Query on loss calculation in word language model
AvisP opened this issue · comments
AvisP commented
In the main.py of word language model, I find that in the evaluate function the total_loss is getting multiplied by length of data
examples/word_language_model/main.py
Line 163 in ca1bd91
However in the train function, total_loss is not getting multiplied by length of data
examples/word_language_model/main.py
Line 194 in ca1bd91
Is this proper?