Language Modeling on WT (test)
42.62PerplexityLSS
Evaluation Results
| Method | Links | |
|---|---|---|
| LSSVariant=mHC-lite, Model Size=L2026.06 | 42.62 | |
| LSS + IdentityVariant=mHC-lite, Model Size=L2026.06 | 42.69 | |
| LSSVariant=mHC, Model Size=L2026.06 | 43.66 | |
| LSS + IdentityVariant=mHC, Model Size=L2026.06 | 43.8 | |
| IdentityVariant=mHC, Model Size=L2026.06 | 43.95 | |
| BaselineVariant=mHC-lite, Model Size=L2026.06 | 43.98 | |
| IdentityVariant=mHC-lite, Model Size=L2026.06 | 44.32 | |
| BaselineVariant=mHC, Model Size=L2026.06 | 44.38 | |
| LSSVariant=mHC-lite, Model Size=M2026.06 | 47.92 | |
| LSS + IdentityVariant=mHC, Model Size=M2026.06 | 48.92 | |
| LSSVariant=mHC, Model Size=M2026.06 | 48.99 | |
| LSS + IdentityVariant=mHC-lite, Model Size=M2026.06 | 49.6 | |
| IdentityVariant=mHC-lite, Model Size=M2026.06 | 49.65 | |
| IdentityVariant=mHC, Model Size=M2026.06 | 50.88 | |
| BaselineVariant=mHC, Model Size=M2026.06 | 50.97 | |
| BaselineVariant=mHC-lite, Model Size=M2026.06 | 51.05 |