PyToune

Provides a high-level Model class that encapsulates training loops, similar to Keras, making PyTorch development more accessible and reducing boilerplate code significantly.

Supports a wide range of callbacks for tasks like early stopping and checkpointing, as shown in the examples, automating common training interventions without manual coding.

ModelBundle class automates checkpointing, logging, and organization, streamlining experiment workflows with minimal setup, as demonstrated in the basic classification example.

Seamlessly integrates with PyTorch Metrics and custom metrics, allowing comprehensive evaluation during training and testing, as seen in the code snippet with AUROC and accuracy.