5 papers
Privacy from Symmetry: Orthogonally Equivariant Transformers for LLM Inference
Alexander Yukhimchuk, Andrey Shulga, Mladen Kolar +1
Running large language models locally is often impractical, pushing inference on sensitive text to third-party providers. Split inference partially mitigates this by keeping tokens…
Gradient Clipping Beyond Vector Norms: A Spectral Approach for Matrix-Valued Parameters
Alexander Yukhimchuk, Mladen Kolar, Martin TakÃ¡Ä +1
Gradient clipping is a standard safeguard for training neural networks under noisy, heavy-tailed stochastic gradients; yet, most clipping rules treat all parameters as vectors and…
Muon with Nesterov Momentum: Heavy-Tailed Noise and (Randomized) Inexact Polar Decomposition
Sayantan Choudhury, Xiaoran Cheng, Martin TakÃ¡Ä +2
Most first-order optimizers treat matrix-valued parameters as vectors, ignoring the intrinsic geometry of hidden-layer weights in neural networks. Muon addresses this mismatch by u…
Adaptive Batch Size Schedules for Distributed Training of Language Models with Data and Model Parallelism
Tim Tsz-Kit Lau, Weijian Li, Chenwei Xu +2
An appropriate choice of batch sizes in large-scale model training is crucial, yet it involves an intrinsic yet inevitable dilemma: large-batch training improves training efficienc…
Communication-Efficient Adaptive Batch Size Strategies for Distributed Local Gradient Methods
Tim Tsz-Kit Lau, Weijian Li, Chenwei Xu +2
Modern deep neural networks often require distributed training with many workers due to their large size. As the number of workers increases, communication overheads become the mai…