CSE244_ML_for_NLP

Code for CSE244 winter 2020 homework 1: Relation Extraction From Text

run.ipynb or run-bert.ipynb

model_utils.py

This file contains all models I have tried
data_utils.py

This contains preparing data with torchtext, for both ngram format and also word embedding format. This file also include wrappers which generate data batches for training.
train_utils.py

This file contains the function used for training and calculating losses during training
evaluation.py

This file contains calculation of f1 scores in various ways
submission.py

This file generate submission file for kaggle. It also includes an Ensemble class to combine results of multiple models into one final result. It also includes basic error anaylysis.
preprocessing.ipynb

This is the notebook for preprocessing: train, validation, test splitting. Converting label text to numerical categories. Among with many minute things I tried.
movie_and_person_names.ipynb

This is the notebook specificially for one type of feature engineering: processing movie and people names into phrases.
run.ipynb and run-bert.ipynb

This should be the starting point. This is where I import all other functions from *.py files and get results.

Name		Name	Last commit message	Last commit date
Latest commit History 37 Commits
data		data
.gitignore		.gitignore
HW1 report.pdf		HW1 report.pdf
README.md		README.md
__init__.py		__init__.py
character_level_model.ipynb		character_level_model.ipynb
data_utils.py		data_utils.py
evaluation.py		evaluation.py
model_utils.py		model_utils.py
movie_and_person_names.ipynb		movie_and_person_names.ipynb
preprocessing.ipynb		preprocessing.ipynb
run-bert.ipynb		run-bert.ipynb
run.ipynb		run.ipynb
submission.py		submission.py
train_utils.py		train_utils.py

MollyZhang/CSE244_ML_for_NLP_HW1