Mathematical Reasoning on Math500 Level 5 (Accuracy)
71.1AccuracyBG-MCTS
Evaluation Results
| Method | Links | |
|---|---|---|
| BG-MCTSBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=30K2026.02 | 71.1 | |
| RepeatedFullBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=30K2026.02 | 70.2 | |
| BG-MCTSBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=20K2026.02 | 69.9 | |
| RepeatedFullBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=20K2026.02 | 66.4 | |
| BG-MCTSBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=10K2026.02 | 66.2 | |
| AB-MCTS-MFullBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=30K2026.02 | 65.9 | |
| MCTSBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=30K2026.02 | 65.9 | |
| AB-MCTS-MFullBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=20K2026.02 | 65.7 | |
| MCTSBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=20K2026.02 | 65.7 | |
| RepeatedFullBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=10K2026.02 | 63.2 | |
| AB-MCTS-MFullBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=10K2026.02 | 62.9 | |
| MCTSBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=10K2026.02 | 61.9 | |
| AB-MCTS-MBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=30K2026.02 | 60 | |
| LiteSearch-Batch.Backbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=30K2026.02 | 55.7 | |
| LiteSearch-Batch.Backbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=20K2026.02 | 55 | |
| AB-MCTS-MBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=20K2026.02 | 54.2 | |
| LiteSearch-Batch.Backbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=10K2026.02 | 52.7 | |
| LiteSearch-Incre.Backbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=30K2026.02 | 51 | |
| LiteSearch-Incre.Backbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=20K2026.02 | 50.5 | |
| RefinementFullBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=10K2026.02 | 49.8 | |
| GreedyBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=10K2026.02 | 49.3 | |
| GreedyBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=20K2026.02 | 49.3 | |
| GreedyBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=30K2026.02 | 49.3 | |
| RefinementFullBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=20K2026.02 | 49.3 | |
| RefinementFullBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=30K2026.02 | 49 | |
| LiteSearch-Incre.Backbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=10K2026.02 | 48.8 | |
| BG-MCTSBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=20K2026.02 | 46.5 | |
| RepeatedFullBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=30K2026.02 | 45 | |
| BG-MCTSBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=30K2026.02 | 44.3 | |
| RepeatedFullBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=20K2026.02 | 43.8 | |
| AB-MCTS-MBackbone LLM=Qwen-2.5-7B-Instruct, Token Budget (B)=10K2026.02 | 43.3 | |
| MCTSBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=30K2026.02 | 43 | |
| MCTSBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=20K2026.02 | 40.6 | |
| RepeatedFullBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=10K2026.02 | 39.3 | |
| BG-MCTSBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=10K2026.02 | 39.3 | |
| AB-MCTS-MFullBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=30K2026.02 | 34.3 | |
| MCTSBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=10K2026.02 | 33.3 | |
| AB-MCTS-MFullBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=20K2026.02 | 32.3 | |
| AB-MCTS-MFullBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=10K2026.02 | 31.1 | |
| LiteSearch-Incre.Backbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=30K2026.02 | 30.4 | |
| LiteSearch-Incre.Backbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=20K2026.02 | 29.1 | |
| LiteSearch-Batch.Backbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=30K2026.02 | 28.1 | |
| LiteSearch-Batch.Backbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=20K2026.02 | 27.1 | |
| RefinementFullBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=10K2026.02 | 24.9 | |
| LiteSearch-Incre.Backbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=10K2026.02 | 24.9 | |
| RefinementFullBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=20K2026.02 | 24.6 | |
| RefinementFullBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=30K2026.02 | 24.1 | |
| LiteSearch-Batch.Backbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=10K2026.02 | 23.6 | |
| GreedyBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=10K2026.02 | 22.4 | |
| GreedyBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=20K2026.02 | 22.4 | |
| GreedyBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=30K2026.02 | 22.4 | |
| AB-MCTS-MBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=30K2026.02 | 20.9 | |
| AB-MCTS-MBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=20K2026.02 | 17.9 | |
| AB-MCTS-MBackbone LLM=Llama-3.1-8B-Instruct, Token Budget (B)=10K2026.02 | 12.4 |