Introducing Harness-1, a 20B search agent trained with a state-externalizing harness. > frontier-level long-horizon search, rivaling Opus-4.6 and outperforming GPT-5.4 > Context-1-level cost and latency > externalizes candidates, evidence, verification, and search history > open-source

CoRank accepted at KDD 2026 🎉🎉 Really grateful to all the collaborators from @dmguiuc — @PatrickXu565299, Bowen @BowenJin13, Prof. Seongku Kang, and Prof. Jiawei Han ❤️❤️

🚨 New Paper Alert! 🚨 Excited to share our latest work: Retrieval is Cheap, Show Me the Code: Executable Multi-Hop Reasoning for Retrieval-Augmented Generation 🐍📚🤖

See you in San Diego for @aclmeeting #ACL2026NLP! 🥂 Thrilled to share that our paper has been accepted. I’m looking forward to sharing more about our research soon!

📰New preprint: How can we build a task-agnostic plug-and-play memory module for LLM agents that supports multiple memory types? We present PlugMem🔌🧠, a plugin memory module that works across tasks by turning heterogeneous experience into knowledge. Evaluated unchanged on long-term dialogue🗣️, multi-hop QA🕵️, and web agents🕸️🤖, PlugMem improves performance while using far fewer memory tokens. 📜Paper: empathyang.github.io/files/P… 🔨Code: github.com/TIMAN-group/PlugM…

Maintaining agent performance over long horizons remains challenging—largely because memory systems fail to associate latent context with intent. 🎉 Introducing our paper: Grounding Agent Memory in Contextual Intent. STITCH achieves 35.6% gains on our new CAME-Bench.

📣 Excited to share #DeepRetrieval - our novel approach using reinforcement learning for query augmentation in information retrieval! 🚀 Our preliminary results (we got on Feb 16) CRUSH previous SOTA: 60.8% vs 24.7% recall on PubMed search engine 70.8% vs 32.1% recall on ClinicalTrial search engine with a SMALLER model (3B vs 7B) 💡NO supervision data: - [no💰] vs [💰💰💰💰...] on creating augmented queries from ChatGPT/Claude! 💻 Github: github.com/pat-jj/DeepRetrie… 📝 Preliminary Technical Report: pat-jj.github.io/assets/pdf/… 🔬 Currently testing on general IR datasets and with dense retrieval methods 📝 Full paper with more results will be released soon. Just created this X account to share this breakthrough - follow for more NLP IR research! #NLP #IR #MachineLearning #LLM #AAAI2025

🚀 Introducing 𝗦𝗲𝗮𝗿𝗰𝗵-𝗥𝟭 – the first 𝗿𝗲𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗼𝗳 𝗗𝗲𝗲𝗽𝘀𝗲𝗲𝗸-𝗥𝟭 (𝘇𝗲𝗿𝗼) for training reasoning and search-augmented LLM agents with reinforcement learning! This is a step towards training an 𝗼𝗽𝗲𝗻-𝘀𝗼𝘂𝗿𝗰𝗲 𝗢𝗽𝗲𝗻𝗔𝗜 “𝗗𝗲𝗲𝗽 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵” via RL. Our 𝟯𝗕 𝗯𝗮𝘀𝗲 𝗟𝗟𝗠𝘀—including not just 𝗤𝘄𝗲𝗻 𝟮.𝟱 but also 𝗟𝗹𝗮𝗺𝗮 𝟯.𝟮—learn to 𝗿𝗲𝗮𝘀𝗼𝗻 and 𝗰𝗮𝗹𝗹 𝘀𝗲𝗮𝗿𝗰𝗵 𝗲𝗻𝗴𝗶𝗻𝗲𝘀 all on their own! Everything will be 𝗳𝘂𝗹𝗹𝘆 𝗼𝗽𝗲𝗻 𝘀𝗼𝘂𝗿𝗰𝗲. Stay tuned! Code: github.com/PeterGriffinJin/S… Experimental logs: wandb.ai/peterjin/Search-R1-… #R1 #deepresearch #deepseek

Super excited that this work has been accepted to the #EMNLP2024 main conference! See you in Miami!🎉

🚀Excited to share "InstructG2I: Synthesizing Images from Multimodal Attributed Graphs" has been accepted by @NeurIPSConf 2024! instructg2i.github.io/ We propose a graph-conditioned stable diffusion model for image generation. GO and PLAY with it! #graph #diffusion #neurips

🎓Successfully defended my Ph.D. thesis! 🎉My deepest gratitude goes to my thesis committee members: Prof. Jiawei Han @dmguiuc, Prof. Tarek Abdelzaher, Prof. Hanghang Tong, Prof. Wei Wang @WeiWang1973, and Dr. Iris Shen!

Happy to announce that TreeInstruct got accepted to EMNLP'24! Excited to discuss the work alongside @wonderingishika as part of a joint collaboration between @dmguiuc and @convai_uiuc. See you all in Miami! #EMNLP2024

Our group has 6 papers accepted to #EMNLP2024 , led by Linyi Ding, SeongKu Kang, @yuz9yuz, @SizheZhou189667 , @priyanka_karg and @Siru_Ouyang respectively. See you in Miami! @emnlpmeeting

Our alumni, Yu Meng, won the #KDD2024 Outstanding Dissertation Award!!! Congratulations on this well-earned distinction, Yu ! 🎉 @yumeng0818 @kdd_news

🚀 Join our tutorial at #KDD2024, Automated Mining of Structured Knowledge from Text with Large Language Models! 👤Presented by @YunyiZhang10, @Siru_Ouyang, Professor Jiawei Han. 📅Aug 25, 10 AM - 1 PM CEST 📍 Room 129-130

📢We have finally turned our "awesome" GitHub repository (290 stars already) into a survey of 𝐒𝐜𝐢𝐞𝐧𝐭𝐢𝐟𝐢𝐜 𝐋𝐋𝐌𝐬 and their applications in 𝐒𝐜𝐢𝐞𝐧𝐭𝐢𝐟𝐢𝐜 𝐃𝐢𝐬𝐜𝐨𝐯𝐞𝐫𝐲! #LLM #AI4Science Paper: arxiv.org/abs/2406.10833 GitHub: github.com/yuzhimanhua/Aweso…

Our group has 2 papers on fine-grained entity typing accepted to #KDD2024, led by @tanaykomarlu and @Siru_Ouyang, respectively. See you in Barcelona! @kdd_news

🚀Excited to share "Language Models as Semantic Indexers" is accepted to ICML 2024! ⭐️We propose to learn document semantic IDs with large language models in a self-supervised fashion. ⭐️The learned semantic IDs can benefit LLM generative recommendation and retrieval. #LLM #IR

Our group has 3 papers (1 main 2 findings) accepted to #ACL2024, led by @XianruiZhong, @Yizhu_Jiao, and @BowenJin13, respectively. See you in Bangkok! @emnlpmeeting

Prof. Jiawei Han received the Distinguished Research Contributions Award at #PAKDD2024! @pakdd_social Photo credit: Prof. Jian Pei @jian_pei