Web Navigation on WebArena (offline)
58.4Shopping Success RatePLUGMEM
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| PLUGMEMEmb=NVE, Agent=Q32+4o, #TokAvg.=301, Info. Density=1.4e-32026.02 | 58.4 | 55.2 | 21.6 | |
| A-MemEmb=NVE, Agent=Q32+4o, #TokAvg.=20516, Info. Density=3.4e-72026.02 | 44.3 | 38.5 | 15.8 | |
| AgentOccamEmb=-, Agent=4o2026.02 | 43.6 | 39.2 | 15.8 | |
| Van. RetrievalEmb=NVE, Agent=Q32+4o, #TokAvg.=8733, Info. Density=2.0e-62026.02 | 42.3 | 41.3 | 18.4 | |
| AWMEmb=-, Agent=4o, #TokAvg.=696, Info. Density=-7.9e-42026.02 | 28.2 | 27.3 | — |