Shakespeare LLM

This LLM was fine-tuned on a curated corpus of Shakespeare’s complete works, including poems, plays, and sonnets. It uses the gpt-2 model from huggingface transformers. The model learns to generate text and complete the user's passage using Shakespeare’s unique linguistic style. The trained model is uploaded to Huggingface Hub and is accessed for inference via FastAPI from my web UI project.

Local Development

To clone repo and re-train model locally:

1. conda torch-nightly environment is recommended
2. Install dependencies in environment.yml. Run check_installation.py to confirm.
3. Run prepare_data.py
4. Run train.py (will take around an hour)
5. Use generate_text.py to interact with your trained model