It was fine-tuned using a curated dataset consisting of public Malayalam news data and carefully generated synthetic examples for fake news detection.

Usage

python

from transformers import AutoTokenizer, AutoModelForSequenceClassification

MODEL_ID = "manasyy/malayalam-fake-news-detector-muril"

tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
model = AutoModelForSequenceClassification.from_pretrained(MODEL_ID)

Intended Use

This model is intended for:

Malayalam fake news detection
NLP research
Academic projects
Fact verification pipelines

Limitations

The model predicts whether text resembles fake or real news but should not be considered a substitute for evidence-based fact checking.
Performance depends on the quality and diversity of the training data.
Real-time claims should still be verified against reliable news sources.

Author

Manasi