distilbert-amazon-polarity-470-final

Model Name and Task

DistilBERT Sentiment Classifier – Amazon Polarity

Task: Binary sentiment analysis for Amazon-style product reviews.

This model predicts whether a product review is positive or negative.

---

Dataset and Training Details

• Dataset: Amazon Polarity (Hugging Face)
• Training data: 4,000 reviews
• Test data: 2,000 reviews
• Base model: distilbert-base-uncased
• Training method: Fine-tuning
• Epochs: 1
• Max sequence length: 256
• Training environment: Google Colab

---

Evaluation Metrics

• Metric: Accuracy
• Result: 0.901000

Accuracy is appropriate because this is a binary classification task (positive vs negative).

---

Intended Uses

This model can be used for:

• Sentiment analysis of Amazon-style product reviews
• Quick review checking for users or students
• Basic customer feedback analysis
• Educational demonstrations of transformer fine-tuning

---

Limitations

• Only supports positive vs negative sentiment (no neutral class)
• May struggle with sarcasm or mixed opinions
• Trained on Amazon-style reviews only
• English language only
• Trained on a small subset due to time and compute limits

---

How to Load and Use the Model

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model_id = "RahimaR/distilbert-amazon-polarity-470-final"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForSequenceClassification.from_pretrained(model_id)

label_names = ["negative", "positive"]

def predict_sentiment(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
    with torch.no_grad():
        logits = model(**inputs).logits
    return label_names[logits.argmax().item()]

print(predict_sentiment("This product works really well!"))
print(predict_sentiment("The item arrived broken and customer service was terrible."))