fix: support quantized MLX safetensors in extract-index#27
Draft
awnion wants to merge 1 commit intochrishayuk:mainfrom
Draft
fix: support quantized MLX safetensors in extract-index#27awnion wants to merge 1 commit intochrishayuk:mainfrom
awnion wants to merge 1 commit intochrishayuk:mainfrom
Conversation
Unsloth Gemma 4 MLX checkpoints store packed U32 weights with separate scales and biases, so extraction skipped embeddings and failed on Apple Silicon. Dequantize those tensors during loading and streaming extraction so quantized MLX models can build vindexes.
Author
|
For now I have something like this. I guess something is off :) Any suggesions? |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Unsloth Gemma 4 MLX checkpoints store packed U32 weights with separate scales and biases, so extraction skipped embeddings and failed on Apple Silicon. Dequantize those tensors during loading and streaming extraction so quantized MLX models can build vindexes.
E.i. I tried this
And it didn't work :)