I am training deepspeech model with about 140h, loss value during training stop at around 3.0 ,is this good?. What is a good value for loss train? Thannk you so much.
Sounds like your model is overfitting. What is the validation loss? Also, consider joining us on Matrix.
My dropout is 0.1.If I increase it the training will take a lot longer. I don’t use validation because there are lots of errors in the data I want to dedicate all the data that looks good for the training. Is that the right thing?
1 Like