[Example] Add fused_linear_cross_entropy example and unit test #342
Conversation
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
meta-cla
bot
added
the
CLA Signed
stack-info: PR: #342, branch: yf225/stack/37
yf225
commented
Jul 21, 2025
| """Fused linear + cross entropy.""" | ||
| loss, grad_input, grad_weight, grad_bias = fused_linear_cross_entropy_forward( | ||
| input_tensor, weight, labels, bias | ||
| ) |
There was a problem hiding this comment.
Other implementation of fused_linear_cross_entropy in tritonbench also include both the Python-level chunking code and the Triton kernel (per-chunk processing) in the benchmark timing measurement. So we do the same here for a fair comparison.
jansel
requested changes
Jul 22, 2025
|
|
||
| # Masked load of block | ||
| mask = block_offsets < vocab_size | ||
| logits_block = torch.where( |
There was a problem hiding this comment.
This masking should be added automatically in helion.
| # Process in blocks like Liger | ||
| for vocab_tile in hl.tile(vocab_size): | ||
| # Create block offsets (like tl.arange in Triton) | ||
| block_offsets = vocab_tile.index |
There was a problem hiding this comment.
Is the alias needed after you remove the extra masking?
| n_total_samples: int, # Total number of samples for mean reduction | ||
| ) -> torch.Tensor: | ||
| # Grid over samples - each program handles one sample | ||
| for program_id in hl.grid(chunk_size): |
There was a problem hiding this comment.
Could this be a hl.tile loop to allow tiling this dimension with block_size>1?
| mask = block_offsets < vocab_size | ||
|
|
||
| # Load block | ||
| logits_block = torch.where( |
There was a problem hiding this comment.
Masking should be automatic. Are you sure this needed?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Stacked PRs:
[Example] Add fused_linear_cross_entropy example and unit test