Crate burn_train
source ·Expand description
A library for training neural networks using the burn crate.
Modules
- The checkpoint module.
- The logger module.
- The metric module.
- Renderer modules to display metrics and training information.
- The trainer module.
Structs
- Simple classification output adapted for multiple metrics.
- Learner struct encapsulating all components necessary to train a Neural Network model.
- Struct to configure and create a learner.
- An early stopping strategy based on a metrics collected during training or validation.
- Multi devices train step.
- Simple regression output adapted for multiple metrics.
- A training epoch.
- A training output.
- A handle that allows aborting the training process early.
- A validation epoch.
Enums
- The condition that early stopping strategies should follow.
Traits
- A strategy that checks if the training should be stopped.
- Trait to be implemented for training models.
- Trait to be implemented for validating models.