🏢 Allen Institute for AI Organization

1 articles First seen: Jul 13, 2026 Last seen: Jul 13

Activity Timeline (90 days)

Co-occurring Entities

Chowdhery1 Feng1 Jiawei Zhou1 Liu1 OpenAI1 Stony Brook University1 University of Chicago1 Vaswani1 Xiao1 Xu1 Yanhong Li1 Zhou1 Zixuan Lan1

Articles (1)

Text or Pixels? It Takes Half: On the Token Efficiency of Visual Text Inputs in Multimodal LLMs

news arxiv.org IrregularChat: Full Stack Dev Jul 13

The paper explores whether long text can be compressed by converting it into an image and feeding that image to a multimodal LLM instead of sending all text tokens directly. The key idea is that visio

🏢 Allen Institute for AI Organization

Submit Link