Hao Peng
Assistant Professor
Studies natural language processing and large language models, including efficient architectures and reasoning.
Hao Peng's lab at UIUC works on large language models, reasoning & agents and accelerators, chips & architecture. 61 papers since 2023, 39 at top venues, mostly ICLR and ICML. 8 PhD students and 3 master's students. Applications close .
At a glance
- papers, 2023–2026
- 16/ year
- at T1 venues
- 64%
- PhD students
- 8
- MS students
- 3
- alumni listed
- —
Papers by yearfrom the lab homepage, 2023–2026
- 26 papers2026 · T1 14 · collecting
- 18 papers2025 · T1 13
- 14 papers2024 · T1 9
- 3 papers2023 · T1 3
T1 venuesOtherVenues: ICLR, ICML, NeurIPS, EMNLP, TMLR
Research focuspaper titles that use each topic's words
- Large language models31 papers
- language19 papers
- Reasoning & agents14 papers
- Accelerators, chips & architecture0 papers
- studies natural0 papers
Paper list
From the lab homepage.
202626 papers
- —ACCORD: Action-Conditioned Contextual Grounding for Language Agents
- —CausaLab: A Scalable Environment for Interactive Causal Discovery Toward AI Scientists
- —Characterizing Rhetorical Misalignment in Decision-Making with Language Models
- EMNLP (T1)Countdown-Code: A Testbed for Studying the Emergence and Generalization of Reward Hacking in RLVR
- —Covert Assistance: Helpful LLM Agents Evade Oversight in Multi-Agent Systems
- ICML (T1)Do We Need Adam? Surprisingly Strong and Sparse Reinforcement Learning with SGD in LLMs
- ICML (T1)Efficient, Property-Aligned Fan-Out Retrieval via RL-Compiled Diffusion
- ICLR (T1)Executable Counterfactuals: Improving LLMs’ Causal Reasoning Through Code
- ICLR (T1)From f(x) and g(x) to f(g(x)): LLMs Learn New Skills in RL by Composing Old Ones
- —Gaming the Judge: Unfaithful Chain-of-Thought Can Undermine Agent Evaluation
- ICLR (T1)Generalization of RLVR Using Causal Reasoning as a Testbed
- ICML (T1)Good SFT Optimizes for SFT, Better SFT Prepares for Reinforcement Learning
202518 papers
- ICLR (T1)A Little Goes a Long Way: Efficient Long Context Training and Inference with Partial Contexts
- —A Survey of Data Attribution: Methods, Applications, and Evaluation in the Era of Generative AI
- —Adaptation of Agentic AI: A Survey of Post-Training, Memory, and Skills
- ICLR (T1)Advancing LLM Reasoning Generalists with Preference Trees
- EMNLP (T1)Context Length Alone Hurts LLM Performance Despite Perfect Retrieval
- ICLR (T1)Eliminating Position Bias of Language Models: A Mechanistic Approach
- EMNLP (T1)FactCheckmate: Preemptively Detecting and Mitigating Hallucinations in LMs
- ICML (T1)Free Process Rewards without Process Labels
- EMNLP (T1)Improving Influence-based Instruction Tuning Data Selection for Balanced Learning of Diverse Capabilities
- —LLMs are Vulnerable to Malicious Prompts Disguised as Scientific Language
- ICLR (T1)OpenHands: An Open Platform for AI Software Developers as Generalist Agents
- —Probing the Critical Point (CritPt) of AI Reasoning: a Frontier Physics Research Benchmark
202414 papers
- TMLR (T1)A Single Transformer for Scalable Vision-Language Modeling
- ACL (T1)ActionIE: Action Extraction from Scientific Literature with Programming Languages
- ICLR (T1)CRAFT: Customizing LLMs by Creating and Retrieving from Specialized Toolsets
- —Data Engineering for Scaling Language Models to 128K Context
- ICML (T1)Executable Code Actions Elicit Better LLM Agents
- NAACL (T1)Language Models Hallucinate, but May Excel at Fact Verification
- NAACL (T1)LeTI: Learning to Generate from Textual Interactions
- NAACL (T1)LM-Infinite: Zero-Shot Extreme Length Generalization for Large Language Models
- Findings of EMNLPMeasuring and Improving Attentiveness to Partial Inputs with Counterfactuals
- ICLR (T1)MINT: Evaluating LLMs in Multi-turn Interaction with Tools and Language Feedback
- —S2-Attention: Hardware-Aware Context Sharding Among Attention Heads
- The Thirty-eight Conference on Neural Information Processing Systems Datasets and Benchmarks TrackSciCode: A Research Coding Benchmark Curated by Scientists
20233 papers
- ICLR (T1)Complexity-Based Prompting for Multi-step Reasoning
- ICML (T1)Specializing Smaller Language Models towards Multi-Step Reasoning
- TACL (T1)Transparency Helps Reveal When Language Models Learn Meaning
Where alumni went
Alumni are not listed on the lab homepage.
Apply to UIUCFall 2027
UIUC Siebel Computer Science PhD
- Round
- Single round
- Opens
- —
- GRE
- optional
- English
- TOEFL iBT / IELTS
- Fee
- USD 90
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USUIUC
Lingming Zhang
- Large language models
- USUMD
Jordan Lee Boyd-Graber
- Knowledge, QA & retrieval
- USMichigan
Dallas Card
- Language & society
- USMichigan
Emily Mower Provost
- Speech recognition & spoken language
- CHETH Zurich
Elliott Ash
- Language & society
- CNPeking
Sujian Li
- Dialogue, generation & summarization
Your notes
Save this lab to keep notes.
Data
Some numbers could not be verified:
- The alumni list could not be found on the lab site.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Not collected
Tiers are venue classifications, not lab rankings.
UIUCcloses