method · first seen 9 Oct 2026
Schedule-Free AdamW
Also: Schedule-Free optimizer, schedule-free optimization
An AdamW variant that removes the need for a learning-rate schedule by combining iterate averaging with momentum.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Towards Easier-to-Tune Training Algorithms: New Results from AlgoPerfPapers · 9 Oct 2026 · Priya Kasimbeg, Ahmed Khaled, Aahlad Chandrabhatta, Yifan Wang, Andy Liu, Niccol…
- Large Diffusion Models Need More Than EMAPapers · 16 Oct 2026 · Chuyu Hu, Huanran Chen, Shitong Shao, Yinpeng Dong
Latest sources
- Large Diffusion Models Need More Than EMAPapers · 16 Oct 2026 · Chuyu Hu, Huanran Chen, Shitong Shao, Yinpeng Dong
- Towards Easier-to-Tune Training Algorithms: New Results from AlgoPerfPapers · 9 Oct 2026 · Priya Kasimbeg, Ahmed Khaled, Aahlad Chandrabhatta, Yifan Wang, Andy Liu, Niccol…