MMBench is a new bilingual benchmark that uses curated questions, CircularEval, and LLM-assisted answer conversion to provide objective, fine-grained evaluation of vision-language models.
Bayesian rationality: The probabilistic approach to human reasoning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2023 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
MMBench: Is Your Multi-modal Model an All-around Player?
MMBench is a new bilingual benchmark that uses curated questions, CircularEval, and LLM-assisted answer conversion to provide objective, fine-grained evaluation of vision-language models.