metadata

license: openrail

Classifier-Bias-SG Model Card

Model Details

Classifier-Bias-SG is a proof of concept model designed to classify texts based on their bias levels. The model categorizes texts into 2 classes: "Biased", and "Non-Biased".

Model Architecture

The model is built upon the distilbert-base-uncased architecture and has been fine-tuned on a custom dataset for the specific task of bias detection.

Dataset

The model was trained on a BABE dataset containing news articles from various sources, annotated with one of the 2 bias levels. The dataset contains:

Biased: 1810 articles
Non-Biased: 1810 articles

Training Procedure

The model was trained using the Adam optimizer for 15 epochs.

Performance

On our validation set, the model achieved:

Accuracy: 78%
F1 Score (Biased): 79%
F1 Score (Non-Biased): 77%

How to Use

To use this model for text classification, use the following code:

from transformers import pipeline
from transformers import AutoTokenizer, AutoModelForSequenceClassification
 
tokenizer = AutoTokenizer.from_pretrained("Social-Media-Fairness/Classifier-Bias-SG")
model = AutoModelForSequenceClassification.from_pretrained("Social-Media-Fairness/Classifier-Bias-SG")

classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)
result = classifier("Women are bad driver.")
print(result)

Developed by Shardul Ghuge