roman-urdu-sentiment-xlmroberta

Roman Urdu Sentiment — XLM-RoBERTa

Fine-tuned xlm-roberta-base for binary sentiment classification of Roman Urdu social media text.

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

tokenizer = AutoTokenizer.from_pretrained("DaniyalCreates/roman-urdu-sentiment-xlmroberta")
model = AutoModelForSequenceClassification.from_pretrained("DaniyalCreates/roman-urdu-sentiment-xlmroberta")
model.eval()

text = "yaar ye movie bohat achi thi"
inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=128)
with torch.no_grad():
    probs = torch.softmax(model(**inputs).logits, dim=1).squeeze()

labels = ["Negative", "Positive"]
print(labels[probs.argmax()], f"{probs.max():.1%}")
# → Positive 99.7%

Training Details

Results

About

Part of an MSc dissertation in Artificial Intelligence & Data Science at the University of Hull (2025).
Author: Daniyal Tariq