Scientific Reasoning on Theorem QA
60.6Pass@1Critique-GRPO (Self-Critique)
Evaluation Results
| Method | Links | |
|---|---|---|
| Critique-GRPO (Self-Critique)w/ External Supervision=true2025.06 | 60.6 | |
| Critique-GRPO (CoT Critique)Backbone=Qwen3-8B2025.06 | 60.1 | |
| Critique-GRPO (Self-Critique & Self-Evaluation)w/ External Supervision=false2025.06 | 59.8 | |
| R1-GRPOw/ External Supervision=true2025.06 | 57.9 | |
| SFTw/ External Supervision=true2025.06 | 48.9 | |
| Qwen3-8B (w/ Think)2025.06 | 46.9 | |
| Qwen3-8BBackbone=Qwen3-8B2025.06 | 46.9 | |
| Critique-GRPO (CoT Critique)Backbone=Qwen2.5-7B-Base2025.06 | 44 | |
| Qwen2.5-7B-BaseBackbone=Qwen2.5-7B-Base2025.06 | 21.6 |