Pretraining Persian AWD-LSTM language model and use the ULMFIT approach to finetune language model.
You can check these two blog posts for getting information about pretraining and fine tunning:
train.py
is the training script and Finetunin Persian Language Model for Text Classification-ULMFiT approach
notebook is demonstrating the finetuning.
Note that the data and the model is accessible through this link: