Add Fused CrossEntropy and Update Vocab Sizes #251
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
added 2 commits
March 23, 2023 23:17
added 2 commits
March 24, 2023 00:01
vchiley
reviewed
Mar 24, 2023
vchiley
reviewed
Mar 24, 2023
vchiley
reviewed
Mar 24, 2023
* mv init * mv export inf file * reorg model file struct * kpm -> attn_mask * fix test; lint * updt init test import * lint
abhi-mosaic
approved these changes
Mar 24, 2023
There was a problem hiding this comment.
Looks good and excited to start using this in all our LLM runs, thank you @nik-mosaic !
vchiley
approved these changes
Mar 24, 2023
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Outdated suggestions cannot be applied.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.
Add Fused Cross Entropy, an (almost) math-equivalent loss function to torch.nn.CrossEntropy.
This loss function's requirements are installed via new requirements in llm/requirements.txt. We default to turning this on, and will throw an error if you are on a CPU and have not turned it off. You can turn it off with a new config
model.loss_fn = torch_crossentropyand turn it on omitting loss_fn or settingmodel.loss_fn = fused_crossentropy.This PR also updates the vocab sizes to be a multiple of 64, accounting for the new tokenizer.