Customize Trainers

FaKnow designed a basic trainer, BaseTrainer, in the training module. This trainer implements the basic functions of model training, with the specified model, loss_func, optimizer, evaluator and scheduler, the algorithm model is trained, and each epoch validates the model. The process is presented in various visual ways, and the trained algorithm model is finally saved.

If the training of the algorithm only requires the most basic work, simply call BaseTrainer; If the training of algorithms requires additional work, Faknow has written specific trainers for these specific algorithms which inherit BaseTrainer, For example, MFANTrainer is a trainer written specifically for MFAN, which minimizes code redundancy and improves code readability.