Instruction Following on wikibios v2
97HALT WinsHALT
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| HALTalpha=0.6, judge=GPT-4o-mini, baseline_model=Llama3-70B2025.06 | 97 | 50 | 3 | 66 |
| Method | Links | ||||
|---|---|---|---|---|---|
| HALTalpha=0.6, judge=GPT-4o-mini, baseline_model=Llama3-70B2025.06 | 97 | 50 | 3 | 66 |