Skip to content

docs: remove retired zero-gradient metric #3267

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
mikasenghaas merged 1 commit into from
Aug 13, 2026
+1 −1
Merged

docs: remove retired zero-gradient metric #3267

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion docs/training.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -199,7 +199,7 @@ Pulled from the console log and mirrored to W&B.
**Stability and optimization:**

- `optim/grad_norm` — spikes precede divergence.
- `optim/lr`, `optim/zero_grad_ratio` — LR schedule and the fraction of params that received zero gradients (high → dead path or wrong loss masking).
- `optim/lr` — LR schedule.
- For MoE: `max_vio/mean` (load-balancing violation), `routing_confidence/mean` — both are logged when non-zero.

**Performance:**
Expand Down