openai-triton

openai-triton

CONTAINERS IMAGES RUN BUILD

CONTAINERS

openai-triton
Requires	L4T >=35
Dependencies	build-essential python cmake cuda cudnn tensorrt numpy onnx pytorch
Dockerfile	Dockerfile
Notes	The openai-triton wheel that's built is saved in the container under /opt. Based on https://cloud.tencent.com/developer/article/2317398, https://zhuanlan.zhihu.com/p/681714973, https://zhuanlan.zhihu.com/p/673525339

RUN CONTAINER

To start the container, you can use the run.sh/autotag helpers or manually put together a docker run command:

# automatically pull or build a compatible container image
./run.sh $(./autotag openai-triton)

# or if using 'docker run' (specify image and mounts/ect)
sudo docker run --runtime nvidia -it --rm --network=host openai-triton:35.2.1

^{run.sh 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:

./run.sh -v /path/on/host:/path/in/container $(./autotag openai-triton)

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

./run.sh $(./autotag openai-triton) my_app --abc xyz

You can pass any options to run.sh 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:

./build.sh openai-triton

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