Textual Equilibrium Propagation for Deep Compound AI Systems
Summary TEP uses local prompt optimization to mitigate exploding and vanishing textual gradients in deep, long-horizon compound AI systems.
Can Textual Gradient Work in Federated Learning?
Summary FedTextGrad studies federated LLM prompt optimization with textual gradients, identifying aggregation challenges and improving prompt summarization.
Local Superior Soups: A Catalyst for Model Merging in Cross-Silo Federated Learning
Summary Local Superior Soups improves client training and pre-trained model adaptation in federated learning through regularized model interpolation.
FedSoup: Improving Generalization and Personalization in Federated Learning via Selective Model Interpolation
Summary FedSoup uses selective model interpolation to balance local and global performance, reduce overfitting, and seek flat minima for improved generalization.
VITA: A Multi-Source Vicinal Transfer Augmentation Method for Out-of-Distribution Generalization
Summary VITA generates diverse on-manifold samples through multi-source vicinal transfer to improve robustness to image corruptions.
Benchmarks for Corruption Invariant Person Re-Identification
Summary The first comprehensive study of five ReID corruption benchmarks compares transformer and CNN robustness and examines cross-dataset generalization.