Commercial use permission request for BS-Roformer-Resurrection-Inst.ckpt in MarksDL
Hello pcunwa / unwa,
I am developing MarksDL, a paid, closed-source macOS and Windows application for local audio stem separation. We would like to use this exact checkpoint:
- File: BS-Roformer-Resurrection-Inst.ckpt
- SHA-256: 16311025a5133ae6411760ccfe9e3e66b31a01d9d8bec0a03fa7ec4bedac7a15
- Source: https://huggingface.co/pcunwa/BS-Roformer-Resurrection
Could you please explicitly confirm whether you authorize MarksDL to:
- use this checkpoint for commercial, paid, closed-source local inference on macOS and Windows;
- let users automatically download the unchanged weights from the upstream source;
- alternatively bundle or rehost the unchanged weights with the app/installer;
- convert the checkpoint format when technically necessary, without changing its learned behavior;
- include it in current and future MarksDL releases on a non-exclusive, worldwide, royalty-free basis?
Proposed attribution:
“BS-RoFormer Resurrection Instrumental — by unwa (pcunwa)”
with a link to the repository above.
Please also confirm:
- whether you have authority to grant these permissions for the checkpoint;
- whether the training data or any third-party material imposes restrictions on commercial use or redistribution;
- any additional credit, notice, or license terms you require.
We will keep this feature private/testing-only and will not publicly distribute the checkpoint unless the commercial permissions are clear.
Thank you very much.
One quick personal introduction: my name is Anthony, and I’m the independent developer behind MarksDL. Thank you again for considering my request.
Best regards,
Anthony
Developer of MarksDL
This model serves as a fine-tuning checkpoint for BS-Roformer SW, which specializes in voc/inst separation.
As for where BS-Roformer SW originally came from, it’s Apple’s Logic Pro.
The model architecture itself has been open-sourced by Lucidrains.
I’m not sure who has the authority to grant permission in this case, but at least I don’t mind.