Reasoning
MMDocRAG
MMDocRAG tests whether models can answer questions about documents by weaving retrieved text and images into a cited answer, which a judge model rates for fluency, citation quality, coherence, reasoning and factuality.
2,000items
68subjects
CC-BY-NC-4.0license
generaldomain
reasoningdomain
imagemodality
textmodality
item-level responses released
Saturation status: No
Response matrix
Loading response matrix…
lowhighUnobserved
Scale: Grades: 0, 0.04, 0.08, 0.12, 0.16, 0.2, 0.24, 0.28, 0.32, 0.36, 0.4, 0.44, 0.48, 0.52, 0.56, 0.6, 0.64, 0.68, 0.72, 0.76, 0.8, 0.84, 0.88, 0.92, 0.96, 1