Here's have two version of generate pre trained transformer, because I want to try it the Traindition Chinese words how will perform the models.
For this situation is not good for the result, but I didn't try it on the same datasets, it's the using some medical records for answer the questions.
ASCII edtion:
- generate_pre-trained_transformer.py
- run_model_as_bot.py
- chat_at_jupyter.py
- load_model.py
Jieba words set edition:
- jieba_and_gpt.js
- jieba_load_model.py
Telegram bot:
Medical question:
