Summarization on MeQSum
47.5ROUGE-LFull FT
Evaluation results
| Method | Links | |
|---|---|---|
| Full FTBase model=gpt-oss 20B, Trainable Parameters=100.0%2026.04 | 47.5 | |
| MPTBase model=gpt-oss 20B, Trainable Parameters=<0.05%2026.04 | 47 | |
| Full FTBase model=Meditron3 8B, Trainable Parameters=100.0%2026.04 | 45.8 | |
| MPTBase model=Meditron3 8B, Trainable Parameters=<0.05%2026.04 | 45.4 | |
| LoRABase model=gpt-oss 20B, Trainable Parameters=~2.50%2026.04 | 45.4 | |
| LoRABase model=Meditron3 8B, Trainable Parameters=~2.50%2026.04 | 43.6 | |
| Full FTBase model=LLaMA 3.1 8B, Trainable Parameters=100.0%2026.04 | 43.5 | |
| MPTBase model=LLaMA 3.1 8B, Trainable Parameters=<0.05%2026.04 | 43.2 | |
| PTBase model=gpt-oss 20B, Trainable Parameters=<0.05%2026.04 | 42 | |
| LoRABase model=LLaMA 3.1 8B, Trainable Parameters=~2.50%2026.04 | 41.2 | |
| PTBase model=Meditron3 8B, Trainable Parameters=<0.05%2026.04 | 40 | |
| PTBase model=LLaMA 3.1 8B, Trainable Parameters=<0.05%2026.04 | 37.4 |