jeongukjae / tf-serving-s3

Custom TensorFlow Serving image with AWS S3 filesystem support

tf-serving-s3

Custom TensorFlow Serving image built with TensorFlow IO for AWS S3 filesystem support.

Check GitHub Packages for the Docker images.

Usage

The usage is exactly the same as the official TensorFlow Serving image. For more details, please refer to the TensorFlow Serving with Docker. But you need to use the image from GitHub Packages instead of the official image.

$ docker run \
    -p 8500:8500 \
    -p 8501:8501 \
    -e AWS_ACCESS_KEY_ID \
    -e AWS_SECRET_ACCESS_KEY \
    -e AWS_DEFAULT_REGION \
    -e MODEL_BASE_PATH=s3://BUCKET_NAME/PATH_TO_MODEL \
    -e MODEL_NAME=MODEL_NAME \
    ghcr.io/jeongukjae/tf-serving-s3:2.11.0

What is the differences/details?

Check this blog post: https://blog.ukjae.io/posts/enabling-s3-filesystem-support-for-tensorflow-serving/

How to build the image?

docker build -t ghcr.io/jeongukjae/tf-serving-s3:2.11.0-devel -f Dockerfile.devel .
docker build -t ghcr.io/jeongukjae/tf-serving-s3:2.11.0 -f Dockerfile .