9 papers
Compute Efficiency and Serial Runtime Tradeoffs for Stochastic Momentum Methods
Depen Morwani, Alexandru Meterez, Pranav Nair +1
Stochastic momentum methods such as heavy ball (HB), Nesterov momentum, and variants of Accelerated SGD (ASGD) [Kidambi et al., 2018] are widely used in modern training, but their…
GenAI for Systems: Recurring Challenges and Design Principles from Software to Silicon
Arya Tschand, Chenyu Wang, Zishen Wan +21
Generative AI is reshaping how computing systems are designed, optimized, and built, yet research remains fragmented across software, architecture, and chip design communities. Thi…
Anytime Pretraining: Horizon-Free Learning-Rate Schedules with Weight Averaging
Alexandru Meterez, Pranav Ajit Nair, Depen Morwani +2
Large language models are increasingly trained in continual or open-ended settings, where the total training horizon is not known in advance. Despite this, most existing pretrainin…
Seesaw: Accelerating Training by Balancing Learning Rate and Batch Size Scheduling
Alexandru Meterez, Depen Morwani, Jingfeng Wu +3
Increasing the batch size during training -- a ''batch ramp'' -- is a promising strategy to accelerate large language model pretraining. While for SGD, doubling the batch size can…
A Simplified Analysis of SGD for Linear Regression with Weight Averaging
Alexandru Meterez, Depen Morwani, Costin-Andrei Oncescu +3
Theoretically understanding stochastic gradient descent (SGD) in overparameterized models has led to the development of several optimization algorithms that are widely used in prac…
Echo Chamber: RL Post-training Amplifies Behaviors Learned in Pretraining
Rosie Zhao, Alexandru Meterez, Sham Kakade +3
Reinforcement learning (RL)-based fine-tuning has become a crucial step in post-training language models for advanced mathematical reasoning and coding. Following the success of fr…
The Optimization Landscape of SGD Across the Feature Learning Strength
Alexander Atanasov, Alexandru Meterez, James B. Simon +1
We consider neural networks (NNs) where the final layer is down-scaled by a fixed hyperparameter . Recent work has identified as controlling the strength of feature learning…
Super Consistency of Neural Network Landscapes and Learning Rate Transfer
Lorenzo Noci, Alexandru Meterez, Thomas Hofmann +1
Recently, there has been growing evidence that if the width and depth of a neural network are scaled toward the so-called rich feature learning limit (\mup and its depth extension)…
Towards Training Without Depth Limits: Batch Normalization Without Gradient Explosion
Alexandru Meterez, Amir Joudaki, Francesco Orabona +3
Normalization layers are one of the key building blocks for deep neural networks. Several theoretical studies have shown that batch normalization improves the signal propagation, b…