This was already reverted due to IMA being detected in CI on all B200 GB200 B300 GB300 systems at this commit 25b324d.
pytest tests/attention/test_trtllm_gen_attention.py -x
But the PR is otherwise prioritized.
Filing this for tracking and auto-triaging attempt.
Go to this PR and repro and analyze the IMA.
This was already reverted due to IMA being detected in CI on all B200 GB200 B300 GB300 systems at this commit 25b324d.
But the PR is otherwise prioritized.
Filing this for tracking and auto-triaging attempt.
Go to this PR and repro and analyze the IMA.