Naihao(Neo) Deng
PhD student
Computer Science of Engineering
University of Michigan
dnaihao@umich.edu
Home | Publications | CV | Services | Google Scholar | Teaching | Mentoring
I am working on Natural Language Processing in the Language and Information Technologies (LIT) lab at University of Michigan, advised by Rada Mihalcea.
News
- 11/2026: Our research on the energy costs of multilingual LLM inference, The Language–Energy Divide, was reported by Michigan Engineering News.
- 8/2026: 3 papers accepted at EMNLP 2026 (main): The Language–Energy Divide: Measuring Energy Costs of Multilingual LLM Inference, One Example Is Enough to Pass Fairness Benchmarks: Rethinking Fairness Evaluation for Aligned LLMs, and AtlasNLP: A Country-Aware Atlas of Dataset Representation in NLP.
- 7/2026: I will give an invited talk at POSTECH on July 28.
- 7/2026: Attending ACL 2026 in San Diego, USA.
- 7/2026: 2 papers accepted at COLM 2026: It Takes One to Bias Them All: Breaking Bad with One-Shot GRPO and One Word at a Time: Incremental Completion Decomposition Breaks LLM Safety.
- 4/2026: Attending Midwest Speech and Language Days (MSLD) 2026 at UIUC in Urbana-Champaign, USA.
- 4/2026: Our challenge SemEval-2026 Task 1: MWAHAHA, Models Write Automatic Humor And Humans Annotate appears at the 20th International Workshop on Semantic Evaluation (SemEval 2026).
- 4/2026: 2 papers accepted at ACL 2026 Findings: SafetyALFRED: Evaluating Safety-Conscious Planning of Multimodal Large Language Models and Beyond Screenshots: Evaluating VLMs' Understanding of UI Animations.
- 3/2026: Attending EACL 2026 in Rabat, Morocco.
- 3/2026: What Really Matters for Table LLMs? A Meta-Evaluation of Model and Data Effects gets accepted to EACL 2026 Findings.
- 9/2025: Domain Generalization: A Tale of Two ERMs gets accepted to Reliable ML@NeurIPS 2025 and DeepMath 2025.
- 8/2025: Benchmarking and Improving LLM Robustness for Personalized Generation gets accepted to EMNLP 2025 Findings.
- 6/2025: Our task proposal on humor generation is accepted to SemEval 2026.
- 6/2025: We release TAMA models in Rethinking Table Instruction Tuning.
- 4/2025: 3 papers accepted at ACL 2025 TRL.
- 4/2025: 3 papers accepted at ACL 2025 Findings.
- 6/2024: Super excited to introduce Chumor 1.0: A Truly Funny and Challenging Chinese Humor Understanding Dataset from Ruo Zhi Ba, stay tuned on our dataset release.
- 5/2024: Start as a research intern at Amazon Inc, New York. Happy to re-connect with my old friends in New York and make new friends here.
- 5/2024: Our paper Tables as Texts or Images: Evaluating the Table Reasoning Ability of LLMs and MLLMs gets accepted at ACL 2024 Findings.
- 3/2024: Our paper Has It All Been Solved? Open NLP Research Questions Not Solved by Large Language Models gets accepted at LREC-COLING 2024.
- 3/2024: Check out our paper SQL-CRAFT: Text-to-SQL through Interactive Refinement and Enhanced Reasoning.
- 2/2024: Check out our paper Tables as Images? Exploring the Strengths and Limitations of LLMs on Multimodal Representations of Tabular Data.
- 10/2023: Poster for You Are What You Annotate: Towards Better Models through Annotator Representations wins the Best Responsible AI Award at 2023 UM AI Symposium.
- 10/2023: 3 papers accepted at EMNLP 2023. 1 Main Conference, 2 Findings.
- 10/2023: Pass my Prelim Exam.
- 9/2023: 1 paper accepted at IJCNLP-AACL 2023.
- 6/2023: 1 paper accepted at ToM workshop at ICML 2023.
- 1/2023: Start as a PhD student at University of Michigan.