facebookresearch/schedule_free
PyTorch library implementing schedule-free optimization algorithms from Facebook Research.
Not currently ranked — collecting fresh signals.
star history
A PyTorch library implementing Schedule-Free optimization algorithms that eliminate the need for learning rate schedules and warmup phases during training. It provides drop-in replacements for standard optimizers like SGD and AdamW, replacing momentum with interpolation and averaging. The approach is derived from the paper ‘The Road Less Scheduled’ (arXiv:2405.15682).
Frequently asked
- What is facebookresearch/schedule_free?
- PyTorch library implementing schedule-free optimization algorithms from Facebook Research.
- Is schedule_free open source?
- Yes — facebookresearch/schedule_free is open source, released under the Apache-2.0 license.
- What language is schedule_free written in?
- facebookresearch/schedule_free is primarily written in Python.
- How popular is schedule_free?
- facebookresearch/schedule_free has 2.3k stars on GitHub.
- Where can I find schedule_free?
- facebookresearch/schedule_free is on GitHub at https://github.com/facebookresearch/schedule_free.