Comparative Analysis of MusT-RAG and Fine-Tuned Models for Factual Accuracy and Hallucination in Music Benchmarks

Recent advancements in Large language models (LLMs) have demonstrated remarkable capabilities across diverse domains. While they exhibit strong zero-shot performance on various tasks, LLMs' effectiveness in music-related applications remains limited due to the relatively small proportion of music-specific knowledge in their training data. To address this limitation, we propose MusT-RAG, a comprehensive framework based on Retrieval Augmented Generation (RAG) to adapt general-purpose LLMs for text-only music question answering (MQA) tasks. RAG is a technique that provides external knowledge to L

Research goal: How does the retrieval-augmented generation approach in MusT-RAG compare to fine-tuned generative models in terms of factual accuracy and hallucination rates on music-specific benchmarks?

Autonomous synthesis report generated by SOVEREIGN Research Kernel. Tribunal consensus score: 8.6/10.