Twitter-NER

General overview:

Kaggle

running inference (this produces test_ans.csv, everything else should be configured)

python inference.py

running training:

python configure.py python train.py

dir breakdown

• assets -> contains diagram
• twitter_ner -> contains model from iteration 1 of system
• tokenizer -> contains tokenizer from iteration 1 of system
• twitter_ner_2 -> contains model from iteration 2 of system
• tokenizer_2 -> contains tokenizer from iteration 2 of system
• configure.py -> creates transformed_train.csv and transformed_val.csv loaded in train.py
• train.py -> used to fine tune BERT models on twitter dataset # Twitter-NER