And there are mainly two datasets demo, one is a json file about poem, another is a conversation demo created by myself.
However I don't recommand to use those demo_datas to train, I prefer use formal datasets.
Funtune method could be found in "Bert_finetune.py", funtune examples mainly include two.
First is the word classify prediction, could be found in ["bert_for_word_classify.py"]{https://codechina.csdn.net/captainAAAjohn/BERT-pytorch/-/blob/main/bert_for_word_classify.py}
Second is the sentences classify prediction, could be found in [" bert_for_sentence_classify.py"]{https://codechina.csdn.net/captainAAAjohn/BERT-pytorch/-/blob/main/bert_for_sentence_classify.py}
Next, I will enrich the language generation as well as conversation process.