Theory

On Statistical Rates of Conditional Diffusion Transformers: Approximation, Estimation and Minimax Optimality ICLR 2025 On the Asymptotic Mean Square Error Optimality of Diffusion Models AISTATS 2025 On the Expressive Power of Sparse Geometric MPNNs ICLR 2025 On the Hardness of Training Deep Neural Networks Discretely AAAI 2025 OpenMIBOOD: Open Medical Imaging Benchmarks for Out-Of-Distribution Detection CVPR 2025 P2 Law: Scaling Law for Post-Training After Model Pruning ACL 2025 Paths and Ambient Spaces in Neural Loss Landscapes AISTATS 2025 Predicting Fine-tuned Performance on Larger Datasets Before Creating Them COLING 2025 Quality over Quantity in Attention Layers: When Adding More Heads Hurts ICLR 2025 Quantitative Approximation for Neural Operators in Nonlinear Parabolic Equations ICLR 2025 Rate of Model Collapse in Recursive Training AISTATS 2025 Relating Piecewise Linear Kolmogorov Arnold Networks to ReLU Networks AISTATS 2025 Rethinking Neural-based Matrix Inversion: Why can’t, and Where can AISTATS 2025 Rethinking the Role of Prompting Strategies in LLM Test-Time Scaling: A Perspective of Probability Theory ACL 2025 Retracing the Past: LLMs Emit Training Data When They Get Lost EMNLP 2025 Revisiting Weak-to-Strong Generalization in Theory and Practice: Reverse KL vs. Forward KL ACL 2025 Scaling Capability in Token Space: An Analysis of Large Vision Language Model JMLR 2025 Scaling Laws Are Unreliable for Downstream Tasks: A Reality Check EMNLP 2025 Scaling Laws for Multilingual Language Models ACL 2025 ScoreFusion: Fusing Score-based Generative Models via Kullback–Leibler Barycenters AISTATS 2025 Seeking Consistent Flat Minima for Better Domain Generalization via Refining Loss Landscapes CVPR 2025 Self-Improvement in Language Models: The Sharpening Mechanism ICLR 2025 Set-Valued Sensitivity Analysis of Deep Neural Networks AAAI 2025 Spectral Insights into Data-Oblivious Critical Layers in Large Language Models ACL 2025 Spectral Scaling Laws in Language Models: emphHow Effectively Do Feed-Forward Networks Use Their Latent Space? EMNLP 2025