Suchir Agarwal

suchir_agarwal.png

I’m a master’s student at Stanford University studying Computer Science with a focus in artificial intelligence. I previously completed my undergraduate degree in computer science and pure mathematics from the University of California, Berkeley.

I’m part of the Stanford Vision and Learning Lab (SVL) under Jiajun Wu and Fei Fei Li. My research focus is visual generation. Previously at Berkeley, I was in Jennifer Listgarten’s ML for protein engineering lab.

research

  1. preprint
    A Language-Informed Loss for Visual Generation
    Suchir Agarwal , Keshigeyan Chandrasegaran, Ziyu Chen, Kyle Sargent, Juan Carlos Niebles, Li Fei-Fei, and Jiajun Wu
    arXiv preprint, 2026

    End-to-end visual generation with pixel-space loss has recently achieved competitive performance without the need for learned image tokenizers. In this work, we study whether language-informed supervision can provide a richer training signal than pixel loss for visual generation. We introduce LILIE (pronounced “lily”), a Language-Informed and Lossless Image Encoder. LILIE is an invertible image transform trained such that a vision-language model can recover captions from noised features. The encoder uses a normalizing flow transformer trained jointly with a vision-language model via next token prediction. We then use the encoder to extract a simple language-informed loss for text-to-image generation. We train JiT models with LILIE loss, reducing gFD by 14 points (66.84 → 53.01) against a pixel-only baseline on the GPIC dataset.

    arXiv Website Code Models
  2. preprint
    gpic.jpeg
    GPIC: A Giant Permissive Image Corpus for Visual Generation
    Keshigeyan Chandrasegaran, Kyle Sargent, Suchir Agarwal , Michael Jang, Michael Poli, Juan Carlos Niebles, Justin Johnson, Jiajun Wu, and Li Fei-Fei
    arXiv preprint, 2026

    Studying scalable methods for visual generative modeling requires large, accessible, and stable datasets. We introduce GPIC, a Giant Permissive Image Corpus of approximately 28 trillion pixels. GPIC comprises diverse internet images captioned by a state-of-the-art vision-language model, including 100M training, 200K validation, and 1M test examples. Moreover, all GPIC images are permissively licensed for both research and commercial use. GPIC is safety-filtered, deduplicated, and centrally hosted on Hugging Face. We provide a benchmarking protocol for generative modeling on GPIC. Finally, we provide a reference baseline for pixel-space flow matching on GPIC.

    arXiv Website Dataset Code