Chenyan Xiong

RePro: Training Language Models to Faithfully Recycle the Web for Pretraining

What Generative Search Engines Like and How to Optimize Web Content Cooperatively

AutoRule: Reasoning Chain-of-Thought Extracted Rule-based Rewards Improve Preference Learning

Native Logical and Hierarchical Representations with Subspace Embeddings

DeepResearchGym: A Free, Transparent, and Reproducible Evaluation Sandbox for Deep Research

Efficient Multi-Agent System Training with Data Influence-Oriented Tree Search

Semi-structured LLM Reasoners Can Be Rigorously Audited

Intercept Cancer: Cancer Pre-Screening with Large Scale Healthcare Foundation Models

Midtraining Bridges Pretraining and Posttraining Distributions

Beneficial Reasoning Behaviors in Agentic Search and Effective Post-training to Obtain Them

Fairshare Data Pricing via Data Valuation for Large Language Models

Group-Level Data Selection for Efficient Pretraining

ParamMute: Suppressing Knowledge-Critical FFNs for Faithful Retrieval-Augmented Generation

Montessori-Instruct: Generate Influential Training Data Tailored for Student Learning

Harnessing Webpage UIs for Text-Rich Visual Understanding

RAG-DDR: Optimizing Retrieval-Augmented Generation Using Differentiable Data Rewards