nvfp4 tensor: refactor weight-only vs dynamic quant #2790
Merged
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
|
Stack from ghstack (oldest at bottom): |
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/2790
Note: Links to docs will display an error until the docs builds have been completed. ⏳ 4 Pending, 1 Unrelated FailureAs of commit ff0ee90 with merge base c120bb7 ( FLAKY - The following job failed but was likely due to flakiness present on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
vkuzo
added a commit
that referenced
this pull request
Aug 18, 2025
Summary: Refactors `NVFP4Tensor` to use `act_quant_kwargs`, to follow the design of recently added `Float8Tensor`. Note that chose not to use `_choose_quant_func_and_quantize_tensor` as we do not support any activation types other than nvfp4. This can be relaxed in the future if needed. This is still not the final API, might need to make more tweaks before we bring out of prototype. Test Plan: Reviewers: Subscribers: Tasks: Tags: ghstack-source-id: f2496ce ghstack-comment-id: 3197771544 Pull-Request: #2790
meta-cla
bot
added
the
CLA Signed
This was referenced Aug 18, 2025
vkuzo
added
the
topic: not user facing
vkuzo
added a commit
that referenced
this pull request
Aug 18, 2025
Summary: Refactors `NVFP4Tensor` to use `act_quant_kwargs`, to follow the design of recently added `Float8Tensor`. Note that chose not to use `_choose_quant_func_and_quantize_tensor` as we do not support any activation types other than nvfp4. This can be relaxed in the future if needed. This is still not the final API, might need to make more tweaks before we bring out of prototype. Test Plan: Reviewers: Subscribers: Tasks: Tags: ghstack-source-id: 02810d1 ghstack-comment-id: 3197771544 Pull-Request: #2790
jerryzh168
reviewed
Aug 18, 2025
| @@ -141,9 +155,11 @@ def to_nvfp4( | |||
| block_size: Block size for quantization (must be 16) | |||
| per_tensor_scale: Optional pre-computed absolute maximum for calibration. | |||
| If provided, uses per-tensor scaling. If None, uses block-wise scaling only. | |||
| mm_config: Matrix multiplication configuration | |||
| per_tensor_scale: Optional pre-computed absolute maximum for calibration for activation | |||
There was a problem hiding this comment.
is this supposed to be act_per_tensor_scale? also is it expected that docstring is the same as the previous item
jerryzh168
approved these changes
Aug 18, 2025
vkuzo
added a commit
that referenced
this pull request
Aug 18, 2025
Summary: Refactors `NVFP4Tensor` to use `act_quant_kwargs`, to follow the design of recently added `Float8Tensor`. Note that chose not to use `_choose_quant_func_and_quantize_tensor` as we do not support any activation types other than nvfp4. This can be relaxed in the future if needed. This is still not the final API, might need to make more tweaks before we bring out of prototype. Test Plan: Reviewers: Subscribers: Tasks: Tags: ghstack-source-id: 61cbdf1 ghstack-comment-id: 3197771544 Pull-Request: #2790
vkuzo
added a commit
that referenced
this pull request
Aug 18, 2025
Summary: Refactors `NVFP4Tensor` to use `act_quant_kwargs`, to follow the design of recently added `Float8Tensor`. Note that chose not to use `_choose_quant_func_and_quantize_tensor` as we do not support any activation types other than nvfp4. This can be relaxed in the future if needed. This is still not the final API, might need to make more tweaks before we bring out of prototype. Test Plan: Reviewers: Subscribers: Tasks: Tags: ghstack-source-id: d56b62a ghstack-comment-id: 3197771544 Pull-Request: #2790
vkuzo
added a commit
that referenced
this pull request
Aug 18, 2025
Summary: Refactors `NVFP4Tensor` to use `act_quant_kwargs`, to follow the design of recently added `Float8Tensor`. Note that chose not to use `_choose_quant_func_and_quantize_tensor` as we do not support any activation types other than nvfp4. This can be relaxed in the future if needed. This is still not the final API, might need to make more tweaks before we bring out of prototype. Test Plan: Reviewers: Subscribers: Tasks: Tags: ghstack-source-id: d56b62a ghstack-comment-id: 3197771544 Pull-Request: #2790
liangel-02
pushed a commit
that referenced
this pull request
Aug 25, 2025
* Update [ghstack-poisoned] * Update [ghstack-poisoned] * Update [ghstack-poisoned] * Update [ghstack-poisoned] * Update [ghstack-poisoned] * Update [ghstack-poisoned] * Update [ghstack-poisoned]
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.
Summary:
Refactors
NVFP4Tensorto useact_quant_kwargs, to follow the designof recently added
Float8Tensor.Note that chose not to use
_choose_quant_func_and_quantize_tensoraswe do not support any activation types other than nvfp4. This can be
relaxed in the future if needed.
This is still not the final API, might need to make more tweaks before
we bring out of prototype.
Test Plan:
Reviewers:
Subscribers:
Tasks:
Tags: