Preference Classification on Math Reasoning (test)
85.4Classification AccuracyBTPO
Evaluation Results
| Method | Links | |
|---|---|---|
| BTPOBackbone=Qwen2.5-7B-Instruct2025.10 | 85.4 | |
| RM-R1-finetunedBackbone=Qwen2.5-7B-Instruct, Fine-tuned=true2025.10 | 55.9 | |
| RM-R1Backbone=Qwen2.5-7B-Instruct2025.10 | 55 | |
| GRPO (pair)Backbone=Qwen2.5-7B-Instruct2025.10 | 53.7 |