Tag: diffusion transformers
Browse our exclusive articles!
MetaSR: Adaptive Metadata for Efficient Super-Resolution
Discover MetaSR, a content-adaptive framework boosting generative super-resolution with efficient metadata orchestration and improved image quality.
Nonlinear Query Projections Boost Transformer Performance
Explore how nonlinear query projections improve transformer models, reducing loss and perplexity without extra parameters.
DiffuMeta: Algebraic Models for Metamaterial Inverse Design
Discover DiffuMeta, a novel framework using algebraic language models and diffusion transformers for efficient inverse design of 3D metamaterials.
Layerwise Dynamics Boost In-Context Classification in Transformers
Discover how layerwise dynamics improve in-context classification and interpretability in transformer models for multi-class tasks.
Spectral Entropy Collapse Signals Delayed Grokking in ML
Discover how spectral entropy collapse predicts delayed generalisation in grokking, revealing key insights into machine learning model behavior and archite...