Notebook for text to image for ds-mii model (#2873)

* Notebook for text to image for ds-mii model * aacs * cli * cli * model name * code quality --------- Co-authored-by: grajguru <[email protected]>
Azure · Dec 22, 2023 · 342e602 · 342e602
1 parent 7f9a32a
commit 342e602
Show file tree

Hide file tree

Showing 5 changed files with 122 additions and 824 deletions.
diff --git a/cli/foundation-models/system/inference/text-to-image/deploy-online.yaml b/cli/foundation-models/system/inference/text-to-image/deploy-online.yaml
@@ -9,4 +9,5 @@ liveness_probe:
   timeout: 299
 request_settings:
   request_timeout_ms: 90000
-
+environment_variables:
+  "WORKER_COUNT": 2
diff --git a/cli/foundation-models/system/inference/text-to-image/text-to-image-online-endpoint.sh b/cli/foundation-models/system/inference/text-to-image/text-to-image-online-endpoint.sh
@@ -32,7 +32,6 @@ az account set -s $subscription_id
 workspace_info="--resource-group $resource_group_name --workspace-name $workspace_name"
 
 # 2. Check if the model exists in the registry
-# Need to confirm model show command works for registries outside the tenant (aka system registry)
 if ! az ml model show --name $model_name --label $model_label --registry-name $registry_name 
 then
     echo "Model $model_name:$model_label does not exist in registry $registry_name"
@@ -44,13 +43,22 @@ model_version=$(az ml model show --name $model_name --label $model_label --regis
 
 # 3. Deploy the model to an endpoint
 # Create online endpoint 
-az ml online-endpoint create --name $endpoint_name $workspace_info  || {
+az ml online-endpoint create --name $endpoint_name $workspace_info || {
     echo "endpoint create failed"; exit 1;
 }
 
+# 3.1 Setup Deployment Parameters
+max_concurrent_request=2  # the maximum number of concurrent requests supported by the endpoint
+
+# Note: We have set the value of `max_concurrent_request` to 2, 
+# as we are utilizing the `Standard_NC6s_v3` SKU for deployment, which has one GPU. 
+# If you are using a larger SKU, please increase this value to get the maximum performance.
+
 # Deploy model from registry to endpoint in workspace
 az ml online-deployment create --file deploy-online.yaml $workspace_info --all-traffic --set \
   endpoint_name=$endpoint_name model=azureml://registries/$registry_name/models/$model_name/versions/$model_version \
+  request_settings.max_concurrent_requests_per_instance=$max_concurrent_request \
+  environment_variables.WORKER_COUNT=$max_concurrent_request \
   instance_type=$deployment_sku || {
     echo "deployment create failed"; exit 1;
 }