activity
20242026
collaborators
Showing cs.LGShow all

5 papers · 1 filter

cs.LG2026

Improving Neural Network Training by Decoupling the Magnitude and Direction of Weight Vectors

Alexander Hägele, Alejandro Hernández-Cano, Atli Kosson +1

Modern neural network training relies on optimizers such as Adam and Muon which act on each weight matrix as a single object. Yet every weight matrix carries two distinct quantitie…

cs.LG2026

Weight Decay may matter more than muP for Learning Rate Transfer in Practice

Atli Kosson, Jeremy Welborn, Yang Liu +2

Transferring the optimal learning rate from small to large neural networks can enable efficient training at scales where hyperparameter tuning is otherwise prohibitively expensive.…

cs.LG2025

Training Dynamics of the Cooldown Stage in Warmup-Stable-Decay Learning Rate Scheduler

Aleksandr Dremov, Alexander Hägele, Atli Kosson +1

Learning rate scheduling is essential in transformer training, where the final annealing plays a crucial role in getting the best performance. However, the mechanisms behind this c…

cs.LG2024

Analyzing & Reducing the Need for Learning Rate Warmup in GPT Training

Atli Kosson, Bettina Messmer, Martin Jaggi

Learning Rate Warmup is a popular heuristic for training neural networks, especially at larger batch sizes, despite limited understanding of its benefits. Warmup decreases the upda…

cs.LG2024

Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations

Alexander Hägele, Elie Bakouch, Atli Kosson +3

Scale has become a main ingredient in obtaining strong machine learning models. As a result, understanding a model's scaling properties is key to effectively designing both the rig…