Realistic Unsupervised CLIP Fine-tuning with Universal Entropy Optimization
Comments ICML 2024 Highlight
期刊&会议
ICML
International Conference on Machine Learning · 会议 · Machine Learning
Sparse-IFT: Sparse Iso-FLOP Transformations for Maximizing Training Efficiency
Comments 14 pages, 5 figures, 6 Tables (Main Paper) + 8 pages (Supplementary Material). Published at ICML 2024
Improving SAM Requires Rethinking its Optimization Formulation
Comments International Conference on Machine Learning (ICML), 2024
Whispering Experts: Neural Interventions for Toxicity Mitigation in Language Models
Comments ICML 2024, 8 pages + appendix
Subequivariant Reinforcement Learning in 3D Multi-Entity Physical Environments
Comments ICML 2024
Private and Federated Stochastic Convex Optimization: Efficient Strategies for Centralized Systems
Comments To be published in ICML 2024
Why Do You Grok? A Theoretical Analysis of Grokking Modular Addition
Comments Accepted by ICML 2024
When can transformers compositionally generalize in-context?
Comments ICML 2024 workshop on Next Generation of Sequence Modeling Architectures
Interpretability in Action: Exploratory Analysis of VPT, a Minecraft Agent
Comments Mechanistic Interpretability Workshop at ICML 2024
Improving AlphaFlow for Efficient Protein Ensembles Generation
Comments Accepted by ICML 2024 AI4Science workshop
Learning High-Frequency Functions Made Easy with Sinusoidal Positional Encoding
Comments 16 pages, Conference, Accepted by ICML 2024
On Stronger Computational Separations Between Multimodal and Unimodal Machine Learning
Comments Appeared in ICML 2024. Camera-ready version
Think Big, Generate Quick: LLM-to-SLM for Fast Autoregressive Decoding
Comments Work presented at the ES-FoMo II Workshop at ICML 2024
Nearest Neighbour Score Estimators for Diffusion Generative Models
Comments 25 pages, 9 figures. To be published in ICML 2024
Fine-tuning Reinforcement Learning Models is Secretly a Forgetting Mitigation Problem
Comments ICML 2024 Spotlight
Non-Vacuous Generalization Bounds for Large Language Models
Comments ICML 2024
Learning in Deep Factor Graphs with Gaussian Belief Propagation
Comments Accepted at ICML 2024
IW-GAE: Importance Weighted Group Accuracy Estimation for Improved Calibration and Model Selection in Unsupervised Domain Adaptation
Comments ICML 2024
Symmetric Replay Training: Enhancing Sample Efficiency in Deep Reinforcement Learning for Combinatorial Optimization
Comments International Conference on Machine Learning
Position: Considerations for Differentially Private Learning with Large-Scale Public Pretraining
Comments Full and unabridged version of paper ICML 2024
Context-Guided Diffusion for Out-of-Distribution Molecular and Protein Design
Comments Published in the Proceedings of the 41st International Conference on Machine Learning (ICML 2024)
Large Language Models as Misleading Assistants in Conversation
Comments Next Generation of AI Safety Workshop, 41st International Conference on Machine Learning (ICML 2024)
Local Feature Selection without Label or Feature Leakage for Interpretable Machine Learning Predictions
Comments Published at ICML 2024
A Theoretical Formulation of Many-body Message Passing Neural Networks
Comments Accepted as a paper at ICML GRaM Workshop, 2024
Fine-Tuning Medical Language Models for Enhanced Long-Contextual Understanding and Domain Expertise
Comments 5 pages, 1 figure. Accepted by the Workshop on Long-Context Foundation Models (LCFM) at ICML 2024
Scaling Exponents Across Parameterizations and Optimizers
Comments 63 pages, International Conference on Machine Learning 2024
Amortizing Pragmatic Program Synthesis with Rankings
Comments icml 2024. This work supersedes and serves as a new version of arXiv:2309.03225
Accelerating Simulation of Two-Phase Flows with Neural PDE Surrogates
Comments Accepted at ICML 2024 AI for Science workshop
Image Clustering with External Guidance
Journal ref ICML 2024 (Oral)
Designing Decision Support Systems Using Counterfactual Prediction Sets
Comments Best paper award in the ICML 2023 AI&HCI Workshop, spotlight paper at ICML 2024
