7 papers
S-SPPO: Semantic-Calibrated Self-Play Preference Optimization
Xiwen Chen, Wenhui Zhu, Jingjing Wang +13
Aligning Large Language Models (LLMs) with human preferences is often formulated via Direct Preference Optimization (DPO). However, the standard Bradley-Terry instantiation of DPO…
Context Pruning for Coding Agents via Multi-Rubric Latent Reasoning
Jingjing Wang, Xiwen Chen, Wenhui Zhu +6
LLM-powered coding agents spend the majority of their token budget reading repository files, yet much of the retrieved code is irrelevant to the task at hand. Existing learned prun…
SODA: Semi On-Policy Black-Box Distillation for Large Language Models
Xiwen Chen, Jingjing Wang, Wenhui Zhu +7
Black-box knowledge distillation for large language models presents a strict trade-off. Simple off-policy methods (e.g., sequence-level knowledge distillation) struggle to correct…
Climate Prompting: Generating the Madden-Julian Oscillation using Video Diffusion and Low-Dimensional Conditioning
Sulian Thual, Feiyang Cai, Jingjing Wang +1
Generative Deep Learning is a powerful tool for modeling of the Madden-Julian oscillation (MJO) in the tropics, yet its relationship to traditional theoretical frameworks remains p…
Integration Flow Models
Jingjing Wang, Dan Zhang, Joshua Luo +2
Ordinary differential equation (ODE) based generative models have emerged as a powerful approach for producing high-quality samples in many applications. However, the ODE-based met…
Unified Directly Denoising for Both Variance Preserving and Variance Exploding Diffusion Models
Jingjing Wang, Dan Zhang, Feng Luo
Previous work has demonstrated that, in the Variance Preserving (VP) scenario, the nascent Directly Denoising Diffusion Models (DDDM) can generate high-quality images in one step w…