Fine-Tuned BERT Sentiment Analysis Application
A complete NLP classification workflow covering exploratory analysis, BERT fine-tuning, evaluation, and live sentiment prediction through Streamlit.
Status:Evaluated portfolio application; reported accuracy and F1 score: 92.09%
View source on GitHub (opens in a new tab)- Services
- Machine Learning
- Stack
- Python, PyTorch, scikit-learn, Hugging Face +2 more
Outcomes
The results
92.09%
Reported accuracy and F1 score
01
Our solution
The project fine-tunes bert-base-uncased on SST-2 with Hugging Face Transformers and PyTorch, evaluates multiple metrics, and serves interactive predictions through Streamlit.
02
The outcome
Shows the full path from dataset exploration to an evaluated transformer model and accessible end-user prediction experience.
03
Best for
Review analysis, customer feedback classification, social listening, support-ticket triage, and NLP model development.
Technology
The stack behind it
- Backend
- Python
- AI & ML
- PyTorch
- scikit-learn
- Hugging Face
- BERT
- Frontend
- Streamlit