Mathematical Reasoning on AMC (Pass@16, Mean@16, Token Usage)
47.06Mean @16ME-ICPO
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ME-ICPOBackbone=Qwen2.5-Math-7B, Time(s/question)=11522026.03 | 47.06 | — | — | — | — | |
| TTRLBackbone=Qwen2.5-Math-7B, Time(s/question)=12532026.03 | 45.18 | — | — | — | — | |
| OptPO-SFTBackbone=Qwen2.5-Math-1.5B2025.12 | 39.7 | 81.9 | 43.4 | — | 12.08 | |
| TTSFTBackbone=Qwen2.5-Math-1.5B2025.12 | 37.9 | 80.7 | 42.2 | — | — | |
| ToT (Maj vote)Backbone=Qwen2.5-Math-7B, Time(s/question)=3632026.03 | 29.37 | — | — | — | — | |
| TTSFTBackbone=Llama-3.1-8B-Instruct2025.12 | 20.1 | 63.9 | 28.9 | — | — | |
| OptPO-SFTBackbone=Llama-3.1-8B-Instruct2025.12 | 18.7 | 60.2 | 19.3 | — | 17.16 | |
| ToT (self eval)Backbone=Qwen2.5-Math-7B, Time(s/question)=7082026.03 | 16.19 | — | — | — | — | |
| MCTRBackbone=Qwen2.5-Math-7B, Time(s/question)=17582026.03 | 1.2 | — | — | — | — |