Shortcuts

Source code for quaterion.train.callbacks.metrics_callback

import pytorch_lightning as pl

try:  # fix for version >= 1.9.0
    from pytorch_lightning import Callback
except ImportError:
    from pytorch_lightning.callbacks.base import Callback


[docs]class MetricsCallback(Callback):
[docs] @staticmethod def reset_metrics(trainable_model): for metric in trainable_model.attached_metrics: metric.reset()
[docs] def on_sanity_check_end( self, trainer: "pl.Trainer", trainable_model: "pl.LightningModule" ) -> None: self.reset_metrics(trainable_model)
[docs] def on_train_epoch_start( self, trainer: "pl.Trainer", trainable_model: "pl.LightningModule" ) -> None: self.reset_metrics(trainable_model)
[docs] def on_train_epoch_end( self, trainer: "pl.Trainer", trainable_model: "pl.LightningModule" ) -> None: self.reset_metrics(trainable_model)
[docs] def on_validation_epoch_start( self, trainer: "pl.Trainer", trainable_model: "pl.LightningModule" ) -> None: self.reset_metrics(trainable_model)
[docs] def on_validation_epoch_end( self, trainer: "pl.Trainer", trainable_model: "pl.LightningModule" ) -> None: self.reset_metrics(trainable_model)

Qdrant

Learn more about Qdrant vector search project and ecosystem

Discover Qdrant

Similarity Learning

Explore practical problem solving with Similarity Learning

Learn Similarity Learning

Community

Find people dealing with similar problems and get answers to your questions

Join Community