C4-Eval shows that multimodal models decode only about half of creatively encoded Chinese idiom puzzles, with open answer search as the main bottleneck.
A causality-awareparadigmforevaluatingcreativityofmultimodallargelanguagemodels.IEEETransactions onPatternAnalysisandMachineIntelligence,47(5):3830–3846,2025
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Can MLLMs Decode the Creative Leap? Introducing C4 for Cross-Concept Understanding
C4-Eval shows that multimodal models decode only about half of creatively encoded Chinese idiom puzzles, with open answer search as the main bottleneck.