Scaling Laws in Patchification: An Image Is Worth 50,176 Tokens And More
Paper
•
2502.03738
•
Published
•
11
Training Theory
-
-
Better Embeddings with Coupled Adam
Paper • 2502.08441 • Published • 2 -
Make LoRA Great Again: Boosting LoRA with Adaptive Singular Values and Mixture-of-Experts Optimization Alignment
Paper • 2502.16894 • Published • 32 -
SALT: Singular Value Adaptation with Low-Rank Transformation
Paper • 2503.16055 • Published • 8 -
Decoupling Angles and Strength in Low-rank Adaptation
Paper • 2503.18225 • Published • 3 -
Entropy-Based Adaptive Weighting for Self-Training
Paper • 2503.23913 • Published • 3 -
Reinforcement Pre-Training
Paper • 2506.08007 • Published • 263 -
DiffusionBlocks: Blockwise Training for Generative Models via Score-Based Diffusion
Paper • 2506.14202 • Published • 2 -
Selective Contrastive Learning for Weakly Supervised Affordance Grounding
Paper • 2508.07877 • Published • 12 -
Why Low-Precision Transformer Training Fails: An Analysis on Flash Attention
Paper • 2510.04212 • Published • 23