A project that analyzes character descriptions and classifies them into 12 archetypes in TV shows and movies using BERT.
Using the Hugging Face Transformers library, the project fine-tunes the distilbert-base-uncased model to achieve high accuracy and deploy a robust classification pipeline.
- Fine-tune a pre-trained model (distilbert-base-uncased) for text classification.
- Hyperparameter optimization using Optuna.
- Deploy model with Gradio.
To set up this project, ensure the following packages are installed.
transformers (4.47.1) datasets (3.2.0) numpy (1.26.4) optuna (4.1.0) torch (2.0.1)