About Me
I am a tenure-track Assistant Professor in the Department of Computer Science at the University of Virginia (UVA). Before joining UVA in 2024, I earned my Ph.D. from the University of Illinois Urbana-Champaign (UIUC), where I was advised by Jiawei Han. During my Ph.D., I also spent time as a visiting researcher with the Princeton NLP Group, working with Danqi Chen.
I am looking for self-motivated PhD students and interns! Please fill out this form if you are interested in working with me. After completing the form, you are also welcome to reach out via email. I will read all submitted forms and emails but I do apologize for not being able to respond to each of them!
Research
My research is dedicated to developing more capable, efficient, and aligned Large Language Models (LLMs). I work across the entire LLM lifecycle, including training paradigms, data and inference efficiency, and the foundations of representations.
Post-Training: Aligning and Enhancing LLMs
My recent work designs better post-training algorithms to improve reasoning, factuality, preference alignment, and model-based evaluation.
- arXiv 2025 Zhu et al. The Surprising Effectiveness of Negative Reinforcement in LLM Reasoning
- arXiv 2025 Chen et al. Do LLM Evaluators Prefer Themselves for a Reason?
- ICLR 2025 Wei et al. InstructRAG: Instructing Retrieval-Augmented Generation via Self-Synthesized Rationales
- NeurIPS 2024 Meng et al. SimPO: Simple Preference Optimization with a Reference-Free Reward
Efficiency: Overcoming Data and Inference Bottlenecks
My research addresses critical bottlenecks in data efficiency and inference efficiency, from synthetic data generation to faster decoding.
- ICML 2025 Wei et al. AdaDecode: Accelerating LLM Decoding with Adaptive Layer Parallelism
- ICML 2023 Meng et al. Tuning Language Models as Training Data Generators for Augmentation-Enhanced Few-Shot Learning
- NeurIPS 2022 Meng et al. Generating Training Data with Language Models: Towards Zero-Shot Language Understanding
Foundations of Representation Learning
My work investigates the core principles of representation learning, uncovers limitations in language model representations, and proposes novel pre-training objectives to build more robust and capable foundation models.
- ICLR 2024 Meng et al. Representation Deficiency in Masked Language Modeling
- ICLR 2022 Meng et al. Pretraining Text Encoders with Adversarial Mixture of Training Signal Generators
- NeurIPS 2021 Meng et al. COCO-LM: Correcting and Contrasting Text Sequences for Language Model Pretraining
News
- 2025 ServiceArea Chair: ICLR, ICML, COLM, NeurIPS. Action Editor: TMLR.
- 2025.05Two papers on Accelerating LLM Decoding (AdaDecode) and Retrieval-Inspired Preference Alignment (LarPO) accepted to ICML 2025!
- 2025.04Honored to be named to the Forbes 30 Under 30 2025 Asia list (Healthcare & Science)!
- 2025.01One paper on Retrieval-Augmented Generation (InstructRAG) accepted to ICLR 2025!
- 2024.09Two papers on Preference Optimization (SimPO) and Contrastive Decoding in MoE (SCMoE) accepted to NeurIPS 2024!
- 2024.09Two papers on Zero-Shot Relation Extraction and LLM Persona Survey accepted to EMNLP 2024 Main Conference/Findings!
- 2024.08My Ph.D. thesis won the ACM SIGKDD 2024 Dissertation Award!
Education
- Ph.D. (2023) Computer Science, University of Illinois Urbana-Champaign
- Thesis: Efficient and Effective Learning of Text Representations Award ACM SIGKDD 2024 Dissertation Award
- M.S. (2019) Computer Science, University of Illinois Urbana-Champaign
- Thesis: Weakly-Supervised Text Classification
- B.S. (2017) Computer Engineering, University of Illinois Urbana-Champaign
- Graduated with Highest Honor & Bronze Tablet
Contact
- Email: yumeng5[at]virginia[dot]edu
- Office: Rice Hall 408, 85 Engineer's Way, Charlottesville, Virginia 22903