kathiravan/TamilAbusiveGPT

Tamil Abusive GPT Classifier

Model Description

This model is a fine-tuned GPT-4o model trained for abusive text detection in Tamil. It classifies text as either Abusive or Non-Abusive.

---

Model Details

• Model Type: GPT-4o (fine-tuned)
• Languages Supported: Tamil
• Training Dataset: Custom dataset containing Tamil abusive and non-abusive text comments
• Framework: PyTorch + Transformers (Hugging Face)

---

Intended Use

This model is intended to: ✅ Detect abusive language in Tamil text.
✅ Assist in content moderation for social media platforms.
✅ Help in identifying harmful or offensive language.

---

Limitations

⚠️ The model may misclassify text with ambiguous or sarcastic language.
⚠️ Since Tamil language data is often code-mixed with English, this model may require additional fine-tuning for such data.

---

How to Use

from transformers import AutoModelForSequenceClassification, AutoTokenizer
import torch

# Load model and tokenizer
model = AutoModelForSequenceClassification.from_pretrained("kathiravan/TamilAbusiveGPT")
tokenizer = AutoTokenizer.from_pretrained("gpt-4o")

# Example Inference
def classify_tamil_text(text):
    inputs = tokenizer(text, return_tensors="pt")
    outputs = model(**inputs)
    label = "Abusive" if outputs.logits.argmax().item() == 1 else "Non-Abusive"
    return label

# Example usage
print(classify_tamil_text("இது மிகவும் மோசமான கருத்து"))