fix: force TileLang-contiguous inputs for Mamba3 MIMO - #994
Open
Chessing234 wants to merge 4 commits into
Open
Conversation
Chessing234
force-pushed
the
fix/mamba3-mimo-seqlen1-contiguous
branch
2 times, most recently
from
July 18, 2026 15:01
6ee2e0c to
8976fb0
Compare
PyTorch treats size-1 trailing dims as contiguous even when stride(-1) != 1, which breaks TileLang kernels at seqlen=1.
Chessing234
force-pushed
the
fix/mamba3-mimo-seqlen1-contiguous
branch
from
July 18, 2026 15:01
8976fb0 to
bb34615
Compare
Keeps the state-spaces#985 stride fix importable without pulling Triton kernels, so CPU tests can cover size-1 trailing dims.
Asserts physical last-stride=1 for size-1 trailing dims without CUDA.
Avoids requiring Triton at collection time for the CPU stride helper tests.
Contributor
Author
|
Follow-up commits on this branch:
Still targeting #985. Ready for another look when you have a minute. |
Chessing234
force-pushed
the
fix/mamba3-mimo-seqlen1-contiguous
branch
from
August 5, 2026 10:24
545d063 to
208be9f
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
.contiguous()before TileLang MIMO kernels.seqlen=1where PyTorch still reports contiguous despitestride(-1) != 1.Fixes #985
Test plan
Mamba3(..., is_mimo=True)(x)withx.shape == (B, 1, D)on CUDA