[TensorFlow][Training][Sagemaker] TensorFlow 2.19.0 Currency Release

dlc_developer_config.toml

@@ -37,12 +37,12 @@ deep_canary_mode = false
 [build]
 # Add in frameworks you would like to build. By default, builds are disabled unless you specify building an image.
 # available frameworks - ["base", "vllm", "autogluon", "huggingface_tensorflow", "huggingface_pytorch", "huggingface_tensorflow_trcomp", "huggingface_pytorch_trcomp", "pytorch_trcomp", "tensorflow", "pytorch", "stabilityai_pytorch"]
-build_frameworks = []
+build_frameworks = ["tensorflow"]
 
 
 # By default we build both training and inference containers. Set true/false values to determine which to build.
 build_training = true
-build_inference = true
+build_inference = false
 
 # Set do_build to "false" to skip builds and test the latest image built by this PR
 # Note: at least one build is required to set do_build to "false"
@@ -120,7 +120,7 @@ use_scheduler = false
 
 # Standard Framework Training
 dlc-pr-pytorch-training = ""
-dlc-pr-tensorflow-2-training = ""
+dlc-pr-tensorflow-2-training = "tensorflow/training/buildspec-2-19-sm.yml"
 dlc-pr-autogluon-training = ""
 
 # ARM64 Training
@@ -176,4 +176,4 @@ dlc-pr-stabilityai-pytorch-inference = ""
 
 # EIA Inference
 dlc-pr-pytorch-eia-inference = ""
-dlc-pr-tensorflow-2-eia-inference = ""
+dlc-pr-tensorflow-2-eia-inference = ""

tensorflow/training/buildspec-2-19-sm.yml

@@ -0,0 +1,68 @@
+account_id: &ACCOUNT_ID <set-$ACCOUNT_ID-in-environment>
+prod_account_id: &PROD_ACCOUNT_ID 763104351884
+region: &REGION <set-$REGION-in-environment>
+framework: &FRAMEWORK tensorflow
+version: &VERSION 2.19.0
+short_version: &SHORT_VERSION "2.19"
+arch_type: x86
+#autopatch_build: "True"
+
+repository_info:
+  training_repository: &TRAINING_REPOSITORY
+    image_type: &TRAINING_IMAGE_TYPE training
+    root: !join [ *FRAMEWORK, "/", *TRAINING_IMAGE_TYPE ]
+    repository_name: &REPOSITORY_NAME !join [pr, "-", *FRAMEWORK, "-", *TRAINING_IMAGE_TYPE]
+    repository: &REPOSITORY !join [ *ACCOUNT_ID, .dkr.ecr., *REGION, .amazonaws.com/, *REPOSITORY_NAME ]
+    release_repository_name: &RELEASE_REPOSITORY_NAME !join [ *FRAMEWORK, "-", *TRAINING_IMAGE_TYPE ]
+    release_repository: &RELEASE_REPOSITORY !join [ *PROD_ACCOUNT_ID, .dkr.ecr., *REGION, .amazonaws.com/,
+                                                    *RELEASE_REPOSITORY_NAME ]
+
+context:
+  training_context: &TRAINING_CONTEXT
+    start_cuda_compat:
+      source: docker/build_artifacts/start_cuda_compat.sh
+      target: start_cuda_compat.sh
+    dockerd-entrypoint:
+      source: docker/build_artifacts/dockerd-entrypoint.py
+      target: dockerd-entrypoint.py
+    dockerd_ec2_entrypoint:
+      source: docker/build_artifacts/dockerd_ec2_entrypoint.sh
+      target: dockerd_ec2_entrypoint.sh
+    deep_learning_container:
+      source: ../../src/deep_learning_container.py
+      target: deep_learning_container.py
+
+images:
+  BuildTensorflowSageMakerCpuPy310TrainingDockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &TENSORFLOW_CPU_TRAINING_PY3 false
+    image_size_baseline: &IMAGE_SIZE_BASELINE 7500
+    device_type: &DEVICE_TYPE cpu
+    python_version: &DOCKER_PYTHON_VERSION py3
+    tag_python_version: &TAG_PYTHON_VERSION py312
+    os_version: &OS_VERSION ubuntu22.04
+    tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *OS_VERSION, "-sagemaker" ]
+    latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *OS_VERSION, "-sagemaker" ]
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /Dockerfile., *DEVICE_TYPE ]
+    # build_tag_override: "pr:2.16.2-cpu-py310-ubuntu20.04-sagemaker-pr-4362-autopatch"
+    target: sagemaker
+    enable_test_promotion: true
+    context:
+      <<: *TRAINING_CONTEXT
+  BuildTensorflowSageMakerGpuPy310Cu125TrainingDockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &TENSORFLOW_GPU_TRAINING_PY3 false
+    image_size_baseline: &IMAGE_SIZE_BASELINE 11998
+    device_type: &DEVICE_TYPE gpu
+    python_version: &DOCKER_PYTHON_VERSION py3
+    tag_python_version: &TAG_PYTHON_VERSION py312
+    cuda_version: &CUDA_VERSION cu125
+    os_version: &OS_VERSION ubuntu22.04
+    tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-", *OS_VERSION, "-sagemaker" ]
+    latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-", *OS_VERSION, "-sagemaker" ]
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /, *CUDA_VERSION, /Dockerfile., *DEVICE_TYPE ]
+    # build_tag_override: "pr:2.16.2-gpu-py310-cu123-ubuntu20.04-sagemaker-pr-4362-autopatch"
+    target: sagemaker
+    enable_test_promotion: true
+    context:
+      <<: *TRAINING_CONTEXT

tensorflow/training/buildspec.yml

@@ -1 +1 @@
-buildspec_pointer: buildspec-2-18-sm.yml
+buildspec_pointer: buildspec-2-19-sm.yml