I've just released a new version of the NAM trainer. The new version is used automatically in the online trainer, and folks training locally on their own computers can update their install from the terminal by typing
pip install --upgrade neural-amp-modeler
Support for different sample rates (CLI trainer only). It's now possible to set up training with a set of audio files that aren't at 48kHz. As always, the model will be trainer to run in the same sample rate as the data provided, and the plugin (as of v0.7.6) will automatically resample to meet the needs of the NAM model it's using. The simplified trainers (online and GUI) will continue to support only the standardized audio files, which are in 48k, since I believe that this is good "bang for the buck" for most users, just like the standardized architectures that are also given to choose from.
Data set class registry. This is the counterpart to the model class registry released in v0.7.3. If you want to play around with what data are fed to the model, then you can make your own data set classes. The model and data set registry features in combination should offer quite a bit of extensibility that can cover a lot of use cases, and I've found them really helpful for playing around with new ideas without having to muck around in the main library.
Bug fixes. Everyone loves bug fixes.