ModernBERT-base-fa-snappfood

Fine-tuned Persian sentiment analysis model based on myrkur/Persian-ModernBert-base for binary sentiment classification on the Snappfood Persian review dataset.

Dataset

The model was trained on the Snappfood Persian reviews dataset available on Hugging Face:

• Snappfood Dataset

Model Details

• Base model: myrkur/Persian-ModernBert-base
• Task: Sentiment Classification
• Language: Persian (Farsi)
• Framework: Transformers + PyTorch

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model_name = "aysangh/ModernBERT-base-fa-snappfood"
model = AutoModelForSequenceClassification.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

def predict_sentiment(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=256, padding=True)
    with torch.no_grad():
        outputs = model(**inputs)
    probs = torch.softmax(outputs.logits, dim=-1)
    pred = torch.argmax(probs, dim=-1).item()
    confidence = probs[0][pred].item()
    return {
    "label": "positive" if pred == 0 else "negative",
    "confidence": confidence
}

text = "سفارش خیلی سریع رسید و کیفیت غذا خوب بود."

result = predict_sentiment(text)
print(result)

Repository

Training and inference scripts are available on GitHub.