1 paper
Yiheng Lyu, Xueying Jiang, Wenhao Li +2
How well can general-purpose multimodal models turn visual understanding and reasoning into embodied manipulation via executable code? We introduce CodeActionBench, a benchmark of…