Instruction Following on WildBench 1.0 (test)
37.98WB-ScoreGPRL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GPRLGroup=Online RL with reward model, RM Size=8B, RM Type=GPM, Iter./Ep.=32026.05 | 37.98 | 11.15 | |
| SimPOGroup=No reward model2026.05 | 37.05 | 9.57 | |
| SPPOGroup=Reward model, RM Size=8B, RM Type=BT, Iter./Ep.=32026.05 | 36.14 | 8.33 | |
| GPOGroup=Reward model, RM Size=8B, RM Type=GPM, Iter./Ep.=32026.05 | 35.98 | 7.42 | |
| DPOGroup=No reward model2026.05 | 35.9 | 8.16 | |
| GPOGroup=Reward model, RM Size=8B, RM Type=BT, Iter./Ep.=32026.05 | 35.87 | 8.25 | |
| GRPOGroup=Online RL with reward model, RM Size=8B, RM Type=BT, Iter./Ep.=32026.05 | 35.41 | 8.62 | |
| SPPOGroup=Reward model, RM Size=8B, RM Type=GPM, Iter./Ep.=32026.05 | 35.21 | 7.18 |