Mathematical Reasoning on AMC 2023 (pass@8)
91.9pass@8VI-CuRL
Evaluation results
| Method | Links | |
|---|---|---|
| VI-CuRLBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Intrinsic Reward Signal=Majority Vote2026.02 | 91.9 | |
| VI-CuRLBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.02 | 88.8 | |
| VI-CuRLBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 85.3 | |
| AdaRFTBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.02 | 85 | |
| AdaRFTBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 84.7 | |
| VCRLBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.02 | 84.7 | |
| No CurriculumBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.02 | 83.4 | |
| VI-CuRLBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 82.5 | |
| BaseBackbone=Qwen2.5-Math-7B, Reward Setting=No RL2026.02 | 81.9 | |
| VI-CuRLBackbone=Qwen2.5-Math-1.5B, Intrinsic Reward Signal=Majority Vote2026.02 | 81.2 | |
| VCRLBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 79.4 | |
| BaseBackbone=Qwen2.5-Math-1.5B, Reward Setting=No RL2026.02 | 79.3 | |
| No CurriculumBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 79.3 | |
| Base (No RL)Backbone=Qwen2.5-Math-1.5B2026.02 | 79.3 | |
| No CurriculumBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 79.2 | |
| VCRLBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 79 | |
| AdaRFTBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.02 | 77.5 | |
| BaseBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=No RL2026.02 | 77.5 | |
| No CurriculumBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.02 | 57.5 | |
| VI-CuRLBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.02 | 56.6 | |
| BaseBackbone=Llama3.2-3B-Instruct, Reward Setting=No RL2026.02 | 54.7 | |
| VCRLBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.02 | 51.5 | |
| AdaRFTBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.02 | 47.5 |