Weiwei Sun

RankGPT. Is ChatGPT Good at Search? LLMs as Re-Ranking Agent [EMNLP 2023 Outstanding Paper Award]

FoldAgent. [ICML'26] Scaling Long-Horizon LLM Agent via Context-Folding

PPP-Agent. Training Proactive and Personalized LLM Agents

OdysSim. Building Foundation Models for Human Behavior Simulation

GenRet. Learning to Tokenize for Generative Retrieval (NeurIPS 2023)

CO-Bench. [AAAI 2026] Benchmarking Language Model Agents in Algorithm Search for Combinatorial Optimization

user-satisfaction-simulation. "Simulating User Satisfaction for the Evaluation of Task-oriented Dialogue Systems" in SIGIR'21

MAIR. MAIR: A Massive Benchmark for Evaluating Instructed Retrieval. Evaluate your retrieval models on 126 diverse tasks. [EMNLP 2024]

MixCL. Contrastive Learning Reduces Hallucination in Conversations

AmbigPrompt. Answering Ambiguous Questions via Iterative Prompting

AirRep. [NeurIPS 2025 Spotlight] Enhancing Training Data Attribution with Representational Optimization

FrontierCO. Python

ZeroGR. [ICLR 2026] ZeroGR: A Generalizable and Scalable Framework for Zero-Shot Generative Retrieval

GenKS. Code for "Generative Knowledge Selection for Knowledge-Grounded Dialogues"

MetaSim. Metaphorical User Simulation for Task-Oriented Dialogue Evaluation

ckgc. "Conversations Powered by Cross-Lingual Knowledge" in SIGIR'21

spanish-reddit-dialogues-corpus. A Spanish Reddit dialogues corpus, constructed using Reddit comments of 2019.

colorful-2048-game. A colorful terminal-based implementation of the classic 2048 game with Python and Colorama

AgentArena. Python

codex-python. A Python port of Codex’s open-source Rust agent runtime.