slimAttn: Experiments section for Whisper, shorter section 1 - #29
Open
Kayvan-Zahiri wants to merge 1 commit into
Open
Kayvan-Zahiri wants to merge 1 commit into
Kayvan-Zahiri wants to merge 1 commit into
Conversation
Moves the Whisper detail out of section 1 into a new Experiments section before the Conclusion and corrects OpenMachine-ai#28: |det W_K| is below 1e-230 in every decoder cross-attention layer of Whisper tiny, small, medium and large-v3 rather than "nowhere near zero", and the merged logit errors came from random encoder output. On LibriSpeech, choosing K or V per layer by condition number gives a larger logit error than the V-cache in every layer in 7 of 8 cases (four models, each on two sets of 4 speakers). Drops the duplicated V-cache equation in favor of appendix C, labels appendices C and F, and adds a LibriSpeech reference.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Per your note: my section 1 text is now one paragraph, and the Whisper details are in a new Experiments section.
Corrections to #28:
|det W_K|is below 1e-230 in every cross-attention layer of Whisper tiny, small, medium and large-v3, not "nowhere near zero".slimAttn_whisper.pyfor its numbers, but the logit errors came from a separate script on random encoder output. The new table uses LibriSpeech.Appendix C already had
W_VK, so I dropped my duplicate equation. Adds a LibriSpeech reference.The logit table and determinant row aren't reproducible from the repo yet. Appendix tables renumber to 10-12,
doc/slimAttn.pdfisn't regenerated, and in my tectonic build the new tables float after the Conclusion, like Table 7.