Deep Learning VM Images is a set ofvirtual machine images optimized for data science and machinelearning tasks. All images come with key ML frameworks and toolspre-installed. You can use them out of the box on instances withGPUs to accelerate your data processing tasks.

Deep Learning VM images are available to support many combinationsof framework and processor. There are currently images supportingTensorFlow Enterprise,TensorFlow, PyTorch, and generic high-performance computing,with versions for both CPU-only and GPU-enabled workflows.

To see a list of frameworks available, seeChoosing animage.

Pre-installed packages

Images are based on the Debian 11 and Ubuntu 22.04operating systems, and these images can beconfigured to include the following:

• Specific frameworks (for example, TensorFlow) andsupporting packages.

• Python 3.10 with the following packages:

  • numpy
  • scipy
  • matplotlib
  • pandas
  • nltk
  • pillow
  • scikit-image
  • opencv-python
  • scikit-learn
  • many more

• JupyterLab notebook environments for quick prototyping

• Nvidia packages with the latest Nvidia driver for GPU-enabled instances:

  • CUDA 11.x and 12.x (the version depends on the framework)
  • CuDNN 7.x and NCCL 2.x (the version depends on the CUDA version)

Updates

Deep Learning VM images are updated regularly with bug fixesand package updates. Check therelease notesfor information about updates.

Community support

Ask a question about Deep Learning VM onStackOverflow or join thegoogle-dl-platformGoogle group to discuss Deep Learning VM.

Learn more about getting support from thecommunity.

What's next

To get started using Deep Learning VM, create a new instanceusing the Cloud Marketplaceorusing the command line.