Language Modeling on MiniPile (train)
30.87Training PerplexityLiMuon
Evaluation Results
| Method | Links | |
|---|---|---|
| LiMuonMemory (GB)=55.15, rank=full rank2025.09 | 30.87 | |
| LiMuonMemory (GB)=54.21, rank=162025.09 | 33.43 | |
| Muon++Memory (GB)=54.302025.09 | 35.16 | |
| MuonMemory (GB)=54.142025.09 | 35.52 | |
| LiMuonMemory (GB)=52.55, rank=82025.09 | 38.86 | |
| SUMOMemory (GB)=52.732025.09 | 41.26 | |
| LiMuonMemory (GB)=50.86, rank=42025.09 | 41.46 | |
| LionMemory (GB)=54.082025.09 | 48.42 | |
| AdamWMemory (GB)=55.262025.09 | 57.97 | |
| AdamMemory (GB)=55.242025.09 | 91.84 |