Lei Li
Associate Professor
Studies large language models, machine translation and AI for science, including efficient LLM inference.
Lei Li's lab at CMU works on large language models, multilingual, translation & linguistics and efficient training & inference. 77 papers since 2023, 52 at top venues, mostly ICML and ICLR. 7 PhD students and 12 master's students. 16 alumni. Applications close .
At a glance
- papers, 2023–2026
- 20/ year
- at T1 venues
- 68%
- PhD students
- 7
- MS students
- 12
- alumni listed
- 16
Papers by yearfrom the lab homepage, 2023–2026
- 13 papers2026 · T1 11 · collecting
- 23 papers2025 · T1 16
- 23 papers2024 · T1 12
- 18 papers2023 · T1 13
T1 venuesOtherVenues: ICML, ICLR, ACL, EMNLP, NAACL
Research focuspaper titles that use each topic's words
- Large language models32 papers
- Multilingual, translation & linguistics19 papers
- Efficient training & inference6 papers
- efficient llm0 papers
- studies large0 papers
Paper list
From the lab homepage.
202613 papers
- IWSLTA Practical Evaluation Method for Long-Form Simultaneous Speech-to-Speech Translation
- ICML (T1)AdaSplash-2: Faster Differentiable Sparse Attention
- ACL (T1)CodeEvo: Interaction-Driven Synthesis of Code-centric Data through Hybrid and Iterative Feedback
- ICLR (T1)HARDTESTGEN: A High-Quality RL Verifier Generation Pipeline for LLM Algorithimic Coding (opens in a new tab)
- ACL (T1)Hierarchical Policy Optimization for Simultaneous Translation of Unbounded Speech
- ICML (T1)InteractScience: Programmatic and Visually-Grounded Evaluation of Interactive Scientific Demonstration Code Generation
- ICML (T1)Is Vibe Coding Safe? Benchmarking Vulnerability of Agent-Generated Code in Real-World Tasks (opens in a new tab)
- ICLR (T1)JanusCoder: Towards a Foundational Visual-Programmatic Interface for Code Intelligence
- ACL-FindingsMending the Holes: Mitigating Reward Hacking in Reinforcement Learning for Multilingual Translation
- ICML (T1)Modeling Attributional Style at Scale: A Dataset and Analysis for Psychological Attribution Assessment and Reframing
- ICML (T1)Sem-Detect: Semantic Level Detection of AI Generated Peer-Reviews
- ICLR (T1)Strategic Planning and Rationalizing on Trees Make LLMs Better Debaters
202523 papers
- Proceedings of Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL) - FindingsA Practical Examination of AI-Generated Text Detectors for Large Language Models
- NAACL (T1)Anticipating Future with Large Language Model for Simultaneous Machine Translation
- the Conference on Empirical Methods in Natural Language Processing (EMNLP) - FindingsBenchMAX: A Comprehensive Multilingual Evaluation Suite for Large Language Models
- the Conference on Empirical Methods in Natural Language Processing (EMNLP): System DemonstrationsBioGraphia: A LLM-Assisted Biological Pathway Graph Annotation Platform
- Proceedings of Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL) - FindingsCA*: Addressing Evaluation Pitfalls in Computation-Aware Latency for Simultaneous Speech Translation
- IWSLTCMU's IWSLT 2025 Simultaneous Speech Translation System (opens in a new tab)
- ICML (T1)DIS-CO: Discovering Copyrighted Content in VLMs Training Data
- ICLR (T1)Diversity Empowers Intelligence: Integrating Expertise of Software Engineering Agents
- Nature Communications (T1)DNALONGBENCH: a benchmark suite for long-range DNA prediction tasks (opens in a new tab)
- ACL (T1)Efficiently Identifying Watermarked Segments in Mixed-Source Texts (opens in a new tab)
- TMLR (T1)Evaluating Durability: Benchmark Insights into Image and Text Watermarking (opens in a new tab)
- COLING (T1)Extrapolating to Unknown Opinions Using LLMs (opens in a new tab)
202423 papers
- EMNLP (T1)A Survey on In-context Learning
- EMNLP (T1)BPO: Supercharging Online Preference Learning by Adhering to the Proximity of Behavior LLM
- IWSLTCMU's IWSLT 2024 Offline Speech Translation System: A Cascaded Approach For Long-Form Robustness
- IWSLTCMU's IWSLT 2024 Simultaneous Speech Translation System
- ICML (T1)DE-COP: Detecting Copyrighted Content in Language Models Training Data
- ICML (T1)Generative Enzyme Design Guided by Functionally Important Sites and Small-Molecule Substrates
- KDD (T1)Global Human-guided Counterfactual Explanations for Molecular Properties via Reinforcement Learning
- ACL-FindingsHire a Linguist!: Learning Endangered Languages in LLMs with In-Context Linguistic Descriptions
- ACL-FindingsHow Vocabulary Sharing Facilitates Multilingualism in LLaMA?
- NeurIPS (T1)Invisible Image Watermarks Are Provably Removable Using Generative AI
- EMNLP (T1)Learning Personalized Alignment for Evaluating Open-ended Text Generation
- the Conference on Empirical Methods in Natural Language Processing (EMNLP) - FindingsLLaMAX: Scaling Linguistic Horizons of LLM by Enhancing Translation Capabilities Beyond 100 Languages
202318 papers
- KDD (T1)Accelerating Antimicrobial Peptide Discovery with Latent Structure
- NeurIPS (T1)ALGO: Synthesizing Algorithmic Programs with Generated Oracle Verifiers
- the Conference on Empirical Methods in Natural Language Processing (EMNLP) - FindingsAutoPlan: Automatic Planning of Interactive Decision-Making Tasks With Large Language Models
- AAAI (T1)Converge to the Truth: Factual Error Correction via Iterative Constrained Editing
- the Conference on Empirical Methods in Natural Language Processing (EMNLP) - FindingsExtrapolating Multilingual Understanding Models as Multilingual Generators
- ICML (T1)Importance Weighted Expectation-Maximization for Protein Sequence Design
- EMNLP (T1)INSTRUCTSCORE: Explainable Text Generation Evaluation with Finegrained Feedback
- EMNLP (T1)Learning from Mistakes via Cooperative Study Assistant for Large Language Models
- ACL-FindingsLego-MT: Learning Detachable Models for Massively Multilingual Machine Translation
- AmericasNLPPlayGround Low Resource Machine Translation System for the 2023 AmericasNLP Shared Task (opens in a new tab)
- ACL (T1)Pre-trained Language Models can be Fully Zero-Shot Learners
- ICML (T1)Protecting Language Generation Models via Invisible Watermarking
Where alumni went16 listed
Apply to CMUFall 2027
CMU SCS Computer Science PhD
- Round
- Single round
- Opens
- GRE
- See program page
- English
- TOEFL / IELTS / Duolingo (F-1/J-1 non-native speakers)
- Fee
- USD 100 (USD 80 by early deadline)
Confirmed on the program page.
Similar labsnearest by research summary
- CNTsinghua
Yiqun Liu
- Knowledge, QA & retrieval
- CNZhejiang
Pei Chen
- Human-AI collaboration & trust
- USUMD
Hernisa Kacorri
- Human-robot interaction & assistive robots
- CNZhejiang
Jian Shao
- Data management & big data
- USUIUC
Kevin Chen-Chuan Chang
- Knowledge, QA & retrieval
- DETUM
Chunyang Chen
- 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.
CMUcloses