Language Modeling on QWEN3-like (val)
2.461LossMACRO-spec
Evaluation Results
| Method | Links | |
|---|---|---|
| MACRO-specParameter Scale=1B2026.05 | 2.461 | |
| MuonH-specParameter Scale=1B2026.05 | 2.464 | |
| MACRO-froParameter Scale=1B2026.05 | 2.467 | |
| MuonH-froParameter Scale=1B2026.05 | 2.468 | |
| MuonParameter Scale=1B2026.05 | 2.473 | |
| SSOParameter Scale=330M2026.05 | 2.712 | |
| MACRO-specParameter Scale=330M2026.05 | 2.714 | |
| MuonH-specParameter Scale=330M2026.05 | 2.716 | |
| MuonH-froParameter Scale=330M2026.05 | 2.717 | |
| MACRO-froParameter Scale=330M2026.05 | 2.718 | |
| FSOParameter Scale=330M2026.05 | 2.726 | |
| MuonParameter Scale=330M2026.05 | 2.736 | |
| FSOParameter Scale=120M2026.05 | 3.001 | |
| MACRO-froParameter Scale=120M2026.05 | 3.005 | |
| MuonH-froParameter Scale=120M2026.05 | 3.007 | |
| SSOParameter Scale=120M2026.05 | 3.011 | |
| MACRO-specParameter Scale=120M2026.05 | 3.017 | |
| MuonParameter Scale=120M2026.05 | 3.019 | |
| MuonH-specParameter Scale=120M2026.05 | 3.019 |