FNP task1

Binary classification task

Text section contains a causal relation => 1, Otherwise => 0

Unbalanced dataset, only about 7% 1s

10837 training data & 2710 testing data

Submit your result on Kaggle

Read data descriptions and download the dataset from kaggle

Code

preprocess.py: Separate the dataset into the training set, validation set, and test set.
Train the models via baseline.py with BERT & Focal Loss and generate the csv file pred.csv.
Generate pseudo-label training data via create_pseudo.py .
Train the models via xlnet_pseudo.py with XLNet, Focal Loss & pseudo-label.
Submit the result final_preditcion.csv.

Name		Name	Last commit message	Last commit date
Latest commit History 13 Commits
data		data
README.md		README.md
baseline.py		baseline.py
create_pseudo.py		create_pseudo.py
eda.py		eda.py
final_prediction.csv		final_prediction.csv
loss.py		loss.py
pred.csv		pred.csv
pred_eda.csv		pred_eda.csv
preprocess.py		preprocess.py
requirements.txt		requirements.txt
xlnet_pseudo.py		xlnet_pseudo.py