Legal Question Answering on DISC-Law-SFT (test)
90.07Win RateLLM-AutoDP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LLM-AutoDPBase Model=Qwen2.5-7B, Judge=GPT-42026.01 | 90.07 | 1.45 | 8.48 | |
| LLM-AutoDPBase Model=Gemma-2-9B, Judge=GPT-42026.01 | 89.41 | 0.24 | 10.35 | |
| LLM-AutoDPBase Model=Llama3.1-8B, Judge=LawLLM2026.01 | 86.64 | 3.94 | 9.42 | |
| LLM-AutoDPBase Model=Llama3.1-8B, Judge=GPT-42026.01 | 84.66 | 0 | 15.34 | |
| LLM-AutoDPBase Model=Qwen2.5-7B, Judge=LawLLM2026.01 | 84.19 | 6.25 | 9.56 | |
| LLM-AutoDPBase Model=Gemma-2-9B, Judge=LawLLM2026.01 | 83.27 | 10.49 | 6.24 |