Legal Evaluation on Unilaw Eval R1
53.7AccuracyLawThinker
Evaluation Results
| Method | Links | |
|---|---|---|
| LawThinkerBackbone=Qwen3-32B2026.02 | 53.7 | |
| Plan-and-ExecuteBackbone=Qwen3-32B2026.02 | 52 | |
| ReActBackbone=Qwen3-32B2026.02 | 51.3 | |
| Plan-and-SolveBackbone=Qwen3-32B2026.02 | 50.3 | |
| Search-o1Backbone=Qwen3-32B2026.02 | 49.7 | |
| Direct Reasoning2026.02 | 48.3 |