Open le1nux opened 3 months ago
As seen here, https://github.com/Modalities/modalities/blob/a599005e646a0ec1db343fbe1ef3be7b53cb75f9/src/modalities/trainer.py#L144, the cumulated_loss_and_gradient_norm is indexed directly in the train function, adding unnecessary complexity to the code.
cumulated_loss_and_gradient_norm
A better solution would be to encapsulate the cumulated_loss_and_gradient_norm within a dedicated class that implements also the reduce operations.
As seen here, https://github.com/Modalities/modalities/blob/a599005e646a0ec1db343fbe1ef3be7b53cb75f9/src/modalities/trainer.py#L144, the
cumulated_loss_and_gradient_normis indexed directly in the train function, adding unnecessary complexity to the code.A better solution would be to encapsulate the
cumulated_loss_and_gradient_normwithin a dedicated class that implements also the reduce operations.