/modelarts_21_0010.html
index 9c3fffb30..bd2653eac 100644
--- a/docs/modelarts/umn/modelarts_21_0010.html
+++ b/docs/modelarts/umn/modelarts_21_0010.html
@@ -46,7 +46,7 @@
 
 
                  Label Set
                  
 
-
                  • Label Name: Enter a label name. The label name can contain only Chinese characters, letters, digits, underscores (_), and hyphens (-), which contains 1 to 32 characters.
                  • Add Label: Click Add Label to add one or more labels.
                  • Set the label color: You need to set label colors for object detection datasets, but you do not need to set label colors for image classification datasets. Select a color from the color palette on the right of a label, or enter the hexadecimal color code to set the color.
                    
+
                    • Label Name: Enter a label name. The label name can contain only Chinese characters, letters, digits, underscores (_), and hyphens (-), which contains 1 to 64 characters.
                    • Add Label: Click Add Label to add one or more labels.
                    • Set the label color: You need to set label colors for object detection datasets, but you do not need to set label colors for image classification datasets. Select a color from the color palette on the right of a label, or enter the hexadecimal color code to set the color.
                      
 
                    
 
 
diff --git a/docs/modelarts/umn/modelarts_21_0012.html b/docs/modelarts/umn/modelarts_21_0012.html
index b913beb78..07772b80d 100644
--- a/docs/modelarts/umn/modelarts_21_0012.html
+++ b/docs/modelarts/umn/modelarts_21_0012.html
@@ -51,10 +51,10 @@
 
                    Instance Flavor
                    
 
 
                    Select the resource specifications used for training. By default, the following specifications are supported:
                    
-
                    • Compute-intensive 1 instance (GPU): This flavor is billed on a pay-per-use basis.
                    
+
                    • Compute-intensive 1 instance (CPU)
                    
 
                    The compute flavors are for reference only. Obtain the flavors on the management console.
                    
 
-
                    ExeML (GPU)
                    
+
                    ExeML (CPU)
                    
 
 
 
diff --git a/docs/modelarts/umn/modelarts_21_0016.html b/docs/modelarts/umn/modelarts_21_0016.html
index fc04f39ec..70cacd4a2 100644
--- a/docs/modelarts/umn/modelarts_21_0016.html
+++ b/docs/modelarts/umn/modelarts_21_0016.html
@@ -2,7 +2,7 @@
 
 
                    Creating a Project
                    
 
                    ModelArts ExeML supports image classification, object detection, and predictive analytics projects. You can create any of them based on your needs. Perform the following operations to create an ExeML project.
                    
-
                    Procedure
                    1. Log in to the ModelArts management console. In the left navigation pane, click ExeML. The ExeML page is displayed. 
                    2. Click Create Project in the box of your desired project. The page for creating an ExeML project is displayed. 
                    3. Enter a project name and set Training Data to the OBS path of the training data. A data file must be specified in the path.
+
                      Procedure
                      1. Log in to the ModelArts management console. In the left navigation pane, click ExeML. The ExeML page is displayed. 
                      2. Click Create Project in the box of your desired project. The page for creating an ExeML project is displayed.
                      3. Enter a project name and set Training Data to the OBS path of the training data. A data file must be specified in the path.
 
                        
-
-
-
-
-
-
-
-
                        Table 1 Parameter description
                        Parameter
                        
                         		
 
                        Description
                        
@@ -12,7 +12,7 @@
 
                        Name
                        
                         		
 
                        Name of an ExeML project
                        
-
                        • Enter a maximum of 20 characters. Only digits, letters, underscores (_), and hyphens (-) are allowed. This parameter is mandatory.
                        • The name must start with a letter.
                        
+
                        • Enter a maximum of 32 characters. Only digits, letters, underscores (_), and hyphens (-) are allowed. This parameter is mandatory.
                        • The name must start with a letter.
                        
                         		
 
                        
 
                        Training Data
                        
diff --git a/docs/modelarts/umn/modelarts_21_0019.html b/docs/modelarts/umn/modelarts_21_0019.html
index 0231b6a05..89143c5db 100644
--- a/docs/modelarts/umn/modelarts_21_0019.html
+++ b/docs/modelarts/umn/modelarts_21_0019.html
@@ -8,43 +8,24 @@
 
                      4. After the model is deployed, view the model deployment status on the Service Deployment page.
                        The deployment takes a certain period of time. If the status in the Version Manager pane changes from Deploying to Running, the deployment is complete.
                        
 
                        5. 
 
-
                        Testing a Service
                        • On the Service Deployment page, select a service type. For example, on the ExeML page, the predictive analytics model is deployed as a real-time service by default. On the Real-Time Services page, click Prediction in the Operation column of the target service to perform a service test. For details, see Testing a Service.
                        • You can also use code to test a service. For details, see Accessing a Real-Time Service.
                        • The following describes the procedure for performing a service test after the predictive analytics model is deployed as a service on the ExeML page.
                          1. After the model is deployed, you can test the model using code. On the ExeML page, click the target project, go to the Deployment Online tab page, select the service version in the Running state, and enter the code in the Code area.
                          2. Click Prediction to perform the test. After the prediction is complete, the result is displayed in the Return Result area on the right. If the model accuracy does not meet your expectation, train and deploy the model again on the Label Data tab page. If you are satisfied with the model prediction result, call the API to access the real-time service as prompted. For details, see Accessing a Real-Time Service.
                            • attr_1 to attr_7 indicate the input data. On the Label Data tab page, the selected label column is attr_7, that is, attr_7 is the target column to be predicted. The value of attr_7 can be set to any value or left blank, which does not affect the prediction result.
                               1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
                              {
-  "data": 
-  {
-    "count": 1,
-    "req_data": 
-	[
-      {
-        "attr_1": "58",
-        "attr_2": "management",
-        "attr_3": "married",
-        "attr_4": "tertiary",
-        "attr_5": "yes",
-        "attr_6": "no",
-        "attr_7": ""
-      }
-    ]
-  }
-}
-
                              
-
                              
+
                              Testing a Service
                              • On the Service Deployment page, select a service type. For example, on the ExeML page, the predictive analytics model is deployed as a real-time service by default. On the Real-Time Services page, click Prediction in the Operation column of the target service to perform a service test. For details, see Testing a Service.
                              • You can also use code to test a service. For details, see Accessing a Real-Time Service.
                              • The following describes the procedure for performing a service test after the predictive analytics model is deployed as a service on the ExeML page.
                                1. After the model is deployed, you can test the model using code. On the ExeML page, click the target project, go to the Deployment Online tab page, select the service version in the Running state, and enter the code in the Code area.
                                2. Click Prediction to perform the test. After the prediction is complete, the result is displayed in the Return Result area on the right. If the model accuracy does not meet your expectation, train and deploy the model again on the Label Data tab page. If you are satisfied with the model prediction result, call the API to access the real-time service as prompted. For details, see Accessing a Real-Time Service.
                                  • attr_1 to attr_7 indicate the input data. On the Label Data tab page, the selected label column is attr_7, that is, attr_7 is the target column to be predicted. The value of attr_7 can be set to any value or left blank, which does not affect the prediction result.
                                    {
+  "data": 
+  {
+    "count": 1,
+    "req_data": 
+	[
+      {
+        "attr_1": "58",
+        "attr_2": "management",
+        "attr_3": "married",
+        "attr_4": "tertiary",
+        "attr_5": "yes",
+        "attr_6": "no",
+        "attr_7": ""
+      }
+    ]
+  }
+}
                                    
 
                                  • In the preceding code snippet, predictioncol is the inference result of label column attr_7.
                                  
 
                                   
                                  A running real-time service keeps consuming the resources. If you do not need to use the real-time service, you are advised to click Stop in the Version Manager pane to stop the service. If you want to use the service again, click Start.
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_21_0038.html b/docs/modelarts/umn/modelarts_21_0038.html
index 013124fce..b0be0d4ee 100644
--- a/docs/modelarts/umn/modelarts_21_0038.html
+++ b/docs/modelarts/umn/modelarts_21_0038.html
@@ -170,403 +170,205 @@
 
                                
 
                              
 
                              Training Script (train_mnist_tf.py)
                              Copy the following code and name the code file train_mnist_tf.py. The code is a training script compiled based on the TensorFlow engine in Python.
                              
-
                               1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
-46
-47
-48
-49
-50
-51
-52
-53
-54
-55
-56
-57
-58
-59
-60
-61
-62
-63
-64
-65
-66
-67
-68
-69
-70
-71
-72
-73
-74
-75
-76
-77
-78
                              from __future__ import absolute_import
-from __future__ import division
-from __future__ import print_function
+
                              from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
 
-import os
+import os
 
-import tensorflow as tf
-from tensorflow.examples.tutorials.mnist import input_data
+import tensorflow as tf
+from tensorflow.examples.tutorials.mnist import input_data
 
-tf.flags.DEFINE_integer('max_steps', 1000, 'number of training iterations.')
-tf.flags.DEFINE_string('data_url', '/home/jnn/nfs/mnist', 'dataset directory.')
-tf.flags.DEFINE_string('train_url', '/home/jnn/temp/delete', 'saved model directory.')
+tf.flags.DEFINE_integer('max_steps', 1000, 'number of training iterations.')
+tf.flags.DEFINE_string('data_url', '/home/jnn/nfs/mnist', 'dataset directory.')
+tf.flags.DEFINE_string('train_url', '/home/jnn/temp/delete', 'saved model directory.')
 
-FLAGS = tf.flags.FLAGS
+FLAGS = tf.flags.FLAGS
 
 
-def main(*args):
-  # Train model
-  print('Training model...')
-  mnist = input_data.read_data_sets(FLAGS.data_url, one_hot=True)
-  sess = tf.InteractiveSession()
-  serialized_tf_example = tf.placeholder(tf.string, name='tf_example')
-  feature_configs = {'x': tf.FixedLenFeature(shape=[784], dtype=tf.float32),}
-  tf_example = tf.parse_example(serialized_tf_example, feature_configs)
-  x = tf.identity(tf_example['x'], name='x')
-  y_ = tf.placeholder('float', shape=[None, 10])
-  w = tf.Variable(tf.zeros([784, 10]))
-  b = tf.Variable(tf.zeros([10]))
-  sess.run(tf.global_variables_initializer())
-  y = tf.nn.softmax(tf.matmul(x, w) + b, name='y')
-  cross_entropy = -tf.reduce_sum(y_ * tf.log(y))
+def main(*args):
+  # Train model
+  print('Training model...')
+  mnist = input_data.read_data_sets(FLAGS.data_url, one_hot=True)
+  sess = tf.InteractiveSession()
+  serialized_tf_example = tf.placeholder(tf.string, name='tf_example')
+  feature_configs = {'x': tf.FixedLenFeature(shape=[784], dtype=tf.float32),}
+  tf_example = tf.parse_example(serialized_tf_example, feature_configs)
+  x = tf.identity(tf_example['x'], name='x')
+  y_ = tf.placeholder('float', shape=[None, 10])
+  w = tf.Variable(tf.zeros([784, 10]))
+  b = tf.Variable(tf.zeros([10]))
+  sess.run(tf.global_variables_initializer())
+  y = tf.nn.softmax(tf.matmul(x, w) + b, name='y')
+  cross_entropy = -tf.reduce_sum(y_ * tf.log(y))
 
-  tf.summary.scalar('cross_entropy', cross_entropy)
+  tf.summary.scalar('cross_entropy', cross_entropy)
 
-  train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy)
+  train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy)
 
-  correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
-  accuracy = tf.reduce_mean(tf.cast(correct_prediction, 'float'))
-  tf.summary.scalar('accuracy', accuracy)
-  merged = tf.summary.merge_all()
-  test_writer = tf.summary.FileWriter(FLAGS.train_url, flush_secs=1)
+  correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
+  accuracy = tf.reduce_mean(tf.cast(correct_prediction, 'float'))
+  tf.summary.scalar('accuracy', accuracy)
+  merged = tf.summary.merge_all()
+  test_writer = tf.summary.FileWriter(FLAGS.train_url, flush_secs=1)
 
-  for step in range(FLAGS.max_steps):
-    batch = mnist.train.next_batch(50)
-    train_step.run(feed_dict={x: batch[0], y_: batch[1]})
-    if step % 10 == 0:
-      summary, acc = sess.run([merged, accuracy], feed_dict={x: mnist.test.images, y_: mnist.test.labels})
-      test_writer.add_summary(summary, step)
-      print('training accuracy is:', acc)
-  print('Done training!')
+  for step in range(FLAGS.max_steps):
+    batch = mnist.train.next_batch(50)
+    train_step.run(feed_dict={x: batch[0], y_: batch[1]})
+    if step % 10 == 0:
+      summary, acc = sess.run([merged, accuracy], feed_dict={x: mnist.test.images, y_: mnist.test.labels})
+      test_writer.add_summary(summary, step)
+      print('training accuracy is:', acc)
+  print('Done training!')
 
-  builder = tf.saved_model.builder.SavedModelBuilder(os.path.join(FLAGS.train_url, 'model'))
+  builder = tf.saved_model.builder.SavedModelBuilder(os.path.join(FLAGS.train_url, 'model'))
 
-  tensor_info_x = tf.saved_model.utils.build_tensor_info(x)
-  tensor_info_y = tf.saved_model.utils.build_tensor_info(y)
+  tensor_info_x = tf.saved_model.utils.build_tensor_info(x)
+  tensor_info_y = tf.saved_model.utils.build_tensor_info(y)
 
-  prediction_signature = (
-      tf.saved_model.signature_def_utils.build_signature_def(
-          inputs={'images': tensor_info_x},
-          outputs={'scores': tensor_info_y},
-          method_name=tf.saved_model.signature_constants.PREDICT_METHOD_NAME))
+  prediction_signature = (
+      tf.saved_model.signature_def_utils.build_signature_def(
+          inputs={'images': tensor_info_x},
+          outputs={'scores': tensor_info_y},
+          method_name=tf.saved_model.signature_constants.PREDICT_METHOD_NAME))
 
-  builder.add_meta_graph_and_variables(
-      sess, [tf.saved_model.tag_constants.SERVING],
-      signature_def_map={
-          'predict_images':
-              prediction_signature,
-      },
-      main_op=tf.tables_initializer(),
-      strip_default_attrs=True)
+  builder.add_meta_graph_and_variables(
+      sess, [tf.saved_model.tag_constants.SERVING],
+      signature_def_map={
+          'predict_images':
+              prediction_signature,
+      },
+      main_op=tf.tables_initializer(),
+      strip_default_attrs=True)
 
-  builder.save()
+  builder.save()
 
-  print('Done exporting!')
+  print('Done exporting!')
 
 
-if __name__ == '__main__':
-  tf.app.run(main=main)
-
                              
-
                              
+if __name__ == '__main__':
+  tf.app.run(main=main)
 
                              
 
                              Inference Code (customize_service.py)
                              Copy the following code and name the code file customize_service.py. The following inference code meets the ModelArts model package specifications.
                              
-
                               1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
                              from PIL import Image
-import numpy as np
-from model_service.tfserving_model_service import TfServingBaseService
+
                              from PIL import Image
+import numpy as np
+from model_service.tfserving_model_service import TfServingBaseService
 
 
-class mnist_service(TfServingBaseService):
-  def _preprocess(self, data):
-    preprocessed_data = {}
+class mnist_service(TfServingBaseService):
+  def _preprocess(self, data):
+    preprocessed_data = {}
 
-    for k, v in data.items():
-      for file_name, file_content in v.items():
-        image1 = Image.open(file_content)
-        image1 = np.array(image1, dtype=np.float32)
-        image1.resize((1, 784))
-        preprocessed_data[k] = image1
+    for k, v in data.items():
+      for file_name, file_content in v.items():
+        image1 = Image.open(file_content)
+        image1 = np.array(image1, dtype=np.float32)
+        image1.resize((1, 784))
+        preprocessed_data[k] = image1
 
-    return preprocessed_data
+    return preprocessed_data
 
-  def _postprocess(self, data):
+  def _postprocess(self, data):
 
-    outputs = {}
-    logits = data['scores'][0]
-    label = logits.index(max(logits))
-    logits = ['%.3f' % logit for logit in logits]
-    outputs['predicted_label'] = str(label)
-    label_list = [str(label) for label in list(range(10))]
-    scores = dict(zip(label_list, logits))
-    scores = sorted(scores.items(), key=lambda item: item[1], reverse=True)[:5]
-    outputs['scores'] = scores
+    outputs = {}
+    logits = data['scores'][0]
+    label = logits.index(max(logits))
+    logits = ['%.3f' % logit for logit in logits]
+    outputs['predicted_label'] = str(label)
+    label_list = [str(label) for label in list(range(10))]
+    scores = dict(zip(label_list, logits))
+    scores = sorted(scores.items(), key=lambda item: item[1], reverse=True)[:5]
+    outputs['scores'] = scores
 
-    return outputs
-
                              
-
                              
+    return outputs
 
                              
 
                              Configuration File (config.json)
                              Copy the following code and name the code file config.json. The configuration file meets the ModelArts model package specifications.
                              
-
                               1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
-46
-47
-48
-49
-50
-51
-52
-53
-54
-55
-56
-57
-58
-59
-60
-61
-62
-63
-64
-65
-66
-67
-68
-69
-70
-71
-72
-73
-74
-75
-76
-77
-78
-79
-80
-81
-82
-83
-84
-85
-86
                              {
-    "model_type":"TensorFlow",
-    "metrics":{
-        "f1":0,
-        "accuracy":0,
-        "precision":0,
-        "recall":0
-    },
-    "dependencies":[
-        {
-            "installer":"pip",
-            "packages":[
-                {
-                    "restraint":"ATLEAST",
-                    "package_version":"1.15.0",
-                    "package_name":"numpy"
-                },
-                {
-                    "restraint":"",
-                    "package_version":"",
-                    "package_name":"h5py"
-                },
-                {
-                    "restraint":"ATLEAST",
-                    "package_version":"1.8.0",
-                    "package_name":"tensorflow"
-                },
-                {
-                    "restraint":"ATLEAST",
-                    "package_version":"5.2.0",
-                    "package_name":"Pillow"
-                }
-            ]
-        }
-    ],
-    "model_algorithm":"image_classification",
-    "apis":[
-        {
-            "procotol":"http",
-            "url":"/",
-            "request":{
-                "Content-type":"multipart/form-data",
-                "data":{
-                    "type":"object",
-                    "properties":{
-                        "images":{
-                            "type":"file"
-                        }
-                    }
-                }
-            },
-            "method":"post",
-            "response":{
-                "Content-type":"multipart/form-data",
-                "data":{
-                    "required":[
-                        "predicted_label",
-                        "scores"
-                    ],
-                    "type":"object",
-                    "properties":{
-                        "predicted_label":{
-                            "type":"string"
-                        },
-                        "scores":{
-                            "items":{
-                                "minItems":2,
-                                "items":[
-                                    {
-                                        "type":"string"
-                                    },
-                                    {
-                                        "type":"number"
-                                    }
-                                ],
-                                "type":"array",
-                                "maxItems":2
-                            },
-                            "type":"array"
-                        }
-                    }
-                }
-            }
-        }
-    ]
-}
-
                              
-
                              
+
                              {
+    "model_type":"TensorFlow",
+    "metrics":{
+        "f1":0,
+        "accuracy":0,
+        "precision":0,
+        "recall":0
+    },
+    "dependencies":[
+        {
+            "installer":"pip",
+            "packages":[
+                {
+                    "restraint":"ATLEAST",
+                    "package_version":"1.15.0",
+                    "package_name":"numpy"
+                },
+                {
+                    "restraint":"",
+                    "package_version":"",
+                    "package_name":"h5py"
+                },
+                {
+                    "restraint":"ATLEAST",
+                    "package_version":"1.8.0",
+                    "package_name":"tensorflow"
+                },
+                {
+                    "restraint":"ATLEAST",
+                    "package_version":"5.2.0",
+                    "package_name":"Pillow"
+                }
+            ]
+        }
+    ],
+    "model_algorithm":"image_classification",
+    "apis":[
+        {
+            "procotol":"http",
+            "url":"/",
+            "request":{
+                "Content-type":"multipart/form-data",
+                "data":{
+                    "type":"object",
+                    "properties":{
+                        "images":{
+                            "type":"file"
+                        }
+                    }
+                }
+            },
+            "method":"post",
+            "response":{
+                "Content-type":"multipart/form-data",
+                "data":{
+                    "required":[
+                        "predicted_label",
+                        "scores"
+                    ],
+                    "type":"object",
+                    "properties":{
+                        "predicted_label":{
+                            "type":"string"
+                        },
+                        "scores":{
+                            "items":{
+                                "minItems":2,
+                                "items":[
+                                    {
+                                        "type":"string"
+                                    },
+                                    {
+                                        "type":"number"
+                                    }
+                                ],
+                                "type":"array",
+                                "maxItems":2
+                            },
+                            "type":"array"
+                        }
+                    }
+                }
+            }
+        }
+    ]
+}
                              
 
                              
 
                        
 
                        
diff --git a/docs/modelarts/umn/modelarts_21_0057.html b/docs/modelarts/umn/modelarts_21_0057.html
index 7e9a37597..58fbbcbb7 100644
--- a/docs/modelarts/umn/modelarts_21_0057.html
+++ b/docs/modelarts/umn/modelarts_21_0057.html
@@ -1,7 +1,7 @@
 
 
 
                        Does ModelArts Support Multiple Projects?
                        
-
                        No. The current ModelArts version does not support multiple projects. Customers can only use it in the default eu-de project. 
                        
+
                        The current version supports multiple projects.
                        
 
                        
 
                        
 
                        
diff --git a/docs/modelarts/umn/modelarts_21_0072.html b/docs/modelarts/umn/modelarts_21_0072.html
index 888e002ff..ee106b9da 100644
--- a/docs/modelarts/umn/modelarts_21_0072.html
+++ b/docs/modelarts/umn/modelarts_21_0072.html
@@ -2,7 +2,7 @@
 
 
                        How Do I View GPU Usage on the Notebook?
                        
 
                        If you select GPU when creating a notebook instance, perform the following operations to view GPU usage:
                        
-
                        1. Log in to the ModelArts management console, and choose DevEnviron > Notebooks.
                        2. In the Operation column of the target notebook instance in the notebook list, click Open to go to the Jupyter page.
                        3. On the Files tab page of the Jupyter page, click New and select Terminal. The Terminal page is displayed.
                        4. Run the following command to view GPU usage:
                          nvidia-smi
                          
+
                          1. Log in to the ModelArts management console, and choose DevEnviron > Notebooks.
                          2. In the Operation column of the target notebook instance in the notebook list, click Open to go to the Jupyter page.
                          3. On the Files tab page of the Jupyter page, click New and select Terminal. The Terminal page is displayed.
                          4. Run the following command to view GPU usage:
                            nvidia-smi
                            
 
                          
 
                        
 
                        
diff --git a/docs/modelarts/umn/modelarts_21_0077.html b/docs/modelarts/umn/modelarts_21_0077.html
index e78da9396..4b960472b 100644
--- a/docs/modelarts/umn/modelarts_21_0077.html
+++ b/docs/modelarts/umn/modelarts_21_0077.html
@@ -4,7 +4,7 @@
 
                        Pay attention to the following when setting training parameters:
                        
 
                        • When setting running parameters for creating a training job, you only need to set the corresponding parameter names and values. See Figure 1.
                          Figure 1 Setting running parameters
                          
 
                        • If a parameter value is an OBS bucket path, use the path to the data. See Figure 2.
                          Figure 2 Configuring an OBS path
                          
-
                        • When creating an OBS folder in code, you need to call a MoXing API as follows:
                          import moxing as mox
+
                        • When creating an OBS folder in code, you need to call a MoXing API as follows:
                          import moxing as mox
 mox.file.make_dirs('obs://bucket_name/sub_dir_0/sub_dir_1')
                          
 
                        
 
                        
diff --git a/docs/modelarts/umn/modelarts_21_0079.html b/docs/modelarts/umn/modelarts_21_0079.html
index 582910228..05ec55462 100644
--- a/docs/modelarts/umn/modelarts_21_0079.html
+++ b/docs/modelarts/umn/modelarts_21_0079.html
@@ -2,7 +2,7 @@
 
 
                        How Do I Check Whether Folder Copy Is Complete During Job Training?
                        
 
                        In the script of the training job boot file, run the following commands to obtain the sizes of the to-be-copied and copied folders. Then determine whether folder copy is complete based on the command output.
                        
-
                        import moxing as mox
+
                        import moxing as mox
 mox.file.get_size('obs://bucket_name/obs_file',recursive=True)
                        
 
                        get_size indicates the size of the file or folder to be obtained. recursive=True indicates that the type is folder. True indicates that the type is folder, and False indicates that the type is file.
                        
 
                        If the command output is consistent, the folder copy is complete. If the command output is inconsistent, the folder copy is not complete.
                        
diff --git a/docs/modelarts/umn/modelarts_21_0080.html b/docs/modelarts/umn/modelarts_21_0080.html
index 52fa7c7fa..5605cea3f 100644
--- a/docs/modelarts/umn/modelarts_21_0080.html
+++ b/docs/modelarts/umn/modelarts_21_0080.html
@@ -3,7 +3,7 @@
 
                        How Do I Obtain Training Job Parameters from the Boot File of the Training Job?
                        
 
                        Training job parameters can be automatically generated in the background or manually entered by users. Perform the following operations to obtain training job parameters:
                        
 
                        1. When a training job is created, train_url in the running parameters of the training job indicates a training output location, and data_url indicates a data source, which is automatically generated in the background. The test parameter is manually entered.
                        2. After the training job is executed, you can click the job name in the training job list to view its details. You can obtain the parameter input mode from logs, as shown in Figure 1.
                          Figure 1 Viewing logs
                          
-
                        3. To obtain the values of train_url, data_url, and test during training, add the following code to the boot file of the training job:
                          import argparse
+
                        4. To obtain the values of train_url, data_url, and test during training, add the following code to the boot file of the training job:
                          import argparse
 parser = argparse.ArgumentParser()
 parser.add_argument('--data_url', type=str, default=None, help='test')
 parser.add_argument('--train_url', type=str, default=None, help='test')
diff --git a/docs/modelarts/umn/modelarts_21_0083.html b/docs/modelarts/umn/modelarts_21_0083.html
index 945358c59..09028c639 100644
--- a/docs/modelarts/umn/modelarts_21_0083.html
+++ b/docs/modelarts/umn/modelarts_21_0083.html
@@ -2,7 +2,7 @@
 
 
                          Only Three Valid Digits Are Retained in a Training Output Log. Can the Value of loss Be Changed?
                          
 
                          In a training job, only three valid digits are retained in a training output log. When the value of loss is too small, the value is displayed as 0.000. Log content is as follows:
                          
-
                          INFO:tensorflow:global_step/sec: 0.382191
+
                          INFO:tensorflow:global_step/sec: 0.382191
 INFO:tensorflow:step: 81600(global step: 81600) sample/sec: 12.098 loss: 0.000
 INFO:tensorflow:global_step/sec: 0.382876
 INFO:tensorflow:step: 81700(global step: 81700) sample/sec: 12.298 loss: 0.000
                          
diff --git a/docs/modelarts/umn/modelarts_21_0084.html b/docs/modelarts/umn/modelarts_21_0084.html
index 2c981cf43..fc190e37a 100644
--- a/docs/modelarts/umn/modelarts_21_0084.html
+++ b/docs/modelarts/umn/modelarts_21_0084.html
@@ -2,7 +2,7 @@
 
 
                          Why Can't I Use os.system ('cd xxx') to Access the Corresponding Folder During Job Training?
                          
 
                          If you cannot access the corresponding folder by using os.system('cd xxx') in the boot script of the training job, you are advised to use the following method:
                          
-
                          import os
+
                          import os
 os.chdir('/home/work/user-job-dir/xxx')
                          
 
                          
 
                          
diff --git a/docs/modelarts/umn/modelarts_21_0085.html b/docs/modelarts/umn/modelarts_21_0085.html
index a83387d50..9544c1982 100644
--- a/docs/modelarts/umn/modelarts_21_0085.html
+++ b/docs/modelarts/umn/modelarts_21_0085.html
@@ -2,7 +2,7 @@
 
 
                          How Do I Invoke a Shell Script in a Training Job to Execute the .sh File?
                          
 
                          ModelArts enables you to invoke a shell script, and you can use Python to invoke .sh. The procedure is as follows:
                          
-
                          1. Upload the .sh script to an OBS bucket. For example, upload the .sh script to /bucket-name/code/test.sh.
                          2. Create the .py file on a local PC, for example, test.py. The background automatically downloads the code directory to the /home/work/user-job-dir/ directory of the container. Therefore, you can invoke the .sh file in the test.py boot file as follows:
                            import os
+
                            1. Upload the .sh script to an OBS bucket. For example, upload the .sh script to /bucket-name/code/test.sh.
                            2. Create the .py file on a local PC, for example, test.py. The background automatically downloads the code directory to the /home/work/user-job-dir/ directory of the container. Therefore, you can invoke the .sh file in the test.py boot file as follows:
                              import os
 os.system('bash /home/work/user-job-dir/code/test.sh')
                              
 
                            3. Upload test.py to OBS. Then the file storage path is /bucket-name/code/test.py.
                            4. When creating a training job, set the code directory to /bucket-name/code/, and the boot file directory to /bucket-name/code/test.py.
                            
 
                            After the training job is created, you can use Python to invoke the .sh file.
                            
diff --git a/docs/modelarts/umn/modelarts_23_0002.html b/docs/modelarts/umn/modelarts_23_0002.html
index 0c662ac7a..013d08038 100644
--- a/docs/modelarts/umn/modelarts_23_0002.html
+++ b/docs/modelarts/umn/modelarts_23_0002.html
@@ -1,124 +1,7 @@
 
 
 
                            Data Management
                            
-
                            ModelArts is easy to use for users with different experience.
                            
-
                            • For service developers without AI development experience, you can use ExeML of ModelArts to build AI models without coding.
                            • For developers who are familiar with code compilation, debugging, and common AI engines, ModelArts provides online code compiling environments as well as AI development lifecycle that covers data preparation, model training, model management, and service deployment, helping the developers build models efficiently and quickly.
                            
-
                            ExeML
                            ExeML is a customized code-free model development tool that helps users start AI application development from scratch with high flexibility. ExeML automates model design, parameter tuning and training, and model compression and deployment with the labeled data. Developers do not need to develop basic and encoding capabilities, but only to upload data and complete model training and deployment as prompted by ExeML.
                            
-
                            For details about how to use ExeML, see Introduction to ExeML.
                            
-
                            
-
                            AI Development Lifecycle
                            The AI development lifecycle on ModelArts takes developers' habits into consideration and provides a variety of engines and scenarios for developers to choose. The following describes the entire process from data preparation to service development using ModelArts.
                            
-
                            Figure 1 Process of using ModelArts
                            
-
-
                            
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
                            Table 1 Process description
                            Task
                            
-
                            Sub Task
                            
-
                            Description
                            
-
                            Reference
                            
-
                            Data preparation
                            
-
                            Creating a dataset
                            
-
                            Create a dataset in ModelArts to manage and preprocess your business data.
                            
-
                            Creating a Dataset
                            
-
                            Labeling data
                            
-
                            Label and preprocess the data in your dataset based on the business logic to facilitate subsequent training. Data labeling affects the model training performance.
                            
-
                            Labeling Data
                            
-
                            Publishing a dataset
                            
-
                            After labeling data, publish the database to generate a dataset version that can be used for model training.
                            
-
                            Publishing a Dataset
                            
-
                            Development
                            
-
                            Creating a notebook instance
                            
-
                            Create a notebook instance as the development environment.
                            
-
                            Creating a Notebook Instance
                            
-
                            Compiling code
                            
-
                            Compile code in an existing notebook to directly build a model.
                            
-
                            Opening a Notebook Instance
                            
-
                            Common Operations on Jupyter Notebook
                            
-
                            Exporting the .py file
                            
-
                            Export the compiled training script as a .py file for subsequent operations, such as model training and management.
                            
-
                            Using the Convert to Python File Function
                            
-
                            Model training
                            
-
                            Creating a training job
                            
-
                            Create a training job, upload and use the compiled training script. After the training is complete, a model is generated and stored in OBS.
                            
-
                            Introduction to Training Jobs
                            
-
                            (Optional) Creating a visualization job
                            
-
                            Create a visualization job (TensorBoard type) to view the model training process, learn about the model, and adjust and optimize the model. Currently, visualization jobs only support the MXNet and TensorFlow engines.
                            
-
                            Managing Visualization Jobs
                            
-
                            Model management
                            
-
                            Compiling inference code and configuration files
                            
-
                            Following the model package specifications provided by ModelArts, compile inference code and configuration files for your model, and save the inference code and configuration files to the training output location.
                            
-
                            Model Package Specifications
                            
-
                            Importing a model
                            
-
                            Import the training model to ModelArts to facilitate service deployment.
                            
-
                            Introduction to Model Management
                            
-
                            Model deployment
                            
-
                            Deploying a model as a service
                            
-
                            Deploy a model as a real-time or batch service.
                            
-                            		
-
                            Accessing the service
                            
-
                            After the service is deployed, access the real-time service, or view the prediction result of the batch service.
                            
-                            		
-
                            
-
                            
-
                            
-
                            
+
                            
 
 
diff --git a/docs/modelarts/umn/modelarts_23_0004.html b/docs/modelarts/umn/modelarts_23_0004.html
index 480de1fe9..44ee5a729 100644
--- a/docs/modelarts/umn/modelarts_23_0004.html
+++ b/docs/modelarts/umn/modelarts_23_0004.html
@@ -2,16 +2,16 @@
 
 
                            Creating a Dataset
                            
 
                            To manage data using ModelArts, create a dataset. Then you can perform operations on the dataset, such as labeling data, importing data, and publishing the dataset.
                            
-
                            Prerequisites
                            • Before using the data management function, you need permissions to access OBS. This function cannot be used if you are not authorized to access OBS. Before using the data management function, go to the Settings page and complete access authorization using an agency.
                            • You have created OBS buckets and folders for storing data. In addition, the OBS buckets and ModelArts are in the same region.
                            • You have uploaded data to be used to OBS. 
                            
+
                            Prerequisites
                            • Before using the data management function, you need permissions to access OBS. This function cannot be used if you are not authorized to access OBS. Before using the data management function, go to the Settings page and complete access authorization using an agency.
                            • You have created OBS buckets and folders for storing data. 
                            • You have uploaded data to be used to OBS. 
                            
 
                            
 
                            Procedure
                            1. Log in to the ModelArts management console. In the left navigation pane, choose Data Management > Datasets. The Datasets page is displayed.
                            2. Click Create Dataset. On the Create Dataset page, create datasets of different types based on the data type and data labeling requirements.
                              1. Set the basic information, the name and description of the dataset.
                                Figure 1 Basic information about a dataset
                                
-
                              2. Select a labeling scene and type as required. For details about the types supported by ModelArts, see Dataset Types.
                                Figure 2 Selecting a labeling scene and type
                                
-
                              3. Set the parameters based on the dataset type. For details, see the parameters of the following dataset types:
+
                              4. Select a labeling scene and type as required. For details about the types supported by ModelArts, see Dataset Types.
                                Figure 2 Selecting a labeling scene and type
                                
+
                              5. Set the parameters based on the dataset type. For details, see the parameters of the following dataset types:
 
                              6. Click Create in the lower right corner of the page.
                                After the dataset is created, the dataset management page is displayed. You can perform the following operations on the dataset: label data, publish dataset versions, manage dataset versions, modify the dataset, import data, and delete the dataset. For details about the operations supported by different types of datasets, see .
                                
 
                              
 
                            
 
                            
-
                            Images (Image Classification, Object Detection, and Image Segmentation)
                            Figure 3 Parameters of datasets for image classification and object detection
                            
+
                            Images (Image Classification, Object Detection, )
                            Figure 3 Parameters of datasets for image classification and object detection
                            
 
 
                            
@@ -37,6 +37,130 @@
 
                          3. Setting label attributes: For an object detection dataset, you can click the plus sign (+) on the right to add label attributes after setting a label color. Label attributes are used to distinguish different attributes of the objects with the same label. For example, yellow kittens and black kittens have the same label cat and their label attribute is color.
                            4. 
+
+
+
+
+
                            Table 1 Dataset parameters
                            Parameter
                            
                             		
 
 
                            Team Labeling
                            
+
                            Enable or disable team labeling. Image segmentation does not support team labeling. Therefore, this parameter is unavailable when you use image segmentation.
                            
+
                            After enabling team labeling, enter the name and type of the team labeling task, and select the labeling team and team members. For details about the parameter settings, see Creating Team Labeling Tasks.
                            
+
                            Before enabling team labeling, ensure that you have added a team and members on the Labeling Teams page. If no labeling team is available, click the link on the page to go to the Labeling Teams page, and add your team and members. For details, see Introduction to Team Labeling.
                            
+
                            After a dataset is created with team labeling enabled, you can view the Team Labeling mark in Labeling Type.
                            
+
                            
+
                            
+
                            
+
                            Audio (Sound Classification, Speech Labeling, and Speech Paragraph Labeling)
                            Figure 4 Parameters of datasets for sound classification, speech labeling, and speech paragraph labeling
                            
+
+
                            
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
                            Parameter
                            
+
                            Description
                            
+
                            Input Dataset Path
                            
+
                            Select the OBS path to the input dataset.
                            
+
                            Output Dataset Path
                            
+
                            Select the OBS path to the output dataset.
                            
+
                             NOTE: 
                            The output dataset path cannot be the same as the input dataset path or cannot be the subdirectory of the input dataset path. Select an empty directory as the Output Dataset Path.
                            
+
                            
+
                            Label Set (Sound Classification)
                            
+
                            Set labels only for datasets of the sound classification type.
                            
+
                            • Label Name: Enter a label name. The label name can contain only letters, digits, underscores (_), and hyphens (-). The name contains 1 to 32 characters.
                            • Add Label: Click Add Label to add more labels.
                            
+
                            Label Management (Speech Paragraph Labeling)
                            
+
                            Only datasets for speech paragraph labeling support multiple labels.
                            
+
                            • Single Label
                              A single label is used to label a piece of audio that has only one class.
                              • Label Name: Enter a label name. The label name can contain contains 1 to 32 characters. Only letters, digits, underscores (_), and hyphens (-) are allowed.
                              • Label Color: Set the label color in the Label Color column. You can select a color from the color palette or enter a hexadecimal color code to set the color.
                              
+
                              
+
                            • Multiple Labels
                              Multiple labels are suitable for multi-dimensional labeling. For example, you can label a piece of audio as both noise and speech. For speech, you can label the audio with different speakers. You can click Add Label Class to add multiple label classes. A label class can contain multiple labels. The label class and name can contain contains 1 to 32 characters. Only letters, digits, underscores (_), and hyphens (-) are allowed.
                              • Label Class: Set a label class.
                              • Label Name: Enter a label name.
                              • Add Label: Click Add Label to add more labels.
                              
+
                              
+
                            
+
                            Speech Labeling (Speech Paragraph Labeling)
                            
+
                            Only datasets for speech paragraph labeling support speech labeling. By default, speech labeling is disabled. If this function is enabled, you can label speech content.
                            
+
                            
+
                            
+
                            
+
                            Text (Text Classification, Named Entity Recognition, and Text Triplet)
                            Figure 5 Parameters of datasets for text classification, named entity recognition, and text triplet
                            
+
+
                            
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
                            Table 2 Dataset parameters
                            Parameter
                            
+
                            Description
                            
+
                            Input Dataset Path
                            
+
                            Select the OBS path to the input dataset.
                            
+
                             NOTE: 
                            Labeled text classification data can be identified only when you import data. When creating a dataset, set an empty OBS directory. After the dataset is created, import the labeled data into it. For details about the format of the data to be imported, see Specifications for Importing Data from an OBS Directory.
                            
+
                            
+
                            Output Dataset Path
                            
+
                            Select the OBS path to the output dataset.
                            
+
                             NOTE: 
                            The output dataset path cannot be the same as the input dataset path or cannot be the subdirectory of the input dataset path. Select an empty directory as the Output Dataset Path.
                            
+
                            
+
                            Label Set (for text classification and named entity recognition)
                            
+
                            • Label Name: Enter a label name. The label name can contain only letters, digits, underscores (_), and hyphens (-). The name contains 1 to 32 characters.
                            • Add Label: Click Add Label to add more labels.
                            • Setting a label color: Select a color from the color palette or enter the hexadecimal color code to set the color.
                              
+
                            
+
                            Label Set (for text triplet)
                            
+
                            For datasets of the text triplet type, set entity labels and relationship labels.
                            
+
                            • Entity Label: Set the label name and label color. You can click the plus sign (+) on the right of the color area to add multiple labels.
                            • Relationship Label: a relationship between two entities. Set the source entity and target entity. Therefore, add at least two entity labels before adding a relationship label.
                            
+
                            
+
                            Team Labeling
                            
+
                            Enable or disable team labeling.
                            
+
                            After enabling team labeling, enter the name and type of the team labeling task, and select the labeling team and team members. For details about the parameter settings, see Creating Team Labeling Tasks.
                            
+
                            Before enabling team labeling, ensure that you have added a team and members on the Labeling Teams page. If no labeling team is available, click the link on the page to go to the Labeling Teams page, and add your team and members. For details, see Introduction to Team Labeling.
                            
+
                            After a dataset is created with team labeling enabled, you can view the Team Labeling mark in Labeling Type.
                            
+
                            
+
                            
+
                            
+
                            Other (Free Format)
                            Figure 6 Parameters of datasets of the free format type
                            
+
+
                            
+
+
+
+
+
+
+
+
+
                            Table 3 Dataset parameters
                            Parameter
                            
+
                            Description
                            
+
                            Input Dataset Path
                            
+
                            Select the OBS path to the input dataset.
                            
+
                            Output Dataset Path
                            
+
                            Select the OBS path to the output dataset.
                            
+
                             NOTE: 
                            The output dataset path cannot be the same as the input dataset path or cannot be the subdirectory of the input dataset path. Select an empty directory as the Output Dataset Path.
                            
+
                            
+
                            
                             
 
                            
 
                            
diff --git a/docs/modelarts/umn/modelarts_23_0006.html b/docs/modelarts/umn/modelarts_23_0006.html
index 9325794b7..8ed4e85f8 100644
--- a/docs/modelarts/umn/modelarts_23_0006.html
+++ b/docs/modelarts/umn/modelarts_23_0006.html
@@ -35,15 +35,6 @@
 
                            Follow the format specifications described in Object Detection.
                            
 
                        		
 
                        Image segmentation
                        
-
                        Supported
                        
-
                        Follow the format specifications described in Image Segmentation.
                        
-
                        Supported
                        
-
                        Follow the format specifications described in Image Segmentation.
                        
-
                        
 
                        Sound classification
                        
                         		
 
                        Supported
                        
@@ -94,14 +85,6 @@
 
                        Follow the format specifications described in Text Triplet.
                        
                         		
 
                        Video
                        
-
                        N/A
                        
-
                        Supported
                        
-
                        Follow the format specifications described in Video Labeling.
                        
-
                        
 
                        Free format
                        
                         		
 
                        N/A
                        
diff --git a/docs/modelarts/umn/modelarts_23_0008.html b/docs/modelarts/umn/modelarts_23_0008.html
index 81552ff17..eeb609ade 100644
--- a/docs/modelarts/umn/modelarts_23_0008.html
+++ b/docs/modelarts/umn/modelarts_23_0008.html
@@ -6,7 +6,7 @@
 
                         
                        To import data from an OBS directory, you must have the read permission on the OBS directory.
                        
 
                        
 
                        Image Classification
                        • Image classification data can be in two modes. The first mode (directory mode) supports only single labels. The second mode (.txt label files) supports multiple labels.
                          • Images with the same label must be stored in the same directory, and the label name is the directory name. If there are multiple levels of directories, the last level is used as the label name.
                            In the following example, Cat and Dog are label names.
                            
-
                            dataset-import-example 
+
                            dataset-import-example 
 ├─Cat 
 │      10.jpg 
 │      11.jpg 
@@ -17,7 +17,7 @@
         2.jpg 
         3.jpg
                            
 
                          • If .txt files exist in the directory, the content in the .txt files is used as the image label. This mode is better than the previous one.
                            In the following example, import-dir-1 and import-dir-2 are the imported subdirectories:
                            
-
                            dataset-import-example 
+
                            dataset-import-example 
 ├─import-dir-1
 │      10.jpg
 │      10.txt    
@@ -38,9 +38,9 @@ Dog
                            
 
                          
 
                        • Only images in JPG, JPEG, PNG, and BMP formats are supported. The size of a single image cannot exceed 5 MB, and the total size of all images uploaded at a time cannot exceed 8 MB.
                        
 
                        
-
                        Object Detection
                        • The simple mode of object detection requires users store labeled objects and their label files (in one-to-one relationship with the labeled objects) in the same directory. For example, if the name of the labeled object file is IMG_20180919_114745.jpg, the name of the label file must be IMG_20180919_114745.xml.
                          The label files for object detection must be in PASCAL VOC format. For details about the format, see Table 8.
                          
+
                          Object Detection
                          • The simple mode of object detection requires users store labeled objects and their label files (in one-to-one relationship with the labeled objects) in the same directory. For example, if the name of the labeled object file is IMG_20180919_114745.jpg, the name of the label file must be IMG_20180919_114745.xml.
                            The label files for object detection must be in PASCAL VOC format. For details about the format, see Table 6.
                            
 
                            Example:
                            
-
                            ├─dataset-import-example 
+
                            ├─dataset-import-example 
 │      IMG_20180919_114732.jpg 
 │      IMG_20180919_114732.xml 
 │      IMG_20180919_114745.jpg 
@@ -48,200 +48,68 @@ Dog
                            
 │      IMG_20180919_114945.jpg 
 │      IMG_20180919_114945.xml
                            
 
                            A label file example is as follows:
                            
-
                             1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
                            <?xml version="1.0" encoding="UTF-8" standalone="no"?>
-<annotation>
-    <folder>NA</folder>
-    <filename>bike_1_1593531469339.png</filename>
-    <source>
-        <database>Unknown</database>
-    </source>
-    <size>
-        <width>554</width>
-        <height>606</height>
-        <depth>3</depth>
-    </size>
-    <segmented>0</segmented>
-    <object>
-        <name>Dog</name>
-        <pose>Unspecified</pose>
-        <truncated>0</truncated>
-        <difficult>0</difficult>
-        <occluded>0</occluded>
-        <bndbox>
-            <xmin>279</xmin>
-            <ymin>52</ymin>
-            <xmax>474</xmax>
-            <ymax>278</ymax>
-        </bndbox>
-    </object>
-    <object>
-        <name>Cat</name>
-        <pose>Unspecified</pose>
-        <truncated>0</truncated>
-        <difficult>0</difficult>
-        <occluded>0</occluded>
-        <bndbox>
-            <xmin>279</xmin>
-            <ymin>198</ymin>
-            <xmax>456</xmax>
-            <ymax>421</ymax>
-        </bndbox>
-    </object>
-</annotation>
-
                            
-
                            
+
                            <?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<annotation>
+    <folder>NA</folder>
+    <filename>bike_1_1593531469339.png</filename>
+    <source>
+        <database>Unknown</database>
+    </source>
+    <size>
+        <width>554</width>
+        <height>606</height>
+        <depth>3</depth>
+    </size>
+    <segmented>0</segmented>
+    <object>
+        <name>Dog</name>
+        <pose>Unspecified</pose>
+        <truncated>0</truncated>
+        <difficult>0</difficult>
+        <occluded>0</occluded>
+        <bndbox>
+            <xmin>279</xmin>
+            <ymin>52</ymin>
+            <xmax>474</xmax>
+            <ymax>278</ymax>
+        </bndbox>
+    </object>
+    <object>
+        <name>Cat</name>
+        <pose>Unspecified</pose>
+        <truncated>0</truncated>
+        <difficult>0</difficult>
+        <occluded>0</occluded>
+        <bndbox>
+            <xmin>279</xmin>
+            <ymin>198</ymin>
+            <xmax>456</xmax>
+            <ymax>421</ymax>
+        </bndbox>
+    </object>
+</annotation>
                            
 
                          
 
                          • Only images in JPG, JPEG, PNG, and BMP formats are supported. The size of a single image cannot exceed 5 MB, and the total size of all images uploaded at a time cannot exceed 8 MB.
                          
 
                          
-
                          Image Segmentation
                          • The simple mode of image segmentation requires users store labeled objects and their label files (in one-to-one relationship with the labeled objects) in the same directory. For example, if the name of the labeled object file is IMG_20180919_114746.jpg, the name of the label file must be IMG_20180919_114746.xml.
                            Fields mask_source and mask_color are added to the label file in PASCAL VOC format. For details about the format, see Table 4.
                            
-
                            Example:
                            
-
                            ├─dataset-import-example 
-│      IMG_20180919_114732.jpg 
-│      IMG_20180919_114732.xml 
-│      IMG_20180919_114745.jpg 
-│      IMG_20180919_114745.xml 
-│      IMG_20180919_114945.jpg 
-│      IMG_20180919_114945.xml
                            
-
                            A label file example is as follows:
                            
-
                             1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
                            <?xml version="1.0" encoding="UTF-8" standalone="no"?>
-<annotation>
-    <folder>NA</folder>
-    <filename>image_0006.jpg</filename>
-    <source>
-        <database>Unknown</database>
-    </source>
-    <size>
-        <width>230</width>
-        <height>300</height>
-        <depth>3</depth>
-    </size>
-    <segmented>1</segmented>
-    <mask_source>obs://xianao/out/dataset-8153-Jmf5ylLjRmSacj9KevS/annotation/V001/segmentationClassRaw/image_0006.png</mask_source>
-    <object>
-        <name>bike</name>
-        <pose>Unspecified</pose>
-        <truncated>0</truncated>
-        <difficult>0</difficult>
-        <mask_color>193,243,53</mask_color>
-        <occluded>0</occluded>
-        <polygon>
-            <x1>71</x1>
-            <y1>48</y1>
-            <x2>75</x2>
-            <y2>73</y2>
-            <x3>49</x3>
-            <y3>69</y3>
-            <x4>68</x4>
-            <y4>92</y4>
-            <x5>90</x5>
-            <y5>101</y5>
-            <x6>45</x6>
-            <y6>110</y6>
-            <x7>71</x7>
-            <y7>48</y7>
-        </polygon>
-    </object>
-</annotation>
-
                            
-
                            
-
                          
-
                          
 
                          Text Classification
                          Text classification supports two import modes.
                          
-
                          • The labeled objects and labels for text classification are in the same text file. You can specify a separator to separate the labeled objects and labels, as well as multiple labeled objects.
                            For example, the following shows an example text file. The Tab key is used to separate the labeled object from the label.
                            It touches good and responds quickly. I don't know how it performs in the future.   positive
+
                            • The labeled objects and labels for text classification are in the same text file. You can specify a separator to separate the labeled objects and labels, as well as multiple labeled objects.
                              For example, the following shows an example text file. The Tab key is used to separate the labeled object from the label.
                              It touches good and responds quickly. I don't know how it performs in the future.   positive
 Three months ago, I bought a very good phone and replaced my old one with it. It can operate longer between charges.  positive
 Why does my phone heat up if I charge it for a while? The volume button stuck after being pressed down.  negative
 It's a gift for Father's Day. The logistics is fast and I received it in 24 hours. I like the earphones because the bass sounds feel good and they would not fall off.  positive
                              
 
                              
 
                            • The labeled objects and label files for text classification are text files, and correspond to each other based on the rows. For example, the first row in a label file indicates the label of the first row in the file of the labeled object.
                              For example, the content of labeled object COMMENTS_20180919_114745.txt is as follows:
                              
-
                              It touches good and responds quickly. I don't know how it performs in the future.
+
                              It touches good and responds quickly. I don't know how it performs in the future.
 Three months ago, I bought a very good phone and replaced my old one with it. It can operate longer between charges.
 Why does my phone heat up if I charge it for a while? The volume button stuck after being pressed down.
 It's a gift for Father's Day. The logistics is fast and I received it in 24 hours. I like the earphones because the bass sounds feel good and they would not fall off.
                              
 
                              The content of label file COMMENTS_20180919_114745_result.txt is as follows:
                              
-
                              positive
+
                              positive
 negative
 negative 
 positive
                              
 
                              The data format requires users to store labeled objects and their label files (in one-to-one relationship with the labeled objects) in the same directory. For example, if the name of the labeled object file is COMMENTS_20180919_114745.txt, the name of the label file must be COMMENTS _20180919_114745_result.txt.
                              
 
                              Example of data file storage:
                              
-
                              ├─dataset-import-example 
+
                              ├─dataset-import-example 
 │      COMMENTS_20180919_114732.txt 
 │      COMMENTS _20180919_114732_result.txt 
 │      COMMENTS _20180919_114745.txt 
@@ -252,7 +120,7 @@ positive
                              
 
                            
 
                            Sound Classification
                            For sound classification, sound files with the same label must be stored in the same directory, and the label name is the directory name.
                            
 
                            Example:
                            
-
                            dataset-import-example 
+
                            dataset-import-example 
 ├─Cat 
 │      10.wav 
 │      11.wav 
diff --git a/docs/modelarts/umn/modelarts_23_0009.html b/docs/modelarts/umn/modelarts_23_0009.html
index 7fab0d5c8..55205abe6 100644
--- a/docs/modelarts/umn/modelarts_23_0009.html
+++ b/docs/modelarts/umn/modelarts_23_0009.html
@@ -5,60 +5,35 @@
 
                             
                            There are many requirements on the Manifest file compilation. Import new data from OBS. Generally, Manifest file import is used for data migration of ModelArts in different regions or using different accounts. If you have labeled data in a region using ModelArts, you can obtain the manifest file of the published dataset from the output path. Then you can import the dataset using the manifest file to ModelArts of other regions or accounts. The imported data carries the labeling information and does not need to be labeled again, improving development efficiency.
                            
 
                            
 
                            The manifest file that contains information about the original file and labeling can be used in labeling, training, and inference scenarios. The manifest file that contains only information about the original file can be used in inference scenarios or used to generate an unlabeled dataset. The manifest file must meet the following requirements:
                            
-
                            • The manifest file uses the UTF-8 encoding format. The source value of text classification can contain Chinese characters. However, Chinese characters are not recommended for other parameters.
                            • The manifest file uses the JSON Lines format (jsonlines.org). A line contains one JSON object.
                              {"source": "/path/to/image1.jpg", "annotation": ... }
+
                              • The manifest file uses the UTF-8 encoding format. The source value of text classification can contain Chinese characters. However, Chinese characters are not recommended for other parameters.
                              • The manifest file uses the JSON Lines format (jsonlines.org). A line contains one JSON object.
                                {"source": "/path/to/image1.jpg", "annotation": ... }
 {"source": "/path/to/image2.jpg", "annotation": ... }
 {"source": "/path/to/image3.jpg", "annotation": ... }
                                
 
                                In the preceding example, the manifest file contains multiple lines of JSON object.
                                
 
                              • The manifest file can be generated by users, third-party tools, or ModelArts Data Labeling. The file name can be any valid file name. To facilitate the internal use of the ModelArts system, the file name generated by the ModelArts Data Labeling function consists of the following character strings: DatasetName-VersionName.manifest. For example, animal-v201901231130304123.manifest.
                              
-
                              Image Classification
                               1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
                              {
-    "source":"s3://path/to/image1.jpg",
-    "usage":"TRAIN",
-    "id":"0162005993f8065ef47eefb59d1e4970",
-    "annotation": [
+
                              Image Classification
                              {
+    "source":"s3://path/to/image1.jpg",
+    "usage":"TRAIN",
+    "id":"0162005993f8065ef47eefb59d1e4970",
+    "annotation": [
         {
-            "type": "modelarts/image_classification",
-            "name": "cat",
-            "property": {
-                "color":"white",
-                "kind":"Persian cat"            
+            "type": "modelarts/image_classification",
+            "name": "cat",
+            "property": {
+                "color":"white",
+                "kind":"Persian cat"            
             },
-            "annotated-by":"human",
-            "creation-time":"2019-01-23 11:30:30"        
+            "annotated-by":"human",
+            "creation-time":"2019-01-23 11:30:30"        
         },
         {
-            "type": "modelarts/image_classification",
-            "name":"animal",
-            "annotated-by":"modelarts/active-learning",
-            "confidence": 0.8,
-            "creation-time":"2019-01-23 11:30:30"        
+            "type": "modelarts/image_classification",
+            "name":"animal",
+            "annotated-by":"modelarts/active-learning",
+            "confidence": 0.8,
+            "creation-time":"2019-01-23 11:30:30"        
         }],
-    "inference-loc":"/path/to/inference-output"
-}
-
                              
-
                              
+    "inference-loc":"/path/to/inference-output"
+}
                              
 
 
                              
@@ -192,151 +167,7 @@
 
                              Table 1 Parameters
                              Parameter
                              
 
                              
 
                              
 
                            
-
                            Image Segmentation
                            {
-    "annotation": [{
-        "annotation-format": "PASCAL VOC",
-        "type": "modelarts/image_segmentation",
-        "annotation-loc": "s3://path/to/annotation/image1.xml",
-        "creation-time": "2020-12-16 21:36:27",
-        "annotated-by": "human"
-    }],
-    "usage": "train",
-    "source": "s3://path/to/image1.jpg",
-    "id": "16d196c19bf61994d7deccafa435398c",
-    "sample-type": 0
-}
                            
-
                            • The parameters such as source, usage, and annotation are the same as those described in Image Classification. For details, see Table 1.
                            • annotation-loc indicates the path for saving the label file. This parameter is mandatory for image segmentation and object detection but optional for other labeling types.
                            • annotation-format indicates the format of the label file. This parameter is optional. The default value is PASCAL VOC. Only PASCAL VOC is supported.
                            • sample-type indicates a sample format. Value 0 indicates image, 1 text, 2 audio, 4 table, and 6 video.
                            
-
-
                            
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
                            Table 4 PASCAL VOC format parameters
                            Parameter
                            
-
                            Mandatory
                            
-
                            Description
                            
-
                            folder
                            
-
                            Yes
                            
-
                            Directory where the data source is located
                            
-
                            filename
                            
-
                            Yes
                            
-
                            Name of the file to be labeled
                            
-
                            size
                            
-
                            Yes
                            
-
                            Image pixel
                            
-
                            • width: image width. This parameter is mandatory.
                            • height: image height. This parameter is mandatory.
                            • depth: number of image channels. This parameter is mandatory.
                            
-
                            segmented
                            
-
                            Yes
                            
-
                            Segmented or not
                            
-
                            mask_source
                            
-
                            No
                            
-
                            Segmentation mask path
                            
-
                            object
                            
-
                            Yes
                            
-
                            Object detection information. Multiple object{} functions are generated for multiple objects.
                            
-
                            • name: class of the labeled content. This parameter is mandatory.
                            • pose: shooting angle of the labeled content. This parameter is mandatory.
                            • truncated: whether the labeled content is truncated (0 indicates that the content is not truncated). This parameter is mandatory.
                            • occluded: whether the labeled content is occluded (0 indicates that the content is not occluded). This parameter is mandatory.
                            • difficult: whether the labeled object is difficult to identify (0 indicates that the object is easy to identify). This parameter is mandatory.
                            • confidence: confidence score of the labeled object. The value ranges from 0 to 1. This parameter is optional.
                            • bndbox: bounding box type. This parameter is mandatory. For details about the possible values, see Table 5.
                            • mask_color: label color, which is represented by the RGB value. This parameter is mandatory.
                            
-
                            
-
                            
-
-
                            
-
-
-
-
-
-
-
-
-
-
                            Table 5 Bounding box types
                            Type
                            
-
                            Shape
                            
-
                            Labeling Information
                            
-
                            polygon
                            
-
                            Polygon
                            
-
                            Coordinates of points
                            
-
                            <x1>100<x1>
                            
-
                            <y1>100<y1>
                            
-
                            <x2>200<x2>
                            
-
                            <y2>100<y2>
                            
-
                            <x3>250<x3>
                            
-
                            <y3>150<y3>
                            
-
                            <x4>200<x4>
                            
-
                            <y4>200<y4>
                            
-
                            <x5>100<x5>
                            
-
                            <y5>200<y5>
                            
-
                            <x6>50<x6>
                            
-
                            <y6>150<y6>
                            
-
                            <x7>100<x7>
                            
-
                            <y7>100<y7>
                            
-
                            
-
                            
-
                            Example:
                            <?xml version="1.0" encoding="UTF-8" standalone="no"?>
-<annotation>
-    <folder>NA</folder>
-    <filename>image_0006.jpg</filename>
-    <source>
-        <database>Unknown</database>
-    </source>
-    <size>
-        <width>230</width>
-        <height>300</height>
-        <depth>3</depth>
-    </size>
-    <segmented>1</segmented>
-    <mask_source>obs://xianao/out/dataset-8153-Jmf5ylLjRmSacj9KevS/annotation/V001/segmentationClassRaw/image_0006.png</mask_source>
-    <object>
-        <name>bike</name>
-        <pose>Unspecified</pose>
-        <truncated>0</truncated>
-        <difficult>0</difficult>
-        <mask_color>193,243,53</mask_color>
-        <occluded>0</occluded>
-        <polygon>
-            <x1>71</x1>
-            <y1>48</y1>
-            <x2>75</x2>
-            <y2>73</y2>
-            <x3>49</x3>
-            <y3>69</y3>
-            <x4>68</x4>
-            <y4>92</y4>
-            <x5>90</x5>
-            <y5>101</y5>
-            <x6>45</x6>
-            <y6>110</y6>
-            <x7>71</x7>
-            <y7>48</y7>
-        </polygon>
-    </object>
-</annotation>
                            
-
                            
-
                            
-
                            Text Classification
                            {
+
                            Text Classification
                            {
     "source": "content://I like this product ",
     "id":"XGDVGS",
     "annotation": [
@@ -349,7 +180,7 @@
 }
                            
 
                            The content parameter indicates the text to be labeled (in UTF-8 encoding format, which can be Chinese). The other parameters are the same as those described in Image Classification. For details, see Table 1.
                            
 
                            
-
                            Named Entity Recognition
                            {
+
                            Named Entity Recognition
                            {
     "source":"content://Michael Jordan is the most famous basketball player in the world.",
     "usage":"TRAIN",
     "annotation":[
@@ -377,35 +208,35 @@
 }
                            
 
                            
 
                            The parameters such as source, usage, and annotation are the same as those described in Image Classification. For details, see Table 1.
                            
-
                            Table 6 describes the property parameters. For example, if you want to extract Michael from "source":"content://Michael Jordan", the value of start_index is 0 and that of end_index is 7.
                            
+
                            Table 4 describes the property parameters. For example, if you want to extract Michael from "source":"content://Michael Jordan", the value of start_index is 0 and that of end_index is 7.
                            
 
-
                            Table 6 Description of property parameters
                            Parameter
                            
+
                            
-
-
-
-
-
-
-
-
                            Table 4 Description of property parameters
                            Parameter
                            
 
                            Data Type
                            
+
                            Data Type
                            
 
                            Description
                            
+
                            Description
                            
                             		
 
                            
 
                            @modelarts:start_index
                            
+
                            @modelarts:start_index
                            
 
                            Integer
                            
+
                            Integer
                            
 
                            Start position of the text. The value starts from 0, including the characters specified by start_index.
                            
+
                            Start position of the text. The value starts from 0, including the characters specified by start_index.
                            
                             		
 
                            @modelarts:end_index
                            
+
                            @modelarts:end_index
                            
 
                            Integer
                            
+
                            Integer
                            
 
                            End position of the text, excluding the characters specified by end_index.
                            
+
                            End position of the text, excluding the characters specified by end_index.
                            
                             		
 
                            
                             
 
                            
 
                            
 
-
                            Text Triplet
                            {
+
                            Text Triplet
                            {
     "source":"content://"Three Body" is a series of long science fiction novels created by Liu Cix.",
     "usage":"TRAIN",
     "annotation":[
@@ -459,46 +290,46 @@
 
                            The parameters such as source, usage, and annotation are the same as those described in Image Classification. For details, see Table 1.
                            
 
                            Table 5 property parameters describes the property parameters. @modelarts:start_index and @modelarts:end_index are the same as those of named entity recognition. For example, when source is set to content://"Three Body" is a series of long science fiction novels created by Liu Cix., Liu Cix is an entity person, Three Body is an entity book, the person is the author of the book, and the book is works of the person.
                            
 
-
                            Table 7 Description of property parameters
                            Parameter
                            
+
                            
-
-
-
-
-
-
-
-
-
-
-
-
-
-
                            Table 5 Description of property parameters
                            Parameter
                            
 
                            Data Type
                            
+
                            Data Type
                            
 
                            Description
                            
+
                            Description
                            
                             		
 
                            
 
                            @modelarts:start_index
                            
+
                            @modelarts:start_index
                            
 
                            Integer
                            
+
                            Integer
                            
 
                            Start position of the triplet entities. The value starts from 0, including the characters specified by start_index.
                            
+
                            Start position of the triplet entities. The value starts from 0, including the characters specified by start_index.
                            
                             		
 
                            @modelarts:end_index
                            
+
                            @modelarts:end_index
                            
 
                            Integer
                            
+
                            Integer
                            
 
                            End position of the triplet entities, excluding the characters specified by end_index.
                            
+
                            End position of the triplet entities, excluding the characters specified by end_index.
                            
                             		
 
                            @modelarts:from
                            
+
                            @modelarts:from
                            
 
                            String
                            
+
                            String
                            
 
                            Start entity ID of the triplet relationship.
                            
+
                            Start entity ID of the triplet relationship.
                            
                             		
 
                            @modelarts:to
                            
+
                            @modelarts:to
                            
 
                            String
                            
+
                            String
                            
 
                            Entity ID pointed to in the triplet relationship.
                            
+
                            Entity ID pointed to in the triplet relationship.
                            
                             		
 
                            
                             
 
                            
 
                            
-
                            Object Detection
                            {
+
                            Object Detection
                            {
     "source":"s3://path/to/image1.jpg",
     "usage":"TRAIN",
     "annotation": [
@@ -512,99 +343,99 @@
 }
                            
 
                            • The parameters such as source, usage, and annotation are the same as those described in Image Classification. For details, see Table 1.
                            • annotation-loc indicates the path for saving the label file. This parameter is mandatory for object detection and image segmentation but optional for other labeling types.
                            • annotation-format indicates the format of the label file. This parameter is optional. The default value is PASCAL VOC. Only PASCAL VOC is supported.
                            
 
-
                            Table 8 PASCAL VOC format parameters
                            Parameter
                            
+
                            
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
                            Table 6 PASCAL VOC format parameters
                            Parameter
                            
 
                            Mandatory
                            
+
                            Mandatory
                            
 
                            Description
                            
+
                            Description
                            
                             		
 
                            
 
                            folder
                            
+
                            folder
                            
 
                            Yes
                            
+
                            Yes
                            
 
                            Directory where the data source is located
                            
+
                            Directory where the data source is located
                            
                             		
 
                            filename
                            
+
                            filename
                            
 
                            Yes
                            
+
                            Yes
                            
 
                            Name of the file to be labeled
                            
+
                            Name of the file to be labeled
                            
                             		
 
                            size
                            
+
                            size
                            
 
                            Yes
                            
+
                            Yes
                            
 
                            Image pixel
                            
+
                            Image pixel
                            
 
                            • width: image width. This parameter is mandatory.
                            • height: image height. This parameter is mandatory.
                            • depth: number of image channels. This parameter is mandatory.
                            
                             		
 
                            segmented
                            
+
                            segmented
                            
 
                            Yes
                            
+
                            Yes
                            
 
                            Segmented or not
                            
+
                            Segmented or not
                            
                             		
 
                            object
                            
+
                            object
                            
 
                            Yes
                            
+
                            Yes
                            
 
                            Object detection information. Multiple object{} functions are generated for multiple objects.
                            
-
                            • name: class of the labeled content. This parameter is mandatory.
                            • pose: shooting angle of the labeled content. This parameter is mandatory.
                            • truncated: whether the labeled content is truncated (0 indicates that the content is not truncated). This parameter is mandatory.
                            • occluded: whether the labeled content is occluded (0 indicates that the content is not occluded). This parameter is mandatory.
                            • difficult: whether the labeled object is difficult to identify (0 indicates that the object is easy to identify). This parameter is mandatory.
                            • confidence: confidence score of the labeled object. The value ranges from 0 to 1. This parameter is optional.
                            • bndbox: bounding box type. This parameter is mandatory. For details about the possible values, see Table 9.
                            
+
                            Object detection information. Multiple object{} functions are generated for multiple objects.
                            
+
                            • name: class of the labeled content. This parameter is mandatory.
                            • pose: shooting angle of the labeled content. This parameter is mandatory.
                            • truncated: whether the labeled content is truncated (0 indicates that the content is not truncated). This parameter is mandatory.
                            • occluded: whether the labeled content is occluded (0 indicates that the content is not occluded). This parameter is mandatory.
                            • difficult: whether the labeled object is difficult to identify (0 indicates that the object is easy to identify). This parameter is mandatory.
                            • confidence: confidence score of the labeled object. The value ranges from 0 to 1. This parameter is optional.
                            • bndbox: bounding box type. This parameter is mandatory. For details about the possible values, see Table 7.
                            
                             		
 
                            
                             
 
                            
 
                            
 
-
                            Table 9 Description of bounding box types
                            Type
                            
+
                            
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
                            Table 7 Description of bounding box types
                            Type
                            
 
                            Shape
                            
+
                            Shape
                            
 
                            Labeling Information
                            
+
                            Labeling Information
                            
                             		
 
                            
 
                            point
                            
+
                            point
                            
 
                            Point
                            
+
                            Point
                            
 
                            Coordinates of a point
                            
+
                            Coordinates of a point
                            
 
                            <x>100<x>
                            
 
                            <y>100<y>
                            
                             		
 
                            line
                            
+
                            line
                            
 
                            Line
                            
+
                            Line
                            
 
                            Coordinates of points
                            
+
                            Coordinates of points
                            
 
                            <x1>100<x1>
                            
 
                            <y1>100<y1>
                            
 
                            <x2>200<x2>
                            
 
                            <y2>200<y2>
                            
                             		
 
                            bndbox
                            
+
                            bndbox
                            
 
                            Rectangle
                            
+
                            Rectangle
                            
 
                            Coordinates of the upper left and lower right points
                            
+
                            Coordinates of the upper left and lower right points
                            
 
                            <xmin>100<xmin>
                            
 
                            <ymin>100<ymin>
                            
 
                            <xmax>200<xmax>
                            
 
                            <ymax>200<ymax>
                            
                             		
 
                            polygon
                            
+
                            polygon
                            
 
                            Polygon
                            
+
                            Polygon
                            
 
                            Coordinates of points
                            
+
                            Coordinates of points
                            
 
                            <x1>100<x1>
                            
 
                            <y1>100<y1>
                            
 
                            <x2>200<x2>
                            
@@ -619,11 +450,11 @@
 
                            <y6>150<y6>
                            
                             		
 
                            circle
                            
+
                            circle
                            
 
                            Circle
                            
+
                            Circle
                            
 
                            Center coordinates and radius
                            
+
                            Center coordinates and radius
                            
 
                            <cx>100<cx>
                            
 
                            <cy>100<cy>
                            
 
                            <r>50<r>
                            
@@ -632,7 +463,7 @@
 
                            
 
                            
 
                            
-
                            Example:
                            <annotation>
+
                            Example:
                            <annotation>
    <folder>test_data</folder>
    <filename>260730932.jpg</filename>
    <size>
@@ -711,7 +542,7 @@
 </annotation>
                            
 
                            
 
                            
-
                            Sound Classification
                            {
+
                            Sound Classification
                            {
 "source":
 "s3://path/to/pets.wav", 
     "annotation": [
@@ -725,7 +556,7 @@
 }
                            
 
                            The parameters such as source, usage, and annotation are the same as those described in Image Classification. For details, see Table 1.
                            
 
                            
-
                            Speech Labeling
                            {
+
                            Speech Labeling
                            {
     "source":"s3://path/to/audio1.wav",
     "annotation":[
         {
@@ -740,7 +571,7 @@
 }
                            
 
                            • The parameters such as source, usage, and annotation are the same as those described in Image Classification. For details, see Table 1.
                            • The @modelarts:content parameter in property indicates speech labeling. The data type is String.
                            
 
                            
-
                            Speech Paragraph Labeling
                            {
+
                            Speech Paragraph Labeling
                            {
     "source":"s3://path/to/audio1.wav",
     "usage":"TRAIN",
     "annotation":[
@@ -771,43 +602,43 @@
         }
     ]
 }
                            
-
                            • The parameters such as source, usage, and annotation are the same as those described in Image Classification. For details, see Table 1.
                            • Table 10 describes the property parameters.
-
                              Table 10 Description of property parameters
                              Parameter
                              
+
                              • The parameters such as source, usage, and annotation are the same as those described in Image Classification. For details, see Table 1.
                              • Table 8 describes the property parameters.
+
                                
-
-
-
-
-
-
-
-
-
-
-
-
-
-
@@ -815,7 +646,7 @@
 
-
                                Video Labeling
                                {
+
                                Video Labeling
                                {
 	"annotation": [{
 		"annotation-format": "PASCAL VOC",
 		"type": "modelarts/object_detection",
@@ -835,132 +666,132 @@
 }
                                
 
                                • The parameters such as source, usage, and annotation are the same as those described in Image Classification. For details, see Table 1.
                                • annotation-loc indicates the path for saving the label file. This parameter is mandatory for object detection but optional for other labeling types.
                                • annotation-format indicates the format of the label file. This parameter is optional. The default value is PASCAL VOC. Only PASCAL VOC is supported.
                                • sample-type indicates a sample format. Value 0 indicates image, 1 text, 2 audio, 4 table, and 6 video.
                                
 
-
                                Table 8 Description of property parameters
                                Parameter
                                
 
                                Data Type
                                
+
                                Data Type
                                
 
                                Description
                                
+
                                Description
                                
                                 		
 
                                
 
                                @modelarts:start_time
                                
+
                                @modelarts:start_time
                                
 
                                String
                                
+
                                String
                                
 
                                Start time of the sound. The format is hh:mm:ss.SSS.
                                
+
                                Start time of the sound. The format is hh:mm:ss.SSS.
                                
 
                                hh indicates the hour, mm indicates the minute, ss indicates the second, and SSS indicates the millisecond.
                                
                                 		
 
                                @modelarts:end_time
                                
+
                                @modelarts:end_time
                                
 
                                String
                                
+
                                String
                                
 
                                End time of the sound. The format is hh:mm:ss.SSS.
                                
+
                                End time of the sound. The format is hh:mm:ss.SSS.
                                
 
                                hh indicates the hour, mm indicates the minute, ss indicates the second, and SSS indicates the millisecond.
                                
                                 		
 
                                @modelarts:source
                                
+
                                @modelarts:source
                                
 
                                String
                                
+
                                String
                                
 
                                Sound source
                                
+
                                Sound source
                                
                                 		
 
                                @modelarts:content
                                
+
                                @modelarts:content
                                
 
                                String
                                
+
                                String
                                
 
                                Sound content
                                
+
                                Sound content
                                
                                 		
 
                                
                                 
 
 
                                Table 11 property parameters
                                Parameter
                                
+
                                
-
-
-
-
-
-
-
-
-
-
-
                                Table 9 property parameters
                                Parameter
                                
 
                                Data Type
                                
+
                                Data Type
                                
 
                                Description
                                
+
                                Description
                                
                                 		
 
                                
 
                                @modelarts:parent_duration
                                
+
                                @modelarts:parent_duration
                                
 
                                Double
                                
+
                                Double
                                
 
                                Duration of the labeled video, in seconds
                                
+
                                Duration of the labeled video, in seconds
                                
                                 		
 
                                @modelarts:time_in_video
                                
+
                                @modelarts:time_in_video
                                
 
                                Double
                                
+
                                Double
                                
 
                                Timestamp of the labeled video frame, in seconds
                                
+
                                Timestamp of the labeled video frame, in seconds
                                
                                 		
 
                                @modelarts:parent_source
                                
+
                                @modelarts:parent_source
                                
 
                                String
                                
+
                                String
                                
 
                                OBS path of the labeled video
                                
+
                                OBS path of the labeled video
                                
                                 		
 
                                
                                 
 
                                
 
                                
 
-
                                Table 12 PASCAL VOC format parameters
                                Parameter
                                
+
                                
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
                                Table 10 PASCAL VOC format parameters
                                Parameter
                                
 
                                Mandatory
                                
+
                                Mandatory
                                
 
                                Description
                                
+
                                Description
                                
                                 		
 
                                
 
                                folder
                                
+
                                folder
                                
 
                                Yes
                                
+
                                Yes
                                
 
                                Directory where the data source is located
                                
+
                                Directory where the data source is located
                                
                                 		
 
                                filename
                                
+
                                filename
                                
 
                                Yes
                                
+
                                Yes
                                
 
                                Name of the file to be labeled
                                
+
                                Name of the file to be labeled
                                
                                 		
 
                                size
                                
+
                                size
                                
 
                                Yes
                                
+
                                Yes
                                
 
                                Image pixel
                                
+
                                Image pixel
                                
 
                                • width: image width. This parameter is mandatory.
                                • height: image height. This parameter is mandatory.
                                • depth: number of image channels. This parameter is mandatory.
                                
                                 		
 
                                segmented
                                
+
                                segmented
                                
 
                                Yes
                                
+
                                Yes
                                
 
                                Segmented or not
                                
+
                                Segmented or not
                                
                                 		
 
                                object
                                
+
                                object
                                
 
                                Yes
                                
+
                                Yes
                                
 
                                Object detection information. Multiple object{} functions are generated for multiple objects.
                                
-
                                • name: class of the labeled content. This parameter is mandatory.
                                • pose: shooting angle of the labeled content. This parameter is mandatory.
                                • truncated: whether the labeled content is truncated (0 indicates that the content is not truncated). This parameter is mandatory.
                                • occluded: whether the labeled content is occluded (0 indicates that the content is not occluded). This parameter is mandatory.
                                • difficult: whether the labeled object is difficult to identify (0 indicates that the object is easy to identify). This parameter is mandatory.
                                • confidence: confidence score of the labeled object. The value ranges from 0 to 1. This parameter is optional.
                                • bndbox: bounding box type. This parameter is mandatory. For details about the possible values, see Table 13.
                                
+
                                Object detection information. Multiple object{} functions are generated for multiple objects.
                                
+
                                • name: class of the labeled content. This parameter is mandatory.
                                • pose: shooting angle of the labeled content. This parameter is mandatory.
                                • truncated: whether the labeled content is truncated (0 indicates that the content is not truncated). This parameter is mandatory.
                                • occluded: whether the labeled content is occluded (0 indicates that the content is not occluded). This parameter is mandatory.
                                • difficult: whether the labeled object is difficult to identify (0 indicates that the object is easy to identify). This parameter is mandatory.
                                • confidence: confidence score of the labeled object. The value ranges from 0 to 1. This parameter is optional.
                                • bndbox: bounding box type. This parameter is mandatory. For details about the possible values, see Table 11.
                                
                                 		
 
                                
                                 
 
                                
 
                                
 
-
                                Table 13 Bounding box types
                                Type
                                
+
                                
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
                                Table 11 Bounding box types
                                Type
                                
 
                                Shape
                                
+
                                Shape
                                
 
                                Labeling Information
                                
+
                                Labeling Information
                                
                                 		
 
                                
 
                                point
                                
+
                                point
                                
 
                                Point
                                
+
                                Point
                                
 
                                Coordinates of a point
                                
+
                                Coordinates of a point
                                
 
                                <x>100<x>
                                
 
                                <y>100<y>
                                
                                 		
 
                                line
                                
+
                                line
                                
 
                                Line
                                
+
                                Line
                                
 
                                Coordinates of points
                                
+
                                Coordinates of points
                                
 
                                <x1>100<x1>
                                
 
                                <y1>100<y1>
                                
 
                                <x2>200<x2>
                                
 
                                <y2>200<y2>
                                
                                 		
 
                                bndbox
                                
+
                                bndbox
                                
 
                                Rectangle
                                
+
                                Rectangle
                                
 
                                Coordinates of the upper left and lower right points
                                
+
                                Coordinates of the upper left and lower right points
                                
 
                                <xmin>100<xmin>
                                
 
                                <ymin>100<ymin>
                                
 
                                <xmax>200<xmax>
                                
 
                                <ymax>200<ymax>
                                
                                 		
 
                                polygon
                                
+
                                polygon
                                
 
                                Polygon
                                
+
                                Polygon
                                
 
                                Coordinates of points
                                
+
                                Coordinates of points
                                
 
                                <x1>100<x1>
                                
 
                                <y1>100<y1>
                                
 
                                <x2>200<x2>
                                
@@ -975,11 +806,11 @@
 
                                <y6>150<y6>
                                
                                 		
 
                                circle
                                
+
                                circle
                                
 
                                Circle
                                
+
                                Circle
                                
 
                                Center coordinates and radius
                                
+
                                Center coordinates and radius
                                
 
                                <cx>100<cx>
                                
 
                                <cy>100<cy>
                                
 
                                <r>50<r>
                                
@@ -988,7 +819,7 @@
 
                                
 
                                
 
                                
-
                                Example:
                                <annotation>
+
                                Example:
                                <annotation>
    <folder>test_data</folder>
    <filename>260730932_t1.473722.jpg.jpg</filename>
    <size>
diff --git a/docs/modelarts/umn/modelarts_23_0010.html b/docs/modelarts/umn/modelarts_23_0010.html
index 63167dd39..42fdc399c 100644
--- a/docs/modelarts/umn/modelarts_23_0010.html
+++ b/docs/modelarts/umn/modelarts_23_0010.html
@@ -9,8 +9,6 @@
 
 
                              • Object Detection
                                
 
                                • 
-
                              • Image Segmentation
                                
-
                                • 
 
                              • Text Classification
                                
 
                                • 
 
                              • Named Entity Recognition
                                
@@ -23,8 +21,6 @@
 
                                • 
 
                              • Speech Paragraph Labeling
                                
 
                                • 
-
                              • Video Labeling
                                
-
                                • 
 
 
 
                                
diff --git a/docs/modelarts/umn/modelarts_23_0011.html b/docs/modelarts/umn/modelarts_23_0011.html
index fd5a51f7a..625d3983e 100644
--- a/docs/modelarts/umn/modelarts_23_0011.html
+++ b/docs/modelarts/umn/modelarts_23_0011.html
@@ -15,7 +15,7 @@
 
                                The following filter criteria are supported. You can set one or more filter criteria.
                                
 
                                • Label: Select All or one or more labels you specified.
                                • Sample Creation Time: Select Within 1 month, Within 1 day, or Custom to customize a time range.
                                • File Name or Path: Filter files by file name or file storage path.
                                • Labeled By: Select the name of the user who performs the labeling operation.
                                
 
                                
-
                                Labeling Images (Manually)
                                The dataset details page displays images on the All, Labeled, and Unlabeled tabs. Images on the All tab page are displayed by default. Click an image to preview it. For the images that have been labeled, the label information is displayed at the bottom of the preview page.
                                
+
                                Labeling Images (Manually)
                                The dataset details page displays images on the All, Labeled, and Unlabeled tabs. Images on the All tab page are displayed by default. Click an image to preview it. For the images that have been labeled, the label information is displayed at the bottom of the preview page.
                                
 
                                1. On the Unlabeled tab page, select the images to be labeled.
                                  • Manual selection: In the image list, click the selection box in the upper left corner of an image to enter the selection mode, indicating that the image is selected. You can select multiple images of the same type and add labels to them together.
                                  • Batch selection: If all the images on the current page of the image list belong to the same type, you can click Select Images on Current Page in the upper right corner to select all the images on the current page.
                                  
 
                                2. Add labels to the selected images.
                                  1. In the label adding area on the right, set the label in the Label text box.
                                    Click the Label text box and select an existing label from the drop-down list. If the existing labels cannot meet the requirements, you can go to the page for modifying the dataset and add labels.
                                    
 
                                  2. Confirm the Labels of Selected Image information and click OK. The selected image is automatically moved to the Labeled tab page. On the Unlabeled and All tab pages, the labeling information is updated along with the labeling process, including the added label names and the number of images for each label.
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0012.html b/docs/modelarts/umn/modelarts_23_0012.html
index d1d6b6002..8d5021922 100644
--- a/docs/modelarts/umn/modelarts_23_0012.html
+++ b/docs/modelarts/umn/modelarts_23_0012.html
@@ -17,11 +17,11 @@
 
 
                                  
 
                                
-
                                Labeling Images (Manually)
                                The dataset details page provides the Labeled and Unlabeled tabs. The All tab page is displayed by default.
                                
+
                                Labeling Images (Manually)
                                The dataset details page provides the Labeled and Unlabeled tabs. The All tab page is displayed by default.
                                
 
                                1. On the Unlabeled tab page, click an image. The image labeling page is displayed. For details about how to use common buttons on the Labeled tab page, see Table 2.
                                2. In the left tool bar, select a proper labeling shape. The default labeling shape is a rectangle. In this example, the rectangle is used for labeling.
                                   
                                  On the left of the page, multiple tools are provided for you to label images. However, you can use only one tool at a time.
                                  
 
                                  
 
-
                                  Table 1 Supported bounding box
                                  Icon
                                  
+
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0013.html b/docs/modelarts/umn/modelarts_23_0013.html
index 80ed9262d..bd97afcc8 100644
--- a/docs/modelarts/umn/modelarts_23_0013.html
+++ b/docs/modelarts/umn/modelarts_23_0013.html
@@ -7,7 +7,7 @@
 
                                  Starting Labeling
                                  1. Log in to the ModelArts management console. In the left navigation pane, choose Data Management > Datasets. The Datasets page is displayed.
                                  2. In the dataset list, select the dataset to be labeled based on the labeling type, and click the dataset name to go to the Dashboard tab page of the dataset.
                                    By default, the Dashboard tab page of the current dataset version is displayed. If you need to label the dataset of another version, click the Versions tab and then click Set to Current Version in the right pane. For details, see Managing Dataset Versions.
                                    
 
                                  3. On the Dashboard page of the dataset, click Label in the upper right corner. The dataset details page is displayed. By default, all data of the dataset is displayed on the dataset details page.
                                  
 
                                  
-
                                  Labeling Content
                                  The dataset details page displays the labeled and unlabeled text files in the dataset. The Unlabeled tab page is displayed by default.
                                  
+
                                  Labeling Content
                                  The dataset details page displays the labeled and unlabeled text files in the dataset. The Unlabeled tab page is displayed by default.
                                  
 
                                  1. On the Unlabeled tab page, the objects to be labeled are listed in the left pane. In the list, click the text object to be labeled, and select a label in the Label Set area in the right pane. Multiple labels can be added to a labeling object.
                                    You can repeat this operation to select objects and add labels to the objects.
                                    
 
                                    Figure 1 Labeling for text classification
                                    
 
                                  2. After all objects are labeled, click Save Current Page at the bottom of the page to complete labeling text files on the Unlabeled tab page.
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0014.html b/docs/modelarts/umn/modelarts_23_0014.html
index 965b46cd4..827b0855b 100644
--- a/docs/modelarts/umn/modelarts_23_0014.html
+++ b/docs/modelarts/umn/modelarts_23_0014.html
@@ -6,7 +6,7 @@
 
                                  Starting Labeling
                                  1. Log in to the ModelArts management console. In the left navigation pane, choose Data Management > Datasets. The Datasets page is displayed.
                                  2. In the dataset list, select the dataset to be labeled based on the labeling type, and click the dataset name to go to the Dashboard tab page of the dataset.
                                    By default, the Dashboard tab page of the current dataset version is displayed. If you need to label the dataset of another version, click the Versions tab and then click Set to Current Version in the right pane. For details, see Managing Dataset Versions.
                                    
 
                                  3. On the Dashboard page of the dataset, click Label in the upper right corner. The dataset details page is displayed. By default, all data of the dataset is displayed on the dataset details page.
                                  
 
                                  
-
                                  Labeling Content
                                  The dataset details page displays the labeled and unlabeled text files in the dataset. The Unlabeled tab page is displayed by default.
                                  
+
                                  Labeling Content
                                  The dataset details page displays the labeled and unlabeled text files in the dataset. The Unlabeled tab page is displayed by default.
                                  
 
                                  
 
                                  1. On the Unlabeled tab page, the objects to be labeled are listed in the left pane. In the list, click the text object to be labeled, select a part of text displayed under Label Set for labeling, and select a label in the Label Set area in the right pane. Multiple labels can be added to a labeling object.
                                    You can repeat this operation to select objects and add labels to the objects.
                                    
 
                                    Figure 1 Labeling for named entity recognition
                                    
diff --git a/docs/modelarts/umn/modelarts_23_0018.html b/docs/modelarts/umn/modelarts_23_0018.html
index a7e998f1e..9421a5826 100644
--- a/docs/modelarts/umn/modelarts_23_0018.html
+++ b/docs/modelarts/umn/modelarts_23_0018.html
@@ -47,7 +47,7 @@
 
                                  
 
                                  Directory Structure of Related Files After the Dataset Is Published
                                  Datasets are managed based on OBS directories. After a new version is published, the directory is generated based on the new version in the output dataset path.
                                  
 
                                  Take an image classification dataset as an example. After the dataset is published, the directory structure of related files generated in OBS is as follows:
                                  
-
                                  |-- user-specified-output-path
+
                                  |-- user-specified-output-path
     |-- DatasetName-datasetId
         |-- annotation
             |-- VersionMame1
@@ -56,7 +56,7 @@
                 ...
             |-- ...
                                  
 
                                  The following uses object detection as an example. If a manifest file is imported to the dataset, the following provides the directory structure of related files after the dataset is published:
                                  
-
                                  |-- user-specified-output-path 
+
                                  |-- user-specified-output-path 
     |-- DatasetName-datasetId 
         |-- annotation 
             |-- VersionMame1 
diff --git a/docs/modelarts/umn/modelarts_23_0033.html b/docs/modelarts/umn/modelarts_23_0033.html
index f3f3a0b10..bb9a23115 100644
--- a/docs/modelarts/umn/modelarts_23_0033.html
+++ b/docs/modelarts/umn/modelarts_23_0033.html
@@ -104,7 +104,7 @@
 
                                  Table 1 Supported bounding box
                                  Icon
                                  
                                   		
 
                                  Description
                                  
 
                                  
 
                                  
 
-
                                  Constraints
                                  • For security purposes, the root permission is not granted to the notebook instances integrated in ModelArts. You can use the non-privileged user jovyan or ma-user (using Multi-Engine) to perform operations. Therefore, you cannot use apt-get to install the OS software.
                                  • Notebook instances support only standalone training under the current AI engine framework. If you need to use distributed training, use ModelArts training jobs and specify multiple nodes in the resource pool.
                                  • ModelArts DevEnviron does not support apt-get. You can use a custom image to train a model.
                                  • Notebook instances do not support GUI-related libraries, such as PyQt.
                                  • Notebook instances created using Ascend specifications cannot be attached to EVS disks.
                                  • Notebook instances cannot be connected to DWS and database services.
                                  • Notebook instances cannot directly read files in OBS. You need to download the files to the local host. To access data in OBS, use  or  for interaction.
                                  • DevEnviron does not support TensorBoard. Use the visualization job function under Training Jobs.
                                  • After a notebook instance is created, you cannot modify its specifications. For example, you cannot change the CPU specifications to GPU specifications or change the work environment. Therefore, select the specifications required by the service when creating a notebook instance, or save your code and data to OBS in a timely manner during development so that you can quickly upload the code and data to a new notebook instance.
                                  • If the code output is still displayed after you close the page and open it again, use Terminal.
                                  
+
                                  Constraints
                                  • For security purposes, the root permission is not granted to the notebook instances integrated in ModelArts. You can use the non-privileged user jovyan or ma-user (using Multi-Engine) to perform operations. Therefore, you cannot use apt-get to install the OS software.
                                  • Notebook instances support only standalone training under the current AI engine framework. If you need to use distributed training, use ModelArts training jobs and specify multiple nodes in the resource pool.
                                  • ModelArts DevEnviron does not support apt-get. You can use a custom image to train a model.
                                  • Notebook instances do not support GUI-related libraries, such as PyQt.
                                  • Notebook instances cannot be connected to DWS and database services.
                                  • Notebook instances cannot directly read files in OBS. You need to download the files to the local host. To access data in OBS, use  or  for interaction.
                                  • DevEnviron does not support TensorBoard. Use the visualization job function under Training Jobs.
                                  • After a notebook instance is created, you cannot modify its specifications. For example, you cannot change the CPU specifications to GPU specifications or change the work environment. Therefore, select the specifications required by the service when creating a notebook instance, or save your code and data to OBS in a timely manner during development so that you can quickly upload the code and data to a new notebook instance.
                                  • If the code output is still displayed after you close the page and open it again, use Terminal.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0034.html b/docs/modelarts/umn/modelarts_23_0034.html
index b86f220ae..51c430aa9 100644
--- a/docs/modelarts/umn/modelarts_23_0034.html
+++ b/docs/modelarts/umn/modelarts_23_0034.html
@@ -58,7 +58,7 @@
 
                                  Instance Flavor
                                  
                                   		
 
                                  If you select a public resource pool, available flavors vary depending on the selected type.
                                  
-
                                  • If you select CPU for Type, available options include 2 vCPUs | 8 GiB and 8 vCPUs | 32 GiB.
                                  • If you select GPU for Type, the available option is GPU: 1 x v100NV32 CPU: 8 vCPUs | 64 GiB.
                                  • If you select Ascend for Type, available options include Ascend: 1 x Ascend 910 CPU: 24 vCPUs | 96 GiB and Ascend: 8 x Ascend 910 CPU: 192 vCPUs | 720 GiB.
                                  
+
                                  • If you select CPU for Type, available options include 2 vCPUs | 8 GiB and 8 vCPUs | 32 GiB.
                                  • If you select GPU for Type, the available option is GPU: 1 x v100NV32 CPU: 8 vCPUs | 64 GiB.
                                  
                                   		
 
                                  
 
                                  Storage
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0039.html b/docs/modelarts/umn/modelarts_23_0039.html
index a6d67ba96..eb01c754f 100644
--- a/docs/modelarts/umn/modelarts_23_0039.html
+++ b/docs/modelarts/umn/modelarts_23_0039.html
@@ -4,70 +4,37 @@
 
                                  In notebook instances, you can use ModelArts SDKs to manage OBS, training jobs, models, and real-time services.
                                  
 
                                  For details about how to use ModelArts SDKs, see ModelArts SDK Reference.
                                  
 
                                  Notebooks carry the authentication (AK/SK) and region information about login users. Therefore, SDK session authentication can be completed without entering parameters.
                                  
-
                                  Example Code
                                  • Creating a training job
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
                                    from modelarts.session import Session
-from modelarts.estimator import Estimator
-session = Session()
-estimator = Estimator(
-                      modelarts_session=session,
-                      framework_type='PyTorch',                                     # AI engine name
-                       framework_version='PyTorch-1.0.0-python3.6',                  # AI engine version
-                      code_dir='/obs-bucket-name/src/',                                      # Training script directory
-                      boot_file='/obs-bucket-name/src/pytorch_sentiment.py',                 # Training startup script directory
-                      log_url='/obs-bucket-name/log/',                                       # Training log directory
-                      hyperparameters=[
-                                       {"label":"classes",
-                                        "value": "10"},    
-                                       {"label":"lr",
-                                        "value": "0.001"}
-                                       ],
-                      output_path='/obs-bucket-name/output/',                                # Training output directory
-                      train_instance_type='modelarts.vm.gpu.p100',                  # Training environment specifications
-                      train_instance_count=1,                                       # Number of training nodes
-                      job_description='pytorch-sentiment with ModelArts SDK')       # Training job description
-job_instance = estimator.fit(inputs='/obs-bucket-name/data/train/', wait=False, job_name='my_training_job')
-
                                    
-
                                    
+
                                    Example Code
                                    • Creating a training job
                                      from modelarts.session import Session
+from modelarts.estimator import Estimator
+session = Session()
+estimator = Estimator(
+                      modelarts_session=session,
+                      framework_type='PyTorch',                                     # AI engine name
+                       framework_version='PyTorch-1.0.0-python3.6',                  # AI engine version
+                      code_dir='/obs-bucket-name/src/',                                      # Training script directory
+                      boot_file='/obs-bucket-name/src/pytorch_sentiment.py',                 # Training startup script directory
+                      log_url='/obs-bucket-name/log/',                                       # Training log directory
+                      hyperparameters=[
+                                       {"label":"classes",
+                                        "value": "10"},    
+                                       {"label":"lr",
+                                        "value": "0.001"}
+                                       ],
+                      output_path='/obs-bucket-name/output/',                                # Training output directory
+                      train_instance_type='modelarts.vm.gpu.p100',                  # Training environment specifications
+                      train_instance_count=1,                                       # Number of training nodes
+                      job_description='pytorch-sentiment with ModelArts SDK')       # Training job description
+job_instance = estimator.fit(inputs='/obs-bucket-name/data/train/', wait=False, job_name='my_training_job')
                                      
 
                                    
-
                                    • Querying a model list
                                      1
-2
-3
-4
                                      from modelarts.session import Session 
-from modelarts.model import Model
-session = Session() 
-model_list_resp = Model.get_model_list(session, model_status="published", model_name="digit", order="desc")
-
                                      
-
                                      
-
                                    • Querying service details
                                      1
-2
-3
-4
-5
                                      from modelarts.session import Session
-from modelarts.model import Predictor
-session = Session()
-predictor_instance = Predictor(session, service_id="input your service_id")
-predictor_info_resp = predictor_instance.get_service_info()
-
                                      
-
                                      
+
                                      • Querying a model list
                                        from modelarts.session import Session 
+from modelarts.model import Model
+session = Session() 
+model_list_resp = Model.get_model_list(session, model_status="published", model_name="digit", order="desc")
                                        
+
                                      • Querying service details
                                        from modelarts.session import Session
+from modelarts.model import Predictor
+session = Session()
+predictor_instance = Predictor(session, service_id="input your service_id")
+predictor_info_resp = predictor_instance.get_service_info()
                                        
 
                                      
 
                                    
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0044.html b/docs/modelarts/umn/modelarts_23_0044.html
index f87d8c883..7e00067ce 100644
--- a/docs/modelarts/umn/modelarts_23_0044.html
+++ b/docs/modelarts/umn/modelarts_23_0044.html
@@ -3,37 +3,44 @@
 
                                  Introduction to Model Training
                                  
 
                                  ModelArts provides model training for you to view the training effect, based on which you can adjust your model parameters. You can select resource pools (CPU or GPU) with different instance flavors for model training. In addition to the models developed by users, ModelArts also provides built-in algorithms. You can directly adjust parameters of the built-in algorithms, instead of developing a model by yourself, to obtain a satisfactory model.
                                  
 
                                  Description of the Model Training Function
                                  
-
                                  
-
-
-
-
-
-
-
                                  Table 1 Function description
                                  Function
                                  
+
                                  
-
-
-
-
-
-
-
+
+
+
+
                                  Table 1 Function description
                                  Function
                                  
 
                                  Description
                                  
+
                                  Description
                                  
                                   		
 
                                  Reference
                                  
                                   		
 
                                  
 
                                  Built-in algorithms
                                  
+
                                  Built-in algorithms
                                  
 
                                  Based on the frequently-used AI engines in the industry, ModelArts provides built-in algorithms to meet a wide range of your requirements. You can directly select the algorithms for training jobs, without concerning model development.
                                  
+
                                  Based on the frequently-used AI engines in the industry, ModelArts provides built-in algorithms to meet a wide range of your requirements. You can directly select the algorithms for training jobs, without concerning model development.
                                  
                                   		
 
                                  Built-in Algorithms
                                  
                                   		
 
                                  Training job management
                                  
+
                                  Training job management
                                  
 
                                  You can create training jobs, manage training job versions, and view details of training jobs, and evaluation details.
                                  
+
                                  You can create training jobs, manage training job versions, and view details of training jobs, and evaluation details.
                                  
                                   		
 
                                  Introduction to Training Jobs
                                  
 
                                  Managing Training Job Versions
                                  
 
                                  Viewing Job Details
                                  
                                   		
 
                                  Job parameter management
                                  
+
                                  Job parameter management
                                  
 
                                  You can save the parameter settings of a training job (including the data source, algorithm source, running parameters, resource pool parameters, and more) as a job parameter, which can be directly used when you create a training job, eliminating the need to set parameters one by one. As such, the configuration efficiency can be greatly improved.
                                  
+
                                  You can save the parameter settings of a training job (including the data source, algorithm source, running parameters, resource pool parameters, and more) as a job parameter, which can be directly used when you create a training job, eliminating the need to set parameters one by one. As such, the configuration efficiency can be greatly improved.
                                  
                                   		
 
                                  Managing Job Parameters
                                  
                                   		
 
                                  Model training visualization
                                  
+
                                  TensorBoard and MindInsight effectively display the computational graph of a model in the running process, the trend of all metrics in time, and the data used in the training.
                                  
+
                                  Managing Visualization Jobs
                                  
+
                                  
                                   
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0051.html b/docs/modelarts/umn/modelarts_23_0051.html
index 0463c178b..651657bc2 100644
--- a/docs/modelarts/umn/modelarts_23_0051.html
+++ b/docs/modelarts/umn/modelarts_23_0051.html
@@ -10,8 +10,6 @@
 
 
                                3. Managing Model Versions
                                  
 
                                  4. 
-
                                4. Model Compression and Conversion
                                  
-
                                  5. 
 
 
 
diff --git a/docs/modelarts/umn/modelarts_23_0052.html b/docs/modelarts/umn/modelarts_23_0052.html
index 309172f40..26aefb708 100644
--- a/docs/modelarts/umn/modelarts_23_0052.html
+++ b/docs/modelarts/umn/modelarts_23_0052.html
@@ -1,26 +1,28 @@
 
 
 
                                  Introduction to Model Management
                                  
-
                                  AI model development and optimization require frequent iterations and debugging. Changes in datasets, training code, or parameters may affect the quality of models. If the metadata of the development process cannot be managed in a unified manner, the optimal model may fail to be reproduced. 
                                  
+
                                  AI model development and optimization require frequent iterations and debugging. Changes in datasets, training code, or parameters may affect the quality of models. If the metadata of the development process cannot be managed in a unified manner, the optimal model may fail to be reproduced.
                                  
 
                                  ModelArts model management allows you to import models generated with all training versions to manage all iterated and debugged models in a unified manner.
                                  
 
                                  Usage Restrictions
                                  • In an automatic learning project, after a model is deployed, the model is automatically uploaded to the model management list. However, models generated by automatic learning cannot be downloaded and can be used only for deployment and rollout.
                                  
 
                                  
+
                                  Methods of Importing a Model
                                  • Importing from Trained Models: You can create a training job on ModelArts and complete model training. After obtaining a satisfactory model, import the model to the Model Management page for model deployment.
                                  • Importing from a Template: Because the configurations of models with the same functions are similar, ModelArts integrates the configurations of such models into a common template. By using this template, you can easily and quickly import models without compiling the config.json configuration file.
                                  • Importing from a Container Image: For AI engines that are not supported by ModelArts, you can import the model you compile to ModelArts using custom images.
                                  • Importing from OBS: If you use a frequently-used framework to develop and train a model locally, you can import the model to ModelArts for model deployment.
                                  
+
                                  
 
                                  Model Management Functions
                                  
-
                                  
-
-
-
                                  Table 1 Model management functions
                                  Supported Function
                                  
+
                                  
-
-
-
-
-
diff --git a/docs/modelarts/umn/modelarts_23_0060.html b/docs/modelarts/umn/modelarts_23_0060.html
index 19ef6604f..a943e1046 100644
--- a/docs/modelarts/umn/modelarts_23_0060.html
+++ b/docs/modelarts/umn/modelarts_23_0060.html
@@ -113,9 +113,20 @@
 
-
-
+
+
+
+
+
+
diff --git a/docs/modelarts/umn/modelarts_23_0061.html b/docs/modelarts/umn/modelarts_23_0061.html
index 449d02d97..59b4c485b 100644
--- a/docs/modelarts/umn/modelarts_23_0061.html
+++ b/docs/modelarts/umn/modelarts_23_0061.html
@@ -201,12 +201,12 @@
 
                                  Table 1 Model management functions
                                  Supported Function
                                  
 
                                  Description
                                  
+
                                  Description
                                  
                                   		
 
                                  
 
                                  Importing a Model
                                  
+
                                  Importing a Model
                                  
 
                                  Import the trained models to ModelArts for unified management. You can import models using four methods. The following provides the operation guide for each method.
                                  
-
+
                                  Import the trained models to ModelArts for unified management. You can import models using four methods. The following provides the operation guide for each method.
                                  
+
                                   		
 
                                  Managing Model Versions
                                  
+
                                  Managing Model Versions
                                  
 
                                  To facilitate source tracing and repeated model tuning, ModelArts provides the model version management function. You can manage models based on versions.
                                  
+
                                  To facilitate source tracing and repeated model tuning, ModelArts provides the model version management function. You can manage models based on versions.
                                  
                                   		
 
                                  
                                   
 
 
                                  CPU: 8 vCPUs | 64 GiB GPU: 1 x V100
                                  
+
                                  ExeML specifications (CPU)
                                  
+
                                  ExeML specifications (GPU)
                                  
 
                                  Suitable for running GPU models.
                                  
+
                                  Only be used by models trained in ExeML projects.
                                  
+
                                  CPU: 2 vCPUs | 8 GiB
                                  
+
                                  Suitable for models with only CPU loads.
                                  
+
                                  CPU: 8 vCPUs | 32 GiB GPU: 1 x T4
                                  
+
                                  Suitable for models requiring CPU and GPU (NVIDIA T4) resources.
                                  
                                   		
 
                                  
 
                                  
 
                                  
 
                                  Pound key (#) indicates that a variable is referenced. The matched character string must be enclosed in single quotation marks.
                                  
-
                                  #{Built-in variable} == 'Character string'
+
                                  #{Built-in variable} == 'Character string'
 #{Built-in variable} matches 'Regular expression'
                                  
 
                                  • Example 1:
                                    If the account name for invoking the inference request is User A, the specified version is matched.
                                    
-
                                    #DOMAIN_NAME == 'User A'
                                    
+
                                    #DOMAIN_NAME == 'User A'
                                    
 
                                  • Example 2:
                                    If the account name in the inference request starts with op, the specified version is matched.
                                    
-
                                    #DOMAIN_NAME matches 'op.*'
                                    
+
                                    #DOMAIN_NAME matches 'op.*'
                                    
 
 
                                    
@@ -264,13 +264,13 @@
 
                                    Figure 1 Traffic distribution by user
                                    
-
                                  • If multiple versions of a real-time service are deployed for dark launch, customized settings can be used to access different versions through the header.
                                    Start with #HEADER_, indicating that the header is referenced as a condition.
                                    #HEADER_{key} == '{value}'
+
                                  • If multiple versions of a real-time service are deployed for dark launch, customized settings can be used to access different versions through the header.
                                    Start with #HEADER_, indicating that the header is referenced as a condition.
                                    #HEADER_{key} == '{value}'
 #HEADER_{key} matches '{value}'
                                    
 
                                    
 
                                    • Example 1:
                                      If the header of an inference HTTP request contains a version and the value is 0.0.1, the condition is met. Otherwise, the condition is not met.
                                      
-
                                      #HEADER_version == '0.0.1'
                                      
+
                                      #HEADER_version == '0.0.1'
                                      
 
                                    • Example 2:
                                      If the header of an inference HTTP request contains testheader and the value starts with mock, the rule is matched.
                                      
-
                                      #HEADER_testheader matches 'mock.*'
                                      
+
                                      #HEADER_testheader matches 'mock.*'
                                      
 
                                      Figure 2 Using the header to access different versions
                                      
 
                                    
 
                                  • If a real-time service version supports different running configurations, you can use Setting Name and Setting Value to specify customized running parameters so that different users can use different running configurations.
                                    Example:
                                    
diff --git a/docs/modelarts/umn/modelarts_23_0063.html b/docs/modelarts/umn/modelarts_23_0063.html
index 5211411c0..493592b60 100644
--- a/docs/modelarts/umn/modelarts_23_0063.html
+++ b/docs/modelarts/umn/modelarts_23_0063.html
@@ -8,7 +8,7 @@
 
                                    For details about how to obtain a token, see Obtaining a User Token.
                                    
 
                                  • On the Body tab page, file input and text input are available.
                                    • File input
                                      Select form-data. Set KEY to the input parameter of the model, for example, images. Set VALUE to an image to be inferred (only one image can be inferred).
                                      
 
                                    • Text input
                                      Select raw and then JSON(application/json). Enter the request body in the text box below. An example request body is as follows:
                                      
-
                                      {
+
                                      {
   "meta": {
     "uuid": "10eb0091-887f-4839-9929-cbc884f1e20e"
   },
@@ -31,11 +31,11 @@
 
 
                                      Method 2: Run the cURL Command to Send an Inference Request
                                      The command for sending inference requests can be input as a file or text.
                                      
 
                                      
-
                                      1. File input
                                        curl -k -F 'images=@Image path' -H 'X-Auth-Token:Token value' -X POST Real-time service URL
                                        
+
                                        1. File input
                                          curl -k -F 'images=@Image path' -H 'X-Auth-Token:Token value' -X POST Real-time service URL
                                          
 
                                          • -k indicates that SSL websites can be accessed without using a security certificate.
                                          • -F indicates file input. In this example, the parameter name is images, which can be changed as required. The image storage path follows @.
                                          • -H indicates the header of the POST command. X-Auth-Token is the KEY value on the Headers page. Token value indicates the obtained token. 
                                          • POST is followed by the API URL of the real-time service.
                                          
 
                                          The following is an example of the cURL command for inference with file input:
                                          
-
                                          curl -k -F 'images=@/home/data/test.png' -H 'X-Auth-Token:MIISkAY***80T9wHQ==' -X POST https://modelarts-infers-1.xxx/v1/infers/eb3e0c54-3dfa-4750-af0c-95c45e5d3e83
                                          
-
                                        2. Text input
                                          curl -k -d '{"data":{"req_data":[{"sepal_length":3,"sepal_width":1,"petal_length":2.2,"petal_width":4}]}}' -H 'X-Auth-Token:MIISkAY***80T9wHQ==' -H 'Content-type: application/json' -X POST https://modelarts-infers-1.xxx/v1/infers/eb3e0c54-3dfa-4750-af0c-95c45e5d3e83
                                          
+
                                          curl -k -F 'images=@/home/data/test.png' -H 'X-Auth-Token:MIISkAY***80T9wHQ==' -X POST https://modelarts-infers-1.xxx/v1/infers/eb3e0c54-3dfa-4750-af0c-95c45e5d3e83
                                          
+
                                        3. Text input
                                          curl -k -d '{"data":{"req_data":[{"sepal_length":3,"sepal_width":1,"petal_length":2.2,"petal_width":4}]}}' -H 'X-Auth-Token:MIISkAY***80T9wHQ==' -H 'Content-type: application/json' -X POST https://modelarts-infers-1.xxx/v1/infers/eb3e0c54-3dfa-4750-af0c-95c45e5d3e83
                                          
 
                                          -d indicates the text input of the request body.
                                          
 
                                        
 
diff --git a/docs/modelarts/umn/modelarts_23_0066.html b/docs/modelarts/umn/modelarts_23_0066.html
index 21a75d860..4f3b0a4d8 100644
--- a/docs/modelarts/umn/modelarts_23_0066.html
+++ b/docs/modelarts/umn/modelarts_23_0066.html
@@ -67,20 +67,20 @@
 
 
                                        Manifest File Specifications
                                        Batch services of the inference platform support the manifest file. The manifest file describes the input and output of data.
                                        
 
                                        
-
                                        Example input manifest file
                                        • File name: test.manifest
                                        • File content:
                                          {"source": "<obs path>/test/data/1.jpg"}
+
                                          Example input manifest file
                                          • File name: test.manifest
                                          • File content:
                                            {"source": "<obs path>/test/data/1.jpg"}
 {"source": "https://infers-data.obs.xxx.com:443/xgboosterdata/data.csv?AccessKeyId=2Q0V0TQ461N26DDL18RB&Expires=1550611914&Signature=wZBttZj5QZrReDhz1uDzwve8GpY%3D&x-obs-security-token=gQpzb3V0aGNoaW5hixvY8V9a1SnsxmGoHYmB1SArYMyqnQT-ZaMSxHvl68kKLAy5feYvLDM..."}
                                            
-
                                          • File requirements:
                                            1. The file name extension must be .manifest.
                                            2. The file content is in JSON format. Each row describes a piece of input data, which must be accurate to a file instead of a folder.
                                            
+
                                          • File requirements:
                                            1. The file name extension must be .manifest.
                                            2. The file content is in JSON format. Each row describes a piece of input data, which must be accurate to a file instead of a folder.
                                            3. A source field must be defined for the JSON content. The field value is the OBS URL of the file: <obs path>/{{Bucket name}}/{{Object name}}.
                                            
 
                                          
 
                                          
 
                                          Example output manifest file
                                          
-
                                          If you use an input manifest file, the output directory will contain an output manifest file.
                                          • Assume that the output path is //test-bucket/test/. The result is stored in the following path:
                                            OBS bucket/directory name
+
                                            If you use an input manifest file, the output directory will contain an output manifest file.
                                            • Assume that the output path is //test-bucket/test/. The result is stored in the following path:
                                              OBS bucket/directory name
 ├── test-bucket
 │   ├── test
 │   │   ├── infer-result-0.manifest
 │   │   ├── infer-result
 │   │   │ ├── 1.jpg_result.txt
 │   │   │ ├── 2.jpg_result.txt
                                              
-
                                            • Content of the infer-result-0.manifest file:
                                              {"source": "<obs path>/obs-data-bucket/test/data/1.jpg",  "inference-loc": "<obs path>/test-bucket/test/infer-result/1.jpg_result.txt"}
+
                                            • Content of the infer-result-0.manifest file:
                                              {"source": "<obs path>/obs-data-bucket/test/data/1.jpg",  "inference-loc": "<obs path>/test-bucket/test/infer-result/1.jpg_result.txt"}
 {"source ": "https://infers-data.obs.xxx.com:443/xgboosterdata/2.jpg?AccessKeyId=2Q0V0TQ461N26DDL18RB&Expires=1550611914&Signature=wZBttZj5QZrReDhz1uDzwve8GpY%3D&x-obs-security-token=gQpzb3V0aGNoaW5hixvY8V9a1SnsxmGoHYmB1SArYMyqnQT-ZaMSxHvl68kKLAy5feYvLDMNZWxzhBZ6Q-3HcoZMh9gISwQOVBwm4ZytB_m8sg1fL6isU7T3CnoL9jmvDGgT9VBC7dC1EyfSJrUcqfB...",  "inference-loc": "obs://test-bucket/test/infer-result/2.jpg_result.txt"}
                                              
 
                                            
 
                                            
@@ -89,72 +89,34 @@
 
                                          
 
                                          Example Mapping
                                          The following example shows the relationship between the configuration file, mapping rule, CSV data, and inference request.
                                          
 
                                          Assume that the apis parameter in the configuration file used by your model is as follows:
                                          
-
                                    • Table 5 Common regular expressions
                                    Character
                                    
                                     		
 
 
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
                                    [
+
                                    [
     {
-        "protocol": "http",
-        "method": "post",
-        "url": "/",
-        "request": {
-            "type": "object",
-            "properties": {
-                "data": {
-                    "type": "object",
-                    "properties": {
-                        "req_data": {
-                            "type": "array",
-                            "items": [
+        "protocol": "http",
+        "method": "post",
+        "url": "/",
+        "request": {
+            "type": "object",
+            "properties": {
+                "data": {
+                    "type": "object",
+                    "properties": {
+                        "req_data": {
+                            "type": "array",
+                            "items": [
                                 {
-                                    "type": "object",
-                                    "properties": {
-                                        "input_1": {
-                                            "type": "number"
+                                    "type": "object",
+                                    "properties": {
+                                        "input_1": {
+                                            "type": "number"
                                         },
-                                        "input_2": {
-                                            "type": "number"
+                                        "input_2": {
+                                            "type": "number"
                                         },
-                                        "input_3": {
-                                            "type": "number"
+                                        "input_3": {
+                                            "type": "number"
                                         },
-                                        "input_4": {
-                                            "type": "number"
+                                        "input_4": {
+                                            "type": "number"
                                         }
                                     }
                                 }
@@ -165,11 +127,9 @@
             }
         }
     }
-]
-
                                    
-
                                    
+]
                                  
 
                                  At this point, the corresponding mapping relationship is shown below. The ModelArts management console automatically resolves the mapping relationship from the configuration file. When calling a ModelArts API, write the mapping relationship by yourself according to the rule.
                                  
-
                                  {
+
                                  {
     "type": "object",
     "properties": {
         "data": {
@@ -206,11 +166,11 @@
     }
 }
                                  
 
                                  The data for inference, that is, the CSV data, is in the following format. The data must be separated by commas (,).
                                  
-
                                  5.1,3.5,1.4,0.2
+
                                  5.1,3.5,1.4,0.2
 4.9,3.0,1.4,0.2
 4.7,3.2,1.3,0.2
                                  
 
                                  Depending on the defined mapping relationship, the inference request is shown below. The format is similar to the format used by the real-time service.
                                  
-
                                  {
+
                                  {
 	"data": {
 		"req_data": [{
 			"input_1": 5.1,
diff --git a/docs/modelarts/umn/modelarts_23_0076.html b/docs/modelarts/umn/modelarts_23_0076.html
index bfcd4f839..d3ee6cc06 100644
--- a/docs/modelarts/umn/modelarts_23_0076.html
+++ b/docs/modelarts/umn/modelarts_23_0076.html
@@ -23,11 +23,6 @@
 
                                  The default value is Dedicated for Service Deployment and cannot be changed.
                                  
                                   		
 
                                  Billing Mode
                                  
-
                                  Select a billing mode, Yearly/Monthly or Pay-per-use. The Yearly/Monthly billing mode is supported only in CN North-Beijing4.
                                  
-
                                  
 
                                  Name
                                  
                                   		
 
                                  Name of a dedicated resource pool.
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0079.html b/docs/modelarts/umn/modelarts_23_0079.html
index 8f5599efc..e50605be6 100644
--- a/docs/modelarts/umn/modelarts_23_0079.html
+++ b/docs/modelarts/umn/modelarts_23_0079.html
@@ -13,7 +13,7 @@
 
 
 
-
                                  Example Policies
                                  • A policy can define a single permission, such as the permission to deny ExeML project deletion.
                                    { 
+
                                    Example Policies
                                    • A policy can define a single permission, such as the permission to deny ExeML project deletion.
                                      { 
       "Version": "1.1", 
       "Statement": [ 
             { 
@@ -24,7 +24,7 @@
             } 
       ] 
 }
                                      
-
                                    • A policy can define multiple permissions, such as the permissions to delete an ExeML version and an ExeML project.
                                      { 
+
                                    • A policy can define multiple permissions, such as the permissions to delete an ExeML version and an ExeML project.
                                      { 
     "Version": "1.1", 
     "Statement": [ 
         { 
diff --git a/docs/modelarts/umn/modelarts_23_0080.html b/docs/modelarts/umn/modelarts_23_0080.html
index df047adaa..8a62d791b 100644
--- a/docs/modelarts/umn/modelarts_23_0080.html
+++ b/docs/modelarts/umn/modelarts_23_0080.html
@@ -8,7 +8,7 @@
 
                                      Precautions
                                      • The permissions to use ModelArts depend on OBS authorization. Therefore, you need to grant OBS system permissions to users.
                                      • A custom policy can contain actions of multiple services that are globally accessible or accessible through region-specific projects.
                                      • To define permissions required to access both global and project-level services, create two custom policies and specify the scope as Global services and Project-level services. Then grant the two policies to the users.
                                      
 
                                      
 
                                      Example Custom Policies of OBS
                                      ModelArts is a project-level service, and OBS is a global service. Therefore, you need to create custom policies for the two services respectively and grant them to users. The permissions to use ModelArts depend on OBS authorization. The following example shows the minimum permissions for OBS, including the permissions for OBS buckets and objects. After being granted the minimum permissions for OBS, users can access OBS from ModelArts without restrictions.
                                      
-
                                      {
+
                                      {
     "Version": "1.1",
     "Statement": [
         {
@@ -35,7 +35,7 @@
 
                                      
 
                                      Example Custom Policies of ModelArts
                                      • Example: Denying ExeML project deletion
                                        A deny policy must be used in conjunction with other policies to take effect. If the permissions assigned to a user contain both Allow and Deny actions, the Deny actions take precedence over the Allow actions.
                                        
 
                                        The following method can be used if you need to assign permissions of the ModelArts FullAccess policy to a user but also forbid the user from deleting ExeML projects. Create a custom policy for denying ExeML project deletion, and assign both policies to the group the user belongs to. Then the user can perform all operations on ModelArts except deleting ExeML projects. The following is an example deny policy:
                                        
-
                                        { 
+
                                        { 
       "Version": "1.1", 
       "Statement": [ 
             { 
@@ -47,7 +47,7 @@
       ] 
 }
                                        
 
                                      • Example: Allowing users to use only development environments
                                        The following is a policy configuration example for this user:
                                        
-
                                        { 
+
                                        { 
     "Version": "1.1", 
     "Statement": [ 
 
diff --git a/docs/modelarts/umn/modelarts_23_0085.html b/docs/modelarts/umn/modelarts_23_0085.html
index 9359b04cd..763f75b86 100644
--- a/docs/modelarts/umn/modelarts_23_0085.html
+++ b/docs/modelarts/umn/modelarts_23_0085.html
@@ -6,7 +6,7 @@
 
                                        Obtain the custom images used by ModelArts for model training and import from the SWR service management list. Upload the custom images you create to SWR.
                                        
 
                                      • Specifications for custom images. For details about how to use a custom image for a training job, see Specifications for Custom Images Used for Training Jobs. For details about how to use a custom image for model import, see Specifications for Custom Images Used for Importing Models.
                                      
 
                                      
-
                                      Creating and Uploading a Custom Image
                                      1. Purchase a cloud server or use a local host to set up the Docker environment.
                                      2. Obtain the basic image from the local environment.
                                      3. Compile a Dockerfile based on your requirements to build a custom image. For details about how to efficiently compile a Dockerfile, see .
                                      1. After customizing an image, upload the image to SWR by referring to .
                                      
+
                                      Creating and Uploading a Custom Image
                                      1. Purchase a cloud server or use a local host to set up the Docker environment.
                                      2. Obtain the basic image from the local environment.
                                      3. Compile a Dockerfile based on your requirements to build a custom image. 
                                      1. After customizing an image, upload the image to SWR by referring to "Uploading an Image Through a Container Engine Client" in Software Repository for Container User Guide. 
                                      
 
                                      
 
                                      
 
                                      
diff --git a/docs/modelarts/umn/modelarts_23_0087.html b/docs/modelarts/umn/modelarts_23_0087.html
index 82487337d..362857d6c 100644
--- a/docs/modelarts/umn/modelarts_23_0087.html
+++ b/docs/modelarts/umn/modelarts_23_0087.html
@@ -2,7 +2,7 @@
 
 
                                      Creating a Training Job Using a Custom Image (GPU)
                                      
 
                                      After creating and uploading a custom image to SWR, you can use the image to create a training job on the ModelArts management console to complete model training.
                                      
-
                                      Prerequisites
                                      • You have created a custom image package based on ModelArts specifications. For details about the specifications you need to comply with when using a custom image to create training jobs, see Specifications for Custom Images Used for Training Jobs.
                                      • You have uploaded the custom image to SWR. For details, see .
                                      
+
                                      Prerequisites
                                      
 
                                      
 
                                      Creating a Training Job
                                      Log in to the ModelArts management console and create a training job according to Creating a Training Job. When using a custom image to create a job, pay attention to the settings of Algorithm Source, Environment Variable, and Resource Pool.
                                      
 
                                      • Algorithm Source
                                        Select Custom.
                                        
diff --git a/docs/modelarts/umn/modelarts_23_0091.html b/docs/modelarts/umn/modelarts_23_0091.html
index c4c1dfbeb..a2c25f386 100644
--- a/docs/modelarts/umn/modelarts_23_0091.html
+++ b/docs/modelarts/umn/modelarts_23_0091.html
@@ -7,21 +7,71 @@
 
                                      
 
                                      ModelArts also provides custom script examples of common AI engines. For details, see Examples of Custom Scripts.
                                      
 
                                      Model Package Example
                                      • Structure of the TensorFlow-based model package
                                        When publishing the model, you only need to specify the ocr directory.
                                        
-
                                        OBS bucket/directory name
+
                                        OBS bucket/directory name
 |── ocr
 |   ├── model (Mandatory) Name of a fixed subdirectory, which is used to store model-related files
 |   │   ├── <<Custom Python package>> (Optional) User's Python package, which can be directly referenced in the model inference code
 |   │   ├── saved_model.pb (Mandatory) Protocol buffer file, which contains the diagram description of the model
-|   │   ├── variables Name of a fixed sub-directory, which contains the weight and deviation rate of the model. It is mandatory for the main file of the *.pb model.
+|   │   ├── variables Name of a fixed sub-directory, which contains the weight and deviation rate of the model. It is mandatory for the main file of the *.pb model.
 |   │   │   ├── variables.index Mandatory
 |   │   │   ├── variables.data-00000-of-00001 Mandatory
-|   │   ├──config.json (Mandatory) Model configuration file. The file name is fixed to config.json. Only one model configuration file is supported.
+|   │   ├──config.json (Mandatory) Model configuration file. The file name is fixed to config.json. Only one model configuration file is supported.
 |   │   ├──customize_service.py (Optional) Model inference code. The file name is fixed to customize_service.py. Only one model inference code file exists. The files on which customize_service.py depends can be directly stored in the model directory.
                                        
-
                                      • Structure of the Image-based model package
                                        When publishing the model, you only need to specify the resnet directory.
                                        
-
                                        OBS bucket/directory name
+
                                      • Structure of the MXNet-based model package
                                        When publishing the model, you only need to specify the resnet directory.
                                        
+
                                        OBS bucket/directory name
 |── resnet
 |   ├── model (Mandatory) Name of a fixed subdirectory, which is used to store model-related files
-|   │  ├──config.json (Mandatory) Model configuration file (the address of the SWR image must be configured). The file name is fixed to config.json. Only one model configuration file is supported.
                                        
+|   │   ├── <<Custom Python package>> (Optional) User's Python package, which can be directly referenced in the model inference code
+|   │   ├── resnet-50-symbol.json (Mandatory) Model definition file, which contains the neural network description of the model
+|   │   ├── resnet-50-0000.params (Mandatory) Model variable parameter file, which contains parameter and weight information
+|   │   ├──config.json (Mandatory) Model configuration file. The file name is fixed to config.json. Only one model configuration file is supported.
+|   │   ├──customize_service.py (Optional) Model inference code. The file name is fixed to customize_service.py. Only one model inference code file exists. The files on which customize_service.py depends can be directly stored in the model directory.
                                      
+
                                    • Structure of the Image-based model package
                                      When publishing the model, you only need to specify the resnet directory.
                                      
+
                                      OBS bucket/directory name
+|── resnet
+|   ├── model (Mandatory) Name of a fixed subdirectory, which is used to store model-related files
+|   │  ├──config.json (Mandatory) Model configuration file (the address of the SWR image must be configured). The file name is fixed to config.json. Only one model configuration file is supported.
                                      
+
                                    • Structure of the PySpark-based model package
                                      When publishing the model, you only need to specify the resnet directory.
                                      
+
                                      OBS bucket/directory name
+|── resnet
+|   ├── model (Mandatory) Name of a fixed subdirectory, which is used to store model-related files
+|   │  ├── <<Custom Python package>> (Optional) User's Python package, which can be directly referenced in the model inference code
+|   │  ├── spark_model (Mandatory) Model directory, which contains the model content saved by PySpark
+|   │  ├──config.json (Mandatory) Model configuration file. The file name is fixed to config.json. Only one model configuration file is supported.
+|   │  ├──customize_service.py (Optional) Model inference code. The file name is fixed to customize_service.py. Only one model inference code file exists. The files on which customize_service.py depends can be directly stored in the model directory.
                                      
+
                                    • Structure of the PyTorch-based model package
                                      When publishing the model, you only need to specify the resnet directory.
                                      
+
                                      OBS bucket/directory name
+|── resnet
+|   ├── model (Mandatory) Name of a fixed subdirectory, which is used to store model-related files
+|   │  ├── <<Custom Python package>> (Optional) User's Python package, which can be directly referenced in the model inference code
+|   │  ├── resnet50.pth (Mandatory) PyTorch model file, which contains variable and weight information and is saved as state_dict
+|   │  ├──config.json (Mandatory) Model configuration file. The file name is fixed to config.json. Only one model configuration file is supported.
+|   │  ├──customize_service.py (Optional) Model inference code. The file name is fixed to customize_service.py. Only one model inference code file exists. The files on which customize_service.py depends can be directly stored in the model directory.
                                      
+
                                    • Structure of the Caffe-based model package
                                      When publishing the model, you only need to specify the resnet directory.
                                      OBS bucket/directory name
+|── resnet
+|   |── model (Mandatory) Name of a fixed subdirectory, which is used to store model-related files
+|   |   |── <<Custom Python package>> (Optional) User's Python package, which can be directly referenced in the model inference code
+|   |   |── deploy.prototxt (Mandatory) Caffe model file, which contains information such as the model network structure
+|   |   |── resnet.caffemodel (Mandatory) Caffe model file, which contains variable and weight information
+|   |   |── config.json (Mandatory) Model configuration file. The file name is fixed to config.json. Only one model configuration file is supported.
+|   |   |── customize_service.py  (Optional) Model inference code. The file name is fixed to customize_service.py. Only one model inference code file exists. The files on which customize_service.py depends can be directly stored in the model directory. 
                                      
+
                                      
+
                                    • Structure of the XGBoost-based model package
                                      When publishing the model, you only need to specify the resnet directory.
                                      OBS bucket/directory name
+|── resnet
+|   |── model (Mandatory) Name of a fixed subdirectory, which is used to store model-related files
+|   |   |── <<Custom Python package>> (Optional) User's Python package, which can be directly referenced in the model inference code
+|   |   |── *.m (Mandatory): Model file whose extension name is .m
+|   |   |── config.json (Mandatory) Model configuration file. The file name is fixed to config.json. Only one model configuration file is supported.
+|   |   |── customize_service.py  (Optional) Model inference code. The file name is fixed to customize_service.py. Only one model inference code file exists. The files on which customize_service.py depends can be directly stored in the model directory. 
                                      
+
                                      
+
                                    • Structure of the Scikit_Learn-based model package
                                      When publishing the model, you only need to specify the resnet directory.
                                      OBS bucket/directory name
+|── resnet
+|   |── model (Mandatory) Name of a fixed subdirectory, which is used to store model-related files
+|   |   |── <<Custom Python package>> (Optional) User's Python package, which can be directly referenced in the model inference code
+|   |   |── *.m (Mandatory): Model file whose extension name is .m
+|   |   |── config.json (Mandatory) Model configuration file. The file name is fixed to config.json. Only one model configuration file is supported.
+|   |   |── customize_service.py  (Optional) Model inference code. The file name is fixed to customize_service.py. Only one model inference code file exists. The files on which customize_service.py depends can be directly stored in the model directory. 
                                      
+
                                      
 
                                    
 
                                    
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0092.html b/docs/modelarts/umn/modelarts_23_0092.html
index 60c79b305..7df81b792 100644
--- a/docs/modelarts/umn/modelarts_23_0092.html
+++ b/docs/modelarts/umn/modelarts_23_0092.html
@@ -411,337 +411,161 @@
 
                                  Example of the Object Detection Model Configuration File
                                  The following code uses the TensorFlow engine as an example. You can modify the model_type parameter based on the actual engine type.
                                  
 
                                  • Model input
                                    Key: images
                                    
 
                                    Value: image files
                                    
-
                                  • Model output
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
                                    ```
-{
-    "detection_classes": [
-        "face",
-        "arm"
-    ],
-    "detection_boxes": [
-        [
-            33.6,
-            42.6,
-            104.5,
-            203.4
-        ],
-        [
-            103.1,
-            92.8,
-            765.6,
-            945.7
-        ]
-    ],
-    "detection_scores": [0.99, 0.73]
-}
-```
-
                                    
-
                                    
-
                                  • Configuration file
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
-46
-47
-48
-49
-50
-51
-52
-53
-54
-55
-56
-57
-58
-59
-60
-61
-62
-63
-64
-65
-66
-67
-68
-69
-70
-71
-72
-73
                                    ```
+
                                  • Model output
                                    ```
 {
-    "model_type": "TensorFlow",
-    "model_algorithm": "object_detection",
-    "metrics": {
-        "f1": 0.345294,
-        "accuracy": 0.462963,
-        "precision": 0.338977,
-        "recall": 0.351852
+    "detection_classes": [
+        "face",
+        "arm"
+    ],
+    "detection_boxes": [
+        [
+            33.6,
+            42.6,
+            104.5,
+            203.4
+        ],
+        [
+            103.1,
+            92.8,
+            765.6,
+            945.7
+        ]
+    ],
+    "detection_scores": [0.99, 0.73]
+}
+```
                                    
+
                                  • Configuration file
                                    ```
+{
+    "model_type": "TensorFlow",
+    "model_algorithm": "object_detection",
+    "metrics": {
+        "f1": 0.345294,
+        "accuracy": 0.462963,
+        "precision": 0.338977,
+        "recall": 0.351852
     },
-    "apis": [{
-        "protocol": "http",
-        "url": "/",
-        "method": "post",
-        "request": {
-            "Content-type": "multipart/form-data",
-            "data": {
-                "type": "object",
-                "properties": {
-                    "images": {
-                        "type": "file"
+    "apis": [{
+        "protocol": "http",
+        "url": "/",
+        "method": "post",
+        "request": {
+            "Content-type": "multipart/form-data",
+            "data": {
+                "type": "object",
+                "properties": {
+                    "images": {
+                        "type": "file"
                     }
                 }
             }
         },
-        "response": {
-            "Content-type": "multipart/form-data",
-            "data": {
-                "type": "object",
-                "properties": {
-                    "detection_classes": {
-                        "type": "array",
-                        "items": [{
-                            "type": "string"
+        "response": {
+            "Content-type": "multipart/form-data",
+            "data": {
+                "type": "object",
+                "properties": {
+                    "detection_classes": {
+                        "type": "array",
+                        "items": [{
+                            "type": "string"
                         }]
                     },
-                    "detection_boxes": {
-                        "type": "array",
-                        "items": [{
-                            "type": "array",
-                            "minItems": 4,
-                            "maxItems": 4,
-                            "items": [{
-                                "type": "number"
+                    "detection_boxes": {
+                        "type": "array",
+                        "items": [{
+                            "type": "array",
+                            "minItems": 4,
+                            "maxItems": 4,
+                            "items": [{
+                                "type": "number"
                             }]
                         }]
                     },
-                    "detection_scores": {
-                        "type": "array",
-                        "items": [{
-                            "type": "number"
+                    "detection_scores": {
+                        "type": "array",
+                        "items": [{
+                            "type": "number"
                         }]
                     }
                 }
             }
         }
     }],
-    "dependencies": [{
-        "installer": "pip",
-        "packages": [{
-                "restraint": "EXACT",
-                "package_version": "1.15.0",
-                "package_name": "numpy"
+    "dependencies": [{
+        "installer": "pip",
+        "packages": [{
+                "restraint": "EXACT",
+                "package_version": "1.15.0",
+                "package_name": "numpy"
             },
             {
-                "restraint": "EXACT",
-                "package_version": "5.2.0",
-                "package_name": "Pillow"
+                "restraint": "EXACT",
+                "package_version": "5.2.0",
+                "package_name": "Pillow"
             }
         ]
     }]
 }
-```
-
                                    • 
-
                                    
+```
 
                                  
 
                                  
 
                                  Example of the Image Classification Model Configuration File
                                  The following code uses the TensorFlow engine as an example. You can modify the model_type parameter based on the actual engine type.
                                  
 
                                  • Model input
                                    Key: images
                                    
 
                                    Value: image files
                                    
-
                                  • Model output
                                    1
-2
-3
-4
-5
-6
-7
-8
-9
                                    ```
-{
-    "predicted_label": "flower",
-    "scores": [
-       ["rose", 0.99],
-       ["begonia", 0.01]
-    ]
-}
-```
-
                                    
-
                                    
-
                                  • Configuration file
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
-46
-47
-48
-49
-50
-51
-52
-53
-54
-55
-56
-57
-58
-59
-60
-61
-62
-63
-64
-65
-66
-67
-68
-69
                                    ```
+
                                  • Model output
                                    ```
 {
-    "model_type": "TensorFlow",
-    "model_algorithm": "image_classification",
-    "metrics": {
-        "f1": 0.345294,
-        "accuracy": 0.462963,
-        "precision": 0.338977,
-        "recall": 0.351852
+    "predicted_label": "flower",
+    "scores": [
+       ["rose", 0.99],
+       ["begonia", 0.01]
+    ]
+}
+```
                                    
+
                                  • Configuration file
                                    ```
+{
+    "model_type": "TensorFlow",
+    "model_algorithm": "image_classification",
+    "metrics": {
+        "f1": 0.345294,
+        "accuracy": 0.462963,
+        "precision": 0.338977,
+        "recall": 0.351852
     },
-    "apis": [{
-        "protocol": "http",
-        "url": "/",
-        "method": "post",
-        "request": {
-            "Content-type": "multipart/form-data",
-            "data": {
-                "type": "object",
-                "properties": {
-                    "images": {
-                        "type": "file"
+    "apis": [{
+        "protocol": "http",
+        "url": "/",
+        "method": "post",
+        "request": {
+            "Content-type": "multipart/form-data",
+            "data": {
+                "type": "object",
+                "properties": {
+                    "images": {
+                        "type": "file"
                     }
                 }
             }
         },
-        "response": {
-            "Content-type": "multipart/form-data",
-            "data": {
-                "type": "object",
-                "properties": {
-                    "predicted_label": {
-                        "type": "string"
+        "response": {
+            "Content-type": "multipart/form-data",
+            "data": {
+                "type": "object",
+                "properties": {
+                    "predicted_label": {
+                        "type": "string"
                     },
-                    "scores": {
-                        "type": "array",
-                        "items": [{
-                            "type": "array",
-                            "minItems": 2,
-                            "maxItems": 2,
-                            "items": [
+                    "scores": {
+                        "type": "array",
+                        "items": [{
+                            "type": "array",
+                            "minItems": 2,
+                            "maxItems": 2,
+                            "items": [
                                 {
-                                    "type": "string"
+                                    "type": "string"
                                 },
                                 {
-                                    "type": "number"
+                                    "type": "number"
                                 }
                             ]
                         }]
@@ -750,236 +574,116 @@
             }
         }
     }],
-    "dependencies": [{
-        "installer": "pip",
-        "packages": [{
-                "restraint": "ATLEAST",
-                "package_version": "1.15.0",
-                "package_name": "numpy"
+    "dependencies": [{
+        "installer": "pip",
+        "packages": [{
+                "restraint": "ATLEAST",
+                "package_version": "1.15.0",
+                "package_name": "numpy"
             },
             {
-                "restraint": "",
-                "package_version": "",
-                "package_name": "Pillow"
+                "restraint": "",
+                "package_version": "",
+                "package_name": "Pillow"
             }
         ]
     }]
 }
-```
-
                                    • 
-
                                    
+```
 
                                  
 
                                  
 
                                  Example of the Predictive Analytics Model Configuration File
                                  The following code uses the TensorFlow engine as an example. You can modify the model_type parameter based on the actual engine type.
                                  
-
                                  • Model input
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
                                    ```
-{
-    "data": {
-        "req_data": [
-            {
-                "buying_price": "high",
-                "maint_price": "high",
-                "doors": "2",
-                "persons": "2",
-                "lug_boot": "small",
-                "safety": "low",
-                "acceptability": "acc"
-            },
-            {
-                "buying_price": "high",
-                "maint_price": "high",
-                "doors": "2",
-                "persons": "2",
-                "lug_boot": "small",
-                "safety": "low",
-                "acceptability": "acc"
-            }
-        ]
-    }
-}
-```
-
                                    
-
                                    
-
                                  • Model output
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
                                    ```
-{
-    "data": {
-        "resp_data": [
-            {
-                "predict_result": "unacc"
-            },
-            {
-                "predict_result": "unacc"
-            }
-        ]
-    }
-}
-```
-
                                    
-
                                    
-
                                  • Configuration file
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
-46
-47
-48
-49
-50
-51
-52
-53
-54
-55
-56
-57
-58
-59
-60
-61
-62
-63
-64
-65
-66
-67
-68
-69
-70
-71
-72
-73
-74
-75
-76
-77
                                    ```
+
                                    • Model input
                                      ```
 {
-    "model_type": "TensorFlow",
-    "model_algorithm": "predict_analysis",
-    "metrics": {
-        "f1": 0.345294,
-        "accuracy": 0.462963,
-        "precision": 0.338977,
-        "recall": 0.351852
+    "data": {
+        "req_data": [
+            {
+                "buying_price": "high",
+                "maint_price": "high",
+                "doors": "2",
+                "persons": "2",
+                "lug_boot": "small",
+                "safety": "low",
+                "acceptability": "acc"
+            },
+            {
+                "buying_price": "high",
+                "maint_price": "high",
+                "doors": "2",
+                "persons": "2",
+                "lug_boot": "small",
+                "safety": "low",
+                "acceptability": "acc"
+            }
+        ]
+    }
+}
+```
                                      
+
                                    • Model output
                                      ```
+{
+    "data": {
+        "resp_data": [
+            {
+                "predict_result": "unacc"
+            },
+            {
+                "predict_result": "unacc"
+            }
+        ]
+    }
+}
+```
                                      
+
                                    • Configuration file
                                      ```
+{
+    "model_type": "TensorFlow",
+    "model_algorithm": "predict_analysis",
+    "metrics": {
+        "f1": 0.345294,
+        "accuracy": 0.462963,
+        "precision": 0.338977,
+        "recall": 0.351852
     },
-    "apis": [
+    "apis": [
         {
-            "protocol": "http",
-            "url": "/",
-            "method": "post",
-            "request": {
-                "Content-type": "application/json",
-                "data": {
-                    "type": "object",
-                    "properties": {
-                        "data": {
-                            "type": "object",
-                            "properties": {
-                                "req_data": {
-                                    "items": [
+            "protocol": "http",
+            "url": "/",
+            "method": "post",
+            "request": {
+                "Content-type": "application/json",
+                "data": {
+                    "type": "object",
+                    "properties": {
+                        "data": {
+                            "type": "object",
+                            "properties": {
+                                "req_data": {
+                                    "items": [
                                         {
-                                            "type": "object",
-                                            "properties": {
+                                            "type": "object",
+                                            "properties": {
                                             }
                                         }],
-                                    "type": "array"
+                                    "type": "array"
                                 }
                             }
                         }
                     }
                 }
             },
-            "response": {
-                "Content-type": "multipart/form-data",
-                "data": {
-                    "type": "object",
-                    "properties": {
-                        "data": {
-                            "type": "object",
-                            "properties": {
-                                "resp_data": {
-                                    "type": "array",
-                                    "items": [
+            "response": {
+                "Content-type": "multipart/form-data",
+                "data": {
+                    "type": "object",
+                    "properties": {
+                        "data": {
+                            "type": "object",
+                            "properties": {
+                                "resp_data": {
+                                    "type": "array",
+                                    "items": [
                                         {
-                                            "type": "object",
-                                            "properties": {
+                                            "type": "object",
+                                            "properties": {
                                             }
                                         }]
                                 }
@@ -989,132 +693,74 @@
                 }
             }
         }],
-    "dependencies": [
+    "dependencies": [
         {
-            "installer": "pip",
-            "packages": [
+            "installer": "pip",
+            "packages": [
                 {
-                    "restraint": "EXACT",
-                    "package_version": "1.15.0",
-                    "package_name": "numpy"
+                    "restraint": "EXACT",
+                    "package_version": "1.15.0",
+                    "package_name": "numpy"
                 },
                 {
-                    "restraint": "EXACT",
-                    "package_version": "5.2.0",
-                    "package_name": "Pillow"
+                    "restraint": "EXACT",
+                    "package_version": "5.2.0",
+                    "package_name": "Pillow"
                 }]
         }]
 }
-```
-
                                    
-
                                    
+```
 
                                  
 
                                  
 
                                  Example of the Custom Image Model Configuration File
                                  The model input and output are similar to those in Example of the Object Detection Model Configuration File.
                                  
-
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
-46
-47
-48
-49
-50
-51
-52
-53
-54
-55
-56
-57
                                  {
-    "model_algorithm": "image_classification",
-    "model_type": "Image",
+
                                  {
+    "model_algorithm": "image_classification",
+    "model_type": "Image",
 
-    "metrics": {
-        "f1": 0.345294,
-        "accuracy": 0.462963,
-        "precision": 0.338977,
-        "recall": 0.351852
+    "metrics": {
+        "f1": 0.345294,
+        "accuracy": 0.462963,
+        "precision": 0.338977,
+        "recall": 0.351852
     },
-    "apis": [{
-        "protocol": "http",
-        "url": "/",
-        "method": "post",
-        "request": {
-            "Content-type": "multipart/form-data",
-            "data": {
-                "type": "object",
-                "properties": {
-                    "images": {
-                        "type": "file"
+    "apis": [{
+        "protocol": "http",
+        "url": "/",
+        "method": "post",
+        "request": {
+            "Content-type": "multipart/form-data",
+            "data": {
+                "type": "object",
+                "properties": {
+                    "images": {
+                        "type": "file"
                     }
                 }
             }
         },
-        "response": {
-            "Content-type": "multipart/form-data",
-            "data": {
-                "type": "object",
-                "required": [
-                    "predicted_label",
-                    "scores"
+        "response": {
+            "Content-type": "multipart/form-data",
+            "data": {
+                "type": "object",
+                "required": [
+                    "predicted_label",
+                    "scores"
                 ],
-                "properties": {
-                    "predicted_label": {
-                        "type": "string"
+                "properties": {
+                    "predicted_label": {
+                        "type": "string"
                     },
-                    "scores": {
-                        "type": "array",
-                        "items": [{
-                            "type": "array",
-                            "minItems": 2,
-                            "maxItems": 2,
-                            "items": [{
-                                    "type": "string"
+                    "scores": {
+                        "type": "array",
+                        "items": [{
+                            "type": "array",
+                            "minItems": 2,
+                            "maxItems": 2,
+                            "items": [{
+                                    "type": "string"
                                 },
                                 {
-                                    "type": "number"
+                                    "type": "number"
                                 }
                             ]
                         }]
@@ -1123,13 +769,11 @@
             }
         }
     }]
-}
-
                                  
-
                                  
+}
 
                                  
 
                                  Example of the Machine Learning Model Configuration File
                                  The following uses XGBoost as an example:
                                  
 
                                  • Model input
                                  
-
                                  {
+
                                  {
 	"data": {
 		"req_data": [{
 			"sepal_length": 5,
@@ -1150,7 +794,7 @@
 	}
 }
                                  
 
                                  • Model output
                                  
-
                                  {
+
                                  {
 	"data": {
 		"resp_data": [{
 			"predict_result": "Iris-setosa"
@@ -1160,7 +804,7 @@
 	}
 }
                                  
 
                                  • Configuration file
                                  
-
                                  {
+
                                  {
   "model_type": "XGBoost",
   "model_algorithm": "xgboost_iris_test",
   "runtime": "python2.7",
@@ -1224,84 +868,34 @@
 
                                  
 
                                  
 
                                  Example of a Model Configuration File Using a Custom Dependency Package
                                  The following example defines the NumPy 1.16.4 dependency environment.
                                  
-
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
-46
-47
-48
-49
-50
-51
                                  {
-     "model_algorithm": "image_classification",
-     "model_type": "TensorFlow",
-     "runtime": "python3.6",
-     "apis": [{
-             "procotol": "http",
-             "url": "/",
-             "method": "post",
-             "request": {
-                 "Content-type": "multipart/form-data",
-                 "data": {
-                     "type": "object",
-                     "properties": {
-                         "images": {
-                             "type": "file"
+
                                  {
+     "model_algorithm": "image_classification",
+     "model_type": "TensorFlow",
+     "runtime": "python3.6",
+     "apis": [{
+             "procotol": "http",
+             "url": "/",
+             "method": "post",
+             "request": {
+                 "Content-type": "multipart/form-data",
+                 "data": {
+                     "type": "object",
+                     "properties": {
+                         "images": {
+                             "type": "file"
                          }
                      }
                  }
              },
-             "response": {
-                 "Content-type": "applicaton/json",
-                 "data": {
-                     "type": "object",
-                     "properties": {
-                         "mnist_result": {
-                             "type": "array",
-             "item": [{
-                "type": "string"
+             "response": {
+                 "Content-type": "applicaton/json",
+                 "data": {
+                     "type": "object",
+                     "properties": {
+                         "mnist_result": {
+                             "type": "array",
+             "item": [{
+                "type": "string"
                          }]
                          }
                      }
@@ -1309,24 +903,22 @@
              }
          }
      ],
-     "metrics": {
-         "f1": 0.124555,
-         "recall": 0.171875,
-         "precision": 0.0023493892851938493,
-         "accuracy": 0.00746268656716417
+     "metrics": {
+         "f1": 0.124555,
+         "recall": 0.171875,
+         "precision": 0.0023493892851938493,
+         "accuracy": 0.00746268656716417
      },
-    "dependencies": [{
-        "installer": "pip",
-        "packages": [{
-                "restraint": "EXACT",
-                "package_version": "1.16.4",
-                "package_name": "numpy"
+    "dependencies": [{
+        "installer": "pip",
+        "packages": [{
+                "restraint": "EXACT",
+                "package_version": "1.16.4",
+                "package_name": "numpy"
             }
         ]
     }]
- }
-
                                  
-
                                  
+ }
 
                                  
 
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0093.html b/docs/modelarts/umn/modelarts_23_0093.html
index d8f3449e5..0561ad9d5 100644
--- a/docs/modelarts/umn/modelarts_23_0093.html
+++ b/docs/modelarts/umn/modelarts_23_0093.html
@@ -101,22 +101,22 @@
 
                                   
                                  • You can choose to rewrite the preprocess and postprocess methods to implement preprocessing of the API input and postprocessing of the inference output.
                                  • Rewriting the init method of the BaseService inheritance class may cause a model to run abnormally.
                                  
 
                                  
 
                                5. The attribute that can be used is the local path where the model resides. The attribute name is self.model_path. In addition, PySpark-based models can use self.spark to obtain the SparkSession object in customize_service.py.
                                   
                                  An absolute path is required for reading files in the inference code. You can obtain the absolute path of the model from the self.model_path attribute.
                                  
-
                                  • When TensorFlow, Caffe, or MXNet is used, self.model_path indicates the path of the model file. See the following example:
                                    # Store the label.json file in the model directory. The following information is read:
+
                                    • When TensorFlow, Caffe, or MXNet is used, self.model_path indicates the path of the model file. See the following example:
                                      # Store the label.json file in the model directory. The following information is read:
 with open(os.path.join(self.model_path, 'label.json')) as f:
     self.label = json.load(f)
                                      
 
                                    
-
                                    • When PyTorch, Scikit_Learn, or PySpark is used, self.model_path indicates the path of the model file. See the following example:
                                      # Store the label.json file in the model directory. The following information is read:
+
                                      • When PyTorch, Scikit_Learn, or PySpark is used, self.model_path indicates the path of the model file. See the following example:
                                        # Store the label.json file in the model directory. The following information is read:
 dir_path = os.path.dirname(os.path.realpath(self.model_path))
 with open(os.path.join(dir_path, 'label.json')) as f:
     self.label = json.load(f)
                                        
 
                                      
 
                                  
-
                                6. Two types of content-type APIs can be used for inputting data: multipart/form-data and application/json
                                  • multipart/form-data request
                                    curl -X POST \
+
                                  • Two types of content-type APIs can be used for inputting data: multipart/form-data and application/json
                                    • multipart/form-data request
                                      curl -X POST \
   <modelarts-inference-endpoint> \
   -F image1=@cat.jpg \
   -F images2=@horse.jpg
                                      
 
                                      The corresponding input data is as follows:
                                      
-
                                      [
+
                                      [
    {
       "image1":{
          "cat.jpg":"<cat..jpg file io>"
@@ -128,81 +128,54 @@ with open(os.path.join(dir_path, 'label.json')) as f:
       }
    }
 ]
                                      
-
                                    • application/json request
                                       curl -X POST \
+
                                    • application/json request
                                       curl -X POST \
    <modelarts-inference-endpoint> \
    -d '{
     "images":"base64 encode image"
     }'
                                      
 
                                      The corresponding input data is python dict.
                                      
-
                                       {
+
                                      {
     "images":"base64 encode image"
 
  }
                                      
 
                                    
 
                                    • 
 
                                  7. 
-
                                  TensorFlow Inference Script Example
                                  The following is an example of TensorFlow MnistService.
                                  • Inference code
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
                                    from PIL import Image
-import numpy as np
-from model_service.tfserving_model_service import TfServingBaseService
+
                                    TensorFlow Inference Script Example
                                    The following is an example of TensorFlow MnistService.
                                    • Inference code
                                      +from PIL import Image
+import numpy as np
+from model_service.tfserving_model_service import TfServingBaseService
 
-class mnist_service(TfServingBaseService):
+class mnist_service(TfServingBaseService):
 
-    def _preprocess(self, data):
-        preprocessed_data = {}
+    def _preprocess(self, data):
+        preprocessed_data = {}
 
-        for k, v in data.items():
-            for file_name, file_content in v.items():
-                image1 = Image.open(file_content)
-                image1 = np.array(image1, dtype=np.float32)
-                image1.resize((1, 784))
-                preprocessed_data[k] = image1
+        for k, v in data.items():
+            for file_name, file_content in v.items():
+                image1 = Image.open(file_content)
+                image1 = np.array(image1, dtype=np.float32)
+                image1.resize((1, 784))
+                preprocessed_data[k] = image1
 
-        return preprocessed_data
+        return preprocessed_data
 
-    def _postprocess(self, data):
+    def _postprocess(self, data):
 
-        infer_output = {}
+        infer_output = {}
 
-        for output_name, result in data.items():
+        for output_name, result in data.items():
 
-            infer_output["mnist_result"] = result[0].index(max(result[0]))
+            infer_output["mnist_result"] = result[0].index(max(result[0]))
 
-        return infer_output
-
                                    
-
                                    
-
                                  • Request
                                    curl -X POST \ Real-time service address \ -F images=@test.jpg
                                    
-
                                  • Response
                                    {"mnist_result": 7}
                                    
+        return infer_output
+
                                  • Request
                                    curl -X POST \ Real-time service address \ -F images=@test.jpg
                                    
+
                                  • Response
                                    {"mnist_result": 7}
                                    
 
                                  
 
                                  
 
                                  The preceding code example resizes images imported to the user's form to adapt to the model input shape. The 32×32 image is read from the Pillow library and resized to 1×784 to match the model input. In subsequent processing, convert the model output into a list for the RESTful API to display.
                                  
 
                                  
-
                                  XGBoost Inference Script Example
                                  # coding:utf-8
+
                                  XGBoost Inference Script Example
                                  # coding:utf-8
 import collections
 import json
 import xgboost as xgb
@@ -239,233 +212,119 @@ class user_Service(XgSklServingBaseService):
 
                                  
 
                                  
 
                                  Inference Script Example of the Custom Inference Logic
                                  First, define a dependency package in the configuration file. For details, see Example of a Model Configuration File Using a Custom Dependency Package. Then, use the following code example to implement the loading and inference of the model in saved_model format.
                                  
-
                                    1
-  2
-  3
-  4
-  5
-  6
-  7
-  8
-  9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
-100
-101
-102
-103
-104
-105
-106
-107
-108
-109
-110
-111
-112
-113
                                  # -*- coding: utf-8 -*-
-import json
-import os
-import threading
+
                                  # -*- coding: utf-8 -*-
+import json
+import os
+import threading
 
-import numpy as np
-import tensorflow as tf
-from PIL import Image
+import numpy as np
+import tensorflow as tf
+from PIL import Image
 
-from model_service.tfserving_model_service import TfServingBaseService
-import logging
+from model_service.tfserving_model_service import TfServingBaseService
+import logging
 
-logger = logging.getLogger(__name__)
+logger = logging.getLogger(__name__)
 
 
-class MnistService(TfServingBaseService):
+class MnistService(TfServingBaseService):
 
-    def __init__(self, model_name, model_path):
-        self.model_name = model_name
-        self.model_path = model_path
-        self.model_inputs = {}
-        self.model_outputs = {}
+    def __init__(self, model_name, model_path):
+        self.model_name = model_name
+        self.model_path = model_path
+        self.model_inputs = {}
+        self.model_outputs = {}
 
-       # The label file can be loaded here and used in the post-processing function.
-        # Directories for storing the label.txt file on OBS and in the model package
+       # The label file can be loaded here and used in the post-processing function.
+        # Directories for storing the label.txt file on OBS and in the model package
 
-        # with open(os.path.join(self.model_path, 'label.txt')) as f:
-        #     self.label = json.load(f)
+        # with open(os.path.join(self.model_path, 'label.txt')) as f:
+        #     self.label = json.load(f)
 
-        # Load the model in saved_model format in non-blocking mode to prevent blocking timeout.
-        thread = threading.Thread(target=self.get_tf_sess)
-        thread.start()
+        # Load the model in saved_model format in non-blocking mode to prevent blocking timeout.
+        thread = threading.Thread(target=self.get_tf_sess)
+        thread.start()
 
-    def get_tf_sess(self):
-        # Load the model in saved_model format.
+    def get_tf_sess(self):
+        # Load the model in saved_model format.
 
-       # The session will be reused. Do not use the with statement.
-        sess = tf.Session(graph=tf.Graph())
-        meta_graph_def = tf.saved_model.loader.load(sess, [tf.saved_model.tag_constants.SERVING], self.model_path)
-        signature_defs = meta_graph_def.signature_def
+       # The session will be reused. Do not use the with statement.
+        sess = tf.Session(graph=tf.Graph())
+        meta_graph_def = tf.saved_model.loader.load(sess, [tf.saved_model.tag_constants.SERVING], self.model_path)
+        signature_defs = meta_graph_def.signature_def
 
-        self.sess = sess
+        self.sess = sess
 
-        signature = []
+        signature = []
 
-        # only one signature allowed
-        for signature_def in signature_defs:
-            signature.append(signature_def)
-        if len(signature) == 1:
-            model_signature = signature[0]
-        else:
-            logger.warning("signatures more than one, use serving_default signature")
-            model_signature = tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY
+        # only one signature allowed
+        for signature_def in signature_defs:
+            signature.append(signature_def)
+        if len(signature) == 1:
+            model_signature = signature[0]
+        else:
+            logger.warning("signatures more than one, use serving_default signature")
+            model_signature = tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY
 
-        logger.info("model signature: %s", model_signature)
+        logger.info("model signature: %s", model_signature)
 
-        for signature_name in meta_graph_def.signature_def[model_signature].inputs:
-            tensorinfo = meta_graph_def.signature_def[model_signature].inputs[signature_name]
-            name = tensorinfo.name
-            op = self.sess.graph.get_tensor_by_name(name)
-            self.model_inputs[signature_name] = op
+        for signature_name in meta_graph_def.signature_def[model_signature].inputs:
+            tensorinfo = meta_graph_def.signature_def[model_signature].inputs[signature_name]
+            name = tensorinfo.name
+            op = self.sess.graph.get_tensor_by_name(name)
+            self.model_inputs[signature_name] = op
 
-        logger.info("model inputs: %s", self.model_inputs)
+        logger.info("model inputs: %s", self.model_inputs)
 
-        for signature_name in meta_graph_def.signature_def[model_signature].outputs:
-            tensorinfo = meta_graph_def.signature_def[model_signature].outputs[signature_name]
-            name = tensorinfo.name
-            op = self.sess.graph.get_tensor_by_name(name)
+        for signature_name in meta_graph_def.signature_def[model_signature].outputs:
+            tensorinfo = meta_graph_def.signature_def[model_signature].outputs[signature_name]
+            name = tensorinfo.name
+            op = self.sess.graph.get_tensor_by_name(name)
 
-            self.model_outputs[signature_name] = op
+            self.model_outputs[signature_name] = op
 
-        logger.info("model outputs: %s", self.model_outputs)
+        logger.info("model outputs: %s", self.model_outputs)
 
-    def _preprocess(self, data):
-        # Two request modes using HTTPS
-        # 1. The request in form-data file format is as follows: data = {"Request key value":{"File name":<File io>}}
-       # 2. Request in JSON format is as follows: data = json.loads("JSON body transferred by the API")
-        preprocessed_data = {}
+    def _preprocess(self, data):
+        # Two request modes using HTTPS
+        # 1. The request in form-data file format is as follows: data = {"Request key value":{"File name":<File io>}}
+       # 2. Request in JSON format is as follows: data = json.loads("JSON body transferred by the API")
+        preprocessed_data = {}
 
-        for k, v in data.items():
-            for file_name, file_content in v.items():
-                image1 = Image.open(file_content)
-                image1 = np.array(image1, dtype=np.float32)
-                image1.resize((1, 28, 28))
-                preprocessed_data[k] = image1
+        for k, v in data.items():
+            for file_name, file_content in v.items():
+                image1 = Image.open(file_content)
+                image1 = np.array(image1, dtype=np.float32)
+                image1.resize((1, 28, 28))
+                preprocessed_data[k] = image1
 
-        return preprocessed_data
+        return preprocessed_data
 
-    def _inference(self, data):
+    def _inference(self, data):
 
-        feed_dict = {}
-        for k, v in data.items():
-            if k not in self.model_inputs.keys():
-                logger.error("input key %s is not in model inputs %s", k, list(self.model_inputs.keys()))
-                raise Exception("input key %s is not in model inputs %s" % (k, list(self.model_inputs.keys())))
-            feed_dict[self.model_inputs[k]] = v
+        feed_dict = {}
+        for k, v in data.items():
+            if k not in self.model_inputs.keys():
+                logger.error("input key %s is not in model inputs %s", k, list(self.model_inputs.keys()))
+                raise Exception("input key %s is not in model inputs %s" % (k, list(self.model_inputs.keys())))
+            feed_dict[self.model_inputs[k]] = v
 
-        result = self.sess.run(self.model_outputs, feed_dict=feed_dict)
-        logger.info('predict result : ' + str(result))
+        result = self.sess.run(self.model_outputs, feed_dict=feed_dict)
+        logger.info('predict result : ' + str(result))
 
-        return result
+        return result
 
-    def _postprocess(self, data):
-        infer_output = {"mnist_result": []}
-        for output_name, results in data.items():
+    def _postprocess(self, data):
+        infer_output = {"mnist_result": []}
+        for output_name, results in data.items():
 
-            for result in results:
-                infer_output["mnist_result"].append(np.argmax(result))
+            for result in results:
+                infer_output["mnist_result"].append(np.argmax(result))
 
-        return infer_output
+        return infer_output
 
-    def __del__(self):
-        self.sess.close()
-
                                  
-
                                  
+    def __del__(self):
+        self.sess.close()
                                  
 
                                  
 
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0100.html b/docs/modelarts/umn/modelarts_23_0100.html
index 242da471f..77464bb43 100644
--- a/docs/modelarts/umn/modelarts_23_0100.html
+++ b/docs/modelarts/umn/modelarts_23_0100.html
@@ -38,7 +38,7 @@
 
                                  
 
                                  
 
                                  The JSON Schema of the inference result is as follows:
                                  
-
                                  {
+
                                  {
 	"type": "object",
 	"properties": {
 		"detection_classes": {
diff --git a/docs/modelarts/umn/modelarts_23_0102.html b/docs/modelarts/umn/modelarts_23_0102.html
index 6e9118948..6fd687e6c 100644
--- a/docs/modelarts/umn/modelarts_23_0102.html
+++ b/docs/modelarts/umn/modelarts_23_0102.html
@@ -42,7 +42,7 @@
 
                                  
 
                                  ReqData is of the Object type and indicates the inference data. The data structure is determined by the application scenario. For models using this mode, the preprocessing logic in the custom model inference code should be able to correctly process the data inputted in the format defined by the mode.
                                  
 
                                  The JSON Schema of a prediction request is as follows:
                                  
-
                                  {
+
                                  {
 	"type": "object",
 	"properties": {
 		"data": {
@@ -101,7 +101,7 @@
 
                                  
 
                                  Similar to ReqData, RespData is also of the Object type and indicates the prediction result. Its structure is determined by the application scenario. For models using this mode, the postprocessing logic in the custom model inference code should be able to correctly output data in the format defined by the mode.
                                  
 
                                  The JSON Schema of a prediction result is as follows:
                                  
-
                                  {
+
                                  {
 	"type": "object",
 	"properties": {
 		"data": {
diff --git a/docs/modelarts/umn/modelarts_23_0157.html b/docs/modelarts/umn/modelarts_23_0157.html
index 3a5563e6b..743c50435 100644
--- a/docs/modelarts/umn/modelarts_23_0157.html
+++ b/docs/modelarts/umn/modelarts_23_0157.html
@@ -3,14 +3,14 @@
 
                                  Requirements on Datasets
                                  
 
                                  The built-in algorithms provided by ModelArts can be used for image classification, object detection, and image semantic segmentation. The requirements for the datasets vary according to the built-in algorithms used for different purposes. Before using a built-in algorithm to create a training job, you are advised to prepare a dataset based on the requirements of the algorithm.
                                  
 
                                  Image Classification
                                  The training dataset must be stored in the OBS bucket. The following shows the OBS path structure of the dataset:
                                  
-
                                  |-- data_url
+
                                  |-- data_url
       |--a.jpg
       |--a.txt
       |--b.jpg
       |--b.txt
       ...
                                  
 
                                  • data_url indicates the folder name. You can customize the folder name. Images and label files cannot be stored in the root directory of an OBS bucket.
                                  • Images and label files must have the same name. The .txt files are label files for image classification. The images can be in JPG, JPEG, PNG, or BMP format.
                                  • The first row of label files for image classification indicates the category name of images, which can be Chinese characters, English letters, or digits. The following provides an example of file content:
                                    cat
                                    
-
                                  • In addition to the preceding files and folders, no other files or folders can exist in the data_url folder.
                                  • You can directly use an existing image classification dataset with published versions in Data Management of ModelArts.
                                  • You can also name sub-folders in the data_url directory by label, as shown in the following:
                                    |-- data_url
+
                                  • In addition to the preceding files and folders, no other files or folders can exist in the data_url folder.
                                  • You can directly use an existing image classification dataset with published versions in Data Management of ModelArts.
                                  • You can also name sub-folders in the data_url directory by label, as shown in the following:
                                    |-- data_url
     |--cat
       |--a.jpg
       |--a.txt
@@ -21,69 +21,42 @@
 
                                  
 
                                  
 
                                  Object Detection and Locating
                                  The training dataset must be stored in the OBS bucket. The following shows the OBS path structure of the dataset:
                                  
-
                                  |-- data_url
+
                                  |-- data_url
       |--a.jpg
       |--a.xml
       |--b.jpg
       |--b.xml
       ...
                                  
-
                                  • data_url indicates the folder name. You can customize the folder name. Images and label files cannot be stored in the root directory of an OBS bucket.
                                  • Images and label files must have the same name. The .xml files are label files for object detection. The images can be in JPG, JPEG, PNG, or BMP format.
                                  • In addition to the preceding files and folders, no other files or folders can exist in the data_url folder.
                                  • You can directly use an existing object detection dataset with published versions in Data Management of ModelArts.
                                  • The following provides a label file for object detection. The key parameters are size (image size), object (object information), and name (label name, which can be Chinese characters, English letters, or digits). Note that the values of xmin, ymin, xmax, and ymax in the bndbox field cannot exceed the value of size. That is, the value of min cannot be less than 0, and the value of max cannot be greater than the value of width or height.
                                     1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
                                    <?xml version="1.0" encoding="UTF-8" standalone="no"?>
-<annotation>
-    <folder>Images</folder>
-    <filename>IMG_20180919_120022.jpg</filename>
-    <source>
-        <database>Unknown</database>
-    </source>
-    <size>
-        <width>800</width>
-        <height>600</height>
-        <depth>1</depth>
-    </size>
-    <segmented>0</segmented>
-    <object>
-        <name>yunbao</name>
-        <pose>Unspecified</pose>
-        <truncated>0</truncated>
-        <difficult>0</difficult>
-        <bndbox>
-            <xmin>216.00</xmin>
-            <ymin>108.00</ymin>
-            <xmax>705.00</xmax>
-            <ymax>488.00</ymax>
-        </bndbox>
-    </object>
-</annotation>
-
                                    
-
                                    
+
                                    • data_url indicates the folder name. You can customize the folder name. Images and label files cannot be stored in the root directory of an OBS bucket.
                                    • Images and label files must have the same name. The .xml files are label files for object detection. The images can be in JPG, JPEG, PNG, or BMP format.
                                    • In addition to the preceding files and folders, no other files or folders can exist in the data_url folder.
                                    • You can directly use an existing object detection dataset with published versions in Data Management of ModelArts.
                                    • The following provides a label file for object detection. The key parameters are size (image size), object (object information), and name (label name, which can be Chinese characters, English letters, or digits). Note that the values of xmin, ymin, xmax, and ymax in the bndbox field cannot exceed the value of size. That is, the value of min cannot be less than 0, and the value of max cannot be greater than the value of width or height.
                                      <?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<annotation>
+    <folder>Images</folder>
+    <filename>IMG_20180919_120022.jpg</filename>
+    <source>
+        <database>Unknown</database>
+    </source>
+    <size>
+        <width>800</width>
+        <height>600</height>
+        <depth>1</depth>
+    </size>
+    <segmented>0</segmented>
+    <object>
+        <name>yunbao</name>
+        <pose>Unspecified</pose>
+        <truncated>0</truncated>
+        <difficult>0</difficult>
+        <bndbox>
+            <xmin>216.00</xmin>
+            <ymin>108.00</ymin>
+            <xmax>705.00</xmax>
+            <ymax>488.00</ymax>
+        </bndbox>
+    </object>
+</annotation>
                                      
 
                                    
 
                                  
 
                                  Image Semantic Segmentation
                                  The training dataset must be stored in the OBS bucket. The following shows the OBS path structure of the dataset:
                                  
-
                                  |-- data_url
+
                                  |-- data_url
        |--Image
               |--a.jpg
               |--b.jpg
@@ -96,7 +69,7 @@
        |--val.txt
                                  
 
                                  Description:
                                  
 
                                  • data_url, Image, and Label indicate the OBS folder names. The Image folder stores images for semantic segmentation, and the Label folder stores labeled images.
                                  • The name and format of the images for semantic segmentation must be the same as those of the corresponding labeled images. Images in JPG, JPEG, PNG, and BMP formats are supported.
                                  • In the preceding code snippet, train.txt and val.txt are two list files. train.txt is the list file of the training set, and val.txt is the list file of the validation set. It is recommended that the ratio of the training set to the validation set be 8:2.
                                    In the list file, the relative paths of images and labels are separated by spaces. Different pieces of data are separated by newline characters. The following gives an example:
                                    
-
                                    Image/a.jpg Label/a.jpg
+
                                    Image/a.jpg Label/a.jpg
 Image/b.jpg Label/b.jpg
 ...
                                    
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0161.html b/docs/modelarts/umn/modelarts_23_0161.html
index c2474992d..090cf33c0 100644
--- a/docs/modelarts/umn/modelarts_23_0161.html
+++ b/docs/modelarts/umn/modelarts_23_0161.html
@@ -8,23 +8,23 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling Model Inference Code.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the TensorFlow-based model package
                                  
 
                                  When publishing the model, you only need to specify the model directory.
                                  
-
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
     ├── <<Custom Python package>>    (Optional) User's Python package, which can be directly referenced in the model inference code
     ├── saved_model.pb        (Mandatory) Protocol buffer file, which contains the diagram description of the model
-    ├── variables             Mandatory for the main file of the *.pb model. The folder must be named variables and contains the weight deviation of the model.
+    ├── variables             Mandatory for the main file of the *.pb model. The folder must be named variables and contains the weight deviation of the model.
         ├── variables.index                   Mandatory
         ├── variables.data-00000-of-00001     Mandatory
-    ├──customize_service.py   (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+    ├──customize_service.py   (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0162.html b/docs/modelarts/umn/modelarts_23_0162.html
index cbd6f0631..d2fce226f 100644
--- a/docs/modelarts/umn/modelarts_23_0162.html
+++ b/docs/modelarts/umn/modelarts_23_0162.html
@@ -8,23 +8,23 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling Model Inference Code.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the TensorFlow-based model package
                                  
 
                                  When publishing the model, you only need to specify the model directory.
                                  
-
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
     ├── <<Custom Python package>>    (Optional) User's Python package, which can be directly referenced in the model inference code
     ├── saved_model.pb        (Mandatory) Protocol buffer file, which contains the diagram description of the model
-    ├── variables             Mandatory for the main file of the *.pb model. The folder must be named variables and contains the weight deviation of the model.
+    ├── variables             Mandatory for the main file of the *.pb model. The folder must be named variables and contains the weight deviation of the model.
         ├── variables.index                   Mandatory
         ├── variables.data-00000-of-00001     Mandatory
-    ├──customize_service.py   (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+    ├──customize_service.py   (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0163.html b/docs/modelarts/umn/modelarts_23_0163.html
index 513c844a1..90ebd0073 100644
--- a/docs/modelarts/umn/modelarts_23_0163.html
+++ b/docs/modelarts/umn/modelarts_23_0163.html
@@ -8,21 +8,21 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling Model Inference Code.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the MXNet-based model package
                                  
 
                                  When publishing the model, you only need to specify the model directory.
                                  
-
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
     ├── <<Custom Python package>>       (Optional) User's Python package, which can be directly referenced in the model inference code
     ├── resnet-50-symbol.json    (Mandatory) Model definition file, which contains the neural network description of the model
     ├── resnet-50-0000.params    (Mandatory) Model variable parameter file, which contains parameter and weight information
-    ├──customize_service.py      (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+    ├──customize_service.py      (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0164.html b/docs/modelarts/umn/modelarts_23_0164.html
index 7c03e2336..18469fc08 100644
--- a/docs/modelarts/umn/modelarts_23_0164.html
+++ b/docs/modelarts/umn/modelarts_23_0164.html
@@ -8,21 +8,21 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling Model Inference Code.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the MXNet-based model package
                                  
 
                                  When publishing the model, you only need to specify the model directory.
                                  
-
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
     ├── <<Custom Python package>>       (Optional) User's Python package, which can be directly referenced in the model inference code
     ├── resnet-50-symbol.json    (Mandatory) Model definition file, which contains the neural network description of the model
     ├── resnet-50-0000.params    (Mandatory) Model variable parameter file, which contains parameter and weight information
-    ├──customize_service.py      (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+    ├──customize_service.py      (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0165.html b/docs/modelarts/umn/modelarts_23_0165.html
index 68c47ec26..a9c0515a5 100644
--- a/docs/modelarts/umn/modelarts_23_0165.html
+++ b/docs/modelarts/umn/modelarts_23_0165.html
@@ -8,20 +8,20 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling Model Inference Code.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the PyTorch-based model package
                                  
 
                                  When publishing the model, you only need to specify the model directory.
                                  
-
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
    ├── <<Custom Python package>>     (Optional) User's Python package, which can be directly referenced in the model inference code
    ├── resnet50.pth           (Mandatory) PyTorch model file, which contains variable and weight information
-   ├──customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+   ├──customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0166.html b/docs/modelarts/umn/modelarts_23_0166.html
index 4a4555cfb..1b4a0e525 100644
--- a/docs/modelarts/umn/modelarts_23_0166.html
+++ b/docs/modelarts/umn/modelarts_23_0166.html
@@ -8,20 +8,20 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling Model Inference Code.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the PyTorch-based model package
                                  
 
                                  When publishing the model, you only need to specify the model directory.
                                  
-
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
    ├── <<Custom Python package>>     (Optional) User's Python package, which can be directly referenced in the model inference code
    ├── resnet50.pth           (Mandatory) PyTorch model file, which contains variable and weight information
-   ├──customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+   ├──customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0167.html b/docs/modelarts/umn/modelarts_23_0167.html
index 45c7c32e6..aa5cef742 100644
--- a/docs/modelarts/umn/modelarts_23_0167.html
+++ b/docs/modelarts/umn/modelarts_23_0167.html
@@ -8,20 +8,20 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling Model Inference Code.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the Caffe-based model package
                                  
-
                                  When publishing the model, you only need to specify the model directory.
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  When publishing the model, you only need to specify the model directory.
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
     |── <<Custom Python package>>      (Optional) User's Python package, which can be directly referenced in the model inference code
     |── deploy.prototxt         (Mandatory) Caffe model file, which contains information such as the model network structure
     |── resnet.caffemodel       (Mandatory) Caffe model file, which contains variable and weight information
-   |── customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+   |── customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0168.html b/docs/modelarts/umn/modelarts_23_0168.html
index 29da8ca1e..6dc451126 100644
--- a/docs/modelarts/umn/modelarts_23_0168.html
+++ b/docs/modelarts/umn/modelarts_23_0168.html
@@ -8,20 +8,20 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling Model Inference Code.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the Caffe-based model package
                                  
-
                                  When publishing the model, you only need to specify the model directory.
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  When publishing the model, you only need to specify the model directory.
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
     |── <<Custom Python package>>      (Optional) User's Python package, which can be directly referenced in the model inference code
     |── deploy.prototxt         (Mandatory) Caffe model file, which contains information such as the model network structure
     |── resnet.caffemodel       (Mandatory) Caffe model file, which contains variable and weight information
-   |── customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+   |── customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0169.html b/docs/modelarts/umn/modelarts_23_0169.html
index 0dcadea73..5d4575a12 100644
--- a/docs/modelarts/umn/modelarts_23_0169.html
+++ b/docs/modelarts/umn/modelarts_23_0169.html
@@ -8,20 +8,20 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling the Model Configuration File.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the Caffe-based model package
                                  
-
                                  When publishing the model, you only need to specify the model directory.
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  When publishing the model, you only need to specify the model directory.
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
     |── <<Custom Python package>>      (Optional) User's Python package, which can be directly referenced in the model inference code
     |── deploy.prototxt         (Mandatory) Caffe model file, which contains information such as the model network structure
     |── resnet.caffemodel       (Mandatory) Caffe model file, which contains variable and weight information
-   |── customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+   |── customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0170.html b/docs/modelarts/umn/modelarts_23_0170.html
index 9b70110f9..0a1d4cfed 100644
--- a/docs/modelarts/umn/modelarts_23_0170.html
+++ b/docs/modelarts/umn/modelarts_23_0170.html
@@ -8,20 +8,20 @@
 
                                  Input and Output Mode
                                  Undefined Mode can be overwritten. That is, you can select another input and output mode during model creation.
                                  
 
                                  
 
                                  Model Package Specifications
                                  • The model package must be stored in the OBS folder named model. Model files and the model inference code file are stored in the model folder.
                                  • The model inference code file is optional. If the file exists, the file name must be customize_service.py. Only one inference code file can exist in the model folder. For details about how to compile the model inference code file, see Specifications for Compiling Model Inference Code.
                                  
-
                                  • The structure of the model package imported using the template is as follows:
                                    model/
+
                                    • The structure of the model package imported using the template is as follows:
                                      model/
 │
 ├── Model file                 //(Mandatory) The model file format varies according to the engine. For details, see the model package example.
 ├── Custom Python package           //(Optional) User's Python package, which can be directly referenced in the model inference code
-├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                      
+├── customize_service.py  //(Optional) Model inference code file. The file name must be customize_service.py. Otherwise, the code is not considered as inference code.
                                    
 
                                  
 
                                  
 
                                  Model Package Example
                                  Structure of the Caffe-based model package
                                  
-
                                  When publishing the model, you only need to specify the model directory.
                                  OBS bucket/directory name
-|── model    (Mandatory) The folder must be named model and is used to store model-related files.
+
                                  When publishing the model, you only need to specify the model directory.
                                  OBS bucket/directory name
+|── model    (Mandatory) The folder must be named model and is used to store model-related files.
     |── <<Custom Python package>>      (Optional) User's Python package, which can be directly referenced in the model inference code
     |── deploy.prototxt         (Mandatory) Caffe model file, which contains information such as the model network structure
     |── resnet.caffemodel       (Mandatory) Caffe model file, which contains variable and weight information
-   |── customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
+   |── customize_service.py    (Optional) Model inference code file. The file must be named customize_service.py. Only one inference code file exists. The .py file on which customize_service.py depends can be directly put in the model directory.
                                  
 
                                  
 
                                  
 
diff --git a/docs/modelarts/umn/modelarts_23_0173.html b/docs/modelarts/umn/modelarts_23_0173.html
index a26e12d7b..dd3cef5e8 100644
--- a/docs/modelarts/umn/modelarts_23_0173.html
+++ b/docs/modelarts/umn/modelarts_23_0173.html
@@ -2,733 +2,369 @@
 
 
                                  TensorFlow
                                  
 
                                  TensorFlow has two types of APIs: Keras and tf. Keras and tf use different code for training and saving models, but the same code for inference.
                                  
-
                                  Training a Model (Keras API)
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
                                  from keras.models import Sequential
-model = Sequential()
-from keras.layers import Dense
-import tensorflow as tf
+
                                  Training a Model (Keras API)
                                  from keras.models import Sequential
+model = Sequential()
+from keras.layers import Dense
+import tensorflow as tf
 
-# Import a training dataset.
-mnist = tf.keras.datasets.mnist
-(x_train, y_train),(x_test, y_test) = mnist.load_data()
-x_train, x_test = x_train / 255.0, x_test / 255.0
+# Import a training dataset.
+mnist = tf.keras.datasets.mnist
+(x_train, y_train),(x_test, y_test) = mnist.load_data()
+x_train, x_test = x_train / 255.0, x_test / 255.0
 
-print(x_train.shape)
+print(x_train.shape)
 
-from keras.layers import Dense
-from keras.models import Sequential
-import keras
-from keras.layers import Dense, Activation, Flatten, Dropout
+from keras.layers import Dense
+from keras.models import Sequential
+import keras
+from keras.layers import Dense, Activation, Flatten, Dropout
 
-# Define a model network.
-model = Sequential()
-model.add(Flatten(input_shape=(28,28)))
-model.add(Dense(units=5120,activation='relu'))
-model.add(Dropout(0.2))
+# Define a model network.
+model = Sequential()
+model.add(Flatten(input_shape=(28,28)))
+model.add(Dense(units=5120,activation='relu'))
+model.add(Dropout(0.2))
 
-model.add(Dense(units=10, activation='softmax'))
+model.add(Dense(units=10, activation='softmax'))
 
-# Define an optimizer and loss functions.
-model.compile(optimizer='adam',
-              loss='sparse_categorical_crossentropy',
-              metrics=['accuracy'])
+# Define an optimizer and loss functions.
+model.compile(optimizer='adam',
+              loss='sparse_categorical_crossentropy',
+              metrics=['accuracy'])
 
-model.summary()
-# Train the model.
-model.fit(x_train, y_train, epochs=2)
-# Evaluate the model.
-model.evaluate(x_test, y_test)
-
                                  
-
                                  
+model.summary()
+# Train the model.
+model.fit(x_train, y_train, epochs=2)
+# Evaluate the model.
+model.evaluate(x_test, y_test)
 
                                  
-
                                  Saving a Model (Keras API)
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
                                  from keras import backend as K  
+
                                  Saving a Model (Keras API)
                                  from keras import backend as K  
 
-# K.get_session().run(tf.global_variables_initializer())
+# K.get_session().run(tf.global_variables_initializer())
 
-# Define the inputs and outputs of the prediction API.
-# The key values of the inputs and outputs dictionaries are used as the index keys for the input and output tensors of the model.
- # The input and output definitions of the model must match the custom inference script.
-predict_signature = tf.saved_model.signature_def_utils.predict_signature_def(
-    inputs={"images" : model.input},
-    outputs={"scores" : model.output}
-)
+# Define the inputs and outputs of the prediction API.
+# The key values of the inputs and outputs dictionaries are used as the index keys for the input and output tensors of the model.
+ # The input and output definitions of the model must match the custom inference script.
+predict_signature = tf.saved_model.signature_def_utils.predict_signature_def(
+    inputs={"images" : model.input},
+    outputs={"scores" : model.output}
+)
 
-# Define a save path.
-builder = tf.saved_model.builder.SavedModelBuilder('./mnist_keras/')
+# Define a save path.
+builder = tf.saved_model.builder.SavedModelBuilder('./mnist_keras/')
 
-builder.add_meta_graph_and_variables(
+builder.add_meta_graph_and_variables(
 
-    sess = K.get_session(),
-    # The tf.saved_model.tag_constants.SERVING tag needs to be defined for inference and deployment.
-    tags=[tf.saved_model.tag_constants.SERVING],
-    """
-    signature_def_map: Only single items can exist, or the corresponding key needs to be defined as follows:
-    tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY
-    """
-    signature_def_map={
-        tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
-            predict_signature
-    }
+    sess = K.get_session(),
+    # The tf.saved_model.tag_constants.SERVING tag needs to be defined for inference and deployment.
+    tags=[tf.saved_model.tag_constants.SERVING],
+    """
+    signature_def_map: Only single items can exist, or the corresponding key needs to be defined as follows:
+    tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY
+    """
+    signature_def_map={
+        tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
+            predict_signature
+    }
 
-)
-builder.save()
-
                                  
-
                                  
+)
+builder.save()
 
                                  
-
                                  Training a Model (tf API)
                                    1
-  2
-  3
-  4
-  5
-  6
-  7
-  8
-  9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
-100
-101
-102
-103
-104
-105
-106
-107
-108
-109
-110
-111
-112
-113
-114
-115
-116
-117
-118
-119
-120
-121
-122
-123
-124
-125
-126
-127
-128
-129
-130
-131
-132
-133
-134
-135
-136
-137
-138
-139
-140
-141
-142
-143
-144
-145
-146
-147
-148
-149
-150
-151
-152
-153
-154
-155
-156
-157
-158
-159
-160
-161
-162
-163
-164
-165
-166
-167
-168
-169
-170
-171
-172
-173
-174
-175
-176
-177
-178
-179
-180
-181
-182
-183
-184
-185
-186
-187
-188
-189
-190
-191
-192
-193
-194
-195
-196
-197
-198
-199
-200
-201
-202
-203
-204
-205
-206
-207
-208
-209
-210
-211
-212
-213
-214
-215
-216
-217
-218
-219
-220
-221
-222
-223
-224
-225
                                  from __future__ import print_function
+
                                  Training a Model (tf API)
                                  from __future__ import print_function
 
-import gzip
-import os
-import urllib
+import gzip
+import os
+import urllib
 
-import numpy
-import tensorflow as tf
-from six.moves import urllib
+import numpy
+import tensorflow as tf
+from six.moves import urllib
 
-# Training data is obtained from the Yann LeCun official website http://yann.lecun.com/exdb/mnist/.
-SOURCE_URL = 'http://yann.lecun.com/exdb/mnist/'
-TRAIN_IMAGES = 'train-images-idx3-ubyte.gz'
-TRAIN_LABELS = 'train-labels-idx1-ubyte.gz'
-TEST_IMAGES = 't10k-images-idx3-ubyte.gz'
-TEST_LABELS = 't10k-labels-idx1-ubyte.gz'
-VALIDATION_SIZE = 5000
+# Training data is obtained from the Yann LeCun official website http://yann.lecun.com/exdb/mnist/.
+SOURCE_URL = 'http://yann.lecun.com/exdb/mnist/'
+TRAIN_IMAGES = 'train-images-idx3-ubyte.gz'
+TRAIN_LABELS = 'train-labels-idx1-ubyte.gz'
+TEST_IMAGES = 't10k-images-idx3-ubyte.gz'
+TEST_LABELS = 't10k-labels-idx1-ubyte.gz'
+VALIDATION_SIZE = 5000
 
 
-def maybe_download(filename, work_directory):
-    """Download the data from Yann's website, unless it's already here."""
-    if not os.path.exists(work_directory):
-        os.mkdir(work_directory)
-    filepath = os.path.join(work_directory, filename)
-    if not os.path.exists(filepath):
-        filepath, _ = urllib.request.urlretrieve(SOURCE_URL + filename, filepath)
-        statinfo = os.stat(filepath)
-        print('Successfully downloaded %s %d bytes.' % (filename, statinfo.st_size))
-    return filepath
+def maybe_download(filename, work_directory):
+    """Download the data from Yann's website, unless it's already here."""
+    if not os.path.exists(work_directory):
+        os.mkdir(work_directory)
+    filepath = os.path.join(work_directory, filename)
+    if not os.path.exists(filepath):
+        filepath, _ = urllib.request.urlretrieve(SOURCE_URL + filename, filepath)
+        statinfo = os.stat(filepath)
+        print('Successfully downloaded %s %d bytes.' % (filename, statinfo.st_size))
+    return filepath
 
 
-def _read32(bytestream):
-    dt = numpy.dtype(numpy.uint32).newbyteorder('>')
-    return numpy.frombuffer(bytestream.read(4), dtype=dt)[0]
+def _read32(bytestream):
+    dt = numpy.dtype(numpy.uint32).newbyteorder('>')
+    return numpy.frombuffer(bytestream.read(4), dtype=dt)[0]
 
 
-def extract_images(filename):
-    """Extract the images into a 4D uint8 numpy array [index, y, x, depth]."""
-    print('Extracting %s' % filename)
-    with gzip.open(filename) as bytestream:
-        magic = _read32(bytestream)
-        if magic != 2051:
-            raise ValueError(
-                'Invalid magic number %d in MNIST image file: %s' %
-                (magic, filename))
-        num_images = _read32(bytestream)
-        rows = _read32(bytestream)
-        cols = _read32(bytestream)
-        buf = bytestream.read(rows * cols * num_images)
-        data = numpy.frombuffer(buf, dtype=numpy.uint8)
-        data = data.reshape(num_images, rows, cols, 1)
-        return data
+def extract_images(filename):
+    """Extract the images into a 4D uint8 numpy array [index, y, x, depth]."""
+    print('Extracting %s' % filename)
+    with gzip.open(filename) as bytestream:
+        magic = _read32(bytestream)
+        if magic != 2051:
+            raise ValueError(
+                'Invalid magic number %d in MNIST image file: %s' %
+                (magic, filename))
+        num_images = _read32(bytestream)
+        rows = _read32(bytestream)
+        cols = _read32(bytestream)
+        buf = bytestream.read(rows * cols * num_images)
+        data = numpy.frombuffer(buf, dtype=numpy.uint8)
+        data = data.reshape(num_images, rows, cols, 1)
+        return data
 
 
-def dense_to_one_hot(labels_dense, num_classes=10):
-    """Convert class labels from scalars to one-hot vectors."""
-    num_labels = labels_dense.shape[0]
-    index_offset = numpy.arange(num_labels) * num_classes
-    labels_one_hot = numpy.zeros((num_labels, num_classes))
-    labels_one_hot.flat[index_offset + labels_dense.ravel()] = 1
-    return labels_one_hot
+def dense_to_one_hot(labels_dense, num_classes=10):
+    """Convert class labels from scalars to one-hot vectors."""
+    num_labels = labels_dense.shape[0]
+    index_offset = numpy.arange(num_labels) * num_classes
+    labels_one_hot = numpy.zeros((num_labels, num_classes))
+    labels_one_hot.flat[index_offset + labels_dense.ravel()] = 1
+    return labels_one_hot
 
 
-def extract_labels(filename, one_hot=False):
-    """Extract the labels into a 1D uint8 numpy array [index]."""
-    print('Extracting %s' % filename)
-    with gzip.open(filename) as bytestream:
-        magic = _read32(bytestream)
-        if magic != 2049:
-            raise ValueError(
-                'Invalid magic number %d in MNIST label file: %s' %
-                (magic, filename))
-        num_items = _read32(bytestream)
-        buf = bytestream.read(num_items)
-        labels = numpy.frombuffer(buf, dtype=numpy.uint8)
-        if one_hot:
-            return dense_to_one_hot(labels)
-        return labels
+def extract_labels(filename, one_hot=False):
+    """Extract the labels into a 1D uint8 numpy array [index]."""
+    print('Extracting %s' % filename)
+    with gzip.open(filename) as bytestream:
+        magic = _read32(bytestream)
+        if magic != 2049:
+            raise ValueError(
+                'Invalid magic number %d in MNIST label file: %s' %
+                (magic, filename))
+        num_items = _read32(bytestream)
+        buf = bytestream.read(num_items)
+        labels = numpy.frombuffer(buf, dtype=numpy.uint8)
+        if one_hot:
+            return dense_to_one_hot(labels)
+        return labels
 
 
-class DataSet(object):
-    """Class encompassing test, validation and training MNIST data set."""
+class DataSet(object):
+    """Class encompassing test, validation and training MNIST data set."""
 
-    def __init__(self, images, labels, fake_data=False, one_hot=False):
-        """Construct a DataSet. one_hot arg is used only if fake_data is true."""
+    def __init__(self, images, labels, fake_data=False, one_hot=False):
+        """Construct a DataSet. one_hot arg is used only if fake_data is true."""
 
-        if fake_data:
-            self._num_examples = 10000
-            self.one_hot = one_hot
-        else:
-            assert images.shape[0] == labels.shape[0], (
-                    'images.shape: %s labels.shape: %s' % (images.shape,
-                                                           labels.shape))
-            self._num_examples = images.shape[0]
+        if fake_data:
+            self._num_examples = 10000
+            self.one_hot = one_hot
+        else:
+            assert images.shape[0] == labels.shape[0], (
+                    'images.shape: %s labels.shape: %s' % (images.shape,
+                                                           labels.shape))
+            self._num_examples = images.shape[0]
 
-            # Convert shape from [num examples, rows, columns, depth]
-            # to [num examples, rows*columns] (assuming depth == 1)
-            assert images.shape[3] == 1
-            images = images.reshape(images.shape[0],
-                                    images.shape[1] * images.shape[2])
-            # Convert from [0, 255] -> [0.0, 1.0].
-            images = images.astype(numpy.float32)
-            images = numpy.multiply(images, 1.0 / 255.0)
-        self._images = images
-        self._labels = labels
-        self._epochs_completed = 0
-        self._index_in_epoch = 0
+            # Convert shape from [num examples, rows, columns, depth]
+            # to [num examples, rows*columns] (assuming depth == 1)
+            assert images.shape[3] == 1
+            images = images.reshape(images.shape[0],
+                                    images.shape[1] * images.shape[2])
+            # Convert from [0, 255] -> [0.0, 1.0].
+            images = images.astype(numpy.float32)
+            images = numpy.multiply(images, 1.0 / 255.0)
+        self._images = images
+        self._labels = labels
+        self._epochs_completed = 0
+        self._index_in_epoch = 0
 
-    @property
-    def images(self):
-        return self._images
+    @property
+    def images(self):
+        return self._images
 
-    @property
-    def labels(self):
-        return self._labels
+    @property
+    def labels(self):
+        return self._labels
 
-    @property
-    def num_examples(self):
-        return self._num_examples
+    @property
+    def num_examples(self):
+        return self._num_examples
 
-    @property
-    def epochs_completed(self):
-        return self._epochs_completed
+    @property
+    def epochs_completed(self):
+        return self._epochs_completed
 
-    def next_batch(self, batch_size, fake_data=False):
-        """Return the next `batch_size` examples from this data set."""
-        if fake_data:
-            fake_image = [1] * 784
-            if self.one_hot:
-                fake_label = [1] + [0] * 9
-            else:
-                fake_label = 0
-            return [fake_image for _ in range(batch_size)], [
-                fake_label for _ in range(batch_size)
-            ]
-        start = self._index_in_epoch
-        self._index_in_epoch += batch_size
-        if self._index_in_epoch > self._num_examples:
-            # Finished epoch
-            self._epochs_completed += 1
-            # Shuffle the data
-            perm = numpy.arange(self._num_examples)
-            numpy.random.shuffle(perm)
-            self._images = self._images[perm]
-            self._labels = self._labels[perm]
-            # Start next epoch
-            start = 0
-            self._index_in_epoch = batch_size
-            assert batch_size <= self._num_examples
-        end = self._index_in_epoch
-        return self._images[start:end], self._labels[start:end]
+    def next_batch(self, batch_size, fake_data=False):
+        """Return the next `batch_size` examples from this data set."""
+        if fake_data:
+            fake_image = [1] * 784
+            if self.one_hot:
+                fake_label = [1] + [0] * 9
+            else:
+                fake_label = 0
+            return [fake_image for _ in range(batch_size)], [
+                fake_label for _ in range(batch_size)
+            ]
+        start = self._index_in_epoch
+        self._index_in_epoch += batch_size
+        if self._index_in_epoch > self._num_examples:
+            # Finished epoch
+            self._epochs_completed += 1
+            # Shuffle the data
+            perm = numpy.arange(self._num_examples)
+            numpy.random.shuffle(perm)
+            self._images = self._images[perm]
+            self._labels = self._labels[perm]
+            # Start next epoch
+            start = 0
+            self._index_in_epoch = batch_size
+            assert batch_size <= self._num_examples
+        end = self._index_in_epoch
+        return self._images[start:end], self._labels[start:end]
 
 
-def read_data_sets(train_dir, fake_data=False, one_hot=False):
-    """Return training, validation and testing data sets."""
+def read_data_sets(train_dir, fake_data=False, one_hot=False):
+    """Return training, validation and testing data sets."""
 
-    class DataSets(object):
-        pass
+    class DataSets(object):
+        pass
 
-    data_sets = DataSets()
+    data_sets = DataSets()
 
-    if fake_data:
-        data_sets.train = DataSet([], [], fake_data=True, one_hot=one_hot)
-        data_sets.validation = DataSet([], [], fake_data=True, one_hot=one_hot)
-        data_sets.test = DataSet([], [], fake_data=True, one_hot=one_hot)
-        return data_sets
+    if fake_data:
+        data_sets.train = DataSet([], [], fake_data=True, one_hot=one_hot)
+        data_sets.validation = DataSet([], [], fake_data=True, one_hot=one_hot)
+        data_sets.test = DataSet([], [], fake_data=True, one_hot=one_hot)
+        return data_sets
 
-    local_file = maybe_download(TRAIN_IMAGES, train_dir)
-    train_images = extract_images(local_file)
+    local_file = maybe_download(TRAIN_IMAGES, train_dir)
+    train_images = extract_images(local_file)
 
-    local_file = maybe_download(TRAIN_LABELS, train_dir)
-    train_labels = extract_labels(local_file, one_hot=one_hot)
+    local_file = maybe_download(TRAIN_LABELS, train_dir)
+    train_labels = extract_labels(local_file, one_hot=one_hot)
 
-    local_file = maybe_download(TEST_IMAGES, train_dir)
-    test_images = extract_images(local_file)
+    local_file = maybe_download(TEST_IMAGES, train_dir)
+    test_images = extract_images(local_file)
 
-    local_file = maybe_download(TEST_LABELS, train_dir)
-    test_labels = extract_labels(local_file, one_hot=one_hot)
+    local_file = maybe_download(TEST_LABELS, train_dir)
+    test_labels = extract_labels(local_file, one_hot=one_hot)
 
-    validation_images = train_images[:VALIDATION_SIZE]
-    validation_labels = train_labels[:VALIDATION_SIZE]
-    train_images = train_images[VALIDATION_SIZE:]
-    train_labels = train_labels[VALIDATION_SIZE:]
+    validation_images = train_images[:VALIDATION_SIZE]
+    validation_labels = train_labels[:VALIDATION_SIZE]
+    train_images = train_images[VALIDATION_SIZE:]
+    train_labels = train_labels[VALIDATION_SIZE:]
 
-    data_sets.train = DataSet(train_images, train_labels)
-    data_sets.validation = DataSet(validation_images, validation_labels)
-    data_sets.test = DataSet(test_images, test_labels)
-    return data_sets
+    data_sets.train = DataSet(train_images, train_labels)
+    data_sets.validation = DataSet(validation_images, validation_labels)
+    data_sets.test = DataSet(test_images, test_labels)
+    return data_sets
 
-training_iteration = 1000
+training_iteration = 1000
 
-modelarts_example_path =  './modelarts-mnist-train-save-deploy-example'
+modelarts_example_path =  './modelarts-mnist-train-save-deploy-example'
 
-export_path = modelarts_example_path + '/model/'
-data_path = './'
+export_path = modelarts_example_path + '/model/'
+data_path = './'
 
-print('Training model...')
-mnist = read_data_sets(data_path, one_hot=True)
-sess = tf.InteractiveSession()
-serialized_tf_example = tf.placeholder(tf.string, name='tf_example')
-feature_configs = {'x': tf.FixedLenFeature(shape=[784], dtype=tf.float32), }
-tf_example = tf.parse_example(serialized_tf_example, feature_configs)
-x = tf.identity(tf_example['x'], name='x')  # use tf.identity() to assign name
-y_ = tf.placeholder('float', shape=[None, 10])
-w = tf.Variable(tf.zeros([784, 10]))
-b = tf.Variable(tf.zeros([10]))
-sess.run(tf.global_variables_initializer())
-y = tf.nn.softmax(tf.matmul(x, w) + b, name='y')
-cross_entropy = -tf.reduce_sum(y_ * tf.log(y))
-train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy)
-values, indices = tf.nn.top_k(y, 10)
-table = tf.contrib.lookup.index_to_string_table_from_tensor(
-    tf.constant([str(i) for i in range(10)]))
-prediction_classes = table.lookup(tf.to_int64(indices))
-for _ in range(training_iteration):
-    batch = mnist.train.next_batch(50)
-    train_step.run(feed_dict={x: batch[0], y_: batch[1]})
-correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
-accuracy = tf.reduce_mean(tf.cast(correct_prediction, 'float'))
-print('training accuracy %g' % sess.run(
-    accuracy, feed_dict={
-        x: mnist.test.images,
-        y_: mnist.test.labels
-    }))
-print('Done training!')
-
                                  
-
                                  
+print('Training model...')
+mnist = read_data_sets(data_path, one_hot=True)
+sess = tf.InteractiveSession()
+serialized_tf_example = tf.placeholder(tf.string, name='tf_example')
+feature_configs = {'x': tf.FixedLenFeature(shape=[784], dtype=tf.float32), }
+tf_example = tf.parse_example(serialized_tf_example, feature_configs)
+x = tf.identity(tf_example['x'], name='x')  # use tf.identity() to assign name
+y_ = tf.placeholder('float', shape=[None, 10])
+w = tf.Variable(tf.zeros([784, 10]))
+b = tf.Variable(tf.zeros([10]))
+sess.run(tf.global_variables_initializer())
+y = tf.nn.softmax(tf.matmul(x, w) + b, name='y')
+cross_entropy = -tf.reduce_sum(y_ * tf.log(y))
+train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy)
+values, indices = tf.nn.top_k(y, 10)
+table = tf.contrib.lookup.index_to_string_table_from_tensor(
+    tf.constant([str(i) for i in range(10)]))
+prediction_classes = table.lookup(tf.to_int64(indices))
+for _ in range(training_iteration):
+    batch = mnist.train.next_batch(50)
+    train_step.run(feed_dict={x: batch[0], y_: batch[1]})
+correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
+accuracy = tf.reduce_mean(tf.cast(correct_prediction, 'float'))
+print('training accuracy %g' % sess.run(
+    accuracy, feed_dict={
+        x: mnist.test.images,
+        y_: mnist.test.labels
+    }))
+print('Done training!')
 
                                  
-
                                  Saving a Model (tf API)
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
                                  # Export the model.
-# The model needs to be saved using the saved_model API.
-print('Exporting trained model to', export_path)
-builder = tf.saved_model.builder.SavedModelBuilder(export_path)
+
                                  Saving a Model (tf API)
                                  # Export the model.
+# The model needs to be saved using the saved_model API.
+print('Exporting trained model to', export_path)
+builder = tf.saved_model.builder.SavedModelBuilder(export_path)
 
-tensor_info_x = tf.saved_model.utils.build_tensor_info(x)
-tensor_info_y = tf.saved_model.utils.build_tensor_info(y)
+tensor_info_x = tf.saved_model.utils.build_tensor_info(x)
+tensor_info_y = tf.saved_model.utils.build_tensor_info(y)
 
-# Define the inputs and outputs of the prediction API.
-# The key values of the inputs and outputs dictionaries are used as the index keys for the input and output tensors of the model.
- # The input and output definitions of the model must match the custom inference script.
-prediction_signature = (
-    tf.saved_model.signature_def_utils.build_signature_def(
-        inputs={'images': tensor_info_x},
-        outputs={'scores': tensor_info_y},
-        method_name=tf.saved_model.signature_constants.PREDICT_METHOD_NAME))
+# Define the inputs and outputs of the prediction API.
+# The key values of the inputs and outputs dictionaries are used as the index keys for the input and output tensors of the model.
+ # The input and output definitions of the model must match the custom inference script.
+prediction_signature = (
+    tf.saved_model.signature_def_utils.build_signature_def(
+        inputs={'images': tensor_info_x},
+        outputs={'scores': tensor_info_y},
+        method_name=tf.saved_model.signature_constants.PREDICT_METHOD_NAME))
 
-legacy_init_op = tf.group(tf.tables_initializer(), name='legacy_init_op')
-builder.add_meta_graph_and_variables(
-    # Set tag to serve/tf.saved_model.tag_constants.SERVING.
-    sess, [tf.saved_model.tag_constants.SERVING],
-    signature_def_map={
-        'predict_images':
-            prediction_signature,
-    },
-    legacy_init_op=legacy_init_op)
+legacy_init_op = tf.group(tf.tables_initializer(), name='legacy_init_op')
+builder.add_meta_graph_and_variables(
+    # Set tag to serve/tf.saved_model.tag_constants.SERVING.
+    sess, [tf.saved_model.tag_constants.SERVING],
+    signature_def_map={
+        'predict_images':
+            prediction_signature,
+    },
+    legacy_init_op=legacy_init_op)
 
-builder.save()
+builder.save()
 
-print('Done exporting!')
-
                                  
-
                                  
+print('Done exporting!')
 
                                  
-
                                  Inference Code (Keras and tf APIs)
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
                                  from PIL import Image
-import numpy as np
-from model_service.tfserving_model_service import TfServingBaseService
+
                                  Inference Code (Keras and tf APIs)
                                  from PIL import Image
+import numpy as np
+from model_service.tfserving_model_service import TfServingBaseService
 
 
-class mnist_service(TfServingBaseService):
+class mnist_service(TfServingBaseService):
 
-    # Match the model input with the user's HTTPS API input during preprocessing.
-    # The model input corresponding to the preceding training part is {"images":<array>}.
-    def _preprocess(self, data):
+    # Match the model input with the user's HTTPS API input during preprocessing.
+    # The model input corresponding to the preceding training part is {"images":<array>}.
+    def _preprocess(self, data):
 
-        preprocessed_data = {}
-        images = []
-        # Iterate the input data.
-        for k, v in data.items():
-            for file_name, file_content in v.items():
-                image1 = Image.open(file_content)
-                image1 = np.array(image1, dtype=np.float32)
-                image1.resize((1,784))
-                images.append(image1)
-        # Return the numpy array.
-        images = np.array(images,dtype=np.float32)
-        # Perform batch processing on multiple input samples and ensure that the shape is the same as that inputted during training.
-        images.resize((len(data), 784))
-        preprocessed_data['images'] = images
-        return preprocessed_data
+        preprocessed_data = {}
+        images = []
+        # Iterate the input data.
+        for k, v in data.items():
+            for file_name, file_content in v.items():
+                image1 = Image.open(file_content)
+                image1 = np.array(image1, dtype=np.float32)
+                image1.resize((1,784))
+                images.append(image1)
+        # Return the numpy array.
+        images = np.array(images,dtype=np.float32)
+        # Perform batch processing on multiple input samples and ensure that the shape is the same as that inputted during training.
+        images.resize((len(data), 784))
+        preprocessed_data['images'] = images
+        return preprocessed_data
 
-    # Processing logic of the inference for invoking the parent class.
+    # Processing logic of the inference for invoking the parent class.
 
-    # The output corresponding to model saving in the preceding training part is {"scores":<array>}.
-    # Postprocess the HTTPS output.
-    def _postprocess(self, data):
-        infer_output = {"mnist_result": []}
-        # Iterate the model output.
-        for output_name, results in data.items():
-            for result in results:
-                infer_output["mnist_result"].append(result.index(max(result)))
-        return infer_output
-
                                  
-
                                  
+    # The output corresponding to model saving in the preceding training part is {"scores":<array>}.
+    # Postprocess the HTTPS output.
+    def _postprocess(self, data):
+        infer_output = {"mnist_result": []}
+        # Iterate the model output.
+        for output_name, results in data.items():
+            for result in results:
+                infer_output["mnist_result"].append(result.index(max(result)))
+        return infer_output
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0175.html b/docs/modelarts/umn/modelarts_23_0175.html
index 7d0d52a88..9f10159c7 100644
--- a/docs/modelarts/umn/modelarts_23_0175.html
+++ b/docs/modelarts/umn/modelarts_23_0175.html
@@ -1,375 +1,190 @@
 
 
 
                                  PyTorch
                                  
-
                                  Training a Model
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
-46
-47
-48
-49
-50
-51
-52
-53
-54
-55
-56
-57
-58
-59
-60
-61
-62
-63
-64
-65
-66
-67
-68
-69
-70
-71
-72
-73
-74
-75
-76
-77
-78
-79
-80
                                  from __future__ import print_function
-import argparse
-import torch
-import torch.nn as nn
-import torch.nn.functional as F
-import torch.optim as optim
-from torchvision import datasets, transforms
+
                                  Training a Model
                                  from __future__ import print_function
+import argparse
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+import torch.optim as optim
+from torchvision import datasets, transforms
 
-# Define a network structure.
-class Net(nn.Module):
-    def __init__(self):
-        super(Net, self).__init__()
-# The second dimension of the input must be 784.
-        self.hidden1 = nn.Linear(784, 5120, bias=False)
-        self.output = nn.Linear(5120, 10, bias=False)
+# Define a network structure.
+class Net(nn.Module):
+    def __init__(self):
+        super(Net, self).__init__()
+# The second dimension of the input must be 784.
+        self.hidden1 = nn.Linear(784, 5120, bias=False)
+        self.output = nn.Linear(5120, 10, bias=False)
 
-    def forward(self, x):
-        x = x.view(x.size()[0], -1)
-        x = F.relu((self.hidden1(x)))
-        x = F.dropout(x, 0.2)
-        x = self.output(x)
-        return F.log_softmax(x)
+    def forward(self, x):
+        x = x.view(x.size()[0], -1)
+        x = F.relu((self.hidden1(x)))
+        x = F.dropout(x, 0.2)
+        x = self.output(x)
+        return F.log_softmax(x)
 
-def train(model, device, train_loader, optimizer, epoch):
-    model.train()
-    for batch_idx, (data, target) in enumerate(train_loader):
-        data, target = data.to(device), target.to(device)
-        optimizer.zero_grad()
-        output = model(data)
-        loss = F.cross_entropy(output, target)
-        loss.backward()
-        optimizer.step()
-        if batch_idx % 10 == 0:
-            print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
-                epoch, batch_idx * len(data), len(train_loader.dataset),
-                       100. * batch_idx / len(train_loader), loss.item()))
+def train(model, device, train_loader, optimizer, epoch):
+    model.train()
+    for batch_idx, (data, target) in enumerate(train_loader):
+        data, target = data.to(device), target.to(device)
+        optimizer.zero_grad()
+        output = model(data)
+        loss = F.cross_entropy(output, target)
+        loss.backward()
+        optimizer.step()
+        if batch_idx % 10 == 0:
+            print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
+                epoch, batch_idx * len(data), len(train_loader.dataset),
+                       100. * batch_idx / len(train_loader), loss.item()))
 
-def test( model, device, test_loader):
-    model.eval()
-    test_loss = 0
-    correct = 0
-    with torch.no_grad():
-        for data, target in test_loader:
-            data, target = data.to(device), target.to(device)
-            output = model(data)
-            test_loss += F.nll_loss(output, target, reduction='sum').item()  # sum up batch loss
-            pred = output.argmax(dim=1, keepdim=True)  # get the index of the max log-probability
-            correct += pred.eq(target.view_as(pred)).sum().item()
+def test( model, device, test_loader):
+    model.eval()
+    test_loss = 0
+    correct = 0
+    with torch.no_grad():
+        for data, target in test_loader:
+            data, target = data.to(device), target.to(device)
+            output = model(data)
+            test_loss += F.nll_loss(output, target, reduction='sum').item()  # sum up batch loss
+            pred = output.argmax(dim=1, keepdim=True)  # get the index of the max log-probability
+            correct += pred.eq(target.view_as(pred)).sum().item()
 
-    test_loss /= len(test_loader.dataset)
+    test_loss /= len(test_loader.dataset)
 
-    print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
-        test_loss, correct, len(test_loader.dataset),
-        100. * correct / len(test_loader.dataset)))
+    print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
+        test_loss, correct, len(test_loader.dataset),
+        100. * correct / len(test_loader.dataset)))
 
-device = torch.device("cpu")
+device = torch.device("cpu")
 
-batch_size=64
+batch_size=64
 
-kwargs={}
+kwargs={}
 
-train_loader = torch.utils.data.DataLoader(
-    datasets.MNIST('.', train=True, download=True,
-                   transform=transforms.Compose([
-                       transforms.ToTensor()
-                   ])),
-    batch_size=batch_size, shuffle=True, **kwargs)
-test_loader = torch.utils.data.DataLoader(
-    datasets.MNIST('.', train=False, transform=transforms.Compose([
-        transforms.ToTensor()
-    ])),
-    batch_size=1000, shuffle=True, **kwargs)
+train_loader = torch.utils.data.DataLoader(
+    datasets.MNIST('.', train=True, download=True,
+                   transform=transforms.Compose([
+                       transforms.ToTensor()
+                   ])),
+    batch_size=batch_size, shuffle=True, **kwargs)
+test_loader = torch.utils.data.DataLoader(
+    datasets.MNIST('.', train=False, transform=transforms.Compose([
+        transforms.ToTensor()
+    ])),
+    batch_size=1000, shuffle=True, **kwargs)
 
-model = Net().to(device)
-optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)
-optimizer = optim.Adam(model.parameters())
+model = Net().to(device)
+optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)
+optimizer = optim.Adam(model.parameters())
 
-for epoch in range(1, 2 + 1):
-    train(model, device, train_loader, optimizer, epoch)
-    test(model, device, test_loader)
-
                                  
-
                                  
+for epoch in range(1, 2 + 1):
+    train(model, device, train_loader, optimizer, epoch)
+    test(model, device, test_loader)
 
                                  
-
                                  Saving a Model
                                  1
-2
                                  # The model must be saved using state_dict and can be deployed remotely.
-torch.save(model.state_dict(), "pytorch_mnist/mnist_mlp.pt")
-
                                  
-
                                  
+
                                  Saving a Model
                                  # The model must be saved using state_dict and can be deployed remotely.
+torch.save(model.state_dict(), "pytorch_mnist/mnist_mlp.pt")
                                  
 
                                  
-
                                  Inference Code
                                    1
-  2
-  3
-  4
-  5
-  6
-  7
-  8
-  9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
-100
                                  from PIL import Image
-import log
-from model_service.pytorch_model_service import PTServingBaseService
-import torch.nn.functional as F
+
                                  Inference Code
                                  from PIL import Image
+import log
+from model_service.pytorch_model_service import PTServingBaseService
+import torch.nn.functional as F
 
-import torch.nn as nn
-import torch
-import json
+import torch.nn as nn
+import torch
+import json
 
-import numpy as np
+import numpy as np
 
-logger = log.getLogger(__name__)
+logger = log.getLogger(__name__)
 
-import torchvision.transforms as transforms
+import torchvision.transforms as transforms
 
-# Define model preprocessing.
-infer_transformation = transforms.Compose([
-    transforms.Resize((28,28)),
-    # Transform to a PyTorch tensor.
-    transforms.ToTensor()
-])
+# Define model preprocessing.
+infer_transformation = transforms.Compose([
+    transforms.Resize((28,28)),
+    # Transform to a PyTorch tensor.
+    transforms.ToTensor()
+])
 
 
-import os
+import os
 
 
-class PTVisionService(PTServingBaseService):
+class PTVisionService(PTServingBaseService):
 
-    def __init__(self, model_name, model_path):
-        # Call the constructor of the parent class.
-        super(PTVisionService, self).__init__(model_name, model_path)
-        # Call the customized function to load the model.
-        self.model = Mnist(model_path)
-         # Load tags.
-        self.label = [0,1,2,3,4,5,6,7,8,9]
-        # Labels can also be loaded by label file.
-        # Store the label.json file in the model directory. The following information is read:
-        dir_path = os.path.dirname(os.path.realpath(self.model_path))
-        with open(os.path.join(dir_path, 'label.json')) as f:
-            self.label = json.load(f)
+    def __init__(self, model_name, model_path):
+        # Call the constructor of the parent class.
+        super(PTVisionService, self).__init__(model_name, model_path)
+        # Call the customized function to load the model.
+        self.model = Mnist(model_path)
+         # Load tags.
+        self.label = [0,1,2,3,4,5,6,7,8,9]
+        # Labels can also be loaded by label file.
+        # Store the label.json file in the model directory. The following information is read:
+        dir_path = os.path.dirname(os.path.realpath(self.model_path))
+        with open(os.path.join(dir_path, 'label.json')) as f:
+            self.label = json.load(f)
 
 
-    def _preprocess(self, data):
+    def _preprocess(self, data):
 
-        preprocessed_data = {}
-        for k, v in data.items():
-            input_batch = []
-            for file_name, file_content in v.items():
-                with Image.open(file_content) as image1:
-                    # Gray processing
-                    image1 = image1.convert("L")
-                    if torch.cuda.is_available():
-                        input_batch.append(infer_transformation(image1).cuda())
-                    else:
-                        input_batch.append(infer_transformation(image1))
-            input_batch_var = torch.autograd.Variable(torch.stack(input_batch, dim=0), volatile=True)
-            print(input_batch_var.shape)
-            preprocessed_data[k] = input_batch_var
+        preprocessed_data = {}
+        for k, v in data.items():
+            input_batch = []
+            for file_name, file_content in v.items():
+                with Image.open(file_content) as image1:
+                    # Gray processing
+                    image1 = image1.convert("L")
+                    if torch.cuda.is_available():
+                        input_batch.append(infer_transformation(image1).cuda())
+                    else:
+                        input_batch.append(infer_transformation(image1))
+            input_batch_var = torch.autograd.Variable(torch.stack(input_batch, dim=0), volatile=True)
+            print(input_batch_var.shape)
+            preprocessed_data[k] = input_batch_var
 
-        return preprocessed_data
+        return preprocessed_data
 
-    def _postprocess(self, data):
-        results = []
-        for k, v in data.items():
-            result = torch.argmax(v[0])
-            result = {k: self.label[result]}
-            results.append(result)
-        return results
+    def _postprocess(self, data):
+        results = []
+        for k, v in data.items():
+            result = torch.argmax(v[0])
+            result = {k: self.label[result]}
+            results.append(result)
+        return results
 
-class Net(nn.Module):
-    def __init__(self):
-        super(Net, self).__init__()
-        self.hidden1 = nn.Linear(784, 5120, bias=False)
-        self.output = nn.Linear(5120, 10, bias=False)
+class Net(nn.Module):
+    def __init__(self):
+        super(Net, self).__init__()
+        self.hidden1 = nn.Linear(784, 5120, bias=False)
+        self.output = nn.Linear(5120, 10, bias=False)
 
-    def forward(self, x):
-        x = x.view(x.size()[0], -1)
-        x = F.relu((self.hidden1(x)))
-        x = F.dropout(x, 0.2)
-        x = self.output(x)
-        return F.log_softmax(x)
+    def forward(self, x):
+        x = x.view(x.size()[0], -1)
+        x = F.relu((self.hidden1(x)))
+        x = F.dropout(x, 0.2)
+        x = self.output(x)
+        return F.log_softmax(x)
 
 
 
-def Mnist(model_path, **kwargs):
-    # Generate a network.
-    model = Net()
-    # Load the model.
-    if torch.cuda.is_available():
-        device = torch.device('cuda')
-        model.load_state_dict(torch.load(model_path, map_location="cuda:0"))
-    else:
-        device = torch.device('cpu')
-        model.load_state_dict(torch.load(model_path, map_location=device))
-    # CPU or GPU mapping
-    model.to(device)
-    # Declare an inference mode.
-    model.eval()
+def Mnist(model_path, **kwargs):
+    # Generate a network.
+    model = Net()
+    # Load the model.
+    if torch.cuda.is_available():
+        device = torch.device('cuda')
+        model.load_state_dict(torch.load(model_path, map_location="cuda:0"))
+    else:
+        device = torch.device('cpu')
+        model.load_state_dict(torch.load(model_path, map_location=device))
+    # CPU or GPU mapping
+    model.to(device)
+    # Declare an inference mode.
+    model.eval()
 
-    return model
-
                                  
-
                                  
+    return model
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0176.html b/docs/modelarts/umn/modelarts_23_0176.html
index ed2bf77a8..b695babd1 100644
--- a/docs/modelarts/umn/modelarts_23_0176.html
+++ b/docs/modelarts/umn/modelarts_23_0176.html
@@ -2,757 +2,382 @@
 
 
                                  Caffe
                                  
 
                                  Training and Saving a Model
                                  lenet_train_test.prototxt file
                                  
-
                                    1
-  2
-  3
-  4
-  5
-  6
-  7
-  8
-  9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
-100
-101
-102
-103
-104
-105
-106
-107
-108
-109
-110
-111
-112
-113
-114
-115
-116
-117
-118
-119
-120
-121
-122
-123
-124
-125
-126
-127
-128
-129
-130
-131
-132
-133
-134
-135
-136
-137
-138
-139
-140
-141
-142
-143
-144
-145
-146
-147
-148
-149
-150
-151
-152
-153
-154
-155
-156
-157
-158
-159
-160
-161
-162
-163
-164
-165
-166
-167
-168
                                  name: "LeNet"
-layer {
-  name: "mnist"
-  type: "Data"
-  top: "data"
-  top: "label"
-  include {
-    phase: TRAIN
-  }
-  transform_param {
-    scale: 0.00390625
-  }
-  data_param {
-    source: "examples/mnist/mnist_train_lmdb"
-    batch_size: 64
-    backend: LMDB
-  }
-}
-layer {
-  name: "mnist"
-  type: "Data"
-  top: "data"
-  top: "label"
-  include {
-    phase: TEST
-  }
-  transform_param {
-    scale: 0.00390625
-  }
-  data_param {
-    source: "examples/mnist/mnist_test_lmdb"
-    batch_size: 100
-    backend: LMDB
-  }
-}
-layer {
-  name: "conv1"
-  type: "Convolution"
-  bottom: "data"
-  top: "conv1"
-  param {
-    lr_mult: 1
-  }
-  param {
-    lr_mult: 2
-  }
-  convolution_param {
-    num_output: 20
-    kernel_size: 5
-    stride: 1
-    weight_filler {
-      type: "xavier"
-    }
-    bias_filler {
-      type: "constant"
-    }
-  }
-}
-layer {
-  name: "pool1"
-  type: "Pooling"
-  bottom: "conv1"
-  top: "pool1"
-  pooling_param {
-    pool: MAX
-    kernel_size: 2
-    stride: 2
-  }
-}
-layer {
-  name: "conv2"
-  type: "Convolution"
-  bottom: "pool1"
-  top: "conv2"
-  param {
-    lr_mult: 1
-  }
-  param {
-    lr_mult: 2
-  }
-  convolution_param {
-    num_output: 50
-    kernel_size: 5
-    stride: 1
-    weight_filler {
-      type: "xavier"
-    }
-    bias_filler {
-      type: "constant"
-    }
-  }
-}
-layer {
-  name: "pool2"
-  type: "Pooling"
-  bottom: "conv2"
-  top: "pool2"
-  pooling_param {
-    pool: MAX
-    kernel_size: 2
-    stride: 2
-  }
-}
-layer {
-  name: "ip1"
-  type: "InnerProduct"
-  bottom: "pool2"
-  top: "ip1"
-  param {
-    lr_mult: 1
-  }
-  param {
-    lr_mult: 2
-  }
-  inner_product_param {
-    num_output: 500
-    weight_filler {
-      type: "xavier"
-    }
-    bias_filler {
-      type: "constant"
-    }
-  }
-}
-layer {
-  name: "relu1"
-  type: "ReLU"
-  bottom: "ip1"
-  top: "ip1"
-}
-layer {
-  name: "ip2"
-  type: "InnerProduct"
-  bottom: "ip1"
-  top: "ip2"
-  param {
-    lr_mult: 1
-  }
-  param {
-    lr_mult: 2
-  }
-  inner_product_param {
-    num_output: 10
-    weight_filler {
-      type: "xavier"
-    }
-    bias_filler {
-      type: "constant"
-    }
-  }
-}
-layer {
-  name: "accuracy"
-  type: "Accuracy"
-  bottom: "ip2"
-  bottom: "label"
-  top: "accuracy"
-  include {
-    phase: TEST
-  }
-}
-layer {
-  name: "loss"
-  type: "SoftmaxWithLoss"
-  bottom: "ip2"
-  bottom: "label"
-  top: "loss"
-}
-
                                  
-
                                  
+
                                  name: "LeNet"
+layer {
+  name: "mnist"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    scale: 0.00390625
+  }
+  data_param {
+    source: "examples/mnist/mnist_train_lmdb"
+    batch_size: 64
+    backend: LMDB
+  }
+}
+layer {
+  name: "mnist"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    scale: 0.00390625
+  }
+  data_param {
+    source: "examples/mnist/mnist_test_lmdb"
+    batch_size: 100
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 20
+    kernel_size: 5
+    stride: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "conv1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 2
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 50
+    kernel_size: 5
+    stride: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "conv2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 2
+    stride: 2
+  }
+}
+layer {
+  name: "ip1"
+  type: "InnerProduct"
+  bottom: "pool2"
+  top: "ip1"
+  param {
+    lr_mult: 1
+  }
+  param {
+    lr_mult: 2
+  }
+  inner_product_param {
+    num_output: 500
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "ip1"
+  top: "ip1"
+}
+layer {
+  name: "ip2"
+  type: "InnerProduct"
+  bottom: "ip1"
+  top: "ip2"
+  param {
+    lr_mult: 1
+  }
+  param {
+    lr_mult: 2
+  }
+  inner_product_param {
+    num_output: 10
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "ip2"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "ip2"
+  bottom: "label"
+  top: "loss"
+}
                                  
 
                                  lenet_solver.prototxt file
                                  
-
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
                                  # The train/test net protocol buffer definition
-net: "examples/mnist/lenet_train_test.prototxt"
-# test_iter specifies how many forward passes the test should carry out.
-# In the case of MNIST, we have test batch size 100 and 100 test iterations,
-# covering the full 10,000 testing images.
-test_iter: 100
-# Carry out testing every 500 training iterations.
-test_interval: 500
-# The base learning rate, momentum and the weight decay of the network.
-base_lr: 0.01
-momentum: 0.9
-weight_decay: 0.0005
-# The learning rate policy
-lr_policy: "inv"
-gamma: 0.0001
-power: 0.75
-# Display every 100 iterations
-display: 100
-# The maximum number of iterations
-max_iter: 1000
-# snapshot intermediate results
-snapshot: 5000
-snapshot_prefix: "examples/mnist/lenet"
-# solver mode: CPU or GPU
-solver_mode: CPU
-
                                  
-
                                  
+
                                  # The train/test net protocol buffer definition
+net: "examples/mnist/lenet_train_test.prototxt"
+# test_iter specifies how many forward passes the test should carry out.
+# In the case of MNIST, we have test batch size 100 and 100 test iterations,
+# covering the full 10,000 testing images.
+test_iter: 100
+# Carry out testing every 500 training iterations.
+test_interval: 500
+# The base learning rate, momentum and the weight decay of the network.
+base_lr: 0.01
+momentum: 0.9
+weight_decay: 0.0005
+# The learning rate policy
+lr_policy: "inv"
+gamma: 0.0001
+power: 0.75
+# Display every 100 iterations
+display: 100
+# The maximum number of iterations
+max_iter: 1000
+# snapshot intermediate results
+snapshot: 5000
+snapshot_prefix: "examples/mnist/lenet"
+# solver mode: CPU or GPU
+solver_mode: CPU
                                  
 
                                  Train the model.
                                  
-
                                  ./build/tools/caffe train --solver=examples/mnist/lenet_solver.prototxt
                                  
+
                                  ./build/tools/caffe train --solver=examples/mnist/lenet_solver.prototxt
                                  
 
                                  The caffemodel file is generated after model training. Rewrite the lenet_train_test.prototxt file to the lenet_deploy.prototxt file used for deployment by modifying input and output layers.
                                  
-
                                    1
-  2
-  3
-  4
-  5
-  6
-  7
-  8
-  9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
-100
-101
-102
-103
-104
-105
-106
-107
-108
-109
-110
-111
-112
-113
-114
-115
-116
-117
-118
-119
-120
-121
-122
-123
-124
-125
-126
-127
-128
-129
                                  name: "LeNet"
-layer {
-  name: "data"
-  type: "Input"
-  top: "data"
-  input_param { shape: { dim: 1 dim: 1  dim: 28 dim: 28 } } 
-}
-layer {
-  name: "conv1"
-  type: "Convolution"
-  bottom: "data"
-  top: "conv1"
-  param {
-    lr_mult: 1
-  }
-  param {
-    lr_mult: 2
-  }
-  convolution_param {
-    num_output: 20
-    kernel_size: 5
-    stride: 1
-    weight_filler {
-      type: "xavier"
-    }
-    bias_filler {
-      type: "constant"
-    }
-  }
-}
-layer {
-  name: "pool1"
-  type: "Pooling"
-  bottom: "conv1"
-  top: "pool1"
-  pooling_param {
-    pool: MAX
-    kernel_size: 2
-    stride: 2
-  }
-}
-layer {
-  name: "conv2"
-  type: "Convolution"
-  bottom: "pool1"
-  top: "conv2"
-  param {
-    lr_mult: 1
-  }
-  param {
-    lr_mult: 2
-  }
-  convolution_param {
-    num_output: 50
-    kernel_size: 5
-    stride: 1
-    weight_filler {
-      type: "xavier"
-    }
-    bias_filler {
-      type: "constant"
-    }
-  }
-}
-layer {
-  name: "pool2"
-  type: "Pooling"
-  bottom: "conv2"
-  top: "pool2"
-  pooling_param {
-    pool: MAX
-    kernel_size: 2
-    stride: 2
-  }
-}
-layer {
-  name: "ip1"
-  type: "InnerProduct"
-  bottom: "pool2"
-  top: "ip1"
-  param {
-    lr_mult: 1
-  }
-  param {
-    lr_mult: 2
-  }
-  inner_product_param {
-    num_output: 500
-    weight_filler {
-      type: "xavier"
-    }
-    bias_filler {
-      type: "constant"
-    }
-  }
-}
-layer {
-  name: "relu1"
-  type: "ReLU"
-  bottom: "ip1"
-  top: "ip1"
-}
-layer {
-  name: "ip2"
-  type: "InnerProduct"
-  bottom: "ip1"
-  top: "ip2"
-  param {
-    lr_mult: 1
-  }
-  param {
-    lr_mult: 2
-  }
-  inner_product_param {
-    num_output: 10
-    weight_filler {
-      type: "xavier"
-    }
-    bias_filler {
-      type: "constant"
-    }
-  }
-}
-layer {
-  name: "prob"
-  type: "Softmax"
-  bottom: "ip2"
-  top: "prob"
-}
-
                                  
-
                                  
+
                                  name: "LeNet"
+layer {
+  name: "data"
+  type: "Input"
+  top: "data"
+  input_param { shape: { dim: 1 dim: 1  dim: 28 dim: 28 } } 
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 20
+    kernel_size: 5
+    stride: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "conv1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 2
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 50
+    kernel_size: 5
+    stride: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "conv2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 2
+    stride: 2
+  }
+}
+layer {
+  name: "ip1"
+  type: "InnerProduct"
+  bottom: "pool2"
+  top: "ip1"
+  param {
+    lr_mult: 1
+  }
+  param {
+    lr_mult: 2
+  }
+  inner_product_param {
+    num_output: 500
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "ip1"
+  top: "ip1"
+}
+layer {
+  name: "ip2"
+  type: "InnerProduct"
+  bottom: "ip1"
+  top: "ip2"
+  param {
+    lr_mult: 1
+  }
+  param {
+    lr_mult: 2
+  }
+  inner_product_param {
+    num_output: 10
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer {
+  name: "prob"
+  type: "Softmax"
+  bottom: "ip2"
+  top: "prob"
+}
                                  
 
                                  
-
                                  Inference Code
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
-46
-47
-48
-49
                                  from model_service.caffe_model_service import CaffeBaseService
+
                                  Inference Code
                                  from model_service.caffe_model_service import CaffeBaseService
 
-import numpy as np
+import numpy as np
 
-import os, json
+import os, json
 
-import caffe
+import caffe
 
-from PIL import Image
+from PIL import Image
 
 
-class LenetService(CaffeBaseService):
+class LenetService(CaffeBaseService):
 
-    def __init__(self, model_name, model_path):
-        # Call the inference method of the parent class.
-        super(LenetService, self).__init__(model_name, model_path)
+    def __init__(self, model_name, model_path):
+        # Call the inference method of the parent class.
+        super(LenetService, self).__init__(model_name, model_path)
 
-        # Configure preprocessing information.
-        transformer = caffe.io.Transformer({'data': self.net.blobs['data'].data.shape})
-        # Transform to NCHW.
-        transformer.set_transpose('data', (2, 0, 1)) 
-        # Perform normalization.
-        transformer.set_raw_scale('data', 255.0)
+        # Configure preprocessing information.
+        transformer = caffe.io.Transformer({'data': self.net.blobs['data'].data.shape})
+        # Transform to NCHW.
+        transformer.set_transpose('data', (2, 0, 1)) 
+        # Perform normalization.
+        transformer.set_raw_scale('data', 255.0)
 
-        # If the batch size is set to 1, inference is supported for only one image.
-        self.net.blobs['data'].reshape(1, 1, 28, 28)
-        self.transformer = transformer
+        # If the batch size is set to 1, inference is supported for only one image.
+        self.net.blobs['data'].reshape(1, 1, 28, 28)
+        self.transformer = transformer
 
-       # Define the class labels.
-        self.label = [0,1,2,3,4,5,6,7,8,9]
+       # Define the class labels.
+        self.label = [0,1,2,3,4,5,6,7,8,9]
 
 
-    def _preprocess(self, data):
+    def _preprocess(self, data):
 
-        for k, v in data.items():
-            for file_name, file_content in v.items():
-                im = caffe.io.load_image(file_content, color=False)
-               # Pre-process the images.
-                self.net.blobs['data'].data[...] = self.transformer.preprocess('data', im)
+        for k, v in data.items():
+            for file_name, file_content in v.items():
+                im = caffe.io.load_image(file_content, color=False)
+               # Pre-process the images.
+                self.net.blobs['data'].data[...] = self.transformer.preprocess('data', im)
 
-                return
+                return
 
-    def _postprocess(self, data):
+    def _postprocess(self, data):
 
-        data = data['prob'][0, :]
-        predicted = np.argmax(data)
-        predicted = {"predicted" : str(predicted) }
+        data = data['prob'][0, :]
+        predicted = np.argmax(data)
+        predicted = {"predicted" : str(predicted) }
 
-        return predicted
-
                                  
-
                                  
+        return predicted
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0177.html b/docs/modelarts/umn/modelarts_23_0177.html
index 5d50c501b..ba260f227 100644
--- a/docs/modelarts/umn/modelarts_23_0177.html
+++ b/docs/modelarts/umn/modelarts_23_0177.html
@@ -1,68 +1,39 @@
 
 
 
                                  XGBoost
                                  
-
                                  Training and Saving a Model
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
                                  import pandas as pd
-import xgboost as xgb
-from sklearn.model_selection import train_test_split
+
                                  Training and Saving a Model
                                  import pandas as pd
+import xgboost as xgb
+from sklearn.model_selection import train_test_split
 
-# Prepare training data and setting parameters
-iris = pd.read_csv('/data/iris.csv')
-X = iris.drop(['virginica'],axis=1)
-y = iris[['virginica']]
-X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=1234565)
-params = {
-    'booster': 'gbtree',
-    'objective': 'multi:softmax',
-    'num_class': 3,
-    'gamma': 0.1,
-    'max_depth': 6,
-    'lambda': 2,
-    'subsample': 0.7,
-    'colsample_bytree': 0.7,
-    'min_child_weight': 3,
-    'silent': 1,
-    'eta': 0.1,
-    'seed': 1000,
-    'nthread': 4,
-}
-plst = params.items()
-dtrain = xgb.DMatrix(X_train, y_train)
-num_rounds = 500
-model = xgb.train(plst, dtrain, num_rounds)
-model.save_model('/tmp/xgboost.m')
-
                                  
-
                                  
-
                                  After the model is saved, it must be uploaded to the OBS directory before being published. The config.json and customize_service.py files must be contained during publishing. For details about the definition method, see Model Package Specifications.
                                  
+# Prepare training data and setting parameters
+iris = pd.read_csv('/home/ma-user/work/iris.csv')
+X = iris.drop(['variety'],axis=1)
+y = iris[['variety']]
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=1234565)
+params = {
+    'booster': 'gbtree',
+    'objective': 'multi:softmax',
+    'num_class': 3,
+    'gamma': 0.1,
+    'max_depth': 6,
+    'lambda': 2,
+    'subsample': 0.7,
+    'colsample_bytree': 0.7,
+    'min_child_weight': 3,
+    'silent': 1,
+    'eta': 0.1,
+    'seed': 1000,
+    'nthread': 4,
+}
+plst = params.items()
+dtrain = xgb.DMatrix(X_train, y_train)
+num_rounds = 500
+model = xgb.train(plst, dtrain, num_rounds)
+model.save_model('/tmp/xgboost.m')
+
                                  Before training, download the iris.csv dataset, decompress it, and upload it to the /home/ma-user/work/ directory of the notebook instance. Download the iris.csv dataset from https://gist.github.com/netj/8836201. 
                                  
+
                                  After the model is saved, it must be uploaded to the OBS directory before being published. The config.json and customize_service.py files must be contained during publishing. For details about the definition method, see Model Package Specifications.
                                  
 
                                  
-
                                  Inference Code
                                  # coding:utf-8
+
                                  Inference Code
                                  # coding:utf-8
 import collections
 import json
 import xgboost as xgb
diff --git a/docs/modelarts/umn/modelarts_23_0178.html b/docs/modelarts/umn/modelarts_23_0178.html
index 7dc8c0d80..a75209775 100644
--- a/docs/modelarts/umn/modelarts_23_0178.html
+++ b/docs/modelarts/umn/modelarts_23_0178.html
@@ -1,146 +1,76 @@
 
 
 
                                  PySpark
                                  
-
                                  Training and Saving a Model
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
                                  from pyspark.ml import Pipeline, PipelineModel
-from pyspark.ml.linalg import Vectors
-from pyspark.ml.classification import LogisticRegression
+
                                  Training and Saving a Model
                                  from pyspark.ml import Pipeline, PipelineModel
+from pyspark.ml.linalg import Vectors
+from pyspark.ml.classification import LogisticRegression
 
-# Prepare training data using tuples.
-# Prepare training data from a list of (label, features) tuples.
-training = spark.createDataFrame([
-    (1.0, Vectors.dense([0.0, 1.1, 0.1])),
-    (0.0, Vectors.dense([2.0, 1.0, -1.0])),
-    (0.0, Vectors.dense([2.0, 1.3, 1.0])),
-    (1.0, Vectors.dense([0.0, 1.2, -0.5]))], ["label", "features"])
+# Prepare training data using tuples.
+# Prepare training data from a list of (label, features) tuples.
+training = spark.createDataFrame([
+    (1.0, Vectors.dense([0.0, 1.1, 0.1])),
+    (0.0, Vectors.dense([2.0, 1.0, -1.0])),
+    (0.0, Vectors.dense([2.0, 1.3, 1.0])),
+    (1.0, Vectors.dense([0.0, 1.2, -0.5]))], ["label", "features"])
 
-# Create a training instance. The logistic regression algorithm is used for training.
-# Create a LogisticRegression instance. This instance is an Estimator.
-lr = LogisticRegression(maxIter=10, regParam=0.01)
+# Create a training instance. The logistic regression algorithm is used for training.
+# Create a LogisticRegression instance. This instance is an Estimator.
+lr = LogisticRegression(maxIter=10, regParam=0.01)
 
-# Train the logistic regression model.
-# Learn a LogisticRegression model. This uses the parameters stored in lr.
-model = lr.fit(training)
+# Train the logistic regression model.
+# Learn a LogisticRegression model. This uses the parameters stored in lr.
+model = lr.fit(training)
 
-# Save the model to a local directory.
-# Save model to local path.
-model.save("/tmp/spark_model")
-
                                  
-
                                  
+# Save the model to a local directory.
+# Save model to local path.
+model.save("/tmp/spark_model")
                                  
 
                                  After the model is saved, it must be uploaded to the OBS directory before being published. The config.json configuration and customize_service.py must be contained during publishing. For details about the definition method, see Model Package Specifications.
                                  
 
                                  
-
                                  Inference Code
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
-33
-34
-35
-36
-37
-38
-39
-40
-41
-42
-43
-44
-45
                                  # coding:utf-8
-import collections
-import json
-import traceback
+
                                  Inference Code
                                  # coding:utf-8
+import collections
+import json
+import traceback
 
-import model_service.log as log
-from model_service.spark_model_service import SparkServingBaseService
-from pyspark.ml.classification import LogisticRegression
+import model_service.log as log
+from model_service.spark_model_service import SparkServingBaseService
+from pyspark.ml.classification import LogisticRegression
 
-logger = log.getLogger(__name__)
+logger = log.getLogger(__name__)
 
 
-class user_Service(SparkServingBaseService):
-    # Pre-process data.
-    def _preprocess(self, data):
-        logger.info("Begin to handle data from user data...")
-        # Read data.
-        req_json = json.loads(data, object_pairs_hook=collections.OrderedDict)
-        try:
-            # Convert data to the spark dataframe format.
-            predict_spdf = self.spark.createDataFrame(pd.DataFrame(req_json["data"]["req_data"]))
-        except Exception as e:
-            logger.error("check your request data does meet the requirements ?")
-            logger.error(traceback.format_exc())
-            raise Exception("check your request data does meet the requirements ?")
-        return predict_spdf
+class user_Service(SparkServingBaseService):
+    # Pre-process data.
+    def _preprocess(self, data):
+        logger.info("Begin to handle data from user data...")
+        # Read data.
+        req_json = json.loads(data, object_pairs_hook=collections.OrderedDict)
+        try:
+            # Convert data to the spark dataframe format.
+            predict_spdf = self.spark.createDataFrame(pd.DataFrame(req_json["data"]["req_data"]))
+        except Exception as e:
+            logger.error("check your request data does meet the requirements ?")
+            logger.error(traceback.format_exc())
+            raise Exception("check your request data does meet the requirements ?")
+        return predict_spdf
 
-    # Perform model inference.
-    def _inference(self, data):
-        try:
-             # Load a model file.
-            predict_model = LogisticRegression.load(self.model_path)
-            # Perform data inference.
-            prediction_result = predict_model.transform(data)
-	except Exception as e:
-            logger.error(traceback.format_exc())
-            raise Exception("Unable to load model and do dataframe transformation.")
-        return prediction_result
+    # Perform model inference.
+    def _inference(self, data):
+        try:
+             # Load a model file.
+            predict_model = LogisticRegression.load(self.model_path)
+            # Perform data inference.
+            prediction_result = predict_model.transform(data)
+	except Exception as e:
+            logger.error(traceback.format_exc())
+            raise Exception("Unable to load model and do dataframe transformation.")
+        return prediction_result
 
-    # Post-process data.
-    def _postprocess(self, pre_data):
-        logger.info("Get new data to respond...")
-        predict_str = pre_data.toPandas().to_json(orient='records')
-        predict_result = json.loads(predict_str)
-        return predict_result
-
                                  
-
                                  
+    # Post-process data.
+    def _postprocess(self, pre_data):
+        logger.info("Get new data to respond...")
+        predict_str = pre_data.toPandas().to_json(orient='records')
+        predict_result = json.loads(predict_str)
+        return predict_result
                                  
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0179.html b/docs/modelarts/umn/modelarts_23_0179.html
index 42fd5caaf..620fcfb6e 100644
--- a/docs/modelarts/umn/modelarts_23_0179.html
+++ b/docs/modelarts/umn/modelarts_23_0179.html
@@ -1,102 +1,55 @@
 
 
 
                                  Scikit Learn
                                  
-
                                  Training and Saving a Model
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
                                  import json
-import pandas as pd
-from sklearn.datasets import load_iris
-from sklearn.model_selection import train_test_split
-from sklearn.linear_model import LogisticRegression
-from sklearn.externals import joblib
-iris = pd.read_csv('/data/iris.csv')
-X = iris.drop(['virginica'],axis=1)
-y = iris[['virginica']]
-# Create a LogisticRegression instance and train model
-logisticRegression = LogisticRegression(C=1000.0, random_state=0)
-logisticRegression.fit(X,y)
-# Save model to local path
-joblib.dump(logisticRegression, '/tmp/sklearn.m')
-
                                  
-
                                  
-
                                  After the model is saved, it must be uploaded to the OBS directory before being published. The config.json and customize_service.py files must be contained during publishing. For details about the definition method, see Model Package Specifications.
                                  
+
                                  Training and Saving a Model
                                  import json
+import pandas as pd
+from sklearn.datasets import load_iris
+from sklearn.model_selection import train_test_split
+from sklearn.linear_model import LogisticRegression
+from sklearn.externals import joblib
+iris = pd.read_csv('/home/ma-user/work/iris.csv')
+X = iris.drop(['variety'],axis=1)
+y = iris[['variety']]
+# Create a LogisticRegression instance and train model
+logisticRegression = LogisticRegression(C=1000.0, random_state=0)
+logisticRegression.fit(X,y)
+# Save model to local path
+joblib.dump(logisticRegression, '/tmp/sklearn.m')
                                  
+
                                  Before training, download the iris.csv dataset, decompress it, and upload it to the /home/ma-user/work/ directory of the notebook instance. Download the iris.csv dataset from https://gist.github.com/netj/8836201. 
                                  
+
                                  After the model is saved, it must be uploaded to the OBS directory before being published. The config.json and customize_service.py files must be contained during publishing. For details about the definition method, see Model Package Specifications.
                                  
 
                                  
-
                                  Inference Code
                                   1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
-31
-32
                                  # coding:utf-8
-import collections
-import json
-from sklearn.externals import joblib
-from model_service.python_model_service import XgSklServingBaseService
+
                                  Inference Code
                                  # coding:utf-8
+import collections
+import json
+from sklearn.externals import joblib
+from model_service.python_model_service import XgSklServingBaseService
 
-class user_Service(XgSklServingBaseService):
+class user_Service(XgSklServingBaseService):
 
-    # request data preprocess
-    def _preprocess(self, data):
-        list_data = []
-        json_data = json.loads(data, object_pairs_hook=collections.OrderedDict)
-        for element in json_data["data"]["req_data"]:
-            array = []
-            for each in element:
-                array.append(element[each])
-                list_data.append(array)
-        return list_data
+    # request data preprocess
+    def _preprocess(self, data):
+        list_data = []
+        json_data = json.loads(data, object_pairs_hook=collections.OrderedDict)
+        for element in json_data["data"]["req_data"]:
+            array = []
+            for each in element:
+                array.append(element[each])
+                list_data.append(array)
+        return list_data
 
-    # predict
-    def _inference(self, data):
-        sk_model = joblib.load(self.model_path)
-        pre_result = sk_model.predict(data)
-        pre_result = pre_result.tolist()
-        return pre_result
+    # predict
+    def _inference(self, data):
+        sk_model = joblib.load(self.model_path)
+        pre_result = sk_model.predict(data)
+        pre_result = pre_result.tolist()
+        return pre_result
 
-    # predict result process
-    def _postprocess(self,data):
-        resp_data = []
-        for element in data:
-            resp_data.append({"predictresult": element})
-        return resp_data
-
                                  
-
                                  
+    # predict result process
+    def _postprocess(self,data):
+        resp_data = []
+        for element in data:
+            resp_data.append({"predictresult": element})
+        return resp_data
 
                                  
 
                                  
 
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0181.html b/docs/modelarts/umn/modelarts_23_0181.html
index abc7eafc3..808230834 100644
--- a/docs/modelarts/umn/modelarts_23_0181.html
+++ b/docs/modelarts/umn/modelarts_23_0181.html
@@ -4,7 +4,7 @@
 
                                  Generally, a small data labeling task can be completed by an individual. However, team work is required to label a large dataset. ModelArts provides the team labeling function. A labeling team can be formed to manage labeling for the same dataset.
                                  
 
                                   
                                  The team labeling function supports only datasets for image classification, object detection, text classification, named entity recognition, text triplet, and speech paragraph labeling.
                                  
 
                                  
-
                                  How to Enable Team Labeling
                                  • When creating a dataset, enable Team Labeling and select a team or task manager.
                                    Figure 1 Enabling during dataset creation
                                    
+
                                    How to Enable Team Labeling
                                    • When creating a dataset, enable Team Labeling and select a team or task manager.
                                      Figure 1 Enabling during dataset creation
                                      
 
                                    • If team labeling is not enabled for a dataset that has been created, create a team labeling task to enable team labeling. For details about how to create a team labeling task, see Creating Team Labeling Tasks.
                                      Figure 2 Creating a team labeling task in a dataset list
                                      
 
                                      Figure 3 Creating a team labeling task
                                      
 
                                      Figure 4 Creating a team labeling task on the dataset details page
                                      
diff --git a/docs/modelarts/umn/modelarts_23_0182.html b/docs/modelarts/umn/modelarts_23_0182.html
index 14773ffc4..dd059163c 100644
--- a/docs/modelarts/umn/modelarts_23_0182.html
+++ b/docs/modelarts/umn/modelarts_23_0182.html
@@ -4,11 +4,13 @@
 
                                      Team labeling is managed in a unit of teams. To enable team labeling for a dataset, a team must be specified. Multiple members can be added to a team.
                                      
 
                                      Background
                                      • An account can have a maximum of 10 teams.
                                      • An account must have at least one team to enable team labeling for datasets. If the account has no team, add a team by referring to Adding a Team.
                                      
 
                                      
-
                                      Adding a Team
                                      1. In the left navigation pane of the ModelArts management console, choose Data Management > Labeling Teams. The Labeling Teams page is displayed.
                                      2. On the Labeling Teams page, click Add Team.
                                      3. In the displayed Add Team dialog box, enter a team name and description and click OK. The labeling team is added.
                                        The new team is displayed on the Labeling Teams page. You can view team details in the right pane. There is no member in the new team. Add members to the new team by referring to Adding a Member.
                                        
+
                                        Adding a Team
                                        1. In the left navigation pane of the ModelArts management console, choose Data Management > Labeling Teams. The Labeling Teams page is displayed.
                                        2. On the Labeling Teams page, click Add Team.
                                        3. In the displayed Add Team dialog box, enter a team name and description and click OK. The labeling team is added.
                                          Figure 1 Adding a team
                                          
+
                                          The new team is displayed on the Labeling Teams page. You can view team details in the right pane. There is no member in the new team. Add members to the new team by referring to Adding a Member.
                                          
 
                                        
 
                                        
 
                                        Deleting a Team
                                        You can delete a team that is no longer used.
                                        
 
                                        On the Labeling Teams page, select the target team and click Delete. In the dialog box that is displayed, click OK.
                                        
+
                                        Figure 2 Deleting a team
                                        
 
                                        
 
                                      
 
                                      
diff --git a/docs/modelarts/umn/modelarts_23_0183.html b/docs/modelarts/umn/modelarts_23_0183.html
index 32c3fcc25..3338cedce 100644
--- a/docs/modelarts/umn/modelarts_23_0183.html
+++ b/docs/modelarts/umn/modelarts_23_0183.html
@@ -17,6 +17,7 @@
 
                                      
 
                                      Deleting Members
                                      • Deleting a single member
                                        In the Team Details area, select the desired member, and click Delete in the Operation column. In the dialog box that is displayed, click OK.
                                        
 
                                      • Batch Deletion
                                        In the Team Details area, select members to be deleted and click Delete. In the dialog box that is displayed, click OK.
                                        
+
                                        Figure 3 Batch deletion
                                        
 
                                      
 
                                      
 
                                      
diff --git a/docs/modelarts/umn/modelarts_23_0187.html b/docs/modelarts/umn/modelarts_23_0187.html
index 00cfa25c4..0d37311b2 100644
--- a/docs/modelarts/umn/modelarts_23_0187.html
+++ b/docs/modelarts/umn/modelarts_23_0187.html
@@ -71,23 +71,6 @@
 
                                  1 minute
                                  
                                   		
 
                                  gpu_mem_usage
                                  
-
                                  GPU Memory Usage
                                  
-
                                  GPU memory usage of ModelArts
                                  
-
                                  Unit: %
                                  
-
                                  ≥ 0%
                                  
-
                                  Measurement object:
                                  
-
                                  ModelArts models
                                  
-
                                  Dimension:
                                  
-
                                  model_id
                                  
-
                                  1 minute
                                  
-
                                  
 
                                  successfully_called_times
                                  
                                   		
 
                                  Number of Successful Calls
                                  
diff --git a/docs/modelarts/umn/modelarts_23_0207.html b/docs/modelarts/umn/modelarts_23_0207.html
index 14f123bf9..d38666bb8 100644
--- a/docs/modelarts/umn/modelarts_23_0207.html
+++ b/docs/modelarts/umn/modelarts_23_0207.html
@@ -2,10 +2,87 @@
 
 
                                  Importing a Meta Model from OBS
                                  
 
                                  In scenarios where frequently-used frameworks are used for model development and training, you can import the model to ModelArts for unified management.
                                  
-
                                  Prerequisites
                                  
+
                                  Prerequisites
                                  • The model has been developed and trained, and the type and version of the AI engine it uses is supported by ModelArts. Common engines supported by ModelArts and their runtime ranges are described as follows:
+
                                    
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
                                    Table 1 Supported AI engines and their runtime
                                    Engine
                                    
+
                                    Runtime
                                    
+
                                    Precautions
                                    
+
                                    TensorFlow
                                    
+
                                    python3.6
                                    
+
                                    python2.7
                                    
+
                                    tf1.13-python2.7-gpu
                                    
+
                                    tf1.13-python2.7-cpu
                                    
+
                                    tf1.13-python3.6-gpu
                                    
+
                                    tf1.13-python3.6-cpu
                                    
+
                                    tf1.13-python3.7-cpu
                                    
+
                                    tf1.13-python3.7-gpu
                                    
+
                                    tf2.1-python3.7
                                    
+
                                    • TensorFlow 1.8.0 is used in python2.7 and python3.6.
                                    • python3.6, python2.7, and tf2.1-python3.7 indicate that the model can run on both CPUs and GPUs. For other runtime values, if the suffix contains cpu or gpu, the model can run only on CPUs or GPUs.
                                    • The default runtime is python2.7.
                                    
+
                                    MXNet
                                    
+
                                    python3.7
                                    
+
                                    python3.6
                                    
+
                                    • MXNet 1.2.1 is used in python3.6 and python3.7.
                                    • python3.6 and python3.7 indicate that the model can run on both CPUs and GPUs.
                                    • The default runtime is python3.6.
                                    
+
                                    Caffe
                                    
+
                                    python3.6
                                    
+
                                    python3.7
                                    
+
                                    python3.6-gpu
                                    
+
                                    python3.7-gpu
                                    
+
                                    python3.6-cpu
                                    
+
                                    python3.7-cpu
                                    
+
                                    • Caffe 1.0.0 is used in python3.6, python3.7, python3.6-gpu, python3.7-gpu, python3.6-cpu, and python3.7-cpu.
                                    • python 3.6 and python3.7 can only be used to run models on CPUs. For other runtime values, if the suffix contains cpu or gpu, the model can run only on CPUs or GPUs. Use the runtime of python3.6-gpu, python3.7-gpu, python3.6-cpu, or python3.7-cpu.
                                    • The default runtime is python3.6.
                                    
+
                                    Spark_MLlib
                                    
+
                                    python3.6
                                    
+
                                    • Spark_MLlib 2.3.2 is used in python3.6.
                                    • python 3.6 can only be used to run models on CPUs.
                                    
+
                                    Scikit_Learn
                                    
+
                                    python3.6
                                    
+
                                    • Scikit_Learn 0.18.1 is used in python3.6.
                                    • python 3.6 can only be used to run models on CPUs.
                                    
+
                                    XGBoost
                                    
+
                                    python3.6
                                    
+
                                    • XGBoost 0.80 is used in python3.6.
                                    • python 3.6 can only be used to run models on CPUs.
                                    
+
                                    PyTorch
                                    
+
                                    python3.6
                                    
+
                                    python3.7
                                    
+
                                    pytorch1.4-python3.7
                                    
+
                                    • PyTorch 1.0 is used in python3.6 and python3.7.
                                    • python3.6, python3.7, and pytorch1.4-python3.7 indicate that the model can run on both CPUs and GPUs.
                                    • The default runtime is python3.6.
                                    
+
                                    
 
                                    
-
                                    Procedure
                                    1. Log in to the ModelArts management console, and choose Model Management > Models in the left navigation pane. The Models page is displayed.
                                    2. Click Import in the upper left corner. The Import page is displayed.
                                    3. On the Import page, set related parameters.
                                      1. Set basic information about the model. For details about the parameters, see Table 1.
-
                                        Table 1 Parameters of basic model information
                                        Parameter
                                        
+
                                      2. The imported model, inference code, and configuration file must comply with the requirements of ModelArts. For details, see Model Package Specifications, Specifications for Compiling the Model Configuration File, and Specifications for Compiling Model Inference Code.
                                      3. The model package that has completed training, inference code, and configuration file have been uploaded to the OBS directory.
                                      4. The OBS directory you use and ModelArts are in the same region.
                                        5. 
+
+
                                        Procedure
                                        1. Log in to the ModelArts management console, and choose Model Management > Models in the left navigation pane. The Models page is displayed.
                                        2. Click Import in the upper left corner. The Import page is displayed.
                                        3. On the Import page, set related parameters.
                                          1. Set basic information about the model. For details about the parameters, see Table 2.
+
                                            
@@ -34,10 +111,10 @@
 
                                            Table 2 Parameters of basic model information
                                            Parameter
                                            
                                             		
 
                                            Description
                                            
                                             		
 
                                            
 
                                            
-
                                          2. Select the meta model source and set related parameters. Meta Model Source has four options based on the scenario.  Set Meta Model Source to OBS. For details about the parameters, see Table 2.
                                            For the meta model imported from OBS, you need to compile the inference code and configuration file by referring to Model Package Specifications and place the inference code and configuration files in the model folder storing the meta model. If the selected directory does not contain the corresponding inference code and configuration files, the model cannot be imported.
                                            
+
                                          3. Select the meta model source and set related parameters. Meta Model Source has four options based on the scenario.  Set Meta Model Source to OBS. For details about the parameters, see Table 3.
                                            For the meta model imported from OBS, you need to compile the inference code and configuration file by referring to Model Package Specifications and place the inference code and configuration files in the model folder storing the meta model. If the selected directory does not contain the corresponding inference code and configuration files, the model cannot be imported.
                                            
 
                                            Figure 1 Setting Meta Model Source to OBS
                                            
 
-
                                            Table 2 Parameters of the meta model source
                                            Parameter
                                            
+
                                            
@@ -82,7 +159,7 @@
 
-
                                            Follow-Up Procedure
                                            • Model Deployment: On the Models page, click the triangle next to a model name to view all versions of the model. Locate the row that contains the target version, click Deploy in the Operation column, and select the deployment type configured when importing the model from the drop-down list. On the Deploy page, set parameters by referring to Introduction to Model Deployment.
                                            
+
                                            Follow-Up Procedure
                                            • Model Deployment: On the Models page, click the triangle next to a model name to view all versions of the model. Locate the row that contains the target version, click Deploy in the Operation column, and select the deployment type configured when importing the model from the drop-down list. On the Deploy page, set parameters by referring to Introduction to Model Deployment .
                                            
 
                                            
 
                                            
diff --git a/docs/modelarts/umn/modelarts_23_0210.html b/docs/modelarts/umn/modelarts_23_0210.html
index 313afda17..67b15205a 100644
--- a/docs/modelarts/umn/modelarts_23_0210.html
+++ b/docs/modelarts/umn/modelarts_23_0210.html
@@ -15,7 +15,6 @@
 
 
                                            On the labeling platform, each member can view the images that are not labeled, to be corrected, rejected, to be reviewed, approved, and accepted. Pay attention to the images rejected by the administrator and the images to be corrected.
                                            
 
                                            If the Reviewer role is assigned for a team labeling task, the labeling result needs to be reviewed. After the labeling result is reviewed, it is submitted to the administrator for acceptance.
                                            
-
                                            Figure 1 Labeling platform
                                            
 
                                            Task Acceptance (Administrator)
                                            • Initiating acceptance
                                              After team members complete data labeling, the dataset creator can initiate acceptance to check labeling results. The acceptance can be initiated only when a labeling member has labeled data. Otherwise, the acceptance initiation button is unavailable.
                                              
 
                                              1. On the Labeling Progress tab page, click Initiate Acceptance to accept tasks.
                                              2. In the displayed dialog box, set Sample Policy to By percentage or By quantity. Click OK to start the acceptance.
                                                By percentage: Sampling is performed based on a percentage for acceptance.
                                                
@@ -40,7 +39,7 @@
 
                                            
-
diff --git a/docs/modelarts/umn/modelarts_23_0211.html b/docs/modelarts/umn/modelarts_23_0211.html
index 3bba11d0c..0db68cc31 100644
--- a/docs/modelarts/umn/modelarts_23_0211.html
+++ b/docs/modelarts/umn/modelarts_23_0211.html
@@ -12,7 +12,7 @@
 
                                            Starting Labeling
                                            1. Log in to the ModelArts management console. In the left navigation pane, choose Data Management > Datasets. The Datasets page is displayed.
                                            2. In the dataset list, select the dataset to be labeled based on the labeling type, and click the dataset name to go to the Dashboard tab page of the dataset.
                                              By default, the Dashboard tab page of the current dataset version is displayed. If you need to label the dataset of another version, click the Versions tab and then click Set to Current Version in the right pane. For details, see Managing Dataset Versions.
                                              
 
                                            3. On the Dashboard page of the dataset, click Label in the upper right corner. The dataset details page is displayed. By default, all data of the dataset is displayed on the dataset details page.
                                            
 
                                            
-
                                            Labeling Content
                                            The dataset details page displays the labeled and unlabeled text objects in the dataset. The Unlabeled tab page is displayed by default.
                                            
+
                                            Labeling Content
                                            The dataset details page displays the labeled and unlabeled text objects in the dataset. The Unlabeled tab page is displayed by default.
                                            
 
                                            1. On the Unlabeled tab page, the objects to be labeled are listed in the left pane. In the list, click a text object, select the corresponding text content on the right pane, and select an entity name from the displayed entity list to label the content.
                                              Figure 3 Labeling an entity
                                              
 
                                            2. After labeling multiple entities, click the source entity and target entity in sequence and select a relationship type from the displayed relationship list to label the relationship.
                                              Figure 4 Labeling a relationship
                                              
 
                                            3. After all objects are labeled, click Save Current Page at the bottom of the page.
                                            
diff --git a/docs/modelarts/umn/modelarts_23_0217.html b/docs/modelarts/umn/modelarts_23_0217.html
index 72164bca9..8e520a03c 100644
--- a/docs/modelarts/umn/modelarts_23_0217.html
+++ b/docs/modelarts/umn/modelarts_23_0217.html
@@ -8,14 +8,14 @@
 
                                            Overview of a Basic Image Package
                                            To facilitate code download, training log output, and log file upload to OBS, ModelArts provides basic image packages for creating custom images. The basic images provided by ModelArts have the following features:
                                            
 
                                            • Some necessary tools are available in the basic image. You need to create a custom image based on the basic images provided by ModelArts.
                                            • ModelArts continuously updates the basic image versions. For compatible updates, after the basic images are updated, you can still use the old images. For incompatible updates, the custom images created based on the old version cannot run on ModelArts, but the approved custom images can still be used.
                                            • If a custom image fails to be approved and the audit log contains an error message indicating that the basic image does not match, you need to use a new basic image to create an image.
                                            
 
                                            Run the following command to obtain a ModelArts image:
                                            
-
                                            docker pull <Address for obtaining a basic image>
                                            
+
                                            docker pull <Address for obtaining a basic image>
                                            
 
                                            After customizing an image, upload it to SWR. Make sure that you have created an organization and obtained the password for logging in to SWR. For details, see .
                                            
-
                                            docker push  swr.<region>.xxx.com/<Organization to which the target image belongs>/<Image name>
                                            
+
                                            docker push  swr.<region>.xxx.com/<Organization to which the target image belongs>/<Image name>
                                            
 
                                            Obtain basic images based on chip requirements:
                                            
 
 
                                            
 
                                            CPU-based Basic Images
                                            Address for obtaining a basic image
                                            
-
                                            swr.<region>.xxx.com/modelarts-job-dev-image/custom-cpu-base:1.3
                                            
+
                                            swr.<region>.xxx.com/modelarts-job-dev-image/custom-cpu-base:1.3
                                            
 
 
                                            Table 3 Parameters of the meta model source
                                            Parameter
                                            
                                             		
 
                                            Description
                                            
                                             		
 
 
 
 
                                            
 
                                            Acceptance Scope
                                            
 
                                            • All: all data that has been labeled by the current team, including Accepted, Pending Acceptance, and Rejected data. It refers to all sample files in the dataset.
                                            • All rejects: rejects all data that has been labeled by the current team. That is, all labeled data is rejected to the labeling personnel.
                                               
                                              
+
                                            • All: all data that has been labeled by the current team, including Accepted, Pending Acceptance, and Rejected data. It refers to all sample files in the dataset.
                                            • All rejects: rejects all data that has been labeled by the current team. That is, all labeled data is rejected to the labeling personnel.
                                              
 
                                            • Accepted and pending acceptance: accepts the data that passes the acceptance or is in the Pending Acceptance state in the sample files and rejects the data that fails the acceptance to the labeling personnel.
                                            • Accepted: accepts the data that has passed the acceptance in the sample files and rejects the data that is in the Pending Acceptance state or fails the acceptance to the labeling personnel.
                                            
                                             		
 
                                            
@@ -81,9 +81,9 @@
 
                                            Table 1 Optional parameters
                                            Parameter
                                            
 
                                            
 
                                            
 
-
                                            GPU-based Basic Images
                                            • Image of the CUDA 10.0, 10.1, or 10.2 version, using Ubuntu 18.04 as the basic image and with MoXing pre-installed by default
                                              swr.<region>.xxx.com/modelarts-job-dev-image/custom-base-<cuda version>-<python version>-<os>-<arch>:<image tag>
                                              
-
                                            • Image of the CUDA 8, 9, or 92 version, with MoXing pre-installed by default
                                              swr.<region>.xxx.com/modelarts-job-dev-image/custom-gpu-<cuda version>-inner-moxing-<python version>:<image tag>
                                              
-
                                            • Image of the CUDA 8, 9, or 92 version
                                              swr.<region>.xxx.com/modelarts-job-dev-image/custom-gpu-<cuda version>-base:<image tag>
                                              
+
                                              GPU-based Basic Images
                                              • Image of the CUDA 10.0, 10.1, or 10.2 version, using Ubuntu 18.04 as the basic image and with MoXing pre-installed by default
                                                swr.<region>.xxx.com/modelarts-job-dev-image/custom-base-<cuda version>-<python version>-<os>-<arch>:<image tag>
                                                
+
                                              • Image of the CUDA 8, 9, or 92 version, with MoXing pre-installed by default
                                                swr.<region>.xxx.com/modelarts-job-dev-image/custom-gpu-<cuda version>-inner-moxing-<python version>:<image tag>
                                                
+
                                              • Image of the CUDA 8, 9, or 92 version
                                                swr.<region>.xxx.com/modelarts-job-dev-image/custom-gpu-<cuda version>-base:<image tag>
                                                
 
 
                                                
diff --git a/docs/modelarts/umn/modelarts_23_0219.html b/docs/modelarts/umn/modelarts_23_0219.html
index 1ce23f918..333c6a609 100644
--- a/docs/modelarts/umn/modelarts_23_0219.html
+++ b/docs/modelarts/umn/modelarts_23_0219.html
@@ -3,11 +3,11 @@
 
                                                Specifications for Custom Images Used for Importing Models
                                                When creating an image using locally developed models, ensure that they meet the specifications defined by ModelArts.
                                                
 
                                                Specifications for Custom Images Used for Model Management
                                                • Custom images cannot contain malicious code.
                                                • The size of a custom image cannot exceed 30 GB.
                                                • External port of images
                                                  The external service port of the image must be 8080. The inference interface must be consistent with the URL defined by apis in the config.json file. The inference interface can be directly accessed when the image is started. The following is an example of accessing the mnist image. The image contains the model trained with the mnist dataset. The model can identify handwritten digits in images. In this example, listen_ip indicates the IP address of the container.
                                                  
-
                                                  • Sample request: curl -X POST \ http://{listen_ip}:8080/ \ -F images=@seven.jpg
                                                  • Sample response
                                                    {"mnist_result": 7}
                                                    
+
                                                    • Sample request: curl -X POST \ http://{listen_ip}:8080/ \ -F images=@seven.jpg
                                                    • Sample response
                                                      {"mnist_result": 7}
                                                      
 
                                                    
 
                                                  • Health check port
                                                    A custom image must provide a health check interface for ModelArts to call. The health check interface is configured in the config.json file. For details, see the model configuration file compilation description. A sample health check interface is as follows:
                                                    
-
                                                    • URI
                                                      GET /health
                                                      
-
                                                    • Sample request: curl -X GET \ http://{listen_ip}:8080/health
                                                    • Sample response
                                                      {"health": "true"}
                                                      
+
                                                      • URI
                                                        GET /health
                                                        
+
                                                      • Sample request: curl -X GET \ http://{listen_ip}:8080/health
                                                      • Sample response
                                                        {"health": "true"}
                                                        
 
                                                      • Status code
 
                                                Table 4 Optional parameters
                                                Parameter
                                                
                                                 		
 
                                                
diff --git a/docs/modelarts/umn/modelarts_23_0238.html b/docs/modelarts/umn/modelarts_23_0238.html
index 3043ea4a3..bb2687866 100644
--- a/docs/modelarts/umn/modelarts_23_0238.html
+++ b/docs/modelarts/umn/modelarts_23_0238.html
@@ -16,9 +16,9 @@
 
-
-
@@ -30,9 +30,9 @@
 
-
-
-
-
-
-
-
-
-
-
-
-
diff --git a/docs/modelarts/umn/modelarts_23_0239.html b/docs/modelarts/umn/modelarts_23_0239.html
index d1554027a..4bba106d0 100644
--- a/docs/modelarts/umn/modelarts_23_0239.html
+++ b/docs/modelarts/umn/modelarts_23_0239.html
@@ -2,7 +2,7 @@
 
 
                                                Using Custom Images to Train Models
                                                If the framework used for algorithm development is not a frequently-used framework, you can build an algorithm into a custom image and use the custom image to create a training job.
                                                
-
                                                Prerequisites
                                                • Data has been prepared. Specifically, you have created an available dataset in ModelArts, or you have uploaded the dataset used for training to the OBS directory.
                                                • If the algorithm source is Custom, create an image and upload the image to SWR. For details, see .
                                                • The training script has been uploaded to the OBS directory.
                                                • At least one empty folder has been created on OBS for storing the training output.
                                                • The account is not in arrears because resources are consumed when training jobs are running.
                                                • The OBS directory you use and ModelArts are in the same region.
                                                
+
                                                Prerequisites
                                                • Data has been prepared. Specifically, you have created an available dataset in ModelArts, or you have uploaded the dataset used for training to the OBS directory.
                                                • If the algorithm source is Custom, create an image and upload the image to SWR. For details, see Creating and Uploading a Custom Image.
                                                • The training script has been uploaded to the OBS directory.
                                                • At least one empty folder has been created on OBS for storing the training output.
                                                • The account is not in arrears because resources are consumed when training jobs are running.
                                                • The OBS directory you use and ModelArts are in the same region.
                                                
 
                                                
 
                                                Precautions
                                                • In the dataset directory specified for a training job, the names of the files (such as the image file, audio file, and label file) containing data used for training contain 0 to 255 characters. If the names of certain files in the dataset directory contain over 255 characters, the training job will ignore these files and use data in the valid files for training. If the names of all files in the dataset directory contain over 255 characters, no data is available for the training job and the training job fails.
                                                • In the training script, the Data Source and Training Output Path parameters must be set to the OBS path. Use the  to perform read and write operations in the path.
                                                
 
                                                
@@ -30,7 +30,7 @@
 
                                                Table 1 Status code
                                                Status Code
                                                
                                                 		
 
                                                System Version
                                                
 
                                                AI Engine and Version
                                                
+
                                                AI Engine and Version
                                                
 
                                                Supported CUDA or Ascend Version
                                                
+
                                                Supported CUDA Version
                                                
                                                 		
 
                                                
                                                 
 
                                                Ubuntu 16.04
                                                
 
                                                TF-1.13.1-python3.6
                                                
+
                                                TF-1.13.1-python3.6
                                                
 
                                                CUDA 10.0
                                                
+
                                                CUDA 10.0
                                                
                                                 		
 
                                                
 
                                                TF-1.8.0-python3.6
                                                
@@ -53,9 +53,9 @@
                                                 		
 
                                                Ubuntu 16.04
                                                
 
                                                Caffe-1.0.0-python2.7
                                                
+
                                                Caffe-1.0.0-python2.7
                                                
 
                                                CUDA 8.0
                                                
+
                                                CUDA 8.0
                                                
                                                 		
 
                                                
 
                                                Spark_MLlib
                                                
@@ -66,9 +66,9 @@
                                                 		
 
                                                Ubuntu 16.04
                                                
 
                                                Spark-2.3.2-python3.6
                                                
+
                                                Spark-2.3.2-python3.6
                                                
 
                                                N/A
                                                
+
                                                N/A
                                                
                                                 		
 
                                                
 
                                                XGBoost-Sklearn
                                                
@@ -79,9 +79,9 @@
                                                 		
 
                                                Ubuntu 16.04
                                                
 
                                                Scikit_Learn-0.18.1-python3.6
                                                
+
                                                Scikit_Learn-0.18.1-python3.6
                                                
 
                                                N/A
                                                
+
                                                N/A
                                                
                                                 		
 
                                                
 
                                                PyTorch
                                                
@@ -92,9 +92,9 @@
                                                 		
 
                                                Ubuntu 16.04
                                                
 
                                                PyTorch-1.3.0-python3.6
                                                
+
                                                PyTorch-1.3.0-python3.6
                                                
 
                                                CUDA 10.0
                                                
+
                                                CUDA 10.0
                                                
                                                 		
 
                                                
 
                                                PyTorch-1.0.0-python3.6
                                                
@@ -110,9 +110,9 @@
                                                 		
 
                                                Ubuntu16.04
                                                
 
                                                MXNet-1.2.1-python3.6
                                                
+
                                                MXNet-1.2.1-python3.6
                                                
 
                                                CUDA 9.0
                                                
+
                                                CUDA 9.0
                                                
                                                 		
 
                                                
                                                 
 
                                                Custom
                                                
                                                 		
 
                                                For details about custom image specifications, see Specifications for Custom Images Used for Training Jobs.
                                                
-
                                                • Image Path: SWR URL after the image is uploaded to SWR. For details about how to upload an image, see Creating and Uploading a Custom Image.
                                                • Code Directory: OBS path for storing the training code file.
                                                
+
                                                • Image Path: SWR URL after the image is uploaded to SWR. For details about how to upload an image, see Creating and Uploading a Custom Image.
                                                • Code Directory: OBS path for storing the training code file.
                                                • Boot Command: Command to boot the training job after the image is started. Set this parameter based on site requirements. If the custom image is based on a basic ModelArts image, set parameters by referring to Creating a Training Job Using a Custom Image (GPU).
                                                
                                                 		
 
                                                
 
                                                Data Source
                                                
diff --git a/docs/modelarts/umn/modelarts_23_0332.html b/docs/modelarts/umn/modelarts_23_0332.html
index aa26609d8..659808aaa 100644
--- a/docs/modelarts/umn/modelarts_23_0332.html
+++ b/docs/modelarts/umn/modelarts_23_0332.html
@@ -5,26 +5,17 @@
 
                                                Step 1: Uploading Files to OBS
                                                Use the OBS API to upload large files because OBS Console has restrictions on the file size and quantity.
                                                
 
                                                
 
                                                Step 2: Downloading Files from OBS to Notebook Instances
                                                A notebook instance can be mounted to OBS or EVS as the storage location. The operation method varies depending on the instance types.
                                                
-
                                                • Downloading files to notebook instances with EVS attached
                                                  • Read an OBS file. For example, if you read the obs://bucket_name/obs_file.txt file, the content is returned as strings.
                                                    1
                                                    file_str = mox.file.read('obs://bucket_name/obs_file.txt')
-
                                                    
-
                                                    
+
                                                    • Downloading files to notebook instances with EVS attached
                                                      • Read an OBS file. For example, if you read the obs://bucket_name/obs_file.txt file, the content is returned as strings.
                                                        file_str = mox.file.read('obs://bucket_name/obs_file.txt')
                                                        
 
                                                        
-
                                                        You can also open the file object and read data from it. Both methods are equivalent.
                                                        1
-2
                                                        with mox.file.File('obs://bucket_name/obs_file.txt', 'r') as f:
-  file_str = f.read()
-
                                                        
-
                                                        
+
                                                        You can also open the file object and read data from it. Both methods are equivalent.
                                                        with mox.file.File('obs://bucket_name/obs_file.txt', 'r') as f:
+  file_str = f.read()
                                                        
 
                                                        
 
                                                      • Use the OBS API in the ModelArts SDK to download data to notebook instances.
                                                         
                                                        If the size of a single file exceeds 5 GB, the file cannot be uploaded in this mode. Use the MoXing API to upload large files.
                                                        
 
                                                        
 
                                                        Sample code:
                                                        
-
                                                        1
-2
-3
                                                        from modelarts.session import Session
-session = Session()
-session.download_data(bucket_path="/bucket-name/dir1/sdk.txt", path="/home/user/sdk/obs.txt")
-
                                                        
-
                                                        
+
                                                        from modelarts.session import Session
+session = Session()
+session.download_data(bucket_path="/bucket-name/dir1/sdk.txt", path="/home/user/sdk/obs.txt")
                                                        
 
                                                      
 
                                                    • Downloading files to notebook instances using OBS for data storage
                                                      Upload files to the OBS path specified during notebook instance creation and synchronize the files from OBS to the notebook instances using Sync OBS.
                                                      
 
                                                    
diff --git a/docs/modelarts/umn/modelarts_23_0333.html b/docs/modelarts/umn/modelarts_23_0333.html
index c321f8ffc..772fd208e 100644
--- a/docs/modelarts/umn/modelarts_23_0333.html
+++ b/docs/modelarts/umn/modelarts_23_0333.html
@@ -4,13 +4,9 @@
 
                                                    Only files within 100 MB in JupyterLab can be downloaded to a local PC. You can perform operations in different scenarios based on the storage location selected when creating a notebook instance.
                                                    
 
                                                    Notebook Instances Using OBS Storage
                                                    For notebook instances that use OBS storage, you can use OBS or the ModelArts SDK to download files from OBS to a local PC.
                                                    
 
                                                    • Method 1: Downloading the files using OBS
                                                      Use OBS to  download the files to the local PC. If you have a large amount of data, use OBS Browser+ to download data or folders.
                                                      
-
                                                    • Method 2: Downloading the files using the ModelArts SDKs
                                                      1. Download and install the ModelArts SDKs on your local PC. For details, see .
                                                      2. Authenticate sessions of the ModelArts SDKs. For details, see .
                                                      3.  from OBS to the local PC. The sample code is as follows:
                                                        1
-2
-3
                                                        from modelarts.session import Session
-session=Session(access_key='***',secret_key='***',project_id='***',region_name='***')
-session.download_data(bucket_path="/bucket_name/obs_file.txt",path="/home/user/obs_file.txt")
-
                                                        
-
                                                        
+
                                                      4. Method 2: Downloading the files using the ModelArts SDKs
                                                        1. Download and install the ModelArts SDKs on your local PC. For details, see .
                                                        2. Authenticate sessions of the ModelArts SDKs. For details, see .
                                                        3.  from OBS to the local PC. The sample code is as follows:
                                                          from modelarts.session import Session
+session=Session(access_key='***',secret_key='***',project_id='***',region_name='***')
+session.download_data(bucket_path="/bucket_name/obs_file.txt",path="/home/user/obs_file.txt")
                                                          
 
                                                        
 
                                                    
 
                                                    
diff --git a/docs/modelarts/umn/modelarts_23_0335.html b/docs/modelarts/umn/modelarts_23_0335.html
index 8981c7cc1..e44ea1982 100644
--- a/docs/modelarts/umn/modelarts_23_0335.html
+++ b/docs/modelarts/umn/modelarts_23_0335.html
@@ -4,70 +4,37 @@
 
                                                    In notebook instances, you can use ModelArts SDKs to manage OBS, training jobs, models, and real-time services.
                                                    
 
                                                    For details about how to use ModelArts SDKs, see ModelArts SDK Reference.
                                                    
 
                                                    Notebooks carry the authentication (AK/SK) and region information about login users. Therefore, SDK session authentication can be completed without entering parameters.
                                                    
-
                                                    Example Code
                                                    • Creating a training job
                                                       1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
                                                      from modelarts.session import Session
-from modelarts.estimator import Estimator
-session = Session()
-estimator = Estimator(
-                      modelarts_session=session,
-                      framework_type='PyTorch',                                     # AI engine name
-                       framework_version='PyTorch-1.0.0-python3.6',                  # AI engine version
-                      code_dir='/obs-bucket-name/src/',                                      # Training script directory
-                      boot_file='/obs-bucket-name/src/pytorch_sentiment.py',                 # Training startup script directory
-                      log_url='/obs-bucket-name/log/',                                       # Training log directory
-                      hyperparameters=[
-                                       {"label":"classes",
-                                        "value": "10"},    
-                                       {"label":"lr",
-                                        "value": "0.001"}
-                                       ],
-                      output_path='/obs-bucket-name/output/',                                # Training output directory
-                      train_instance_type='modelarts.vm.gpu.p100',                  # Training environment specifications
-                      train_instance_count=1,                                       # Number of training nodes
-                      job_description='pytorch-sentiment with ModelArts SDK')       # Training job description
-job_instance = estimator.fit(inputs='/obs-bucket-name/data/train/', wait=False, job_name='my_training_job')
-
                                                      
-
                                                      
+
                                                      Example Code
                                                      • Creating a training job
                                                        from modelarts.session import Session
+from modelarts.estimator import Estimator
+session = Session()
+estimator = Estimator(
+                      modelarts_session=session,
+                      framework_type='PyTorch',                                     # AI engine name
+                       framework_version='PyTorch-1.0.0-python3.6',                  # AI engine version
+                      code_dir='/obs-bucket-name/src/',                                      # Training script directory
+                      boot_file='/obs-bucket-name/src/pytorch_sentiment.py',                 # Training startup script directory
+                      log_url='/obs-bucket-name/log/',                                       # Training log directory
+                      hyperparameters=[
+                                       {"label":"classes",
+                                        "value": "10"},    
+                                       {"label":"lr",
+                                        "value": "0.001"}
+                                       ],
+                      output_path='/obs-bucket-name/output/',                                # Training output directory
+                      train_instance_type='modelarts.vm.gpu.p100',                  # Training environment specifications
+                      train_instance_count=1,                                       # Number of training nodes
+                      job_description='pytorch-sentiment with ModelArts SDK')       # Training job description
+job_instance = estimator.fit(inputs='/obs-bucket-name/data/train/', wait=False, job_name='my_training_job')
                                                        
 
                                                      
-
                                                      • Querying a model list
                                                        1
-2
-3
-4
                                                        from modelarts.session import Session 
-from modelarts.model import Model
-session = Session() 
-model_list_resp = Model.get_model_list(session, model_status="published", model_name="digit", order="desc")
-
                                                        
-
                                                        
-
                                                      • Querying service details
                                                        1
-2
-3
-4
-5
                                                        from modelarts.session import Session
-from modelarts.model import Predictor
-session = Session()
-predictor_instance = Predictor(session, service_id="input your service_id")
-predictor_info_resp = predictor_instance.get_service_info()
-
                                                        
-
                                                        
+
                                                        • Querying a model list
                                                          from modelarts.session import Session 
+from modelarts.model import Model
+session = Session() 
+model_list_resp = Model.get_model_list(session, model_status="published", model_name="digit", order="desc")
                                                          
+
                                                        • Querying service details
                                                          from modelarts.session import Session
+from modelarts.model import Predictor
+session = Session()
+predictor_instance = Predictor(session, service_id="input your service_id")
+predictor_info_resp = predictor_instance.get_service_info()
                                                          
 
                                                        
 
                                                      
 
                                                    
diff --git a/docs/modelarts/umn/modelarts_23_0336.html b/docs/modelarts/umn/modelarts_23_0336.html
index 021513368..562d41ab3 100644
--- a/docs/modelarts/umn/modelarts_23_0336.html
+++ b/docs/modelarts/umn/modelarts_23_0336.html
@@ -3,7 +3,7 @@
 
                                                    Synchronizing Files with OBS
                                                    
 
                                                    If you specify Storage Path during notebook instance creation, your compiled code will be automatically stored in your specified OBS bucket. If code invocation among different .ipynb files is required, you can use the Sync OBS function.
                                                    
 
                                                    The Sync OBS function is used to synchronize the objects selected in the list of notebook instance files from the OBS bucket to the current container directory ~/work.
                                                    
-
                                                    Precautions
                                                    • The maximum size of files to be synchronized at a time is 500 MB, and the maximum number of files to be synchronized at a time is 1,024.
                                                    • The total size of objects to be synchronized cannot exceed 5 GB. For example, if 2 GB files exist in the ~/work container directory, you can use Sync OBS to synchronize a maximum of 3 GB files.
                                                    • The Sync OBS function only takes effect on notebook instances for which Storage is OBS. For notebook instances whose Storage is not OBS, all files are read and written in the ~/work container directory.
                                                    • Notebook instances created using Ascend resources do not support the Sync OBS function.
                                                    
+
                                                    Precautions
                                                    • The maximum size of files to be synchronized at a time is 500 MB, and the maximum number of files to be synchronized at a time is 1,024.
                                                    • The total size of objects to be synchronized cannot exceed 5 GB. For example, if 2 GB files exist in the ~/work container directory, you can use Sync OBS to synchronize a maximum of 3 GB files.
                                                    • The Sync OBS function only takes effect on notebook instances for which Storage is OBS. For notebook instances whose Storage is not OBS, all files are read and written in the ~/work container directory.
                                                    
 
                                                    
 
                                                    Procedure
                                                    The Sync OBS function can be used in JupyterLab. The following describes how to use the function.
                                                    
 
                                                    For example, Example-1.ipynb needs to invoke module in Example-2.ipynb. In File Browser on the left, select the two .ipynb files and click Sync OBS. In the dialog box that is displayed, click YES to complete synchronization. After the synchronization is successful, the code can be called mutually.
                                                    
diff --git a/docs/modelarts/umn/public_sys-resources/commonrtl.css b/docs/modelarts/umn/public_sys-resources/commonrtl.css
index 947a9a033..f261da752 100644
--- a/docs/modelarts/umn/public_sys-resources/commonrtl.css
+++ b/docs/modelarts/umn/public_sys-resources/commonrtl.css
@@ -1,2 +1,2 @@
-.msgph{font-family:Courier New}.unresolved{background-color:#87ceeb}.noTemplate{background-color:#ff0}.base{background-color:#fff}/*!  Add space for top level topics */.nested0,.p{margin-top:1em}/*!  div with class=p is used for paragraphs that contain blocks, to keep the XHTML valid *//*!  Default of italics to set apart figure captions */.figcap,.italic,.var{font-style:italic}.figdesc{font-style:normal}/*!  Use @frame to create frames on figures */.figborder{padding-left:3px;padding-right:3px;margin-top:1em;border:2px solid Silver}.figsides{margin-top:1em;padding-left:3px;padding-right:3px;border-left:2px solid Silver;border-right:2px solid Silver}.figtop{border-top:2px solid Silver;margin-top:1em}.figbottom{border-bottom:2px solid Silver}.figtopbot{border-top:2px solid Silver;border-bottom:2px solid Silver;margin-top:1em}/*!  Most link groups are created with 
                                                    . Ensure they have space before and after. */.ullinks,ul.simple{list-style-type:none}.attention,.danger,.ddexpand,.dlexpand,.example,.fastpath,.important,.liexpand,.linklist,.note,.notice,.olchildlink,.relconcepts,.relinfo,.relref,.reltasks,.remember,.restriction,.section,.sliexpand,.stepexpand,.substepexpand,.tip,.ulchildlink,.warning{margin-top:1em;margin-bottom:1em}.linklistwithchild,.sublinklist{margin-top:1em;margin-right:1.5em;margin-bottom:1em}.breadcrumb{font-size:smaller;margin-bottom:1em}.prereq{margin-right:20px}/*!  Set heading sizes, getting smaller for deeper nesting */.topictitle1{font-size:1.34em;margin-top:0;margin-bottom:.1em}.topictitle2,.topictitle3,.topictitle4,.topictitle5,.topictitle6,.sectiontitle{font-size:1.17em}.topictitle2{margin-top:1pc;margin-bottom:.45em}.topictitle3{margin-top:1pc;margin-bottom:.17em;font-weight:700}.topictitle4{margin-top:.83em;font-weight:700}.topictitle5{font-weight:700}.topictitle6{font-style:italic}.sectiontitle{margin-top:1em;margin-bottom:0;color:#000;font-weight:700}/*!  All note formats have the same default presentation */.attentiontitle,.bold,.cautiontitle,.dangertitle,.dlterm,.fastpathtitle,.firstcol,.importanttitle,.notelisttitle,.notetitle,.noticetitle,.parmname,.remembertitle,.restrictiontitle,.tiptitle,.uicontrol,.warningtitle{font-weight:700}.caution{font-weight:700;margin-bottom:1em}/*!  Simple lists do not get a bullet *//*!  Used on the first column of a table, when rowheader="firstcol" is used *//*!  Various basic phrase styles */.boldItalic{font-weight:700;font-style:italic}.shortcut,.underlined{text-decoration:underline}/*!  2008-10-27 keyword采用跟随上下文的样式
+/*! Copyright (c) Huawei Technologies Co., Ltd. 2020-2022. All rights reserved. */.msgph{font-family:Courier New}.unresolved{background-color:#87ceeb}.noTemplate{background-color:#ff0}.base{background-color:#fff}/*!  Add space for top level topics */.nested0,.p{margin-top:1em}/*!  div with class=p is used for paragraphs that contain blocks, to keep the XHTML valid *//*!  Default of italics to set apart figure captions */.figcap,.italic,.var{font-style:italic}.figdesc{font-style:normal}/*!  Use @frame to create frames on figures */.figborder{padding-left:3px;padding-right:3px;margin-top:1em;border:2px solid Silver}.figsides{margin-top:1em;padding-left:3px;padding-right:3px;border-left:2px solid Silver;border-right:2px solid Silver}.figtop{border-top:2px solid Silver;margin-top:1em}.figbottom{border-bottom:2px solid Silver}.figtopbot{border-top:2px solid Silver;border-bottom:2px solid Silver;margin-top:1em}/*!  Most link groups are created with 
                                                    . Ensure they have space before and after. */.ullinks,ul.simple{list-style-type:none}.attention,.danger,.ddexpand,.dlexpand,.example,.fastpath,.important,.liexpand,.linklist,.note,.notice,.olchildlink,.relconcepts,.relinfo,.relref,.reltasks,.remember,.restriction,.section,.sliexpand,.stepexpand,.substepexpand,.tip,.ulchildlink,.warning{margin-top:1em;margin-bottom:1em}.linklistwithchild,.sublinklist{margin-top:1em;margin-right:1.5em;margin-bottom:1em}.breadcrumb{font-size:smaller;margin-bottom:1em}.prereq{margin-right:20px}/*!  Set heading sizes, getting smaller for deeper nesting */.topictitle1{font-size:1.34em;margin-top:0;margin-bottom:.1em}.topictitle2,.topictitle3,.topictitle4,.topictitle5,.topictitle6,.sectiontitle{font-size:1.17em}.topictitle2{margin-top:1pc;margin-bottom:.45em}.topictitle3{margin-top:1pc;margin-bottom:.17em;font-weight:700}.topictitle4{margin-top:.83em;font-weight:700}.topictitle5{font-weight:700}.topictitle6{font-style:italic}.sectiontitle{margin-top:1em;margin-bottom:0;color:#000;font-weight:700}/*!  All note formats have the same default presentation */.attentiontitle,.bold,.cautiontitle,.dangertitle,.dlterm,.fastpathtitle,.firstcol,.importanttitle,.notelisttitle,.notetitle,.noticetitle,.parmname,.remembertitle,.restrictiontitle,.tiptitle,.uicontrol,.warningtitle{font-weight:700}.caution{font-weight:700;margin-bottom:1em}/*!  Simple lists do not get a bullet *//*!  Used on the first column of a table, when rowheader="firstcol" is used *//*!  Various basic phrase styles */.boldItalic{font-weight:700;font-style:italic}.shortcut,.underlined{text-decoration:underline}/*!  2008-10-27 keyword采用跟随上下文的样式
 *//*!  Default of bold for definition list terms *//*!  Use CSS to expand lists with @compact="no" */.dltermexpand{font-weight:700;margin-top:1em}[compact="yes"]>li{margin-top:0}[compact="no"]>li{margin-top:.53em}/*!  Align images based on @align on topic/image */div.imageleft,.text-align-left{text-align:left}div.imagecenter,.text-align-center{text-align:center}div.imageright,.text-align-right{text-align:right}div.imagejustify,.text-align-justify{text-align:justify}.cellrowborder{border-right:0;border-top:0;border-left:1px solid;border-bottom:1px solid}.row-nocellborder{border-left:hidden;border-right:0;border-top:0;border-bottom:1px solid}.cell-norowborder{border-top:0;border-bottom:hidden;border-right:0;border-left:1px solid}.nocellnorowborder{border:0;border-left:hidden;border-bottom:hidden}pre.codeblock,pre.screen{padding:5px;border:outset;background-color:#ccc;margin-top:2px;margin-bottom:2px;white-space:pre}
\ No newline at end of file