Code Generation on MBPP (Accuracy, Loss)
43AccuracyMuon
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MuonOptimizer=Muon, Model Size=3B2026.04 | 43 | 1.951 | |
| AdamWOptimizer=AdamW, Model Size=3B2026.04 | 38 | 1.922 | |
| Adam+NexusOptimizer=Adam+Nexus, Model Size=3B2026.04 | 38 | 1.921 |