Adversarial Robustness in Statute Prediction on Legal Statute Dataset (Adversarial)
0.4988ASRLexGuard
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| LexGuardBase Model=GPT-5.22026.05 | 0.4988 | 0.1571 | 0.2602 | 55.88 | |
| LLM-J&H-Few-shotBase Model=GPT-5.2, Prompting Strategy=Few-shot2026.05 | 0.5357 | 0.0238 | 0.0735 | 44.93 | |
| LLM-J&H-CoTBase Model=GPT-5.2, Prompting Strategy=Chain-of-Thought2026.05 | 0.5417 | 0.01 | 0.0401 | 42.09 | |
| LLMBase Model=GPT-5.22026.05 | 0.5625 | 0.0082 | 0.0573 | 40.13 |