How to use wikitext-103 without fine-tuning?

Hi,

I want to compare the type of text generated by the language model with and without fine-tuning on my custom data. Can somebody please guide me about how I can generate some text with the pre-trained wt103 model provided by fastai?

I don’t want to provide any custom data to the wt103 model and I just want to look at the text generated by this pre-trained model.

Thanks