Reward Performance on Webshop Agent High-data
45Test-Time RewardMC
Evaluation Results
| Method | Links | |
|---|---|---|
| MCtraining_regime=BOOST2026.05 | 45 | |
| MCtraining_regime=Bilevel Only2026.05 | 43 | |
| ILQLtraining_regime=Bilevel Only2026.05 | 41 | |
| MCtraining_regime=Vanilla2026.05 | 38 | |
| ILQLtraining_regime=BOOST2026.05 | 36 | |
| MCtraining_regime=Vanilla+Syn2026.05 | 31 | |
| ILQLtraining_regime=Vanilla2026.05 | 31 | |
| ILQLtraining_regime=Vanilla+Syn2026.05 | 27 |