Save network?

[GorIvanov]

Is there a way to save trained network and load it from?

There is no sense in Neural Network lib if the trained network can't be saved and loaded.

[vxern]

The library currently does not support saving and loading its state.

There was a PR made to add that functionality with saving to and reading from a path a year and a half ago, but, regrettably and somewhat irresponsibly, I never got around to actually integrating it. I later removed the dev branch that the PR was targeting after making some changes of my own, and the PR got auto-closed.

Although this package isn't under active development, and was just a side-project at one point, if you would like to have a go at implementing this functionality yourself, I would be happy to review your changes and add it to the library.