Web Agent Navigation on WebArena Lite
47.9Success RateCATTS (entropy, best)
Evaluation Results
| Method | Links | |
|---|---|---|
| CATTS (entropy, best)N/P=10, K=1, Calls/Step=~72026.02 | 47.9 | |
| Arbiter (K=1)N/P=10, K=1, Calls/Step=112026.02 | 44 | |
| RSAN/P=16, K=4, T=4, Calls/Step=802026.02 | 43.6 | |
| Majority VoteN/P=10, Calls/Step=102026.02 | 43.2 | |
| RSAN/P=16, K=4, T=2, Calls/Step=482026.02 | 42 | |
| RSAN/P=8, K=2, T=2, Calls/Step=242026.02 | 41 | |
| RSAN/P=8, K=2, T=4, Calls/Step=402026.02 | 41 | |
| PlanRSA (arbiter select)N/P=8, K=4, T=2, Calls/Step=402026.02 | 35.2 | |
| PlanRSA (random select)N/P=8, K=4, T=2, Calls/Step=402026.02 | 35.2 |