Statistical Learning
Task-Robust Pre-Training for Worst-Case Downstream Adaptation Jianghui Wang, Y ang Chen
Pre-training has achieved remarkable success when transferred to downstream tasks. In machine learning, we care about not only the good performance of a model but also its behavior under reasonable shifts of condition.
White-Box Transformers via Sparse Rate Reduction
In Section 2.2 we show, using an idealized model for the token distribution, that if one iteratively
Co-AdaptationofAlgorithmicand ImplementationalInnovationsin Inference-basedDeepReinforcementLearning
Recently many algorithms were devised for reinforcement learning (RL) with functionapproximation.
