Skip to content

Latest commit

 

History

History

pytorch

README.md

pytorch

CONTAINERS IMAGES RUN BUILD

Containers for PyTorch with CUDA support. Note that the l4t-pytorch containers also include PyTorch, torchvision, and torchaudio.

CONTAINERS
pytorch:2.0
   Aliases torch:2.0
   Builds pytorch-20_jp51
   Requires L4T ['==35.*']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dependants torchaudio:2.0.1 torchaudio:2.0.1-builder torchvision:0.15.1
   Dockerfile Dockerfile.pip
   Images dustynv/pytorch:2.0-r35.2.1 (2023-12-06, 5.4GB)
dustynv/pytorch:2.0-r35.3.1 (2023-12-14, 5.4GB)
dustynv/pytorch:2.0-r35.4.1 (2023-10-07, 5.4GB)
pytorch:2.0-builder
   Aliases torch:2.0-builder
   Requires L4T ['==35.*']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dockerfile Dockerfile.pip
pytorch:2.1
   Aliases torch:2.1
   Builds pytorch-21_jp60 pytorch-21_jp51
   Requires L4T ['>=35']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dependants torchaudio:2.1.0 torchaudio:2.1.0-builder torchvision:0.16.2
   Dockerfile Dockerfile.pip
   Images dustynv/pytorch:2.1-r35.2.1 (2023-12-11, 5.4GB)
dustynv/pytorch:2.1-r35.3.1 (2023-12-14, 5.4GB)
dustynv/pytorch:2.1-r35.4.1 (2023-11-05, 5.4GB)
dustynv/pytorch:2.1-r36.2.0 (2023-12-14, 7.2GB)
pytorch:2.1-builder
   Aliases torch:2.1-builder
   Requires L4T ['>=35']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dockerfile Dockerfile.pip
pytorch:2.2
   Aliases torch:2.2 pytorch torch
   Requires L4T ['>=35']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dependants audiocraft auto_awq:0.2.4 auto_gptq:0.7.1 awq:0.1.0 bitsandbytes bitsandbytes:builder efficientvit exllama:0.0.14 exllama:0.0.15 faiss_lite flash-attention:2.5.6 flash-attention:2.5.6-builder flash-attention:2.5.7 flash-attention:2.5.7-builder gptq-for-llama jetson-inference l4t-diffusion l4t-ml l4t-pytorch langchain langchain:samples llama-index llava minigpt4 mlc:0.1.0 mlc:0.1.0-builder mlc:0.1.1 mlc:0.1.1-builder nanodb nanoowl nanosam nemo openai-triton openai-triton:builder optimum raft sam stable-diffusion stable-diffusion-webui tam tensorrt_llm:0.10.dev0 tensorrt_llm:0.10.dev0-builder tensorrt_llm:0.5 tensorrt_llm:0.5-builder text-generation-inference text-generation-webui:1.7 text-generation-webui:6a7cd01 text-generation-webui:main torch2trt torch_tensorrt torchaudio:2.2.2 torchaudio:2.2.2-builder torchvision:0.17.2 transformers transformers:git transformers:nvgpt tvm voicecraft whisper whisperx xformers:0.0.26 xformers:0.0.26-builder xtts
   Dockerfile Dockerfile.pip
pytorch:2.2-builder
   Aliases torch:2.2-builder pytorch:builder torch:builder
   Requires L4T ['>=35']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dockerfile Dockerfile.pip
pytorch:2.3
   Aliases torch:2.3
   Requires L4T ['==36.*']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dependants torchaudio:2.3.0 torchaudio:2.3.0-builder torchvision:0.18.0
   Dockerfile Dockerfile.pip
pytorch:2.3-builder
   Aliases torch:2.3-builder
   Requires L4T ['==36.*']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dockerfile Dockerfile.pip
pytorch:1.10
   Aliases torch:1.10
   Builds pytorch-110_jp46
   Requires L4T ['==32.*']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dependants torchaudio:0.10.0 torchaudio:0.10.0-builder torchvision:0.11.1
   Dockerfile Dockerfile
   Images dustynv/pytorch:1.10-r32.7.1 (2023-12-14, 1.1GB)
pytorch:1.9
   Aliases torch:1.9
   Builds pytorch-19_jp46
   Requires L4T ['==32.*']
   Dependencies build-essential cuda cudnn python numpy cmake onnx
   Dependants torchaudio:0.9.0 torchaudio:0.9.0-builder torchvision:0.10.0
   Dockerfile Dockerfile
   Images dustynv/pytorch:1.9-r32.7.1 (2023-12-14, 1.0GB)
CONTAINER IMAGES
Repository/Tag Date Arch Size
  dustynv/pytorch:1.10-r32.7.1 2023-12-14 arm64 1.1GB
  dustynv/pytorch:1.11-r35.2.1 2023-11-05 arm64 5.4GB
  dustynv/pytorch:1.11-r35.3.1 2023-12-14 arm64 5.4GB
  dustynv/pytorch:1.11-r35.4.1 2023-12-11 arm64 5.4GB
  dustynv/pytorch:1.12-r35.2.1 2023-12-14 arm64 5.5GB
  dustynv/pytorch:1.12-r35.3.1 2023-08-29 arm64 5.5GB
  dustynv/pytorch:1.12-r35.4.1 2023-11-03 arm64 5.5GB
  dustynv/pytorch:1.13-r35.2.1 2023-08-29 arm64 5.5GB
  dustynv/pytorch:1.13-r35.3.1 2023-12-12 arm64 5.5GB
  dustynv/pytorch:1.13-r35.4.1 2023-12-14 arm64 5.5GB
  dustynv/pytorch:1.9-r32.7.1 2023-12-14 arm64 1.0GB
  dustynv/pytorch:2.0-r35.2.1 2023-12-06 arm64 5.4GB
  dustynv/pytorch:2.0-r35.3.1 2023-12-14 arm64 5.4GB
  dustynv/pytorch:2.0-r35.4.1 2023-10-07 arm64 5.4GB
  dustynv/pytorch:2.1-r35.2.1 2023-12-11 arm64 5.4GB
  dustynv/pytorch:2.1-r35.3.1 2023-12-14 arm64 5.4GB
  dustynv/pytorch:2.1-r35.4.1 2023-11-05 arm64 5.4GB
  dustynv/pytorch:2.1-r36.2.0 2023-12-14 arm64 7.2GB

Container images are compatible with other minor versions of JetPack/L4T:
    • L4T R32.7 containers can run on other versions of L4T R32.7 (JetPack 4.6+)
    • L4T R35.x containers can run on other versions of L4T R35.x (JetPack 5.1+)

RUN CONTAINER

To start the container, you can use jetson-containers run and autotag, or manually put together a docker run command:

# automatically pull or build a compatible container image
jetson-containers run $(autotag pytorch)

# or explicitly specify one of the container images above
jetson-containers run dustynv/pytorch:2.1-r36.2.0

# or if using 'docker run' (specify image and mounts/ect)
sudo docker run --runtime nvidia -it --rm --network=host dustynv/pytorch:2.1-r36.2.0

jetson-containers run forwards arguments to docker run with some defaults added (like --runtime nvidia, mounts a /data cache, and detects devices)
autotag finds a container image that's compatible with your version of JetPack/L4T - either locally, pulled from a registry, or by building it.

To mount your own directories into the container, use the -v or --volume flags:

jetson-containers run -v /path/on/host:/path/in/container $(autotag pytorch)

To launch the container running a command, as opposed to an interactive shell:

jetson-containers run $(autotag pytorch) my_app --abc xyz

You can pass any options to it that you would to docker run, and it'll print out the full command that it constructs before executing it.

BUILD CONTAINER

If you use autotag as shown above, it'll ask to build the container for you if needed. To manually build it, first do the system setup, then run:

jetson-containers build pytorch

The dependencies from above will be built into the container, and it'll be tested during. Run it with --help for build options.