fix: fast-dllm dream batch decoding #57
Open
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
…ask in Dream for batch decoding
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Outdated suggestions cannot be applied.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.
Problem
It seems that when batch size > 1 ( see
dream_generate.py), fast-dllm Dream meets an exception:This is caused by wrong kv-cache replacing in
modeling_dream.py( LLaDA's version is correct).After Handling batched replace_position correctly, it still doesn't work, because the shape of attention mask mismatches with Q and K's shapes.
By modifying
generation_utils_block.py, now it works well.Solution
See changes in
modeling_dream.py:and changes in
generation_utils_block.py: