Preference Prediction on RM-Bench
87.8AccuracyC2
Evaluation results
| Method | Links | |
|---|---|---|
| C2Backbone=Qwen3-8B2026.04 | 87.8 | |
| Reasoning RM + External-Rubric (32B)Backbone=Qwen3-8B2026.04 | 84.6 | |
| Reasoning RMBackbone=Qwen3-8B2026.04 | 81.3 | |
| Reasoning RM + Self-RubricBackbone=Qwen3-8B2026.04 | 81.3 | |
| Base ModelBackbone=Qwen3-8B2026.04 | 80.1 | |
| Reasoning RM + External-Rubric (32B)Backbone=Tulu3-8B-SFT2026.04 | 77.7 | |
| C2Backbone=Tulu3-8B-SFT2026.04 | 65.6 | |
| Reasoning RMBackbone=Tulu3-8B-SFT2026.04 | 64.9 | |
| Reasoning RM + Self-RubricBackbone=Tulu3-8B-SFT2026.04 | 64.2 | |
| Base ModelBackbone=Tulu3-8B-SFT2026.04 | 56.1 |