This project uses the GPT-2 model to generate cooking recipes automatically. Leveraging a dataset from Hugging Face's datasets library, specifically "darkraipro/recipe-instructions", this notebook explores the ability of the GPT-2 model to understand and produce detailed recipe instructions.
The objective of this project is to showcase the power of language models like GPT-2 in generating coherent and creative cooking recipes. By fine-tuning GPT-2 on a targeted dataset containing structured culinary instructions, we delve into the intersection of AI and culinary arts.
The dataset employed is the "recipe-instructions" from Hugging Face, curated by darkraipro. It consists of detailed cooking recipes, providing a rich source of culinary vocabulary and structured instructions.
The model utilized is GPT-2, a renowned text generation model developed by OpenAI. For more details about GPT-2, visit its official model page on Hugging Face.
To execute the Jupyter notebook in this repository, ensure you have Python 3 and the following packages installed:
- jupyter
- transformers
- datasets
You can install these packages using pip with the following command:
pip install jupyter transformers datasetsTo view and run the notebook:
- Clone this repository to your local machine.
- Navigate to the cloned directory.
- Start the Jupyter Notebook server:
jupyter notebook- Open gen-ai-fine-tuning-llms.ipynb in the Jupyter interface that opens in your browser.
Contributions to this project are welcome! To contribute, please fork the repository, make your changes, and submit a pull request.
- Thanks to darkraipro for creating and sharing the "recipe-instructions" dataset.
- Thanks to OpenAI for developing the GPT-2 model.