Open the user inputs

#9
by wayneworkman2012 - opened

Hello, can you open up the user inputs of the Arena please? And preferably put a permissive license on it like Apache2

All conversations are logged to a public bucket. It's not under a license right now but I'll probably license it under Apache or similar.

Ah ok, I found it. Wasn't immediately obvious where it was. for anyone looking: https://huggingface.co/buckets/CodeSoft/slm-arena-data

I'd suggest putting a link to the bucket somewhere on the SLM Arena interface, probably somewhere along the bottom.

if you can put a license on the user inputs, I'd really like to use them.

I already disclose that all submitted prompts will be shared publicly, but I didn’t specify an explicit license for them. I’d be happy to make them reusable though. Do you know the best way to formalize that? Ideally I’d like to use CC0, or something similarly permissive.

This is what Claude suggests, seems like the best option to me:

Put a LICENSE file and a README.md at the bucket root. The Hub renders README.md below the file list, so state the license there using its standard short ID (e.g. CC-BY-4.0, Apache-2.0) and link to the LICENSE file. This is readable by people only. The Hub won't read YAML frontmatter in it as metadata.

Okay. Is CC0 a good license for this?

Yeah that seems fine. Normally I opt for apache2 because it ensures if anyone gets a patent or copyright on derivative work, you automatically get a license for usage - and importantly - they wouldn't be able to sue you for using their work.

Is what it is. Lawyers ruin the world. I'd suggest apache2 for literally everything but if you accept the liabilities of CC0 then that works too.

Done

Sign up or log in to comment