Fix training gradient underflow in quantization tests#13539
Open
jiqing-feng wants to merge 3 commits intohuggingface:mainfrom
Open
Fix training gradient underflow in quantization tests#13539jiqing-feng wants to merge 3 commits intohuggingface:mainfrom
jiqing-feng wants to merge 3 commits intohuggingface:mainfrom
Conversation
Change autocast dtype from float16 to bfloat16 in _test_quantization_training. Float16's limited dynamic range causes gradients to underflow to zero when passing through quantized tensor subclass operations.
github-actions
Bot
added
size/S
github-actions
Bot
added
size/S
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does this PR do?
Changes autocast dtype from
float16tobfloat16in_test_quantization_training. Float16's limited dynamic range (max ~65504, min subnormal ~5.96e-8) causes gradients to underflow to zero when passing through quantized tensor subclass operations; bfloat16 shares float32's exponent range and avoids this.