fp8 bwd #108
Draft
fp8 bwd #108
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
feat: added fp32 output to input_helper passing feat: fp8 tests. small amount of error added fp8e5m2 type note: RuntimeError: "abs_cuda" not implemented for 'Float8_e4m3fnuz' enabled fp8 GEMMs fix: error down to < 0.1 added another fp8 dtype best accuracy is with no scaling improved accuracy to within < 0.02. issue related to torch side casting fix: passes if we allow v to be fp16 instead of fp8. otherwise we have error < 0.1 all error is < 0.07 feat: added per head scaling tensors progress towards implementing scaling tensors in kernel save issue: error caused by acc += tl.dot(p.to(v.type.element_ty), v)
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Outdated suggestions cannot be applied.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.
No description provided.