DiffScore is a bidirectional masked-diffusion evaluation framework that measures text recoverability across masking rates and outperforms autoregressive baselines on ten benchmarks.
Weinberger, and Yoav Artzi
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 2verdicts
UNVERDICTED 2roles
background 1polarities
background 1representative citing papers
Introduces YesBut benchmark showing state-of-the-art multimodal models lag humans on interpreting humorous contradictions in comics.
citing papers explorer
-
DiffScore: Text Evaluation Beyond Autoregressive Likelihood
DiffScore is a bidirectional masked-diffusion evaluation framework that measures text recoverability across masking rates and outperforms autoregressive baselines on ten benchmarks.
-
Cracking the Code of Juxtaposition: Can AI Models Understand the Humorous Contradictions
Introduces YesBut benchmark showing state-of-the-art multimodal models lag humans on interpreting humorous contradictions in comics.