Weight Divergence Across Ranks
Weight divergence silently corrupts distributed training. Denpex detects divergence by comparing per-rank weight snapshots.
Weight divergence silently corrupts distributed training.
What this failure is
Weight Divergence Across Ranks is a Training Stability failure seen during ML training runs. Weight divergence silently corrupts distributed training. Denpex detects divergence by comparing per-rank weight snapshots. Common tags: Weight Divergence, Ddp","Fsdp, Determinism, Distributed.
Is this what broke your run? Paste your log.
You're reading about Weight Divergence Across Ranks. Paste your own crash log or traceback below and get the real root cause for YOUR run, not this generic entry. No account, no card. Logs are masked at ingress and never saved to account history.
Want 14 days on the Scale plan?
Request an evaluation code. A verified workplace organization activates up to 50 diagnoses a day, alerts, history, and follow-up questions. No credit card or automatic subscription.
Why it happens (the mechanism)
Asymmetric gradient accumulation. Non-deterministic operations across devices. BN running stats not synchronized. Random seed not synchronized. Taken together, these mechanisms explain why the failure is reproducible, why it tends to surface on specific workloads or scales, and why generic mitigation attempts often fall short without addressing the underlying cause.
What you'll observe
- Model metrics degrade without explicit error
- Loss improves on some ranks while degrading on others
- Validation accuracy diverges between runs
Common symptoms and what they mean
| Symptom | Why it happens |
|---|---|
| Per-rank loss values diverge by >10% | Asymmetric gradient accumulation |
| Gradient norm histograms are bimodal | Non-deterministic operations across devices |
| Checkpoint weights differ beyond precision tolerance | BN running stats not synchronized |
Which systems are affected
- DDP and FSDP training
- Pipeline and tensor parallel training
- Multi-node distributed training
How to confirm this is the problem
Use this checklist to test the hypothesis against a small reproduction. No single line proves the root cause, so preserve the preceding events and compare one variable at a time.
- ✓Reproduce the failure from a clean checkpoint/seed: the symptom must appear without warm-up state from a previous run.
- ✓Verified signal present: Per-rank loss values diverge by >10%
- ✓Verified signal present: Gradient norm histograms are bimodal
- ✓Verified signal present: Checkpoint weights differ beyond precision tolerance
- ✓A targeted fix from the "How to fix it" section eliminates or substantially reduces the symptom within one validation pass.
The fix and the prevention pattern
The root cause is on this page and stays free. A free account adds the exact remediation steps, saved history, and the fix on every entry in the encyclopedia.
Sign up free. Unlock the full analysisNo credit card. Daily allowance follows verified trust tier. Instant access.
Diagnose this failure in VS Code
Select the traceback or open the failed terminal, then run Denpex locally to see the initiating rank, collateral failures, exact fix, and verification command without uploading the log.
Install the free VS Code extensionRelated failures to investigate next
Root cause
- Asymmetric gradient accumulation
- Non-deterministic operations across devices
- BN running stats not synchronized
- Random seed not synchronized
The fix and how to prevent it
Evaluate Denpex on your own logs
Request a Scale evaluation code. A verified workplace organization activates 14 days with up to 50 diagnoses a day. Every account keeps its current diagnosis allowance and gets a verification path. No card or automatic subscription.
Don't just read the fix, diagnose your run
The encyclopedia tells you what went wrong. Denpex tells you what went wrong in YOUR training run. With your logs, your config, and your stack.