review flash/sdpa arg #25
Merged
Conversation
This file contains 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
Closed
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Outdated suggestions cannot be applied.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.
replace the model setting self_attn_type by a RunningConfig setting self_attn_backend = "flash, pytorch"
In fact:
At training, when using Rotary or Legacy Position Encoding, using flash or pytroch sdpa is almost the same. if alibi or max_relative_positions then it will use "manual" matmul anyway.
At inference: using "flash" instead of "pytorch" will trigger the use of flash_func_with_kvcache which is much faster and not implemented in pytorch 2.3 yet