VTC-R1: Vision-Text Compression for Efficient Long-Context Reasoning
IntermediateYibo Wang, Yongcheng Jing et al.Jan 29arXiv
This paper shows a new way to help AI think through long problems faster by turning earlier text steps into small pictures the AI can reread.
#vision-text compression#optical memory#iterative reasoning