RADIO1D: Elastic Representations for Condensed Vision Modeling
Greg Heinrich, Mike Ranzinger, Collin McCarthy, Natan Bagrov, Eugene Khvedchenya, Bryan Catanzaro, Jan Kautz, Andrew Tao
Abstract
This paper challenges the assumption that vision-language models (VLMs) require fixed patch-based 2D vision features. Analyzing fine-tuned vision encoders, we find that representations become increasingly abstract and less spatially coherent during VLM training. Notably, models trained with image-text alignment (such as SigLIP2) develop a small number of specialized tokens that effectively summarize global image content. Building on this, we introduce RADIO1D, which compresses images into a compact, variable-length 1D token sequence using multi-teacher knowledge distillation and an autoencoder design. The resulting representations exhibit strong hierarchical summarization, enabling accurate scene understanding–even with a single token–and support improved composition-aware image retrieval. In VLMs, RADIO1D provides flexible accuracy-efficiency tradeoffs through adjustable token counts, delivering competitive performance on diverse multimodal benchmarks with lower computational overhead and better accuracy. We release our models under a permissive license.
BibTeX
@inproceedings{
heinrich2026radiod,
title={{RADIO}1D: Elastic Representations for Condensed Vision Modeling},
author={Greg Heinrich and Mike Ranzinger and Collin McCarthy and Natan Bagrov and Eugene Khvedchenya and Bryan Catanzaro and Jan Kautz and Andrew Tao and Pavlo Molchanov},
booktitle={Forty-third International Conference on Machine Learning},
year={2026},
url={https://openreview.net/forum?id=971mgKp9g7}
}