ResearchBenchmarksContinual-memory deployment on ALFWorldFollow60.21Success RateAuto-Dreamer29.42637.41845.4153.402May 20, 2026Evaluation ResultsMethodMethodLinksSuccess RateSuccess Rate (95% CI Lower Bound)Auto-Dreamer2026.0560.2154.65UMEM2026.0558.4353.15Mem-α2026.0557.452.04Reflexion2026.0549.1743.59ExpeL2026.0533.5928.79AWM2026.0532.7528.16Memp2026.0531.427.09LightMem2026.0531.1626.72ReasoningBank2026.0531.1126.56No memory2026.0530.8326.4Mem02026.0530.6125.81