This is the github repo for the NLSea meetup Aug.27.2019.
In this session, we discussed about the pytorch implementation of BERT model purposed Nov. 2018.
The session mainly focused on 4 aspects:
- What is BERT and word embedding?
- What is a tokenizer and contextual level embedding?
- How to get word, sentence, and paragraph embeddings from token level embeddings?
- How to fine tune your own BERT model?
Link to pytorch: https://pytorch.org/
Link to hugging face repo: https://github.com/huggingface/pytorch-transformers
Link to a nicely written tutorial: https://mccormickml.com/2019/05/14/BERT-word-embeddings-tutorial/
Link to code for further classification models: https://colab.research.google.com/drive/1ywsvwO6thOVOrfagjjfuxEf6xVRxbUNO#scrollTo=DEfSbAA4QHas
Link to the BERT paper: https://arxiv.org/abs/1810.04805