Industry
Mesa-Extrapolation: A Weave Position Encoding Method for Enhanced Extrapolation in LLMs
LLMs sharply declines beyond their max training lengths.
Classifier Clustering and Feature Alignment for Federated Learning under Distributed Concept Drift
Motivated by the above findings, we propose FedCCFA, a federated learning framework with classifier clustering and feature alignment.
The Implicit Bias of Gradient Descent on Separable Multiclass Data
However, there is a noticeable gap in analysis for multiclass classification, with only a handful of results which themselves are restricted to the cross-entropy loss.
3D-Aware Visual Question Answering about Parts, Poses and Occlusions
Visual question answering ( VQA) is a challenging task that requires an in-depth understanding of vision and language, as well as multi-modal reasoning.
Joint Attribute and Model Generalization Learning for Privacy-Preserving Action Recognition
In this case, accurately recognizing the action requires richer contextual information from the background and surroundings.
Stable Diffusion is Unstable
Recently, text-to-image models have been thriving. Despite their powerful generative capacity, our research has uncovered a lack of robustness in this generation process.
