Question Answering on ARC Challenge DE
66.9Normalized Log AccuracyHATified-SFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HATified-SFTMode=SFT, Shots=25-shot2026.03 | 66.9 | — | |
| Llama-InstructMode=SFT, Shots=25-shot2026.03 | 65.3 | — | |
| T-Freeshots=25-shot2026.03 | 64 | 6.57 | |
| HATifiedshots=25-shot2026.03 | 59.4 | 6.57 | |
| Llamashots=25-shot2026.03 | 51.5 | 3.68 | |
| Tülushots=25-shot2026.03 | 36.3 | 3.68 |