Web/tool tasks on Tool API 900 (test)
86.9Success RateCVT-RL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CVT-RLDescription=per-seed CVT-RL success, Split (train/dev/test)=5400/700/900, Individual Seeds=86.0/87.8/86.5/87.3/86.92026.06 | 86.9 | 4.5 | |
| IMDescription=information-matched counterfactual-process RL, Split (train/dev/test)=5400/700/9002026.06 | 82.4 | — | |
| CMDescription=compute-matched non-causal RL, Split (train/dev/test)=5400/700/9002026.06 | 78.5 | — |