When VLMs 'Fix' Students: Identifying and Penalizing Over-Correction in the Evaluation of Multi-line Handwritten Math OCR 文章

ArXiv CS.CV2026-05-27NEWSen作者: Jin Seong, Wencke Liermann, Minho Kim, Jong-hun Shin, Soojong Lim
大语言模型
查看原文 →
关系图谱

When VLMs 'Fix' Students: Identifying and Penalizing Over-Correction in the Evaluation of Multi-line Handwritten Math OCR · 相关技术

相关技术

LLMTTEODEPLAlanguage modelevalStraight-Through EstimatorSSESEMReferring expression comprehension (REC)Parts-of-Speech (POS) tagsOCRVision-Language ModelsVLMLMMHISForEffort Metric AttentionENAARGANN