Yuwei Niu

A novel decoupled autoencoder design unlocks 32x spatial compression for diffusion models while simultaneously boosting reconstruction fidelity and accelerating training convergence.

Native unified modelling is position as a promising path towards systems that perceive, reason and create within a fully end-to-end framework through SenseNova-U1.5, an 8B-MoT native unified multimodal model that understands, reasons about, and generates visual content within an encoder-free and VAE-free architecture.

Layer-native design transforms how generative models create and edit images, leading to significant performance gains in visual tasks.

Raw context outperforms compact memory designs, revealing that memory structure is crucial for effective video generation in action-conditioned models.

Ditching modular architectures unlocks surprisingly competitive vision-language performance, proving that end-to-end pixel-to-word models can rival traditional approaches at scale.