Hypothesis Identification in Dialogues on BarExamQA Legal domain (test)
66.5Success RateINFOGATHERER
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| INFOGATHERERModel=gpt-5-nano2026.03 | 66.5 | 5.9 | |
| INFOGATHERERModel=Qwen3 32B2026.03 | 64.3 | 5.4 | |
| IG BayesianModel=gpt-5-nano2026.03 | 61.3 | 6.2 | |
| UoTModel=Qwen3 32B2026.03 | 60.8 | 5.1 | |
| IG BayesianModel=Qwen3 32B2026.03 | 57.9 | 5.7 | |
| UoTModel=gpt-5-nano2026.03 | 32.8 | 4.9 | |
| MediQModel=gpt-5-nano2026.03 | 27.2 | 9.4 | |
| MediQModel=Qwen3 32B2026.03 | 25.7 | 8.9 | |
| AoPModel=gpt-5-nano2026.03 | 17.1 | 2.6 | |
| AoPModel=Qwen3 32B2026.03 | 15.4 | 2.2 |