Why is there no test_loss

[Zzz-L]

Hi @jonnykira, thanks very much for sharing this! why is no test_loss in train_mLSTM, only find the train_loss

[jonny-d]

The code only trains a language model on the training set, if you would like to split your data into training/test set and measure loss on the test set, it should be fairly straightforward to modify the code.

[Zzz-L]

I mean，Don't you need to go through the test loss evaluation language model，can you just use the train loss to measure the model ? beacuse from the paper "Generating Reviews and Discovering Sentiment." , there need split data into training/test set. look forward to your reply , thanks