Write your own LOTR story with LORA BLOOM-3B (2023) (github.com)

A new AI assistant, LORA BLOOM-3B, has been developed to help users craft their own stories in the style of "The Lord of the Rings". This project involved extracting text from the original novels, preparing a specialized dataset, and fine-tuning the BLOOM-3B language model using Low-Rank Adapters (LoRA). The model has been deployed using Hugging Face's Inference Endpoints, while the user interface for the story creation app was built with Streamlit and is hosted on the Streamlit cloud. This innovation is significant for the AI/ML community as it represents a creative application of advanced language modeling techniques to generate narrative content in a beloved literary style. By utilizing techniques such as LoRA for efficient fine-tuning and deploying advanced models like BLOOM-3B, the project showcases the potential for AI in personalized storytelling. Key technical details include a context length of 2048, specific LoRA hyperparameters (e.g., lora_r = 16, lora_alpha = 32), and a learning rate of 2e-4, all of which contribute to the model's capability to generate coherent and engaging content within a familiar narrative framework.