Shengbang Tong

Early unification in multimodal training can prevent models from becoming overly reliant on language, unlocking new efficiencies in generative performance.

Vision models are far more data-hungry than language models, but Mixture-of-Experts can harmonize this asymmetry for truly unified multimodal models.

Text-to-image models wash away the unique stylistic fingerprints of their captioning counterparts, revealing a surprising disconnect between text and image generation.