add FP8 sweep and step_size flag #758
Draft
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
Signed-off-by: Fridah-nv <[email protected]>
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
realAsma
reviewed
Jan 9, 2026
| for step, multiplier in enumerate(multipliers): | ||
| candidate_amax = self._initial_amax * multiplier | ||
| for step, candidate in enumerate(candidates): | ||
| if self._fp8_scale_sweep: |
realAsma
reviewed
Jan 9, 2026
Comment on lines
149
to
152
| # For FP8 scale sweep, use FP8 values as multipliers of initial_amax | ||
| # This ensures we search in a reasonable range relative to max calibration | ||
| multiplier = candidate | ||
| candidate_amax = self._initial_amax * multiplier |
Contributor
There was a problem hiding this comment.
is not in this case candidate_amax = (fp8_by_448 * 6.0)?
Suggested change
| # For FP8 scale sweep, use FP8 values as multipliers of initial_amax | |
| # This ensures we search in a reasonable range relative to max calibration | |
| multiplier = candidate | |
| candidate_amax = self._initial_amax * multiplier | |
| candidate_amax = (candidate * global_amax).view_as(self._initial_amax) |
Contributor
There was a problem hiding this comment.
why?
fp8_scale = FP8(block_eqv_amax / 6.0 * (448/(global_amax/6.0))
so if we reverse calculate block_eqv_amax_from_fp8, we get:
block_eqv_amax = (fp8_scale / 448.0) * global_amax
candidate_amax = candidate * global_amax
Signed-off-by: Fridah-nv <[email protected]>
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Outdated suggestions cannot be applied.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.
What does this PR do?
Type of change: ?
Overview: ?
Usage
# Add a code snippet demonstrating how to use thisTesting
Before your PR is "Ready for review"
Additional Information