OpticalRAG: Pixel-Space Compression for Token-Efficient Retrieval-Augmented Generation
A framework that renders documents as page images, encodes them with a frozen vision encoder, and distills the retrieved visual tokens into as few as $16$ injected tokens before passing them to the LLM, which achieves the best compression-aware accuracy on all three benchmarks.
Sen-Hao Liu, Yu-Heng Zhang, Chun-Yu Wei et al.
· 0 citations