Accuracy on AMC Math Reasoning
80AccuracyLLM-J
Evaluation Results
| Method | Links | |
|---|---|---|
| LLM-JBackbone=GPT-OSS-120B2025.11 | 80 | |
| LLM-JBackbone=GPT-OSS-120B2025.11 | 80 | |
| LLM-JBackbone=Gemma-3-27B2025.11 | 73.3 | |
| LLM-JBackbone=Gemma-3-27B2025.11 | 73.3 | |
| APRMBackbone=GPT-OSS-120B2025.11 | 70.7 | |
| APRMBackbone=GPT-OSS-120B2025.11 | 70.7 | |
| Hard Neg.Backbone=GPT-OSS-120B2025.11 | 68.5 | |
| APRMBackbone=GPT-OSS-20B2025.11 | 68 | |
| APRMBackbone=GPT-OSS-20B2025.11 | 68 | |
| ReST-MCTSBackbone=GPT-OSS-120B2025.11 | 67.4 | |
| ReST-MCTSBackbone=GPT-OSS-120B2025.11 | 67.4 | |
| rStar MathBackbone=GPT-OSS-120B2025.11 | 67 | |
| ToTBackbone=GPT-OSS-120B2025.11 | 66.7 | |
| ToTBackbone=GPT-OSS-20B2025.11 | 66.7 | |
| ToTBackbone=GPT-OSS-120B2025.11 | 66.7 | |
| ToTBackbone=GPT-OSS-20B2025.11 | 66.7 | |
| ReST-MCTSBackbone=GPT-OSS-20B2025.11 | 66 | |
| ReST-MCTSBackbone=GPT-OSS-20B2025.11 | 66 | |
| rStar MathBackbone=GPT-OSS-20B2025.11 | 65.5 | |
| AutoPSVBackbone=GPT-OSS-120B2025.11 | 65 | |
| ARMBackbone=GPT-OSS-120B2025.11 | 63.3 | |
| Math S.Backbone=GPT-OSS-120B2025.11 | 62 | |
| ReST-MCTSBackbone=Gemma-3-27B2025.11 | 61.2 | |
| ReST-MCTSBackbone=Gemma-3-27B2025.11 | 61.2 | |
| Verify SbS.Backbone=GPT-OSS-120B2025.11 | 61 | |
| APRMBackbone=Gemma-3-27B2025.11 | 60.8 | |
| APRMBackbone=Gemma-3-27B2025.11 | 60.8 | |
| rStar MathBackbone=Gemma-3-27B2025.11 | 60.5 | |
| CoT-SCBackbone=GPT-OSS-120B2025.11 | 60 | |
| LLM-JBackbone=GPT-OSS-20B2025.11 | 60 | |
| ToTBackbone=Gemma-3-27B2025.11 | 60 | |
| Hard Neg.Backbone=Gemma-3-27B2025.11 | 60 | |
| CoT-SCBackbone=GPT-OSS-120B2025.11 | 60 | |
| LLM-JBackbone=GPT-OSS-20B2025.11 | 60 | |
| ToTBackbone=Gemma-3-27B2025.11 | 60 | |
| AutoPSVBackbone=GPT-OSS-20B2025.11 | 50 | |
| Math S.Backbone=GPT-OSS-20B2025.11 | 48 | |
| Verify SbS.Backbone=GPT-OSS-20B2025.11 | 45 | |
| AutoPSVBackbone=Gemma-3-27B2025.11 | 40 | |
| AutoMRBackbone=Qwen, Skeleton Structure=DAG2025.10 | 38.6 | |
| Meta-ReasonerBackbone=Qwen, Skeleton Structure=Sequential2025.10 | 36.1 | |
| HILAType=MA, Backbone=LLaMA3-8B2026.03 | 35.83 | |
| Math S.Backbone=Gemma-3-27B2025.11 | 35 | |
| CoTBackbone=Qwen, Skeleton Structure=Sequential2025.10 | 34.9 | |
| MRPBackbone=Qwen, Skeleton Structure=-2025.10 | 33.7 | |
| MaASBackbone=Qwen, Skeleton Structure=Sequential2025.10 | 33.7 | |
| CoT-SCBackbone=GPT-OSS-20B2025.11 | 33.3 | |
| CoT-SCBackbone=GPT-OSS-20B2025.11 | 33.3 | |
| rStarBackbone=Qwen, Skeleton Structure=Tree2025.10 | 32.5 | |
| AutoMRBackbone=LLaMA, Skeleton Structure=DAG2025.10 | 30.1 | |
| ARMBackbone=Gemma-3-27B2025.11 | 30 | |
| Verify SbS.Backbone=Gemma-3-27B2025.11 | 30 | |
| Meta-ReasonerBackbone=LLaMA, Skeleton Structure=Sequential2025.10 | 26.5 | |
| MRPBackbone=LLaMA, Skeleton Structure=-2025.10 | 25.3 | |
| MaASBackbone=LLaMA, Skeleton Structure=Sequential2025.10 | 24.1 | |
| CoT-SCBackbone=Gemma-3-12B2025.11 | 23.3 | |
| CoT-SCBackbone=Gemma-3-12B2025.11 | 23.3 | |
| APRMBackbone=Gemma-3-12B2025.11 | 21.6 | |
| APRMBackbone=Gemma-3-12B2025.11 | 21.6 | |
| CoTBackbone=LLaMA, Skeleton Structure=Sequential2025.10 | 21.2 | |
| G-DebateType=MA, Backbone=LLaMA3-8B2026.03 | 20.48 | |
| ReST-MCTSBackbone=Gemma-3-12B2025.11 | 20.2 | |
| ReST-MCTSBackbone=Gemma-3-12B2025.11 | 20.2 | |
| AutoPSVBackbone=Gemma-3-12B2025.11 | 20.1 | |
| Math S.Backbone=Gemma-3-12B2025.11 | 20.1 | |
| rStar MathBackbone=Gemma-3-12B2025.11 | 20.1 | |
| Verify SbS.Backbone=Gemma-3-12B2025.11 | 20.05 | |
| LLM-JBackbone=Gemma-3-12B2025.11 | 20 | |
| ToTBackbone=Gemma-3-12B2025.11 | 20 | |
| LLM-JBackbone=Gemma-3-12B2025.11 | 20 | |
| ToTBackbone=Gemma-3-12B2025.11 | 20 | |
| DyLANType=MA, Backbone=LLaMA3-8B2026.03 | 19.68 | |
| DebateType=MA, Backbone=LLaMA3-8B2026.03 | 19.28 | |
| A-PruneType=MA, Backbone=LLaMA3-8B2026.03 | 16.47 | |
| rStarBackbone=LLaMA, Skeleton Structure=Tree2025.10 | 15.7 | |
| PHPType=MA, Backbone=LLaMA3-8B2026.03 | 15.66 | |
| G-SwarmType=MA, Backbone=LLaMA3-8B2026.03 | 15.66 | |
| CoT-SCBackbone=Gemma-3-27B2025.11 | 13.3 | |
| CoT-SCBackbone=Gemma-3-27B2025.11 | 13.3 | |
| SCType=SA, Backbone=LLaMA3-8B2026.03 | 12.45 | |
| AFlowType=MA, Backbone=LLaMA3-8B2026.03 | 12.05 | |
| Direct-I/OBackbone=LLaMA, Skeleton Structure=-2025.10 | 12 | |
| CoTType=SA, Backbone=LLaMA3-8B2026.03 | 11.65 | |
| Direct-I/OBackbone=Qwen, Skeleton Structure=-2025.10 | 8.4 | |
| VanillaType=SA, Backbone=LLaMA3-8B2026.03 | 8.03 | |
| CoScale-RL2026.01 | 7.6 | |
| COMPASS2026.01 | 6.8 | |
| SpeedControl2026.01 | 5.5 | |
| GRPO2026.01 | 5.1 | |
| SAPO2026.01 | 5.1 | |
| Scale-RL2026.01 | 4 | |
| Pretrain + RL2026.01 | 2.7 | |
| SFTed (w/o Data Scaling)Data Scaling=No2026.01 | 2.3 | |
| Qwen2.5-0.5B Instruct2026.01 | 2.1 | |
| SFTed (w/ Data Scaling)Data Scaling=Yes2026.01 | 1.4 |