We show that adaptive optimizers like RMSProp lead to fairer minima than SGD, both theoretically and empirically.

A comprehensive study into verbal confidence in LLMs and its general robustness as well as its use as the objective for adversarial attacks.

Introduced Refined Regularized Preference Optimization with a self-alignment framework to enable fine-grained alignment of large video language models by learning from their own errors.

NormFit is a novel fine-tuning framework designed specifically for few-shot federated learning scenarios characterized by heterogeneous and imbalanced data distributions