metadata

license: mit
datasets:
  - yassiracharki/Amazon_Reviews_for_Sentiment_Analysis_fine_grained_5_classes
language:
  - en
metrics:
  - accuracy
tags:
  - text-classification
  - sentiment-classification
  - BERT
  - Roberta
  - mini-roberta

RoBERTa-mini: Sentiment Classifier

Model Name: dilip025/RoBERTa-mini
Task: Sentiment Classification
Labels: Very Negative, Negative, Neutral, Positive, Very Positive

A compact RoBERTa like model trained from scratch for sentiment classification.

Example Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

tokenizer = AutoTokenizer.from_pretrained("dilip025/RoBERTa-mini")
model = AutoModelForSequenceClassification.from_pretrained("dilip025/RoBERTa-mini", trust_remote_code=True)

id2label = {
    0: "Very Negative",
    1: "Negative",
    2: "Neutral",
    3: "Positive",
    4: "Very Positive"
}

def predict_sentiment(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=128)
    with torch.no_grad():
        outputs = model(**inputs)
        probs = torch.softmax(outputs["logits"], dim=1)
        pred_class = torch.argmax(probs, dim=1).item()
    return {
        "text": text,
        "class_id": pred_class,
        "label": id2label[pred_class],
        "probabilities": probs.tolist()[0]
    }

# Example
result = predict_sentiment("I absolutely hate this product.")
print(result)

Model Card

Architecture: RoBERTa (custom small version)
Training Dataset: Amazon Reviews Dataset
Use Case: Sentiment classification for customer feedback, reviews, etc.
Input Format: Plain text (string)
Output: Dictionary with class ID, label, and class probabilities

License

This model is licensed under the MIT License. You are free to use, modify, and distribute it with attribution.

Author

Developed and Trained by Dilip Pokhrel