4 papers · 1 filter
One-Step Gradient Delay is Not a Barrier for Large-Scale Asynchronous Pipeline Parallel LLM Pretraining
Philip Zmushko, Egor Petrov, Nursultan Abdullaev +2
Modern large-scale LLM pretraining benefits from utilizing Pipeline Parallelism; however, synchronous implementations leave GPUs idle during pipeline bubbles, wasting computational…
Sign-SGD via Parameter-Free Optimization
Daniil Medyakov, Sergey Stanko, Gleb Molodtsov +4
Large language models have achieved major advances across domains, yet training them remains extremely resource-intensive. We revisit Sign-SGD, which serves both as a memory-effici…
Closing the Curvature Gap: Full Transformer Hessians
Egor Petrov, Nikita Kiselev, Vladislav Meshkov +1
The optimization landscape of Transformer models remains poorly understood despite their widespread adoption. While recent studies have derived curvature properties for isolated se…
Leveraging Coordinate Momentum in SignSGD and Muon: Memory-Optimized Zero-Order
Egor Petrov, Grigoriy Evseev, Aleksey Antonov +4
Fine-tuning Large Language Models (LLMs) is essential for adapting pre-trained models to downstream tasks. Yet traditional first-order optimizers such as Stochastic Gradient Descen…