Merge branch 'main' into ui-flow-v11

Azure · Feb 25, 2025 · f4c385e · f4c385e
2 parents f085188 + c6e4cf3
commit f4c385e
Show file tree

Hide file tree

Showing 10 changed files with 30 additions and 23 deletions.
diff --git a/assets/models/system/Phi-4/spec.yaml b/assets/models/system/Phi-4/spec.yaml
@@ -8,7 +8,7 @@ properties:
   inference-recommended-sku: Standard_NC24ads_A100_v4, Standard_NC48ads_A100_v4, Standard_NC96ads_A100_v4, Standard_ND96asr_v4, Standard_ND96amsr_A100_v4
   finetuning-tasks: chat-completion
   finetune-min-sku-spec: 24|1|220|64
-  finetune-recommended-sku: Standard_NC24ads_A100_v4, Standard_NC48ads_A100_v4, Standard_NC96ads_A100_v4, Standard_ND96asr_v4, Standard_ND96amsr_A100_v4
+  finetune-recommended-sku: Standard_NC24ads_A100_v4, Standard_NC48ads_A100_v4, Standard_NC96ads_A100_v4, Standard_ND96asr_v4, Standard_ND96amsr_A100_v4, Standard_NC40ads_H100_v5, Standard_NC80adis_H100_v5, Standard_ND96isr_H100_v5
   languages: en
   SharedComputeCapacityEnabled: true
 
@@ -52,7 +52,10 @@ tags:
       Standard_NC48ads_A100_v4,
       Standard_NC96ads_A100_v4,
       Standard_ND96asr_v4,
-      Standard_ND96amsr_A100_v4
+      Standard_ND96amsr_A100_v4,
+      Standard_NC40ads_H100_v5,
+      Standard_NC80adis_H100_v5,
+      Standard_ND96isr_H100_v5
     ]
   model_specific_defaults:
     apply_deepspeed: "true"

diff --git a/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu-ptca/context/Dockerfile b/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu-ptca/context/Dockerfile
@@ -1,4 +1,4 @@
-FROM mcr.microsoft.com/aifx/acpt/stable-ubuntu2004-cu121-py310-torch222:{{latest-image-tag}}
+FROM mcr.microsoft.com/aifx/acpt/stable-ubuntu2204-cu118-py310-torch222:{{latest-image-tag}}
 
 USER root:root
 
@@ -15,7 +15,7 @@ RUN pip install torch-ort==1.18.0 && TORCH_CUDA_ARCH_LIST="5.2;6.0;7.0;8.0;8.6;9
 
 RUN pip uninstall -y onnxruntime
 
-RUN pip install transformers==4.38.2
+RUN pip install transformers==4.48.0
 RUN pip install optimum==1.21.4
 RUN pip install accelerate==0.33.0
 RUN pip install deepspeed~=0.15.1

diff --git a/...s/training/automl/environments/ai-ml-automl-dnn-text-gpu-ptca/tests/automl_sample_test.py b/...s/training/automl/environments/ai-ml-automl-dnn-text-gpu-ptca/tests/automl_sample_test.py
@@ -9,7 +9,7 @@
 import time
 from pathlib import Path
 from azure.ai.ml import command, Input, MLClient
-from azure.ai.ml._restclient.models import JobStatus
+from azure.ai.ml.operations._run_history_constants import JobStatus
 from azure.ai.ml.entities import Environment, BuildContext
 from azure.identity import AzureCliCredential
 
@@ -54,7 +54,7 @@ def test_azure_ai_ml_automl():
         compute=os.environ.get("gpu_cluster"),
         display_name="sklearn-diabetes-example",
         description="A test run of the ai-ml-automl-dnn-text-gpu-ptca curated environment",
-        experiment_name="sklearnExperiment"
+        experiment_name="sklearnExperiment",
     )
 
     returned_job = ml_client.create_or_update(job)
@@ -70,9 +70,11 @@ def test_azure_ai_ml_automl():
         time.sleep(30)  # sleep 30 seconds
     else:
         # Timeout
-        ml_client.jobs.cancel(returned_job.name)
-        raise Exception(f"Test aborted because the job took longer than {TIMEOUT_MINUTES} minutes. "
-                        f"Last status was {status}.")
+        ml_client.jobs.begin_cancel(returned_job.name)
+        raise Exception(
+            f"Test aborted because the job took longer than {TIMEOUT_MINUTES} minutes. "
+            f"Last status was {status}."
+        )
 
     if status == JobStatus.FAILED:
         ml_client.jobs.download(returned_job.name)

diff --git a/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu-ptca/tests/requirements.txt b/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu-ptca/tests/requirements.txt
@@ -1,2 +1,2 @@
-azure-ai-ml==1.2.0
+azure-ai-ml==1.25.0
 azure.identity==1.10.0
diff --git a/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu/context/Dockerfile b/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu/context/Dockerfile
@@ -1,4 +1,4 @@
-FROM mcr.microsoft.com/aifx/acpt/stable-ubuntu2004-cu121-py310-torch222:{{latest-image-tag}}
+FROM mcr.microsoft.com/aifx/acpt/stable-ubuntu2204-cu118-py310-torch222:{{latest-image-tag}}
 
 ENV AZUREML_CONDA_ENVIRONMENT_PATH /azureml-envs/azureml-automl-dnn-text-gpu
 # Prepend path to AzureML conda environment
@@ -62,7 +62,7 @@ RUN HOROVOD_WITH_PYTORCH=1 pip install --no-cache-dir git+https://github.com/hor
 # by fixing dependencies in the base packages
 RUN pip list && \
     pip install pyarrow==14.0.2 \
-                'transformers[sentencepiece,torch]==4.37.2' \
+                'transformers[sentencepiece,torch]==4.48.0' \
                 aiohttp==3.10.2
 
 # end pip install

diff --git a/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu/tests/automl_sample_test.py b/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu/tests/automl_sample_test.py
@@ -9,7 +9,7 @@
 import time
 from pathlib import Path
 from azure.ai.ml import command, Input, MLClient
-from azure.ai.ml._restclient.models import JobStatus
+from azure.ai.ml.operations._run_history_constants import JobStatus
 from azure.ai.ml.entities import Environment, BuildContext
 from azure.identity import AzureCliCredential
 
@@ -54,7 +54,7 @@ def test_azure_ai_ml_automl():
         compute=os.environ.get("gpu_cluster"),
         display_name="sklearn-diabetes-example",
         description="A test run of the ai-ml-automl-dnn-text-gpu curated environment",
-        experiment_name="sklearnExperiment"
+        experiment_name="sklearnExperiment",
     )
 
     returned_job = ml_client.create_or_update(job)
@@ -70,9 +70,11 @@ def test_azure_ai_ml_automl():
         time.sleep(30)  # sleep 30 seconds
     else:
         # Timeout
-        ml_client.jobs.cancel(returned_job.name)
-        raise Exception(f"Test aborted because the job took longer than {TIMEOUT_MINUTES} minutes. "
-                        f"Last status was {status}.")
+        ml_client.jobs.begin_cancel(returned_job.name)
+        raise Exception(
+            f"Test aborted because the job took longer than {TIMEOUT_MINUTES} minutes. "
+            f"Last status was {status}."
+        )
 
     if status == JobStatus.FAILED:
         ml_client.jobs.download(returned_job.name)

diff --git a/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu/tests/requirements.txt b/assets/training/automl/environments/ai-ml-automl-dnn-text-gpu/tests/requirements.txt
@@ -1,2 +1,2 @@
-azure-ai-ml==1.2.0
+azure-ai-ml==1.25.0
 azure.identity==1.10.0
diff --git a/assets/training/forecasting_demand/environments/automl-gpu/context/Dockerfile b/assets/training/forecasting_demand/environments/automl-gpu/context/Dockerfile
@@ -1,4 +1,4 @@
-FROM mcr.microsoft.com/azureml/openmpi4.1.0-cuda11.6-cudnn8-ubuntu20.04:{{latest-image-tag}}
+FROM mcr.microsoft.com/azureml/openmpi4.1.0-cuda11.8-cudnn8-ubuntu22.04:{{latest-image-tag}}
 
 ENV AZUREML_CONDA_ENVIRONMENT_PATH /azureml-envs/automl
 # Prepend path to AzureML conda environment

diff --git a/assets/training/forecasting_demand/environments/automl-gpu/context/conda_dependencies.yaml b/assets/training/forecasting_demand/environments/automl-gpu/context/conda_dependencies.yaml
@@ -8,13 +8,13 @@ dependencies:
 - numpy~=1.22.3
 - pandas~=1.5.3
 - py-xgboost=1.3.3
-- pyopenssl=24.2.1
-- cryptography=42.0.2
+- pyopenssl=25.0.0
+- cryptography=44.0.1
 - 'psutil>=5.2.2,<6.0.0'
 - tqdm
 - setuptools=72.1.0
 - wheel=0.44.0
-- openssl=3.0.15
+- openssl=3.4.0
 - pip:
   - inference-schema
   - azureml-core=={{latest-pypi-version}}

diff --git a/assets/training/model_management/environments/model-management/environment.yaml b/assets/training/model_management/environments/model-management/environment.yaml
@@ -9,4 +9,4 @@ image:
     - requirements.txt
   publish:
     location: mcr
-    visibility: unlisted
+    visibility: public