Aggregate Performance on All Average
40.3AccuracyAdam+Nexus
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Adam+NexusOptimizer=Adam+Nexus, Model Size=3B2026.04 | 40.3 | 1.601 | |
| MuonOptimizer=Muon, Model Size=3B2026.04 | 39.4 | 1.606 | |
| AdamWOptimizer=AdamW, Model Size=3B2026.04 | 37.1 | 1.633 |