block-sparse-attention

block-sparse-attention

CONTAINERS IMAGES RUN BUILD

CONTAINERS

block_sparse_attn:0.0.1
Aliases	block_sparse_attn
Requires	L4T ['>=35']
Dependencies	build-essential pip_cache:cu126 cuda:12.6 cudnn python numpy cmake onnx pytorch:2.8 torchvision torchaudio triton torchao huggingface_hub rust transformers diffusers xformers cuda-python cutlass flash-attention
Dependants	xattention:0.0.1
Dockerfile	Dockerfile

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 block-sparse-attention)

# or if using 'docker run' (specify image and mounts/ect)
sudo docker run --runtime nvidia -it --rm --network=host block-sparse-attention:36.4.0

^{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 block-sparse-attention)

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

jetson-containers run $(autotag block-sparse-attention) 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 block-sparse-attention

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