Skip to content

Conversation

@baogorek
Copy link
Collaborator

@baogorek baogorek commented Jan 21, 2026

Summary

  • Add verbose weight distribution buckets during training: <0.01, 0.01-0.1, 0.1-1, 1-10, 10-1000, >1000
  • Add use_gates parameter to SparseCalibrationWeights to optionally disable L0 gates (defaults to True)

Test plan

  • All existing tests pass
  • Manual verification of weight distribution output during calibration

🤖 Generated with Claude Code

baogorek and others added 2 commits January 20, 2026 21:51
- Add verbose weight distribution buckets: <0.01, 0.01-0.1, 0.1-1, 1-10, 10-1000, >1000
- Add use_gates parameter to disable L0 gates (default True)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
@baogorek baogorek merged commit 727723c into main Jan 21, 2026
4 checks passed
@baogorek baogorek deleted the large-weight-diagnostics branch January 21, 2026 03:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants