Skip to content

Commit 14fde5e

Browse files
LokiiiiiiLokiiiiii
committed
Testing MWMS in TF 2.9.1 with TF Model Garden
1 parent 30fed8f commit 14fde5e

9 files changed

+95
-5
lines changed

‎buildspec-dlc-cpu-tests.yml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@ version: 0.2
22

33
env:
44
variables:
5-
FRAMEWORK_VERSION: '2.7.1'
5+
FRAMEWORK_VERSION: '2.9.1'
66
CPU_INSTANCE_TYPE: 'ml.c4.xlarge'
77
ECR_REPO: 'sagemaker-test'
88

‎buildspec-dlc-gpu-tests.yml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@ version: 0.2
22

33
env:
44
variables:
5-
FRAMEWORK_VERSION: '2.7.1'
5+
FRAMEWORK_VERSION: '2.9.1'
66
GPU_INSTANCE_TYPE: 'ml.p3.2xlarge'
77
ECR_REPO: 'sagemaker-test'
88
GITHUB_REPO: 'sagemaker-tensorflow-training-toolkit'

‎buildspec-gen-cpu-tests.yml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@ version: 0.2
22

33
env:
44
variables:
5-
FRAMEWORK_VERSION: '2.7.1'
5+
FRAMEWORK_VERSION: '2.9.1'
66
CPU_INSTANCE_TYPE: 'ml.c4.xlarge'
77
ECR_REPO: 'sagemaker-test'
88

‎buildspec-gen-gpu-tests.yml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@ version: 0.2
22

33
env:
44
variables:
5-
FRAMEWORK_VERSION: '2.7.1'
5+
FRAMEWORK_VERSION: '2.9.1'
66
GPU_INSTANCE_TYPE: 'ml.p3.16xlarge'
77
ECR_REPO: 'sagemaker-test'
88
GITHUB_REPO: 'sagemaker-tensorflow-training-toolkit'

‎test/container/2.9.1/Dockerfile.dlc.cpu

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,6 @@
1+
ARG region
2+
FROM 763104351884.dkr.ecr.$region.amazonaws.com/tensorflow-training:2.9.1-cpu-py39-ubuntu20.04-sagemaker
3+
4+
COPY dist/sagemaker_tensorflow_training-*.tar.gz /sagemaker_tensorflow_training.tar.gz
5+
RUN pip install --upgrade --no-cache-dir /sagemaker_tensorflow_training.tar.gz && \
6+
rm /sagemaker_tensorflow_training.tar.gz

‎test/container/2.9.1/Dockerfile.dlc.gpu

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,6 @@
1+
ARG region
2+
FROM 763104351884.dkr.ecr.$region.amazonaws.com/tensorflow-training:2.9.1-gpu-py39-cu112-ubuntu20.04-sagemaker
3+
4+
COPY dist/sagemaker_tensorflow_training-*.tar.gz /sagemaker_tensorflow_training.tar.gz
5+
RUN pip install --upgrade --no-cache-dir /sagemaker_tensorflow_training.tar.gz && \
6+
rm /sagemaker_tensorflow_training.tar.gz

‎test/container/2.9.1/Dockerfile.tf.cpu

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,9 @@
1+
FROM tensorflow/tensorflow:2.9.1
2+
3+
ENV SAGEMAKER_TRAINING_MODULE sagemaker_tensorflow_container.training:main
4+
5+
COPY dist/sagemaker_tensorflow_training-*.tar.gz /sagemaker_tensorflow_training.tar.gz
6+
RUN pip install --upgrade --no-cache-dir /sagemaker_tensorflow_training.tar.gz && \
7+
rm /sagemaker_tensorflow_training.tar.gz
8+
RUN pip install --no-cache-dir tensorflow-io
9+
RUN apt-get update && apt-get install -y --no-install-recommends openssh-server && mkdir -p /var/run/sshd

‎test/container/2.9.1/Dockerfile.tf.gpu

Lines changed: 13 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,13 @@
1+
FROM tensorflow/tensorflow:2.9.1-gpu
2+
3+
ENV SAGEMAKER_TRAINING_MODULE sagemaker_tensorflow_container.training:main
4+
5+
COPY dist/sagemaker_tensorflow_training-*.tar.gz /sagemaker_tensorflow_training.tar.gz
6+
RUN pip install --upgrade --no-cache-dir /sagemaker_tensorflow_training.tar.gz && \
7+
rm /sagemaker_tensorflow_training.tar.gz
8+
RUN pip install --no-cache-dir tensorflow-io
9+
RUN apt-key del 7fa2af80 \
10+
&& rm /etc/apt/sources.list.d/nvidia-ml.list \
11+
&& apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub \
12+
&& apt-get update \
13+
&& apt-get install -y --no-install-recommends openssh-server && mkdir -p /var/run/sshd

‎test/integration/sagemaker/test_multi_worker_mirrored.py

Lines changed: 57 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -21,7 +21,9 @@
2121
RESOURCE_PATH = os.path.join(os.path.dirname(__file__), "..", "..", "resources")
2222

2323

24-
def test_multi_node(sagemaker_session, instance_type, image_uri, tmpdir, framework_version, capsys):
24+
def test_keras_example(
25+
sagemaker_session, instance_type, image_uri, tmpdir, framework_version, capsys
26+
):
2527
estimator = TensorFlow(
2628
entry_point=os.path.join(RESOURCE_PATH, "multi_worker_mirrored", "train_dummy.py"),
2729
role="SageMakerRole",
@@ -40,3 +42,57 @@ def test_multi_node(sagemaker_session, instance_type, image_uri, tmpdir, framewo
4042
logs = captured.out + captured.err
4143
assert "Running distributed training job with multi_worker_mirrored_strategy setup" in logs
4244
assert "TF_CONFIG=" in logs
45+
46+
47+
def test_tf_model_garden(
48+
sagemaker_session, instance_type, image_uri, tmpdir, framework_version, capsys
49+
):
50+
epochs = 10
51+
batch_size = 512
52+
train_steps = int(1024 * epochs / batch_size)
53+
steps_per_loop = train_steps // 10
54+
overrides = (
55+
f"runtime.enable_xla=False,"
56+
f"runtime.num_gpus=1,"
57+
f"runtime.distribution_strategy=multi_worker_mirrored,"
58+
f"runtime.mixed_precision_dtype=float16,"
59+
f"task.train_data.global_batch_size={batch_size},"
60+
f"task.train_data.input_path=/opt/ml/input/data/training/validation*,"
61+
f"task.train_data.cache=True,"
62+
f"trainer.train_steps={train_steps},"
63+
f"trainer.steps_per_loop={steps_per_loop},"
64+
f"trainer.summary_interval={steps_per_loop},"
65+
f"trainer.checkpoint_interval={train_steps},"
66+
f"task.model.backbone.type=resnet,"
67+
f"task.model.backbone.resnet.model_id=50"
68+
)
69+
estimator = TensorFlow(
70+
git_config={
71+
"repo": "https://github.com/tensorflow/models.git",
72+
"branch": "v2.9.2",
73+
},
74+
source_dir=".",
75+
entry_point="official/vision/train.py",
76+
model_dir=False,
77+
instance_type=instance_type,
78+
instance_count=2,
79+
image_uri=image_uri,
80+
hyperparameters={
81+
"sagemaker_multi_worker_mirrored_strategy_enabled": True,
82+
"experiment": "resnet_imagenet",
83+
"config_file": "official/vision/configs/experiments/image_classification/imagenet_resnet50_gpu.yaml",
84+
"mode": "train",
85+
"model_dir": "/opt/ml/model",
86+
"params_override": overrides,
87+
},
88+
max_run=60 * 60 * 1, # 1 hour
89+
role="SageMakerRole",
90+
)
91+
estimator.fit(
92+
inputs="s3://collection-of-ml-datasets/Imagenet/TFRecords/validation",
93+
job_name=unique_name_from_base("test-tf-mwms"),
94+
)
95+
captured = capsys.readouterr()
96+
logs = captured.out + captured.err
97+
assert "Running distributed training job with multi_worker_mirrored_strategy setup" in logs
98+
assert "TF_CONFIG=" in logs

0 commit comments

Comments
 (0)