Computational creativity - virtual muse generative poetry

Abstract

The convergence of artificial intelligence with creative expression has given rise to the emerging field of Computational Creativity, a domain that explores the capacity of machines to generate innovative and aesthetically pleasing content. This BTech project navigates through the intriguing world of poetry and Bollywood song lyrics generation, leveraging the power of Large Language Models (LLMs) such as GPT-2 and Llama. With a particular emphasis on fine-tuning these models, the project surpasses traditional methodologies by incorporating personalized datasets and pioneering efforts with the Devanagari script to craft poetry and songs in Hindi. The journey through the evolution of LLMs highlights their transformative impact in the field of natural language processing, culminating in their current capability to produce content that rivals human creativity. This research focuses on tailoring these advanced models to personalized datasets, significantly extending the scope of computational creativity into the realm of Hindi language, thereby enriching cultural and linguistic dimensions within the creative process. By exploring the nuances of language, culture, and expression, this project aims to demonstrate the vast potential of LLMs to generate poetry and songs that deeply resonate with human emotions and cultural intricacies. Ultimately, this project stands at the intersection of linguistic innovation, cultural expression, and artificial intelligence, aspiring to extend the limits of what technology can achieve in the captivating field of creative automation.