4 papers
Generalized Gradient Norm Clipping & Non-Euclidean -Smoothness
Thomas Pethick, Wanyun Xie, Mete Erdogan +3
This work introduces a hybrid non-Euclidean optimization method which generalizes gradient norm clipping by combining steepest descent and conditional gradient approaches. The meth…
Training Deep Learning Models with Norm-Constrained LMOs
Thomas Pethick, Wanyun Xie, Kimon Antonakopoulos +3
In this work, we study optimization methods that leverage the linear minimization oracle (LMO) over a norm-ball. We propose a new stochastic family of algorithms that uses the LMO…
Multi-Step Alignment as Markov Games: An Optimistic Online Gradient Descent Approach with Convergence Guarantees
Yongtao Wu, Luca Viano, Yihang Chen +4
Reinforcement Learning from Human Feedback (RLHF) has been highly successful in aligning large language models with human preferences. While prevalent methods like DPO have demonst…
Layer-wise Quantization for Quantized Optimistic Dual Averaging
Anh Duc Nguyen, Ilia Markov, Frank Zhengqing Wu +4
Modern deep neural networks exhibit heterogeneity across numerous layers of various types such as residuals, multi-head attention, etc., due to varying structures (dimensions, acti…