Batch notebooks update for txt2img to address environment error (Azur…

…e#2953) * Pass custom environment in batch txt2img notebooks * cli example * cli update * code quality --------- Co-authored-by: grajguru <[email protected]>
samuel100 · Jan 17, 2024 · 581112f · 581112f
1 parent a5c29c3
commit 581112f
Show file tree

Hide file tree

Showing 13 changed files with 737 additions and 18 deletions.
diff --git a/.../aacs-scoring-files/docker_env/Dockerfile → ...image/scoring-files/docker_env/Dockerfile b/.../aacs-scoring-files/docker_env/Dockerfile → ...image/scoring-files/docker_env/Dockerfile
diff --git a/...-files/docker_env/conda_dependencies.yaml → ...-files/docker_env/conda_dependencies.yaml b/...-files/docker_env/conda_dependencies.yaml → ...-files/docker_env/conda_dependencies.yaml
diff --git a/...les/score/parallel_run_step.settings.json → ...les/score/parallel_run_step.settings.json b/...les/score/parallel_run_step.settings.json → ...les/score/parallel_run_step.settings.json
diff --git a/...e/aacs-scoring-files/score/score_batch.py → ...-image/scoring-files/score/score_batch.py b/...e/aacs-scoring-files/score/score_batch.py → ...-image/scoring-files/score/score_batch.py
@@ -36,6 +36,7 @@ def init():
     global g_logger
     global g_file_loader_dictionary
     global aacs_client
+    global aacs_enabled
 
     g_logger = logging.getLogger("azureml")
     g_logger.setLevel(logging.INFO)
@@ -60,10 +61,15 @@ def init():
         ".pqt": load_parquet,
     }
 
-    endpoint = os.environ.get("CONTENT_SAFETY_ENDPOINT")
-    key = os.environ.get("CONTENT_SAFETY_KEY")
+    endpoint = os.environ.get("CONTENT_SAFETY_ENDPOINT", None)
+    key = os.environ.get("CONTENT_SAFETY_KEY", None)
     # Create an Content Safety client
-    aacs_client = ContentSafetyClient(endpoint, AzureKeyCredential(key))
+    if endpoint is not None and key is not None:
+        aacs_client = ContentSafetyClient(endpoint, AzureKeyCredential(key))
+        aacs_enabled = True
+    else:
+        aacs_enabled = False
+        g_logger.warn("Azure AI Content Safety (aacs) is disabled.")
 
 
 def get_input_schema(model_path):
@@ -372,13 +378,18 @@ def run(batch_input):
     predict_result = []
     try:
         aacs_threshold = int(os.environ.get("CONTENT_SAFETY_THRESHOLD", default=1))
-        blocked_input = analyze_data(
-            batch_input, aacs_threshold, blocked_input=None, is_input=True
-        )
+        if aacs_enabled:
+            blocked_input = analyze_data(
+                batch_input, aacs_threshold, blocked_input=None, is_input=True
+            )
         predict_result = g_model.predict(batch_input)
-        _ = analyze_data(
-            predict_result, aacs_threshold, blocked_input=blocked_input, is_input=False
-        )
+        if aacs_enabled:
+            _ = analyze_data(
+                predict_result,
+                aacs_threshold,
+                blocked_input=blocked_input,
+                is_input=False,
+            )
     except Exception as e:
         g_logger.error("Processing mini batch failed with exception: " + str(e))
         g_logger.error(traceback.format_exc())

diff --git a/cli/foundation-models/system/inference/text-to-image/text-to-image-batch-endpoint.sh b/cli/foundation-models/system/inference/text-to-image/text-to-image-batch-endpoint.sh
@@ -52,7 +52,7 @@ else
 fi
 
 # 4. Submit a sample request to endpoint
-data_path="./text_to_image_batch_data/batch_data"
+data_path="./text_to_image_batch_data"
 python utils/prepare_data.py --payload-path $data_path --mode "batch"
 # Path where the processes csvs are dumped. This is the input to the endpoint
 processed_data_path="./text_to_image_batch_data/processed_batch_data"
@@ -73,11 +73,32 @@ az ml batch-endpoint create --name $endpoint_name $workspace_info  || {
     echo "endpoint create failed"; exit 1;
 }
 
+# create a environment for batch deployment
+
+environment_name="text-to-image-model-env"
+environment_label="latest"
+
+if ! az ml environment show --name $environment_name --label $environment_label $workspace_info
+then
+    echo "Environment $environment_name:$environment_label does not exist in Workspace."
+    echo "---Creating environment---"
+    az ml environment create --name $environment_name  --build-context "./scoring-files/docker_env" \
+    $workspace_info || {
+    echo "environment create failed"; exit 1;
+}
+    exit 1
+fi
+
+environment_version=$(az ml environment show --name $environment_name --label $environment_label $workspace_info --query version --output tsv)
+
 # deploy model from registry to endpoint in workspace
 az ml batch-deployment create --file batch-deploy.yml $workspace_info --set \
   endpoint_name=$endpoint_name \
   name=$deployment_name \
   compute=$deployment_compute \
+  environment=azureml:$environment_name:$environment_version \
+  code_configuration.code="scoring-files/score" \
+  code_configuration.scoring_script="score_batch.py" \
   model=azureml://registries/$registry_name/models/$model_name/versions/$model_version || {
     echo "deployment create failed"; exit 1;
 }

diff --git a/...oundation-models/system/inference/text-to-image/safe-text-to-image-batch-deployment.ipynb b/...oundation-models/system/inference/text-to-image/safe-text-to-image-batch-deployment.ipynb
@@ -329,7 +329,7 @@
     "    print(\"---Creating environment---\")\n",
     "    env = Environment(\n",
     "        name=environment_name,\n",
-    "        build=BuildContext(path=\"./aacs-scoring-files/docker_env\"),\n",
+    "        build=BuildContext(path=\"./scoring-files/docker_env\"),\n",
     "    )\n",
     "    ml_client.environments.create_or_update(env)\n",
     "    env = ml_client.environments.get(environment_name, label=\"latest\")\n",
@@ -447,7 +447,7 @@
     "    model=model,\n",
     "    environment=env,\n",
     "    code_configuration=CodeConfiguration(\n",
-    "        code=\"aacs-scoring-files/score\",\n",
+    "        code=\"scoring-files/score\",\n",
     "        scoring_script=\"score_batch.py\",\n",
     "    ),\n",
     "    compute=compute_name,\n",
@@ -516,7 +516,7 @@
     "import pandas as pd\n",
     "\n",
     "# Specify the folder where your CSV files are located\n",
-    "data_path = \"aacs-scoring-files/text-to-image-batch-data\"\n",
+    "data_path = \"scoring-files/text-to-image-batch-data\"\n",
     "\n",
     "# Use glob to get a list of CSV files in the folder\n",
     "csv_files = glob.glob(os.path.join(data_path, \"*.csv\"))\n",
@@ -537,7 +537,7 @@
     "from pathlib import Path\n",
     "\n",
     "# Specify the folder where your CSV files should be saved\n",
-    "processed_dataset_parent_dir = \"aacs-scoring-files/processed-text-to-image-batch-data\"\n",
+    "processed_dataset_parent_dir = \"scoring-files/processed-text-to-image-batch-data\"\n",
     "os.makedirs(processed_dataset_parent_dir, exist_ok=True)\n",
     "batch_input_file = \"batch_input.csv\"\n",
     "\n",
@@ -712,7 +712,7 @@
     "    names=[\n",
     "        \"row_number_per_file\",\n",
     "        \"prompt\",\n",
-    "        \"image_file_name\",\n",
+    "        \"generated_image\",\n",
     "        \"nsfw_content_detected\",\n",
     "        \"file_name\",\n",
     "    ],\n",

diff --git a/...thon/foundation-models/system/inference/text-to-image/scoring-files/docker_env/Dockerfile b/...thon/foundation-models/system/inference/text-to-image/scoring-files/docker_env/Dockerfile
@@ -0,0 +1,18 @@
+FROM mcr.microsoft.com/azureml/openmpi4.1.0-ubuntu20.04:20230620.v1
+
+ENV CONDA_ENVIRONMENT_PATH /azureml-envs/text-to-image
+
+# Prepend path to AzureML conda environment
+ENV PATH $CONDA_ENVIRONMENT_PATH/bin:$PATH
+
+# Create conda environment
+COPY conda_dependencies.yaml .
+RUN conda env create -p $CONDA_ENVIRONMENT_PATH -f conda_dependencies.yaml -q && \
+    rm conda_dependencies.yaml && \
+    conda run -p $CONDA_ENVIRONMENT_PATH pip cache purge && \
+    conda clean -a -y
+
+RUN pip freeze
+
+# This is needed for mpi to locate libpython
+ENV LD_LIBRARY_PATH $CONDA_ENVIRONMENT_PATH/lib:$LD_LIBRARY_PATH
diff --git a/...on-models/system/inference/text-to-image/scoring-files/docker_env/conda_dependencies.yaml b/...on-models/system/inference/text-to-image/scoring-files/docker_env/conda_dependencies.yaml
@@ -0,0 +1,19 @@
+channels:
+- conda-forge
+dependencies:
+- python=3.8.16
+- pip<=23.1.2
+- pip:
+  - mlflow==2.3.2
+  - torch==1.13.0
+  - transformers==4.29.1
+  - diffusers==0.23.0
+  - accelerate==0.22.0
+  - azureml-core==1.52.0
+  - azureml-mlflow==1.52.0
+  - azure-ai-contentsafety==1.0.0b1
+  - aiolimiter==1.1.0
+  - azure-ai-mlmonitoring==0.1.0a3
+  - azure-mgmt-cognitiveservices==13.4.0
+  - azure-identity==1.13.0
+name: mlflow-env
diff --git a/...models/system/inference/text-to-image/scoring-files/score/parallel_run_step.settings.json b/...models/system/inference/text-to-image/scoring-files/score/parallel_run_step.settings.json
@@ -0,0 +1 @@
+{"append_row": {"pandas.DataFrame.to_csv": {"sep": ",","index": true}}}