metadata
language:
- en
license: mit
tags:
- text-classification
- sentiment-analysis
- distilbert
- amazon-reviews
- nlp
datasets:
- amazon_polarity
metrics:
- accuracy
- f1
🛍️ pranalyzer — Sentiment Analysis Model
Fine-tuned DistilBERT on Amazon product reviews for binary sentiment classification (POSITIVE / NEGATIVE).
Part of the pranalyzer end-to-end NLP pipeline.
📊 Model Performance
| Metric | Score |
|---|---|
| Accuracy | 93.00% |
| F1 Score | 0.9299 |
| Loss | 0.1923 |
- Dataset:
amazon_polarity(5,000 train / 1,000 test samples) - Hardware: T4 GPU (Google Colab)
- Epochs: 3
- Batch size: 32
- Learning rate: 2e-5
🚀 Quick Start
from transformers import pipeline
classifier = pipeline(
"text-classification",
model="Ved2001/pranalyzer"
)
result = classifier("This product is absolutely amazing!")
print(result)
# [{'label': 'POSITIVE', 'score': 0.98}]
🏋️ Training Details
Base model: distilbert-base-uncased
Task: Binary sentiment classification
Labels: NEGATIVE (0), POSITIVE (1)
Dataset: Amazon Polarity — 3.6M reviews (sampled 5K for fine-tuning)
🔍 Part of pranalyzer Pipeline
This model is the sentiment component of a 4-model pipeline:
| Task | Model |
|---|---|
| Sentiment | Ved2001/pranalyzer (this model) |
| Category | facebook/bart-large-mnli |
| Aspects | cross-encoder/nli-roberta-base |
| Summary | facebook/bart-large-xsum |
👤 Author
Vedant Nagarkar
GitHub • HuggingFace