numpy constraint - can it be lifted?

[rsxdalv]

https://github.com/LAION-AI/CLAP/blame/8e558817d853808486768004fa1b61ac9d69f2a2/pyproject.toml#L24

Most of the TTS/Audio projects I have use numpy >= constraints, except CLAP with ==, which is a requirement of Stable Audio Tools.

How can I test if it's still relevant, and perhaps lift this constraint?

[lukewys]

Hi, Thanks for raising this issue. I think you can clone from git, change the pyproject.toml and install from there. Please let us know if the constraint can be lifted. Thanks! Best, Yusong Roberts Slisans ***@***.***> 于2024年8月4日周日 14:57写道：

…

https://github.com/LAION-AI/CLAP/blame/8e558817d853808486768004fa1b61ac9d69f2a2/pyproject.toml#L24 Most of the TTS/Audio projects I have use numpy >= constraints, except CLAP with ==, which is a requirement of Stable Audio Tools. How can I test if it's still relevant, and perhaps lift this constraint? — Reply to this email directly, view it on GitHub <#160>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGXJZ76TDLGNJTZUDVOKXA3ZPZ2SLAVCNFSM6AAAAABL7CC6ISVHI2DSMVQWIX3LMV43ASLTON2WKOZSGQ2DOMRVG4YDAOA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[pharmapsychotic]

This is an issue for me as well resolving dependencies with other libraries that pin numpy==1.26.4 while laion-clap pins numpy==1.23.5. It would be helpful for libraries to specify the range of versions they work with so a mutually compatible version can be selected.

[rsxdalv]

I think the downstream project that I'm using works fine with newer numpy versions. However I still have not done a direct test of the full library API.