FORGE : F ine-grained multim o dal evaluation for manufactu r in g sc e narios

Xiangru Jian, Hao Xu, Wei Pang, Xinjian Zhao, Chengyu Tao, Qixin Zhang, Xikun Zhang, Chao Zhang, Guanzhi Deng, Alex Xue, Juan Du, Tianshu Yu, Garth Tarr, Linqi Song Qiuzhuang Sun, Dacheng Tao · 2026

Read FORGE : F ine-grained multim o dal evaluation for manufactu r in g sc e narios by Xiangru Jian, Hao Xu, Wei Pang, Xinjian Zhao, Chengyu Tao, Qixin Zhang, Xikun Zhang, Chao Zhang, Guanzhi Deng, Alex Xue, Juan Du, Tianshu Yu, Garth Tarr, Linqi Song Qiuzhuang Sun, Dacheng Tao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-07413.txt.

Description

Jian and 15 colleagues introduce FORGE, a fine-grained multimodal test for manufacturing settings. Current tests miss factory rigor, hampered by scarce data and weak domain semantics. FORGE pairs real 2D images with 3D point clouds annotated with fine details like exact model numbers, then tests 18 leading multimodal models on three tasks. Visual grounding is not the main bottleneck; missing domain knowledge is. Tuning a compact 3-billion-parameter model on FORGE data lifts held-out accuracy by up to 90.8 percent relatively.

Catalog record

CREATORXiangru Jian, Hao Xu, Wei Pang, Xinjian Zhao, Chengyu Tao, Qixin Zhang, Xikun Zhang, Chao Zhang, Guanzhi Deng, Alex Xue, Juan Du, Tianshu Yu, Garth Tarr, Linqi Song Qiuzhuang Sun, Dacheng Tao
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE201403
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.07413

Browse the full library · About the Tower of Babel