[PyTorch][Training][EC2][SageMaker]PyTorch 2.7.0 Currency Release

dlc_developer_config.toml

@@ -36,13 +36,14 @@ 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", "autogluon", "huggingface_tensorflow", "huggingface_pytorch", "huggingface_tensorflow_trcomp", "huggingface_pytorch_trcomp", "pytorch_trcomp", "tensorflow", "pytorch", "stabilityai_pytorch"]
-build_frameworks = []
+build_frameworks = ["pytorch"]
 
 
 # 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"
@@ -104,7 +105,7 @@ use_scheduler = false
 ### TRAINING PR JOBS ###
 
 # Standard Framework Training
-dlc-pr-pytorch-training = ""
+dlc-pr-pytorch-training = "pytorch/training/buildspec-2-7-sm.yml"
 dlc-pr-tensorflow-2-training = ""
 dlc-pr-autogluon-training = ""
 

miscellaneous_scripts/dlc_template.py

@@ -1,14 +1,31 @@
-import os
-
-try:
-    if os.path.exists("/usr/local/bin/deep_learning_container.py") and (
-        os.getenv("OPT_OUT_TRACKING") is None or os.getenv("OPT_OUT_TRACKING", "").lower() != "true"
-    ):
-        import threading
-
-        cmd = "python /usr/local/bin/deep_learning_container.py --framework {FRAMEWORK} --framework-version {FRAMEWORK_VERSION} --container-type {CONTAINER_TYPE} &>/dev/null"
-        x = threading.Thread(target=lambda: os.system(cmd))
-        x.setDaemon(True)
-        x.start()
-except Exception:
-    pass
+def main():
+    import os
+
+    if os.getenv("OPT_OUT_TRACKING", "").lower() == "true":
+        return
+
+    try:
+        if os.path.exists("/usr/local/bin/deep_learning_container.py"):
+            import sys
+            import subprocess
+
+            subprocess.Popen(
+                [
+                    sys.executable,
+                    "/usr/local/bin/deep_learning_container.py",
+                    "--framework",
+                    "{FRAMEWORK}",
+                    "--framework-version",
+                    "{FRAMEWORK_VERSION}",
+                    "--container-type",
+                    "{CONTAINER_TYPE}",
+                ],
+                stdout=subprocess.DEVNULL,
+                stderr=subprocess.DEVNULL,
+                start_new_session=True,
+            )
+    except:
+        pass
+
+
+main()

pytorch/training/buildspec-2-7-ec2.yml

@@ -0,0 +1,72 @@
+account_id: &ACCOUNT_ID <set-$ACCOUNT_ID-in-environment>
+prod_account_id: &PROD_ACCOUNT_ID 763104351884
+region: &REGION <set-$REGION-in-environment>
+framework: &FRAMEWORK pytorch
+version: &VERSION 2.7.0
+short_version: &SHORT_VERSION "2.7"
+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.sh
+      target: dockerd_entrypoint.sh
+    changehostname:
+      source: docker/build_artifacts/changehostname.c
+      target: changehostname.c
+    start_with_right_hostname:
+      source: docker/build_artifacts/start_with_right_hostname.sh
+      target: start_with_right_hostname.sh
+    example_mnist_file:
+      source: docker/build_artifacts/mnist.py
+      target: mnist.py
+    deep_learning_container:
+      source: ../../src/deep_learning_container.py
+      target: deep_learning_container.py
+
+images:
+  BuildEC2CPUPTTrainPy3DockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &PYTORCH_CPU_TRAINING_PY3 false
+    image_size_baseline: 6500
+    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, "-ec2" ]
+    latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *OS_VERSION, "-ec2" ]
+    # build_tag_override: "True"
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /Dockerfile., *DEVICE_TYPE ]
+    target: ec2
+    context:
+      <<: *TRAINING_CONTEXT
+  BuildEC2GPUPTTrainPy3cu128DockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &PYTORCH_GPU_TRAINING_PY3 false
+    image_size_baseline: 24000
+    device_type: &DEVICE_TYPE gpu
+    python_version: &DOCKER_PYTHON_VERSION py3
+    tag_python_version: &TAG_PYTHON_VERSION py312
+    cuda_version: &CUDA_VERSION cu128
+    os_version: &OS_VERSION ubuntu22.04
+    tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-", *OS_VERSION, "-ec2" ]
+    latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-", *OS_VERSION, "-ec2" ]
+    # build_tag_override: "True"
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /, *CUDA_VERSION, /Dockerfile.,
+                         *DEVICE_TYPE ]
+    target: ec2
+    context:
+      <<: *TRAINING_CONTEXT

pytorch/training/buildspec-2-7-sm.yml

@@ -0,0 +1,72 @@
+account_id: &ACCOUNT_ID <set-$ACCOUNT_ID-in-environment>
+prod_account_id: &PROD_ACCOUNT_ID 763104351884
+region: &REGION <set-$REGION-in-environment>
+framework: &FRAMEWORK pytorch
+version: &VERSION 2.7.0
+short_version: &SHORT_VERSION "2.7"
+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.sh
+      target: dockerd_entrypoint.sh
+    changehostname:
+      source: docker/build_artifacts/changehostname.c
+      target: changehostname.c
+    start_with_right_hostname:
+      source: docker/build_artifacts/start_with_right_hostname.sh
+      target: start_with_right_hostname.sh
+    example_mnist_file:
+      source: docker/build_artifacts/mnist.py
+      target: mnist.py
+    deep_learning_container:
+      source: ../../src/deep_learning_container.py
+      target: deep_learning_container.py
+
+images:
+  BuildSageMakerCPUPTTrainPy3DockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &PYTORCH_CPU_TRAINING_PY3 false
+    image_size_baseline: 6200
+    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" ]
+    # build_tag_override: "True"
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /Dockerfile., *DEVICE_TYPE ]
+    target: sagemaker
+    context:
+      <<: *TRAINING_CONTEXT
+  BuildSageMakerGPUPTTrainPy3DockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &PYTORCH_GPU_TRAINING_PY3 false
+    image_size_baseline: 24000
+    device_type: &DEVICE_TYPE gpu
+    python_version: &DOCKER_PYTHON_VERSION py3
+    tag_python_version: &TAG_PYTHON_VERSION py312
+    cuda_version: &CUDA_VERSION cu128
+    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" ]
+    # build_tag_override: "True"
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /, *CUDA_VERSION, /Dockerfile.,
+                         *DEVICE_TYPE ]
+    target: sagemaker
+    context:
+      <<: *TRAINING_CONTEXT

pytorch/training/buildspec.yml

@@ -1 +1 @@
-buildspec_pointer: buildspec-2-6-sm.yml
+buildspec_pointer: buildspec-2-7-ec2.yml