Natural Language Inference on SciNLI
87.1F1 ScoreFull FT
Evaluation results
| Method | Links | |
|---|---|---|
| Full FTBase model=gpt-oss 20B, Trainable Parameters=100.0%2026.04 | 87.1 | |
| MPTBase model=gpt-oss 20B, Trainable Parameters=<0.05%2026.04 | 86.5 | |
| LoRABase model=gpt-oss 20B, Trainable Parameters=~2.50%2026.04 | 85.3 | |
| Full FTBase model=Meditron3 8B, Trainable Parameters=100.0%2026.04 | 84.9 | |
| MPTBase model=Meditron3 8B, Trainable Parameters=<0.05%2026.04 | 84.3 | |
| Full FTBase model=LLaMA 3.1 8B, Trainable Parameters=100.0%2026.04 | 83.6 | |
| LoRABase model=Meditron3 8B, Trainable Parameters=~2.50%2026.04 | 83 | |
| MPTBase model=LLaMA 3.1 8B, Trainable Parameters=<0.05%2026.04 | 82.9 | |
| LoRABase model=LLaMA 3.1 8B, Trainable Parameters=~2.50%2026.04 | 81.5 | |
| PTBase model=gpt-oss 20B, Trainable Parameters=<0.05%2026.04 | 81 | |
| PTBase model=Meditron3 8B, Trainable Parameters=<0.05%2026.04 | 78.5 | |
| PTBase model=LLaMA 3.1 8B, Trainable Parameters=<0.05%2026.04 | 76.8 |