This is terge, an easy-to-use library I built for merging PyTorch models of any number, size and architecture.

I've already used it with my own custom Transformers and LSTMs to produce models stronger than any of their constituent members.

It only takes a single line of code to get started, which puts its usability ahead of most other libraries, which tend to require complex, often poorly documented YAML configurations.

If you end up finding the library useful, consider starring it on GitHub to let others know.

PRs are also always welcome