This repository contains both deep sequence tagging model which uses word level + Char level embedding with CRF. Also it contains feature level CRF model which uses pycrfsuite with additional features added.
- To run CRF Model which uses pycrfsuite.
- pip install pycrfsuite
- And then run python Notebook NLP_Assignment3_NER.ipynb
- To run LSTM Word+Char CRF Model
- pip install tensorflow
- To train model run file generate_model.sh which is present in LSTM_Word+Char_Model folder
- Then to evaluate model on private test file run python evaluate_test.py "testfilepath" or evaluate_model.sh "testfilepath" in LSTM_Word+Char_Model. Since model is already there you don't need to train it. When you will run this command it generate a file "testfilepath.labelled" as given in training format.