Kc2fresh / Extracting-Counterfactual-data

Code for paper

Extracting-Counterfactual-data

Code for paper
This repository contains the Code and a helpful guide for reproducing the results of the tasks as described in the paper for the Semantic evaluation task Counterfactual Detection meets Transfer Learning

Task1 : I suggest all experiments are performed in googlr colab so as to use the free gpu and save bandwith for downloading the model. but if you have other options, thats okay too. put the path of the csv train file and csv test file as required, it will first train the model, and then predict over the test data. it will save the labels as a dataframe and you can download it from google drive(If you are using google colab), or else you can download it in anyway you choose.

task 2: there are three processes in task 2:

1. preprocess2
2. train_and_inference2
3. post_process2