metadata
license: mit
datasets:
- ucirvine/sms_spam
metrics:
- accuracy
BERT Email Spam Classifier
This repository contains a fine-tuned BERT model for SMS/email spam detection using the SMS Spam Collection Dataset.
Accuracy:
Loss: 0.012 Accuracy: 0.98
Model Details
- Architecture: BERT (Bidirectional Encoder Representations from Transformers)
- Task: Binary classification (spam vs. ham)
- Base Model:
bert-base-uncased - Framework: Hugging Face Transformers (PyTorch)
- License: MIT
Usage
Load the model and tokenizer using Hugging Face Transformers:
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("SGHOSH1999/bert-email-spam-classifier_tuned")
model = AutoModelForSequenceClassification.from_pretrained("SGHOSH1999/bert-email-spam-classifier_tuned")
Training Data
- SMS Spam Collection Dataset
- The dataset contains 5,574 SMS messages labeled as 'ham' (legitimate) or 'spam'.
How to Cite
If you use this model, please cite the original dataset and this repository.
License
MIT