espnet2.cls.lightning_callbacks.MultilabelAUPRCCallback

source

class espnet2.cls.lightning_callbacks.MultilabelAUPRCCallback

Bases: Callback

Computes and logs Multilabel AUPRC (mAP) at the end of each validation epoch.

To use this callback, you must implement a update_mAP method in the espnet model that accepts a MultilabelAUPRC object and calls its update method with predictions and targets. For example:

``

`

python class MyESPnetModel(AbsESPnetModel):

def update_mAP(self, mAP_function: MultilabelAUPRC): : … mAP_function.update(predictions, targets) …

``

` The model should also have a get_vocab_size() function that specifies the number of labels/classes.

compute_mAP(trainer)

Computes the mAP.

on_train_batch_end(trainer, pl_module, outputs, batch, batch_idx)

Called when the train batch ends.

NOTE

The value outputs["loss"] here will be the normalized value w.r.t accumulate_grad_batches of the loss returned from training_step.

on_train_start(trainer, pl_module)

Called when the train begins.

on_validation_batch_end(trainer, pl_module, outputs, batch, batch_idx, dataloader_idx=0)

Called when the validation batch ends.

on_validation_epoch_end(trainer, pl_module)

Called when the val epoch ends.

on_validation_epoch_start(trainer, pl_module)

Called when the val epoch begins.

on_validation_start(trainer, pl_module)

Called when the validation loop begins.

setup_mAP(model)