lucidrains/ema-pytorch
A PyTorch library for maintaining Exponential Moving Average versions of neural network models during training.
Not currently ranked — collecting fresh signals.
star history
This repository provides a wrapper to track and update an EMA version of a PyTorch model alongside the primary model during training. It includes warmup logic to delay EMA updates, configurable update frequencies to save compute, and support for post-hoc synthesized EMA as proposed by Karras et al. for improved model checkpointing.
Frequently asked
- What is lucidrains/ema-pytorch?
- A PyTorch library for maintaining Exponential Moving Average versions of neural network models during training.
- Is ema-pytorch open source?
- Yes — lucidrains/ema-pytorch is open source, released under the MIT license.
- What language is ema-pytorch written in?
- lucidrains/ema-pytorch is primarily written in Python.
- How popular is ema-pytorch?
- lucidrains/ema-pytorch has 654 stars on GitHub.
- Where can I find ema-pytorch?
- lucidrains/ema-pytorch is on GitHub at https://github.com/lucidrains/ema-pytorch.