Scientific Verification on GPQA DIAMOND
65.15EMPRISM-MCTS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PRISM-MCTSModel=Qwen3-30B-A3B-Instruct-25072026.04 | 65.15 | 66.26 | |
| PRISM-MCTSModel=GPT-4.1-mini2026.04 | 65.08 | 65.16 | |
| MCTS-RAGModel=GPT-4.1-mini2026.04 | 64.65 | 64.8 | |
| MCTS-RAGModel=Qwen3-30B-A3B-Instruct-25072026.04 | 63.64 | 64.26 | |
| Search-O1Model=Qwen3-30B-A3B-Instruct-25072026.04 | 61.11 | 53.76 | |
| ReST-MCTS*Model=GPT-4.1-mini2026.04 | 60.61 | 60.53 | |
| ReST-MCTS*Model=Qwen3-30B-A3B-Instruct-25072026.04 | 58.08 | 60.45 | |
| Search-O1Model=GPT-4.1-mini2026.04 | 57.07 | 57.16 | |
| ReActModel=GPT-4.1-mini2026.04 | 55.05 | 60.75 | |
| ReActModel=Qwen3-30B-A3B-Instruct-25072026.04 | 48.99 | 53.49 | |
| Zero-shotModel=GPT-4.1-mini2026.04 | 47.55 | 46.55 | |
| Zero-shotModel=Qwen3-30B-A3B-Instruct-25072026.04 | 45.8 | 45.29 |