Requested Package Name and Version:

flashinfer-cubin

Brief Explanation for request

Right now, the flashinfer cubins are downloaded and installed manually:

https://gitlab.com/redhat/rhel-ai/rhaiis/containers/-/blob/main/Containerfile.cuda-ubi9?ref_type=heads#L35-39

It's not ideal and AIPCC should build the wheel instead, which would be shipped in the RHAIIS collection.

QE user acceptance tests

The wheel can be installed and cubins can correctly be used from VLLM.

Package License

Apache 2.0: https://pypi.org/project/flashinfer-cubin/