Yoon Kim
Associate Professor
Studies natural language processing and machine learning, including efficient training and deployment of large-scale models.
Yoon Kim's lab at MIT works on efficient training & inference. 60 papers since 2023, 45 at top venues, mostly ICLR and ICML. 9 PhD students. 6 alumni. Applications close .
At a glance
- papers, 2023–2026
- 16/ year
- at T1 venues
- 75%
- PhD students
- 9
- MS students
- 0
- alumni listed
- 6
Papers by yearfrom the lab homepage, 2023–2026
- 12 papers2026 · T1 11 · collecting
- 10 papers2025 · T1 10
- 25 papers2024 · T1 16
- 13 papers2023 · T1 8
T1 venuesOtherVenues: ICLR, ICML, ACL, EMNLP, Findings of EMNLP
Research focuspaper titles that use each topic's words
- Efficient training & inference10 papers
- training5 papers
- efficient training1 paper
- studies natural0 papers
- large-scale0 papers
Paper list
From the lab homepage.
202612 papers
- ICLR (T1)Beyond Binary Rewards: Training LMs to Reason About Their Uncertainty
- ICLR (T1)Distilling to Hybrid Attention Models via KL-Guided Layer Selection
- ICML (T1)DR Tulu: Reinforcement Learning with Evolving Rubrics for Deep Research
- ICML (T1)Fast KV Compaction via Attention Matching
- ICML (T1)FlowBot: Inducing LLM Workflows with Bilevel Optimization and Textual Gradients
- ACL (T1)Implicit Representations of Grammaticality in Language Models
- ICLR (T1)Learning to Interpret Weight Differences in Language Models
- ACL (T1)LLM as a Risk Manager: LLM Semantic Filtering for Lead-Lag Trading in Prediction Markets
- ICLR (T1)Log-Linear Attention
- Findings of ACLParallel-SFT: Improving Zero-Shot Cross-Programming-Language Transfer for Code RL
- ICML (T1)Physiology as Language: Translating Respiration to Sleep EEG
- ICML (T1)Reaching Beyond the Mode: RL for Distributional Reasoning in Language Models
202510 papers
- ICML (T1)Ladder-residual: parallelism-aware architecture for accelerating large model inference with communication overlapping
- ICML (T1)On the Duality between Gradient Transformations and Adapters
- EMNLP (T1)On the Same Wavelength? Evaluating Pragmatic Reasoning in Language Models across Broad Concepts
- NeurIPS (T1)PaTH Attention: Position Encoding via Accumulating Householder Transformations
- EMNLP (T1)reWordBench: Benchmarking and Improving the Robustness of Reward Models with Transformed Inputs
- NeurIPS (T1)Self-Adapting Language Models
- ICLR (T1)The Semantic Hub Hypothesis: Language Models Share Semantic Representations Across Languages and Modalities
- ICML (T1)The Surprising Effectiveness of Test-Time Training for Few-Shot Learning
- ICLR (T1)Training-Free Activation Sparsity in Large Language Models
- CVPR (T1)Vision-Language Models Do Not Understand Negation
202425 papers
- Findings of ACLCan You Learn Semantics Through Next-Word Prediction? The Case of Entailment
- Findings of ACLCHAMP: A Competition-level Dataset for Fine-Grained Analyses of LLMs' Mathematical Reasoning Capabilities
- ICML (T1)Data Engineering for Scaling Language Models to 128K Context
- ICLR (T1)DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Models
- Findings of EMNLPFast Matrix Multiplications for Lookup Table-Quantized LLMs
- ICML (T1)Gated Linear Attention Transformers with Hardware-Efficient Training
- EMNLP (T1)Improving Dialogue Agents by Decomposing One Global Explicit Annotation with Local Implicit Multimodal Feedback
- ICML (T1)In-Context Language Learning: Architectures and Algorithms
- Findings of NAACLLangNav: Language as a Perceptual Representation for Navigation
- ACL (T1)Learning to Decode Collaboratively with Multiple Language Models
- ACL (T1)Leveraging Large Language Models for Learning Complex Legal Concepts through Storytelling
- COLMLLM economicus? Mapping the Behavioral Biases of LLMs via Utility Theory
202313 papers
- NeurIPS (T1)Aging with GRACE: Lifelong Model Editing with Discrete Key-Value Adaptors
- CoNLLAlignment via Mutual Information
- ASRUAudio-Visual Neural Syntax Acquisition
- ACL (T1)Deriving Language Models from Masked Language Models
- ACL (T1)Entailment as Robust Self-Learner
- Findings of EMNLPExplain-then-Translate: An Analysis on Improving Program Translation with Self-generated Explanations
- ICLR (T1)Federated Learning as Variational Inference: A Scalable Expectation Propagation Approach
- NeurIPS (T1)Grammar Prompting for Domain-Specific Language Generation with Large Language Models
- ICLR (T1)Learning to Grow Pretrained Models for Efficient Transformer Training
- ICLR (T1)Multitask Prompt Tuning Enables Parameter-efficient Transfer Learning
- Findings of EMNLPSAIL: Search-Augmented Instruction Learning
- Findings of EMNLPSimple Hardware-Efficient PCFGs with Independent Left and Right Productions
Where alumni went6 listed
Apply to MITFall 2027
MIT EECS Computer Science PhD
- Round
- Single round
- Opens
- —
- GRE
- See program page
- English
- TOEFL/IELTS
- Fee
- USD 90
Confirmed on the program page.
Similar labsnearest by research summary
- GBEdinburgh
Jeff Dalton
- Knowledge, QA & retrieval
- USGeorgia Tech
Alan Ritter
- Language & society
- SGNTU
Wei Lu
- Large language models
- USCMU
Chenyan Xiong
- Large language models
- USUMD
Douglas Oard
- Knowledge, QA & retrieval
- CNSJTU
Lu 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.
MITcloses