jl.Junru Lu

TENCENT YOUTU LAB · SHANGHAI

Junru Lu陆俊如

Foundation Model Researcher

I work on foundation models, efficient attention, and memory for long-horizon agents.

My research spans model architecture, pre-training, post-training, and evaluation. I received my Ph.D. in Computer Science from the University of Warwick, advised by Prof. Yulan He.

Curriculum vitaeEnglish /中文
Portrait of Junru Lu
Junru Lu 陆俊如Researcher, Tencent

01 / RESEARCH

Selected work

From training models
to making them more efficient.

01 / FOUNDATION MODELSTechnical report

Youtu-LLM

A lightweight language model pre-trained from scratch, with a Commonsense–STEM–Agent curriculum and native reasoning and planning capabilities.

Explore the work
02 / EFFICIENT ATTENTIONICML 2026

SSA

Bidirectional alignment of full and sparse attention outputs during training, addressing the mismatch between the two attention modes.

Explore the work
03 / AGENT MEMORYEMNLP 2026

ContextPilot

Proactive context management for long-horizon agents, with planning, memory, soft offloading, and action-level credit assignment.

Explore the work

Recent updates

Public releases & research

WRITING / TECHNICAL ESSAYS

Ideas, beyond the paper.

All writing & notes

02 / PUBLICATIONS

Papers & technical reports

Full list on Scholar Hugging Face

23 publications & technical reports

2026

Self-Distillation with Verbal Regularization

Jiazheng Li, Yawei Li, Junru Lu, Zhenyi Shen, Di Yin, Lin Gui, Xing Sun, David Rügamer, Xinpeng Wang, Yulan He

PreprintAgents & alignment
Paper
2026

Blind Men and the Elephant: Probing the Epistemic Myopia of LLMs under Long-Tail Divergent Knowledge

Zhuoshi Pan, Junru Lu, Yan Qian, H. Vicky Zhao, Di Yin, Xing Sun

PreprintAgents & alignment
Paper Code
2026

ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL

Zhuoshi Pan, Qizhi Pei, Junru Lu, Honglin Lin, H. Vicky Zhao, Di Yin, Xing Sun

EMNLP 2026 · MainAgents & alignment
Paper Code
2026

Toward Native Multimodal Modeling: A Roadmap

Siyu An, Junru Lu, Junnan Dong, et al.

EMNLP 2026Foundation models
Paper Project
2026

Code-as-Room: Generating 3D Rooms from Top-Down View Images via Agentic Code Synthesis

Yixuan Yang, Zhen Luo, Wanshui Gan, Jinkun Hao, Junru Lu, Jinghao Yan, Zhaoyang Lyu, Xudong Xu

PreprintOther research
Paper
2026

Youtu-VL: Unleashing Visual Potential via Unified Vision-Language Supervision

Zhixiang Wei, Yi Li, Zhehan Kan, et al. (including Junru Lu)

Technical reportFoundation models
Paper
2026

SSA: Sparse Sparse Attention by Aligning Full and Sparse Attention Outputs in Feature Space

Zhenyi Shen, Junru Lu, Lin Gui, Jiazheng Li, Yulan He, Di Yin, Xing Sun

ICML 2026Efficient attention
Paper Code
2025

Youtu-LLM: Unlocking the Native Agentic Potential for Lightweight Large Language Models

Junru Lu, Jiarui Qin, Lingfeng Qiao, et al.

Technical reportFoundation models
Paper
2025

OptiScene: LLM-driven Indoor Scene Layout Generation via Scaled Human-aligned Data Synthesis and Multi-Stage Preference Optimization

Yixuan Yang, Zhen Luo, Tongsheng Ding, Junru Lu, Mingqi Gao, Jinyu Yang, Victor Sanchez, Feng Zheng

NeurIPS 2025Other research
Paper Project
2025

Two Heads Are Better Than One: Dual-Model Verbal Reflection at Inference-Time

Jiazheng Li, Yuxiang Zhou, Junru Lu, Gladys Tyen, Lin Gui, Cesare Aloisi, Yulan He

EMNLP 2025Agents & alignment
Paper Code
2025

RoleMRC: A Fine-Grained Composite Benchmark for Role-Playing and Instruction-Following

Junru Lu, Jiazheng Li, Guodong Shen, Lin Gui, Siyu An, Yulan He, Di Yin, Xing Sun

ACL 2025 · FindingsAgents & alignment
Paper Code
2025

Tell Me What You Don’t Know: Enhancing Refusal Capabilities of Role-Playing Agents via Representation Space Analysis and Editing

Wenhao Liu, Siyu An, Junru Lu, et al.

ACL 2025 · FindingsAgents & alignment
Paper
2025

LLplace: The 3D Indoor Scene Layout Generation and Editing via Large Language Model

Yixuan Yang, Junru Lu, Zixiang Zhao, Zhen Luo, James J.Q. Yu, Victor Sanchez, Feng Zheng

IROS 2025Other research
Paper
2025

FIPO: Free-form Instruction-oriented Prompt Optimization with Preference Dataset and Modular Fine-tuning Schema

Junru Lu, Siyu An, Min Zhang, Yulan He, Di Yin, Xing Sun

COLING 2025Agents & alignment
Paper Code
2024

Advancing Video Anomaly Detection: A Bi-Directional Hybrid Framework for Enhanced Single- and Multi-Task Approaches

Guodong Shen, Yuqi Ouyang, Junru Lu, Yixuan Yang, Victor Sanchez

IEEE Transactions on Image ProcessingOther research
Paper Code
2024

SNFinLLM: Systematic and Nuanced Financial Domain Adaptation of Chinese Large Language Models

Shujuan Zhao, Lingfeng Qiao, Kangyang Luo, Qian-Wen Zhang, Junru Lu, Di Yin

PreprintFoundation models
Paper
2024

Eliminating Biased Length Reliance of Direct Preference Optimization via Down-Sampled KL Divergence

Junru Lu, Jiazheng Li, Siyu An, Meng Zhao, Yulan He, Di Yin, Xing Sun

EMNLP 2024Agents & alignment
Paper Code
2024

FinVerse: An Autonomous Agent System for Versatile Financial Analysis

Siyu An, Qin Li, Junru Lu, Di Yin, Xing Sun

PreprintAgents & alignment
Paper Demo
2023

MemoChat: Tuning LLMs to Use On-the-fly Text Memos for Consistent Long-Range Conversations

Junru Lu, Siyu An, Mingbao Lin, Gabriele Pergola, Yulan He, Di Yin, Xing Sun, Yunsheng Wu

PreprintAgents & alignment
Paper Code
2023

Event Knowledge Incorporation with Posterior Regularization for Event-Centric Question Answering

Junru Lu, Gabriele Pergola, Lin Gui, Yulan He

PreprintOther research
Paper Code
2023

NapSS: Paragraph-level Medical Text Simplification via Narrative Prompting and Sentence-matching Summarization

Junru Lu, Jiazheng Li, Byron Wallace, Yulan He, Gabriele Pergola

EACL 2023 · FindingsOther research
Paper Code
2022

Event-Centric Question Answering via Contrastive Learning and Invertible Event Transformation

Junru Lu, Xingwei Tan, Gabriele Pergola, Lin Gui, Yulan He

EMNLP 2022 · FindingsOther research
Paper Code
2020

CHIME: Cross-passage Hierarchical Memory Network for Generative Review Question Answering

Junru Lu, Gabriele Pergola, Lin Gui, Binyang Li, Yulan He

COLING 2020Agents & alignment
Paper

03 / BACKGROUND

Experience & education

View my CV

EXPERIENCE

Tencent YouTu Lab

Foundation Model Researcher

May 2023 – present · Shanghai, China

Pre-training data pipelines, sparse attention and native multimodal architectures, GPU-aware acceleration kernels, large-scale MoE training, and model evaluation.

LLM / MLLMSparse attentionAgent memory

EDUCATION

  • University of Warwick

    Ph.D. in Computer Science

    Advisor: Prof. Yulan He
    2019–2024

  • New York University

    M.S. in Applied Urban Science & Informatics

    2018–2019

  • Peking University

    Double Bachelor of Economics

    2015–2018

  • University of International Relations

    B.Eng. in Information Management and Information System

    Advisor: Prof. Binyang Li
    2014–2018
Earlier notes2017–2019

Earlier notes on NLP, question answering, algorithms, and learning.

Browse the archive