PC: Add support for Starcoder2#518
Closed
Blaizzy wants to merge 12 commits intoml-explore:mainfrom
Blaizzy:pc/starcoder2
Closed
PC: Add support for Starcoder2#518Blaizzy wants to merge 12 commits intoml-explore:mainfrom Blaizzy:pc/starcoder2
Blaizzy wants to merge 12 commits intoml-explore:mainfrom
Blaizzy:pc/starcoder2
Conversation
Contributor
|
Nice work! Just curious, does this work with the FIM prompt? |
Merged
mzbac
reviewed
Mar 3, 2024
| scores = (queries * self.scale) @ keys.transpose(0, 1, 3, 2) | ||
| if mask is not None: | ||
| scores += mask | ||
| scores = mx.softmax(scores, axis=-1).astype(values.dtype) |
Contributor
There was a problem hiding this comment.
Maybe we need to keep scores.astype(mx.float32) in case softmax overflows on float16 values.
Member
Contributor
Author
Thanks! It should work without a problem. If not I can investigate. |
Contributor
Author
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Hey @awni and @Muhtasham,
I was super excited about integrating starcoder 2 into mlx. However, while working on the mlx-langchain integration, I realized that @Muhtasham beat me to the punch and submitted a PR. Kudos to you!
I took a look at the original draft PR a few days ago and noticed a few areas that could use some improvement (mlp and others). When I tested the code with the quantized models uploaded to the hub, it didn't quite work as expected. But not to worry, I've fixed it! ✨ ✅
This implementation is working well and matches the original model using transformers in full precision and quantized configurations.
python -m mlx_lm.generate --model mlx-community/starcoder2-3b-4bit --prompt "Write a quick sort in C++" --colorize --max-tokens 200Output: