tensorflow/tensorflow repository overview

⁠TensorFlow Runtime Docker Images

These containers are a quick way to run or try TensorFlow. The source is available on GitHub⁠. For building TensorFlow or extensions for TensorFlow, please see the TensorFlow Build Dockerfiles⁠.

These images are based on TensorFlow's official Python binaries, which require a CPU with AVX support. Most modern CPUs do support AVX, so it's unlikely that you will have a problem with this. See also https://github.com/tensorflow/tensorflow/issues/19584⁠

The tags described below are accurate for all releases starting with TF 1.13. Older releases are still tagged using the older format and images. See the full list of tags⁠ for the available images.

⁠Base Image Tags

Images built after Sept 2021 are based on Ubuntu 20.04. Earlier images are based on Ubuntu 18.04 or 16.04.

• 1.xx-, latest-, and nightly- tags come with TensorFlow pre-installed. Versioned tags contain their version, the latest- tags contain the latest release (excluding pre-releases like release candidates, alphas, and betas), and the nightly images come with the latest TensorFlow nightly Python package.
• devel and custom-op tags are no longer supported. Please use the TensorFlow SIG Build Dockerfiles⁠ instead.

⁠Optional Features

• Versioned images <= 1.15.0 (1.x) and <= 2.1.0 (2.x) have Python 3 (3.5 for Ubuntu 16-based images; 3.6 for Ubuntu 18-based images; 3.8 for Ubuntu 20-based images) in images tagged "-py3" and Python 2.7 in images without "py" in the tag. All newer images are Python 3 only. Tags containing -py3 are deprecated.
• -gpu tags are based on Nvidia CUDA⁠. You need nvidia-docker⁠ to run them. NOTE: GPU versions of TensorFlow 1.13 and above (this includes the latest- tags) require an NVidia driver that supports CUDA 10. See NVidia's support matrix⁠.
• -jupyter tags include Jupyter and some TensorFlow tutorial notebooks.. They start a Jupyter notebook server on boot. Mount a volume to /tf/notebooks to work on your own notebooks.

⁠Running Containers

$ docker run -it --rm tensorflow/tensorflow bash

Start a CPU-only container

$ docker run -it --rm --runtime=nvidia tensorflow/tensorflow:latest-gpu python

Start a GPU container, using the Python interpreter.

$ docker run -it --rm -v $(realpath ~/notebooks):/tf/notebooks -p 8888:8888 tensorflow/tensorflow:latest-jupyter

Run a Jupyter notebook server with your own notebook directory (assumed here to be ~/notebooks). To use it, navigate to localhost:8888 in your browser.