quantizer: reject non-finite source tensors - #741
Open
exochard wants to merge 1 commit into
Open
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
The DeepSeek quantizer converts source tensors to F32 before applying the selected quantization scheme. Non-finite source values can otherwise be encoded into a GGUF and only surface much later as corrupted routing or activations.
Change
Validate every converted regular HF tensor and every converted routed-expert tensor before imatrix lookup and quantization. On the first NaN or Inf, fail with the source tensor name and element index.
Validation
The quantizer tool builds cleanly with no compiler warnings:
git diff --checkis clean. This change is independent of the CUDA runtime and can be reviewed or merged separately from the router and SSD-streaming changes.