EXt1/ThaiFakeNews-BERT

Thai-FakeNews-BERT

A Thai-language BERT model fine-tuned for fake news detection. This model is part of a Senior Project by CPE35 students from King Mongkut's University of Technology Thonburi (KMUTT).

Model Description

• Base model: monsoon-nlp/bert-base-thai
• Dataset: EXt1/Thai-True-Fake-News
• Model Size: 105M parameters
• Task: Text Classification
• Language: Thai
• Labels:
  • 0: True News
  • 1: Fake News

Evaluation Results

• Loss: 0.3976
• Accuracy: 85% on the test set

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

tokenizer = AutoTokenizer.from_pretrained("EXt1/ThaiFakeNews-BERT")
model = AutoModelForSequenceClassification.from_pretrained("EXt1/ThaiFakeNews-BERT")


text = "เตรียมรับมือ พายุฤดูร้อนพัดถล่ม 26-28 เม.ย. 68"

inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)


with torch.no_grad():
    logits = model(**inputs).logits

predicted_class = torch.argmax(logits, dim=1).item()


if predicted_class == 1:
    print("ข่าวปลอม")
else:
    print("ข่าวจริง")