1.8.0 image

Dockerfile.gpu

@@ -6,7 +6,11 @@ WORKDIR /usr/src
 
 RUN python3 setup.py sdist
 
-FROM tensorflow-base:1.4.1-gpu-py3
+FROM tensorflow/tensorflow:1.8.0-gpu-py3
+
+#tensorflow-serving-api-python3==1.7.0 
+RUN pip3 list && pip3 install numpy boto3 six awscli flask==0.11 Jinja2==2.9 gevent gunicorn keras==2.1.3 pillow h5py \
+    && pip3 list
 
 WORKDIR /root
 

Dockerfile_base.gpu

@@ -1,29 +0,0 @@
-FROM nvidia/cuda:8.0-cudnn6-devel-ubuntu16.04
-
-RUN apt-get update && apt-get install -y --no-install-recommends \
-    python3-pip python3-dev python3-setuptools \
-    && \
-    apt-get clean && \
-    rm -rf /var/lib/apt/lists/* \
-    && pip3 install tensorflow-gpu==1.4.1
-
-RUN pip3 list && pip3 install numpy boto3 six awscli flask==0.11 Jinja2==2.9 gevent gunicorn keras==2.1.3 pillow h5py \
-    && pip3 list
-
-# Configure the build for our CUDA configuration.
-ENV LD_LIBRARY_PATH=/usr/local/cuda/lib64/stubs:$LD_LIBRARY_PATH
-ENV CI_BUILD_PYTHON=python \
-    LD_LIBRARY_PATH=/usr/local/cuda/extras/CUPTI/lib64:$LD_LIBRARY_PATH \
-    TF_NEED_CUDA=1 \
-    TF_CUDA_VERSION=8.0 \
-    TF_CUDNN_VERSION=6 \
-    TF_CUDA_COMPUTE_CAPABILITIES=3.7,6.1
-
-# Fix paths so that CUDNN can be found
-# See https://github.com/tensorflow/tensorflow/issues/8264
-RUN ls -lah /usr/local/cuda/lib64/*
-RUN mkdir /usr/lib/x86_64-linux-gnu/include/ && \
-  ln -s /usr/lib/x86_64-linux-gnu/include/cudnn.h /usr/lib/x86_64-linux-gnu/include/cudnn.h && \
-  ln -s /usr/include/cudnn.h /usr/local/cuda/include/cudnn.h && \
-  ln -s /usr/lib/x86_64-linux-gnu/libcudnn.so /usr/local/cuda/lib64/libcudnn.so && \
-  ln -s /usr/lib/x86_64-linux-gnu/libcudnn.so.6 /usr/local/cuda/lib64/libcudnn.so.6

Readme.md

@@ -4,16 +4,10 @@ Run DIY Robocars model training as Sagemaker (https://aws.amazon.com/fr/sagemake
 
 # Build images
 
-- Build base image:
-
-``` 
-docker build -t robocars-base:1.4.1-gpu-py3 -f Dockerfile_base.gpu .
-```
-
 - Build model image:
 
 ``` 
-docker build -t robocars:1.4.1-gpu-py3 -f Dockerfile.gpu .
+docker build -t robocars:1.8.0-gpu-py3 -f Dockerfile.gpu .
 ```
 
 # Prepare training (once)
@@ -22,9 +16,9 @@ docker build -t robocars:1.4.1-gpu-py3 -f Dockerfile.gpu .
 - Create an AWS docker registry and push your model image to it. Docker hub registry is not supported
 
 ``` 
-docker tag robocars:1.4.1-gpu-py <replace_me>.dkr.ecr.eu-west-1.amazonaws.com/robocars:1.4.1-gpu-py3
+docker tag robocars:1.8.0-gpu-py <replace_me>.dkr.ecr.eu-west-1.amazonaws.com/robocars:1.8.0-gpu-py3
 # you should have AWS SDK installed and login to docker
-docker push <replace_me>.dkr.ecr.eu-west-1.amazonaws.com/robocars:1.4.1-gpu-py3
+docker push <replace_me>.dkr.ecr.eu-west-1.amazonaws.com/robocars:1.8.0-gpu-py3
 ``` 
 
 # Run training
@@ -47,7 +41,7 @@ echo 'Creating training job '$1
 aws sagemaker create-training-job \
     --training-job-name $job_name \
     --hyper-parameters '{ "sagemaker_region": "\"eu-west-1\"", "with_slide": "true" }' \
-    --algorithm-specification TrainingImage="<replace_me>.dkr.ecr.eu-west-1.amazonaws.com/robocars:1.4.1-gpu-py3",TrainingInputMode=File \
+    --algorithm-specification TrainingImage="<replace_me>.dkr.ecr.eu-west-1.amazonaws.com/robocars:1.8.0-gpu-py3",TrainingInputMode=File \
     --role-arn "<your_iam_sagemaker_role>" \
     --input-data-config '[{ "ChannelName": "train", "DataSource": { "S3DataSource": { "S3DataType": "S3Prefix", "S3Uri": "s3://<your_input_bucket>", "S3DataDistributionType": "FullyReplicated" }} }]' \
     --output-data-config S3OutputPath=s3://<your_output_bucket> \

build_base_gpu.sh

@@ -1 +0,0 @@
-docker build -t tensorflow-base:1.4.1-gpu-py3 -f Dockerfile_base_pip.gpu .

build_gpu.sh

@@ -1 +1 @@
-docker build -t tensorflow:1.4.1-gpu-py3 -f Dockerfile.gpu .
+docker build -t tensorflow:1.8.0-gpu-py3 -f Dockerfile.gpu .

create_job.sh

@@ -8,7 +8,7 @@ then
 fi 
 echo 'Creating training job '$1
 
-training_image="<replace_me>.dkr.ecr.eu-west-1.amazonaws.com/robocars:1.4.1-gpu-py3"
+training_image="<replace_me>.dkr.ecr.eu-west-1.amazonaws.com/robocars:1.8.0-gpu-py3"
 iam_role_arn="arn:aws:iam::<replace_me>:role/service-role/<replace_me>"
 
 aws sagemaker create-training-job \