Rank #864 · on radar since 2026-07-03
MoEWatch
👻 MoEWatch: A lightweight diagnostic and audit library for Mixture-of-Experts (MoE) models in HuggingFace Transformers — the pytest for MoE models.
Visit homepage ↗qwenmoewatchmixtralmixture-of-experts+16
Momentum
42.2
24h–7d–
Why it's ranked
Every score decomposes into published factors — the same math for every tool, paid or not. Read the methodology →
| Velocity (weighted, cohort-normalized) | 0.438 |
| Signal decay | 0.995 |
| Corroboration | 1.000 |
| Quality gate | 1.000 |
Raw signals (30 days)
github · forks0 latest · 2 snapshots
github · stars5 latest · 2 snapshots