Zhiyuan Liu
Professor
Works on large language models, knowledge graphs and semantic computing, and computational social science.
Zhiyuan Liu's lab at Tsinghua works on large language models, knowledge, QA & retrieval and graph mining & networks. 203 papers since 2023, 133 at top venues, mostly ACL and ICLR. 21 PhD students and 3 master's students. 27 alumni, most often to ByteDance and Tencent. Applications close .
At a glance
- papers, 2023–2026
- 53/ year
- at T1 venues
- 66%
- PhD students
- 21
- MS students
- 3
- alumni listed
- 27
Papers by yearfrom DBLP, 2023–2026
- 28 papers2026 · T1 17 · collecting
- 66 papers2025 · T1 45
- 63 papers2024 · T1 43
- 46 papers2023 · T1 28
T1 venuesOtherVenues: ACL, ICLR, Findings of ACL, NeurIPS, EMNLP
Research focuspaper titles that use each topic's words
- Large language models93 papers
- Knowledge, QA & retrieval28 papers
- Graph mining & networks6 papers
- social science0 papers
- computational social0 papers
Paper list
From DBLP.
202628 papers
- Trans. Mach. Learn. Res.ACDiT: Interpolating Autoregressive Conditional Modeling and Diffusion Transformer (opens in a new tab)
- ACL (T1)APB-V: Accelerating Long-Video Understanding via Sequence-Parallelism-aware Approximate Attention (opens in a new tab)
- ACL (T1)AutoReproduce: Automatic AI Experiment Reproduction with Paper Lineage (opens in a new tab)
- ECCV (T1)Cheers[inline-graphic not available: see fulltext] : Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation (opens in a new tab)
- Frontiers Comput. Sci.ColdChat: benchmarking large language model personalization using limited real-user interaction history (opens in a new tab)
- Findings of ACLEnhancing Long-Chain Reasoning Distillation through Error-Aware Self-Reflection (opens in a new tab)
- AAAI (T1)FANoise: Singular Value-Adaptive Noise Modulation for Robust Multimodal Representation Learning (opens in a new tab)
- ICLR (T1)From f(x) and g(x) to f(g(x)): LLMs Learn New Skills in RL by Composing Old Ones (opens in a new tab)
- J. Comput. Sci. Technol.From MOOC to MAIC: Reimagine Online Teaching and Learning Through LLM-Driven Agents (opens in a new tab)
- Findings of ACLGROLE: Instance-Level Group Relative Optimization for LoRA Experts in Incremental Learning (opens in a new tab)
- ICLR (T1)Hierarchical Semantic-Acoustic Modeling via Semi-Discrete Residual Representations for Expressive End-to-End Speech Synthesis (opens in a new tab)
- ICLR (T1)How Far Can Unsupervised RLVR Scale LLM Training? (opens in a new tab)
202566 papers
- NeurIPS (T1)3D-GSRD: 3D Molecular Graph Auto-Encoder with Selective Re-mask Decoding (opens in a new tab)
- ICLR (T1)A Multi-Power Law for Loss Curve Prediction Across Learning Rate Schedules (opens in a new tab)
- NeurIPS (T1)A*-Thought: Efficient Reasoning via Bidirectional Compression for Low-Resource Settings (opens in a new tab)
- ICLR (T1)Advancing LLM Reasoning Generalists with Preference Trees (opens in a new tab)
- EMNLP (T1)AgentCPM-GUI: Building Mobile-Use Agents with Reinforcement Fine-Tuning (opens in a new tab)
- ACL (T1)AgentRM: Enhancing Agent Generalization with Reward Modeling (opens in a new tab)
- Int. J. Appl. Earth Obs. GeoinformationAnomalous far field ground subsidence induced by the 2023 Mw 6.8 Morocco earthquake (opens in a new tab)
- ACL (T1)APB: Accelerating Distributed Long-Context Inference by Passing Compressed Context Blocks across GPUs (opens in a new tab)
- NAACL (T1)AutoClean: LLMs Can Prepare Their Training Corpus (opens in a new tab)
- KDD (T1)Awaking the Slides: A Tuning-free and Knowledge-regulated AI Tutoring System via Language Model Coordination (opens in a new tab)
- SCBurstEngine: An efficient distributed framework for training transformers On extremely Long sequences of over 1M tokens (opens in a new tab)
- ACL (T1)ChartCoder: Advancing Multimodal Large Language Model for Chart-to-Code Generation (opens in a new tab)
202463 papers
- CSCWDABCF: An Adaptive Balanced Multimodal Website Classification Framework (opens in a new tab)
- ECCV (T1)AdaNAT: Exploring Adaptive Policy for Token-Based Image Generation (opens in a new tab)
- ICLR (T1)AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors (opens in a new tab)
- Findings of EMNLPBeyond Natural Language: LLMs Leveraging Alternative Formats for Enhanced Reasoning and Communication (opens in a new tab)
- EMNLP (T1)Beyond the Turn-Based Game: Enabling Real-Time Conversations with Duplex Models (opens in a new tab)
- NeurIPS (T1)Can Large Language Models Analyze Graphs like Professionals? A Benchmark, Datasets and Models (opens in a new tab)
- ICLR (T1)Chain of Thought Empowers Transformers to Solve Inherently Serial Problems (opens in a new tab)
- ACL (T1)ChatDev: Communicative Agents for Software Development (opens in a new tab)
- ICLR (T1)ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate (opens in a new tab)
- Bioinform.ChatMol: interactive molecular discovery with natural language (opens in a new tab)
- EMNLP (T1)Controllable Preference Optimization: Toward Controllable Multi-Objective Alignment (opens in a new tab)
- Mach. Intell. Res.Correction to: Red Alarm for Pre-trained Models: Universal Vulnerability to Neuron-level Backdoor Attacks (opens in a new tab)
202346 papers
- ACL (T1)A Close Look into the Calibration of Pre-trained Language Models (opens in a new tab)
- ACL (T1)Augmentation-Adapted Retriever Improves Generalization of Language Models as Generic Plug-In (opens in a new tab)
- EMNLP (T1)Beat LLMs at Their Own Game: Zero-Shot LLM-Generated Text Detection via Querying ChatGPT (opens in a new tab)
- Findings of EMNLPBoosting Inference Efficiency: Unleashing the Power of Parameter-Shared Pre-trained Language Models (opens in a new tab)
- Findings of EMNLPCREATOR: Tool Creation for Disentangling Abstract and Concrete Reasoning of Large Language Models (opens in a new tab)
- ACL (T1)Decoder Tuning: Efficient Language Understanding as Decoding (opens in a new tab)
- Findings of ACLEmergent Modularity in Pre-trained Transformers (opens in a new tab)
- EMNLP (T1)Enhancing Chat Language Models by Scaling High-quality Instructional Conversations (opens in a new tab)
- TKDE (T1)Event Detection With Dynamic Word-Trigger-Argument Graph Neural Networks (opens in a new tab)
- ACL (T1)Exploring Lottery Prompts for Pre-trained Language Models (opens in a new tab)
- EMNLP (T1)Exploring the Impact of Model Scaling on Parameter-Efficient Tuning (opens in a new tab)
- Findings of ACLFew-shot Classification with Hypersphere Modeling of Prototypes (opens in a new tab)
Where alumni went27 listed, 56% to industry
Industry 10Academia 8
Most often to
- ByteDance
- Tencent
- Alibaba
Also to
- Inner Mongolia University of Finance and Economics
- Nanjing University
Apply to TsinghuaFall 2027
Computer Science PhD (international applicants)
- Round
- Main
- Opens
- GRE
- See program page
- English
- Per school catalog
- Fee
- RMB 800
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USUIUC
Heng Ji
- Knowledge, QA & retrieval
- CNSJTU
Hao He
- Knowledge, QA & retrieval
- USCMU
Chenyan Xiong
- Large language models
- CNSJTU
Xiaodong Gu
- Large language models
- GBEdinburgh
Jeff Dalton
- Knowledge, QA & retrieval
- USUSC
Morteza Dehghani
- Language & society
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- DBLP
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
Tsinghuacloses