How to get the pre-trained language model loss for an input text?

Part 2 & Alumni (2018)
1

After pre-training the LM model (AWD-LSTM) on Wikitext-103, I was wondering how can we get
the LM loss for an input text. Any idea?