WorldMemArena is a new benchmark that diagnoses multimodal agent memory failures at the stage level by comparing long-context, RAG, external, and harness-based systems on evolving tasks with gold annotations.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
WorldMemArena: Evaluating Multimodal Agent Memory Through Action-World Interaction
WorldMemArena is a new benchmark that diagnoses multimodal agent memory failures at the stage level by comparing long-context, RAG, external, and harness-based systems on evolving tasks with gold annotations.