This repository contains the Gen-AI model developed for the Elevate app.

The model is hosted on the Hugging Face model hub: lordgrim18/llama2-elevate-story-3

The Gen-AI produces a short story based on the genre given. You can also add a context if needed.

• scraper.py: Python script for scraping stories from Archive of Our Own.
• data.csv: CSV file containing extracted stories including its url, title and content.

• initial_cleaning.ipynb: Jupyter Notebook for initial data cleaning - removing rows with empty fields.
• cleaned_data.csv: CSV file containing cleaned data after the initial cleaning process.

• Basic_llama_working.ipynb: Jupyter Notebook demonstrating basic usage of the Llama model.
• data_augmentation.ipynb: Jupyter Notebook for data augmentation using the Llama model - creating new content, genre, context.
• output.csv: CSV file containing augmented data.

• final_touches.ipynb: Jupyter Notebook for final data preparation which is then uploaded into huggingface - lordgrim18/story-2 .
• train.csv: CSV file containing the final training data which is in the form that is used to fine tune the - NousResearch/Llama-2-7b-hf.

• training_saving.ipynb: Jupyter Notebook for training the Gen-AI model and saving the trained model.

• testing.ipynb: Jupyter Notebook for testing the Gen-AI model.

You can use this repository to explore and understand the development and deployment, from obtaining the data and creating a dataset to saving and testing a short story generating Gen-AI model.