Mathematical Reasoning on Minerva Math (pass@8)
28.3Pass@8 RateNo Curriculum
Evaluation Results
| Method | Links | |
|---|---|---|
| No CurriculumBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.02 | 28.3 | |
| VI-CuRLBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.02 | 28.3 | |
| VCRLBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.02 | 24.3 | |
| AdaRFTBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.02 | 22.8 | |
| AdaRFTBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 21.7 | |
| VCRLBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 18.7 | |
| AdaRFTBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 17.6 | |
| VI-CuRLBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 17.1 | |
| VI-CuRLBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 16.6 | |
| No CurriculumBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 16.2 | |
| BaseBackbone=Qwen2.5-Math-1.5B, Reward Setting=No RL2026.02 | 15.5 | |
| No CurriculumBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 15.5 | |
| BaseBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=No RL2026.02 | 14 | |
| VI-CuRLBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.02 | 14 | |
| BaseBackbone=Qwen2.5-Math-7B, Reward Setting=No RL2026.02 | 11.8 | |
| VCRLBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 10.8 | |
| AdaRFTBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.02 | 8.8 | |
| No CurriculumBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.02 | 8.4 | |
| VCRLBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.02 | 8.1 | |
| BaseBackbone=Llama3.2-3B-Instruct, Reward Setting=No RL2026.02 | 4.8 |