This fix addresses an issue where a TypeError occurred during training.
The problem was that during training, when tracking the best loss (self.best_loss), there were cases where the loss value unexpectedly came in the form of a dictionary. This dictionary contained multiple values, such as "train_loss" and "eval_loss," causing subsequent comparisons to raise a TypeError.
This fix consists of two steps:
Updating the self.best_loss variable with ch["model_loss"] to ensure it contains the training loss value. Correcting the self.best_loss variable to only contain the "train_loss" value. The aim of this fix is to accurately track the best training loss value during the training process.
Fix for TypeError During Training
This fix addresses an issue where a TypeError occurred during training.
The problem was that during training, when tracking the best loss (self.best_loss), there were cases where the loss value unexpectedly came in the form of a dictionary. This dictionary contained multiple values, such as "train_loss" and "eval_loss," causing subsequent comparisons to raise a TypeError.
This fix consists of two steps:
Updating the self.best_loss variable with ch["model_loss"] to ensure it contains the training loss value. Correcting the self.best_loss variable to only contain the "train_loss" value. The aim of this fix is to accurately track the best training loss value during the training process.