Open Sllambias opened 6 months ago
https://github.com/facebookresearch/schedule_free/blob/main/schedulefree/adamw_schedulefree.py
https://github.com/Lightning-AI/pytorch-lightning/discussions/19759
Also see caveats, regarding choice of hyperparams, batch norm etc: https://github.com/facebookresearch/schedule_free?tab=readme-ov-file#caveats
https://github.com/facebookresearch/schedule_free/blob/main/schedulefree/adamw_schedulefree.py