Science
FGBench
Molecular property reasoning at the functional-group level. This curation links individual model outputs to the 7,146-question test bank, preserving classification correctness and numeric squared errors.
7,142items
8subjects
MITlicense
sciencedomain
reasoningdomain
textmodality
item-level responses released
Response matrix
Loading response matrix…
lowhighUnobserved
Scale: Item-specific grading scales