Relevance Prediction on In-the-Wild W-Eval
52.77Class 1 MetricTaoSR-AGRL
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| TaoSR-AGRLBackbone=TbStar-DPO2025.10 | 52.77 | 64.34 | 46.22 | 82.79 | 86.6 | 61.53 | 74.28 | |
| GRPO-PRBackbone=TbStar-DPO2025.10 | 45.49 | 63.56 | 45.08 | 82.76 | 86.59 | 59.22 | 73.95 | |
| GRPOBackbone=TbStar-DPO2025.10 | 45.41 | 62.95 | 45.93 | 82.72 | 86.49 | 59.25 | 73.83 | |
| TbStar-DPOBackbone=TbStar (42B MoE)2025.10 | 44.49 | 63.26 | 45.95 | 82.59 | 86.38 | 59.07 | 73.7 |