5 papers
i1: A Simple and Fully Open Recipe for Strong Text-to-Image Models
Boya Zeng, Tianze Luo, Shu Pu +4
Diffusion models have consistently driven progress in text-to-image generation. However, it is challenging to attribute recent progress to specific modeling and data choices: state…
One Model, Two Roles: Emergent Specialization in a Shared Recurrent Transformer
Jucheng Shen, Barbara Su, Anastasios Kyrillidis
Can a shared-weight recurrent Transformer develop distinct internal roles without being partitioned into separate modules? We study this in Asymmetric Input Recurrence (AIR), a min…
Improving the Throughput of Diffusion-based Large Language Models via a Training-Free Confidence-Aware Calibration
Jucheng Shen, Gaurav Sarkar, Yeonju Ro +4
We present CadLLM, a training-free method to accelerate the inference throughput of diffusion-based LLMs (dLLMs). We first investigate the dynamic nature of token unmasking confide…
Beyond Static Cutoffs: One-Shot Dynamic Thresholding for Diffusion Language Models
Jucheng Shen, Yeonju Ro
Masked diffusion language models (MDLMs) are becoming competitive with their autoregressive counterparts but typically decode with fixed steps and sequential unmasking. To accelera…
SuperGen: An Efficient Ultra-high-resolution Video Generation System with Sketching and Tiling
Fanjiang Ye, Zepeng Zhao, Yi Mu +11
Diffusion models have recently achieved remarkable success in generative tasks (e.g., image and video generation), and the demand for high-quality content (e.g., 2K/4K videos) is r…