Papers

12094 papers

ICML2025

Scaling Sparse Feature Circuits For Studying In-Context Learning

Summary pending...

SAEICLSFC
ICML2025

NTK-DFL: Enhancing Decentralized Federated Learning in Heterogeneous Settings via Neural Tangent Kernel

Summary pending...

Federated LearningDecentralized Federated LearningNeural Tangent Kernel
ICML2025

Tokenized Bandit for LLM Decoding and Alignment

Summary pending...

Contextual BanditMulti-armed BanditLarge Language Model
ICML2025

Accurate Identification of Communication Between Multiple Interacting Neural Populations

Summary pending...

Computational NeuroscienceDynamical SystemsInterpretability
ICML2025

Fully Dynamic Embedding into $\ell_p$ Spaces

Summary pending...

Dynamic algorithmsEmbedding
ICML2025

TuCo: Measuring the Contribution of Fine-Tuning to Individual Responses of LLMs

Summary pending...

Large Language ModelsInterpretabilityAI Safety
ICML2025

Provably Efficient RL for Linear MDPs under Instantaneous Safety Constraints in Non-Convex Feature Spaces

Summary pending...

Reinforcement LearningEpisodic Linear MDPConstrained RL
ICML2025

Permutation-based Rank Test in the Presence of Discretization and Application in Causal Discovery with Mixed Data

Summary pending...

Rank TestDiscretizationCausal Discovery
ICML2025

Grokking at the Edge of Linear Separability

Summary pending...

GrokkingLogistic RegressionInterpolation Threshold
ICML2025

A Certified Unlearning Approach without Access to Source Data

Summary pending...

Certified machine unlearningprivacysurrogate data
ICML2025

Learning Multi-Level Features with Matryoshka Sparse Autoencoders

Summary pending...

sparse autoencodersmechanistic interpretability
ICML2025

A Bayesian Model Selection Criterion for Selecting Pretraining Checkpoints

Summary pending...

transfer learningBayesian model selectionefficient fine-tuning
ICML2025

Optimal Transfer Learning for Missing Not-at-Random Matrix Completion

Summary pending...

Transfer LearningActive LearningMatrix Completion
ICML2025

On the Importance of Gaussianizing Representations

Summary pending...

mutual information gamegaussianizationpower transform
ICML2025

Wait-Less Offline Tuning and Re-solving for Online Decision Making

Summary pending...

Online Linear ProgrammingSequential Decision MakingResource Allocation
ICML2025

Integer Programming for Generalized Causal Bootstrap Designs

Summary pending...

design uncertaintycausal bootstrapinteger programming
ICML2025

Learnings from Scaling Visual Tokenizers for Reconstruction and Generation

Summary pending...

Auto-EncoderVisual CompressionLatent Diffusion Models
ICML2025

Can Transformers Reason Logically? A Study in SAT Solving

Summary pending...

TransformersLogical ReasoningSAT-Solving
ICML2025

Minimax Optimal Regret Bound for Reinforcement Learning with Trajectory Feedback

Summary pending...

reinforcement learningonline learningtrajectory feedback
ICML2025

Star Attention: Efficient LLM Inference over Long Sequences

Summary pending...

Large Language ModelsTransformersAttention Mechanisms