Optical reasoning treats images as a medium for language and multimodal tasks
Optical Reasoning: Rethinking Images as an Expressive Reasoning Medium Beyond Text
Hugging FaceTrending 5
Summary
The authors propose optical reasoning, which uses images instead of text as the reasoning medium. Across mathematical, scientific, and multimodal benchmarks, optical reasoning reduces reasoning tokens by an average of 28.57% on language tasks and 16% on multimodal tasks while maintaining or exceeding performance, achieving 1.96 times the token efficiency of text reasoning.