M2CA-Bench
Cognitive-structured Multimodal Agent for Multimodal Understanding, Generation, and Editing
Feng Wang, Canmiao Fu, Zhipeng Huang, Chen Li, Jing Lyu, Ge Li · 2026-07-09
- Modality
- Image
- Data
- Synthetic
- Task · ground truth
- Ret./Gen. · Ev.+Ans.
- Scope
- Memory-specific benchmark
What it tests
- CECross-modal evidenceDirectly evaluated
- LELong-range evidenceDirectly evaluated
- SCState changesPartially evaluated
- CFConflicting memoriesNot evaluated
- UQUnanswerable queriesNot evaluated
- MAMemory-grounded actionNot evaluated
- FGFine-grained visual detailPartially evaluated
- ETEntity trackingDirectly evaluated
- VCVisual state changePartially evaluated
Survey notes
Held-out multimodal dialogues evaluate memory-image selection and the faithfulness of responses using retrieved images.