Fake News Detection Distilbert

This model was trained on 35,900 news articles from CLÉMENT BISAILLON's dataset on Kaggle. The goal is to classify fake news from real news.

Dataset label structure:

0 : Fake News, 1 : Real News

About Dataset

ISOT Fake News detection dataset

Dataset separated in two files:

Fake.csv (23502 fake news article) True.csv (21417 true news article)

Dataset columns

Title: title of news article

Text: body text of news article

Subject: subject of news article

Date: publish date of news article

Data columns used for fine-tuning:

text: body text of news article

Labels: (0s and 1s)

About Model

Model Description: This model is a fine-tune checkpoint of distilbert-base-uncased, fine-tuned on ISOT Fake News Dataset.

Fine-tuning hyper-parameters

learning_rate = 2e-5 batch_size = 16 warmup = 600 max_seq_length = 128 weight_decay=0.01 num_train_epochs = 3.0

Sources

Dataset used: https://www.kaggle.com/datasets/clmentbisaillon/fake-and-real-news-dataset

Base Model (Distilbert): https://huggingface.co/distilbert/distilbert-base-uncased