Software & Data Downloads — SMITIN

Self-Monitored Inference-Time INtervention for Generative Music Transformers for controlling an autoregressive generative music transformer using classifier probes.

PyTorch code for Self-Monitored Inference-Time INtervention (SMITIN), an approach for controlling an autoregressive generative music transformer using classifier probes.

    •  Koo, J., Wichern, G., Germain, F.G., Khurana, S., Le Roux, J., "SMITIN: Self-Monitored Inference-Time INtervention for Generative Music Transformers", IEEE Open Journal of Signal Processing, DOI: 10.1109/​OJSP.2025.3534686, Vol. 6, pp. 266-275, January 2025.
      BibTeX TR2025-012 PDF Software
      • @article{Koo2025jan,
      • author = {Koo, Junghyun and Wichern, Gordon and Germain, François G and Khurana, Sameer and {Le Roux}, Jonathan},
      • title = {{SMITIN: Self-Monitored Inference-Time INtervention for Generative Music Transformers}},
      • journal = {IEEE Open Journal of Signal Processing},
      • year = 2025,
      • volume = 6,
      • pages = {266--275},
      • month = jan,
      • doi = {10.1109/OJSP.2025.3534686},
      • issn = {2644-1322},
      • url = {https://www.merl.com/publications/TR2025-012}
      • }

    Access software at https://github.com/merlresearch/smitin.