{"type":"ScholarlyArticle","title":"NotaGen: Advancing Musicality in Symbolic Music Generation with Large Language Model Training Paradigms","authors":[{"name":"Yashan Wang","slug":"yashan-wang","given":"Yashan","family":"Wang","affiliation":""},{"name":"Shangda Wu","slug":"shangda-wu","given":"Shangda","family":"Wu","affiliation":""},{"name":"Jianhuai Hu","slug":"jianhuai-hu","given":"Jianhuai","family":"Hu","affiliation":""},{"name":"Xingjian Du","slug":"xingjian-du","given":"Xingjian","family":"Du","affiliation":""},{"name":"Yueqi Peng","slug":"yueqi-peng","given":"Yueqi","family":"Peng","affiliation":""},{"name":"Yongxin Huang","slug":"yongxin-huang","given":"Yongxin","family":"Huang","affiliation":""},{"name":"Shuai Fan","slug":"shuai-fan","given":"Shuai","family":"Fan","affiliation":""},{"name":"Xiaobing Li","slug":"xiaobing-li","given":"Xiaobing","family":"Li","affiliation":""},{"name":"Feng Yu","slug":"feng-yu","given":"Feng","family":"Yu","affiliation":""},{"name":"Maosong Sun","slug":"maosong-sun","given":"Maosong","family":"Sun","affiliation":""}],"abstract":"We introduce NotaGen, a symbolic music generation model aiming to explore the potential of producing high-quality classical sheet music. Inspired by the success of Large Language Models (LLMs), NotaGen adopts pre-training, fine-tuning, and reinforcement learning paradigms (henceforth referred to as the LLM training paradigms). It is pre-trained on 1.6M pieces of music in ABC notation, and then fine-tuned on approximately 9K high-quality classical compositions conditioned on \"period-composer-instrumentation\"  prompts. For reinforcement learning, we propose the CLaMP-DPO method, which further enhances generation quality and controllability without requiring human annotations or predefined rewards. Our experiments demonstrate the efficacy of CLaMP-DPO in symbolic music generation models with different architectures and encoding schemes. Furthermore, subjective A/B tests show that NotaGen outperforms baseline models against human compositions, greatly advancing musical aesthetics in symbolic music generation.","doi":"10.24963/ijcai.2025/1134","url":"https://ijcai.org/proceedings/2025/1134","pdf":"https://ijcai.org/proceedings/2025/1134.pdf","markdown":"https://ijcai.org/proceedings/2025/1134.md","bibtex":"https://ijcai.org/proceedings/2025/bibtex/1134","volume":{"slug":"2025","title":"Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2025","url":"https://ijcai.org/proceedings/2025/","year":2025,"location":"Montreal, Canada","dates":"16-22 August 2025","editors":"James Kwok","isbn":"978-1-956792-06-5","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2025"},"section":"AI, Arts & Creativity","track":"AI, Arts & Creativity","number":1134,"pages":{"first":10207,"last":10215},"keywords":["Theory and philosophy of arts and creativity in AI systems: Autonomous creative or artistic AI","Methods and resources: Machine learning, deep learning, neural models, reinforcement learning","Application domains: Music and sound","Methods and resources: Techniques for modeling and simulation of creativity"],"videos":[],"references":[],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}