NLP Course Project

Implementation of DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter

Three ipynb files for the three datasets mentioned in the paper.
BERT_RTE.py is to train RTE dataset, CustomInLawBert_RR.py, DistilBert_RR.py are to train rhetorical role predicition dataset.
Json Files indicates the Rhetorical Role prediction dataset.

The KD code is in a different repo-

Name		Name	Last commit message	Last commit date
Latest commit History 15 Commits
BERT_RTE.py		BERT_RTE.py
CustomInLawBERT_RR.py		CustomInLawBERT_RR.py
DistilBert_RR.py		DistilBert_RR.py
DistilBert_RTE.ipynb		DistilBert_RTE.ipynb
DistilBert_sst2.ipynb		DistilBert_sst2.ipynb
DistilBert_stsb.ipynb		DistilBert_stsb.ipynb
README.md		README.md
dev.json		dev.json
train.json		train.json