Reward Modeling on Blessing
99.4Agreement RateAC-GenRM
Evaluation Results
| Method | Links | |
|---|---|---|
| AC-GenRMBackbone=Qwen3-8B2026.04 | 99.4 | |
| AC-GenRMBackbone=Qwen3-4B2026.04 | 99.3 | |
| AC-GenRMBackbone=Qwen3-1.7B2026.04 | 99.1 | |
| AC-GenRMBackbone=Qwen3-0.6B2026.04 | 98.9 | |
| DeepSeek-R1Category=Baselines2026.04 | 97.7 | |
| GPT-4.1-MiniCategory=Baselines2026.04 | 96.9 | |
| Claude-3.5-HaikuCategory=Baselines2026.04 | 94.9 | |
| BaseBackbone=Qwen3-8B2026.04 | 94.4 | |
| BaseBackbone=Qwen3-4B2026.04 | 93.8 | |
| GPT-4.1Category=Baselines2026.04 | 92.3 | |
| DeepSeek-V3Category=Baselines2026.04 | 90.6 | |
| Claude-Sonnet-3.7Category=Baselines2026.04 | 90.2 | |
| BaseBackbone=Qwen3-1.7B2026.04 | 84.9 | |
| BaseBackbone=Qwen3-0.6B2026.04 | 64.3 |