Web/tool tasks on WebShop 1000 (test)
81.2Success RateCVT-RL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CVT-RLDescription=per-seed CVT-RL success, Split (train/dev/test)=6800/900/1000, Individual Seeds=80.8/81.9/82.4/80.7/80.22026.06 | 81.2 | 3.5 | |
| IMDescription=information-matched counterfactual-process RL, Split (train/dev/test)=6800/900/10002026.06 | 77.7 | — | |
| CMDescription=compute-matched non-causal RL, Split (train/dev/test)=6800/900/10002026.06 | 74 | — |