Continual-memory deployment on ScienceWorld benchmark leaderboard · SOTA2 Research

Continual-memory deployment on ScienceWorld

41.07Success RateAuto-Dreamer

Evaluation Results

Method	Links
Auto-Dreamer2026.05		41.07	37.53
UMEM2026.05		34.07	30.51
ReasoningBank2026.05		30.86	27.32
Memp2026.05		30.36	26.9
AWM2026.05		30.18	26.61
Mem-α2026.05		29.97	26.43
Reflexion2026.05		29.55	26.07
No memory2026.05		28.69	25.18
ExpeL2026.05		28.33	24.97
LightMem2026.05		28.08	23.9
Mem02026.05		26.79	23.39