Atli Kosson - RankMe

19 papers A* 6Journal 12Unranked 1

Year	Rank	Type	Title / Venue / Authors
2025	J	`jnl`	Training Dynamics of the Cooldown Stage in Warmup-Stable-Decay Learning Rate Scheduler. CoRR Aleksandr Dremov, Alexander Hägele, Atli Kosson, Martin Jaggi
2025	J	`jnl`	Training Dynamics of the Cooldown Stage in Warmup-Stable-Decay Learning Rate Scheduler. Trans. Mach. Learn. Res. Aleksandr Dremov, Alexander Hägele, Atli Kosson, Martin Jaggi
2025	J	`jnl`	Weight Decay may matter more than muP for Learning Rate Transfer in Practice. CoRR Atli Kosson, Jeremy Welborn, Yang Liu, Martin Jaggi, Xi Chen
2024	A*	`conf`	Analyzing & Reducing the Need for Learning Rate Warmup in GPT Training. NeurIPS Atli Kosson, Bettina Messmer, Martin Jaggi
2024	J	`jnl`	Analyzing & Reducing the Need for Learning Rate Warmup in GPT Training. CoRR Atli Kosson, Bettina Messmer, Martin Jaggi
2024	A*	`conf`	Ghost Noise for Regularizing Deep Neural Networks. AAAI Atli Kosson, Dongyang Fan, Martin Jaggi
2024	A*	`conf`	Rotational Equilibrium: How Weight Decay Balances Learning Across Neural Networks. ICML Atli Kosson, Bettina Messmer, Martin Jaggi
2024	A*	`conf`	Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations. NeurIPS Alexander Hägele, Elie Bakouch, Atli Kosson, Loubna Ben Allal, Leandro von Werra, Martin Jaggi
2024	J	`jnl`	Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations. CoRR Alexander Hägele, Elie Bakouch, Atli Kosson, Loubna Ben Allal, Leandro von Werra, Martin Jaggi
2023	J	`jnl`	Ghost Noise for Regularizing Deep Neural Networks. CoRR Atli Kosson, Dongyang Fan, Martin Jaggi
2023	J	`jnl`	Hardware-Efficient Transformer Training via Piecewise Affine Operations. CoRR Atli Kosson, Martin Jaggi
2023	J	`jnl`	Memory Efficient Mixed-Precision Optimizers. CoRR Basile Lewandowski, Atli Kosson
2023	A*	`conf`	Multiplication-Free Transformer Training via Piecewise Affine Operations. NeurIPS Atli Kosson, Martin Jaggi
2023	J	`jnl`	Rotational Optimizers: Simple & Robust DNN Training. CoRR Atli Kosson, Bettina Messmer, Martin Jaggi
2021	—	`conf`	Pipelined Backpropagation at Scale: Training Large Models without Batches. MLSys Atli Kosson, Vitaliy Chiley, Abhinav Venigalla, Joel Hestness, Urs Köster
2020	J	`jnl`	Adaptive Braking for Mitigating Gradient Delay. CoRR Abhinav Venigalla, Atli Kosson, Vitaliy Chiley, Urs Köster
2020	J	`jnl`	Pipelined Backpropagation at Scale: Training Large Models without Batches. CoRR Atli Kosson, Vitaliy Chiley, Abhinav Venigalla, Joel Hestness, Urs Köster
2019	A*	`conf`	Online Normalization for Training Neural Networks. NeurIPS Vitaliy Chiley, Ilya Sharapov, Atli Kosson, Urs Köster, Ryan Reece, Sofia Samaniego de la Fuente, Vishal Subbiah, Michael James
2019	J	`jnl`	Online Normalization for Training Neural Networks. CoRR Vitaliy Chiley, Ilya Sharapov, Atli Kosson, Urs Köster, Ryan Reece, Sofia Samaniego de la Fuente, Vishal Subbiah, Michael James