List your company

Lifelong Agent Interaction on AlfWorld

71.28Success RatePROACTAGENT
25.946437.715749.48561.2543Apr 22, 2026

Evaluation results

MethodLinks
PROACTAGENTRegime=Ours2026.0471.2812.73
GRPO (online)Regime=Online baselines2026.0468.7113.99
GRPORegime=Offline baselines2026.0467.6917.84
GRPO+ReflexionRegime=Online baselines2026.0467.1816.42
SFTRegime=Offline baselines2026.0454.3626.78
ReflexionRegime=Online baselines2026.0436.9238.57
Mem0Regime=Online baselines2026.0433.8532.11
MemoryBankRegime=Online baselines2026.0432.3137.14
AWMRegime=Online baselines2026.0431.2833.56
BaselineRegime=Offline baselines2026.0427.6932.85