He He
Associate Professor of Computer Science and Data Science
Studies natural language processing and machine learning, focusing on robustness, reasoning and the trustworthiness of language models.
He He's lab at NYU works on interpretability, evaluation & safety of LMs, large language models and trustworthy & safe ML. 39 papers since 2023, 27 at top venues, mostly ICLR and ACL. 7 PhD students and 1 master's student. 14 alumni, most often to Anthropic and Google. Applications close .
At a glance
- papers, 2023–2026
- 10/ year
- at T1 venues
- 69%
- PhD students
- 7
- MS students
- 1
- alumni listed
- 14
Papers by yearfrom the lab homepage, 2023–2026
- 6 papers2026 · T1 6 · collecting
- 8 papers2025 · T1 4
- 14 papers2024 · T1 11
- 11 papers2023 · T1 6
T1 venuesOtherVenues: ICLR, ACL, ICML, NeurIPS, EACL
Research focuspaper titles that use each topic's words
- Large language models21 papers
- Interpretability, evaluation & safety of LMs1 paper
- Trustworthy & safe ML0 papers
- trustworthiness0 papers
- studies natural0 papers
Paper list
From the lab homepage.
20266 papers
- ICML (T1)Estimating Tail Risks in Language Model Output Distributions
- ICLR (T1)Is It Thinking or Cheating? Detecting Implicit Reward Hacking by Measuring Reasoning Effort
- ICLR (T1)Jailbreak Transferability Emerges from Shared Representations
- ICLR (T1)Measuring LLM Novelty As The Frontier Of Original And High-Quality Output
- ICLR (T1)Monitoring Decomposition Attacks in LLMs with Lightweight Sequential Monitors
- ICML (T1)The Unlearnability Phenomenon in RLVR for Language Models
20258 papers
- ICLR (T1)Adaptive Deployment of Untrusted LLMs Reduces Distributed Threats
- COLMHyperparameter Loss Surfaces Are Simple Near their Optima
- ICLR (T1)Language Models Learn to Mislead Humans via RLHF
- ICLR WorkshopMonitoring LLM Agents for Sequentially Contextual Harm
- NeurIPS (T1)Predicting Empirical AI Research Outcomes with Language Models
- COLMReasoning Models Know When They're Right: Probing Hidden States for Self-Verification
- ICLR (T1)Transformers Struggle to Learn to Search
- Findings of the Association for Computational LinguisticsWhen Benchmarks Talk: Re-Evaluating Code LLMs with Interactive Feedback
202414 papers
- ICML (T1)Do Models Explain Themselves? Counterfactual Simulatability of Natural Language Explanations
- ICLR (T1)Does Writing with Language Models Reduce Content Diversity?
- TMLR (T1)Foundational Challenges in Assuring Alignment and Safety of Large Language Models
- AAAI WorkshopImproving Multi-Hop Reasoning in LLMs by Learning from Rich Human Feedback
- NeurIPS (T1)Iterative Reasoning Preference Optimization
- EACLLeveraging Implicit Feedback from Deployment Data in Dialogue
- EMNLP (T1)LLMs Are Prone to Fallacies in Causal Inference
- TMLR (T1)Nuisances via Negativa: Adjusting for Spurious Correlations via Data Augmentation
- ACL (T1)Parallel Structures in Pre-training Data Yield In-Context Learning
- EMNLP (T1)Personas as a Way to Model Truthfulness in Language Models
- NAACL (T1)Show Your Work with Confidence: Confidence Bands for Tuning Curves
- Nature (T1)Solving Olympiad Geometry without Human Demonstrations
202311 papers
- TVCG\ARGUS\: Visualization of \AI\-Assisted Task Guidance in \AR\
- ACL (T1)Efficient Shapley Values Estimation by Amortization for Text Classification
- ICML (T1)Extrapolative Controlled Sequence Generation via Iterative Refinement
- Findings of the European Chapter of the Association for Computational LinguisticsHow do decoding algorithms distribute information in dialogue responses?
- ICLR (T1)Language Models are Greedy Reasoners: A Systematic Formal Analysis of Chain-of-Thought
- ACL (T1)Measuring Inductive Biases of In-Context Learning with Underspecified Demonstrations
- Findings of the Empirical Methods in Natural Language ProcessingOn the Relation between Sensitivity and Accuracy in In-context Learning
- ML4HPragmatic Radiology Report Generation
- ACL (T1)Reward Gaming in Conditional Text Generation
- EACLRobustification of Multilingual Language Models to Real-world Noise in Crosslingual Zero-shot Settings with Robust Contrastive Pretraining
- NeurIPS (T1)Testing the General Deductive Reasoning Capacity of Large Language Models Using OOD Examples
Where alumni went14 listed, 44% to industry
Industry 4Academia 5
Most often to
- Anthropic
- Google DeepMind
Also to
- NYU
- ETH Zurich
- George Washington University
Apply to NYUFall 2027
Computer Science PhD (Courant / Tandon)
- Round
- Single round (December)
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USUSC
Swabha Swayamdipta
- Interpretability, evaluation & safety of LMs
- GBImperial
Marek Rei
- Language & society
- CNZhejiang
Zhixuan Chu
- Large language models
- CNPeking
Huishuai Zhang
- Large language models
- USUT Austin
Yizhong Wang
- Large language models
- SGNTU
Soujanya Poria
- Large language models
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
NYUcloses