Repository files navigation

Tensorflow GPU with Docker in WSL2 using VS Code Dev Containers

• Windows 11
• Windows Subsystem for Linux (WSL2)
• LatestNVIDIA Windows GPU Driver
• Docker Desktop WSL2
• Visual Studio Code withDev Containers extension

• Create a new repository from the template or clone the code in WSL
• InDockerfile:
  • Update theTensorFlow Docker image: runnvidia-smi command in your Terminal, and get the CUDA version. Then find the Tensorflow version compatible with the CUDA versionhere
  • UpdateUSERNAMEARG to create a non-root user
  • UpdateUSER_UID andUSER_GIDARGs (useid command in your Terminal to get them)
  • Update required packages (optional)
• Update pip packages inrequirements.txt (optional)
• Open your project in WSL with VS Code and reopen in Container: (F1) use theDev Containers: Reopen in Container command

Runpyhton main.py to test it.