Knowledge on BBH (3-shot)
49.3Normalized Log AccuracyLlama-3.1-8B-TFree-HAT-SFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Llama-3.1-8B-TFree-HAT-SFTTraining=SFT, Few-shot=3-shot2026.03 | 49.3 | 4.67 | |
| Llama-3.1-Tulu-3-8B-SFTTraining=SFT, Few-shot=3-shot2026.03 | 46.2 | 3.79 |