Fake News Detection with DistilBERT

Model Description

Fine-tuned DistilBERT model for classifying news articles as Real or Fake.

Training Data

  • Total: 18,281 samples
  • Train: 12,796 (70%)
  • Validation: 1,829 (10%)
  • Test: 3,656 (20%)

Performance

  • Test Accuracy: 96.36%
  • Test Loss: 0.194

Classification Report

| | Precision | Recall | F1-Score | Support | |----------|-----------|--------|----------|---------|| | Real | 0.97 | 0.96 | 0.97 | 2,072 | | Fake | 0.95 | 0.96 | 0.96 | 1,584 |

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained('Aakash22134/fake_news_DistilBert')
model = AutoModelForSequenceClassification.from_pretrained('Aakash22134/fake_news_DistilBert')
Downloads last month
-
Safetensors
Model size
67M params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Space using Aakash22134/fake_news_DistilBert 1