5 papers
PreDiff-LM: Pretrained Discrete Masked Diffusion Language Modeling with Hybrid Attention
Zhengtao Yao, Runhao Li, Xupeng Chen +12
Discrete masked diffusion language models support bidirectional generation and infilling, but adapting pretrained autoregressive (AR) transformers requires reconciling causal pretr…
Less Data, Better Alignment: Data-Centric Multi-Evaluator Agreement for Preference Optimization
Zhengtao Yao, Runhao Li, Xupeng Chen +12
Research on preference optimization often varies the training objective while holding the data fixed. We instead ask whether a small, high-confidence set of on-policy responses can…
TimeROME-DLM: Temporal Causal Tracing and Low-Rank Inference-Time Knowledge Editing for Masked Diffusion Language Models
Zhengtao Yao, Liuyang Song, Hongbo Zhang +4
Masked diffusion language models (MDLMs) such as LLaDA now rival autoregressive (AR) LLMs, but every existing knowledge-editing and unlearning method (ROME, MEMIT, etc.) targets AR…
Inversion-DPO: Precise and Efficient Post-Training for Diffusion Models
Zejian Li, Yize Li, Chenye Meng +7
Recent advancements in diffusion models (DMs) have been propelled by alignment methods that post-train models to better conform to human preferences. However, these approaches typi…
LAION-SG: An Enhanced Large-Scale Dataset for Training Complex Image-Text Models with Structural Annotations
Zejian Li, Chenye Meng, Yize Li +9
Recent advances in text-to-image (T2I) generation have shown remarkable success in producing high-quality images from text. However, existing T2I models show decayed performance in…