def validation_epoch_end(self, outputs):
if self.logger is not None and self.logger.experiment is not None:
waveglow_log_to_tb_func(
self.logger.experiment,
outputs[0].values(),
self.global_step,
tag="eval",
mel_fb=self.audio_to_melspec_precessor.fb,
)
avg_loss = torch.stack([x['val_loss'] for x in outputs]).mean()
self.log('val_loss', avg_loss)
def validation_epoch_end(self, outputs):
if self.logger is not None and self.logger.experiment is not None:
waveglow_log_to_tb_func(
self.logger.experiment,
tuple(outputs[0].values())[:-1],
self.global_step,
tag="eval",
mel_fb=self.audio_to_melspec_precessor.fb,
)
avg_loss = torch.stack([x['val_loss'] for x in outputs]).mean()
avg_stoi = torch.FloatTensor([x['stoi'] for x in outputs]).mean()
tensorboard_logs = {'val_loss': avg_loss, 'stoi': avg_stoi}
logging.info(
f"Validation summary | Epoch {self.current_epoch} | NLL {avg_loss:.2f} | STOI: {avg_stoi:.2f}"
)
return {'val_loss': avg_loss, 'log': tensorboard_logs}
def validation_epoch_end(self, outputs):
if self.logger is not None and self.logger.experiment is not None:
tb_logger = self.logger.experiment
if isinstance(self.logger, LoggerCollection):
for logger in self.logger:
if isinstance(logger, TensorBoardLogger):
tb_logger = logger.experiment
break
waveglow_log_to_tb_func(
tb_logger,
outputs[0].values(),
self.global_step,
tag="eval",
mel_fb=self.audio_to_melspec_precessor.fb,
)
avg_loss = torch.stack([x['val_loss'] for x in outputs]).mean()
self.log('val_loss', avg_loss)
