Beast212004/nlp-sentiment-models

NLP Sentiment Analysis Models

Pre-trained deep learning models for sentiment analysis on Amazon product reviews.

Models Included

• CNN (TensorFlow): Convolutional Neural Network with learned embeddings
• CNN + GloVe (TensorFlow): CNN with pre-trained GloVe embeddings
• CNN (PyTorch): Convolutional Neural Network with learned embeddings
• CNN + GloVe (PyTorch): CNN with pre-trained GloVe embeddings
• BiLSTM + GloVe (PyTorch): Bidirectional LSTM with pre-trained GloVe embeddings

Dataset

Trained on Amazon Reviews 2023 dataset with ~47K product reviews across multiple categories.

Performance

• CNN models: ~85-87% accuracy
• BiLSTM models: ~86-88% accuracy

Usage

These models are automatically downloaded by the Streamlit dashboard when deployed on Streamlit Cloud.

Manual Download

from huggingface_hub import hf_hub_download

model_path = hf_hub_download(
    repo_id="Beast212004/nlp-sentiment-models",
    filename="cnn_pytorch_pretrained.pt"
)

Repository

GitHub: https://github.com/YOUR_USERNAME/nlp-sentiment-analysis