#Songrater
16 messages ยท Page 1 of 1 (latest)
there was a Korean group which hosted a music ML conference recently: https://ismir2025.ismir.net/
i hear rumors in various discords they are working on a music version of LMArena. potential colab opportunity with LAION, or at least might be worth a few friendly emails along the way - songrater is already amazing, just hoping to add N+1 to those who love music and research
more information on their work in the music domain:
https://music-ir.org/mirex/wiki/MIREX_HOME
I get kicked out of the session every 2-3 minutes.
it should be fixed, just uploaded a new version
@gritty iris @potent tulip the tab playback should be fixed now (if you move from another tab back to your music playing tab, the music should keep playing as intended, also for longer sessions)
Ughh, a bit too niche I would say
Also LMArena for Music models is an overkill as most models don't have a sense of what they are generating. That's why we are launching this benchmark
Love the idea and want to help. I am still getting deauthorized every few minutes. I signed in using webauthn on Firefox but can test on a chromium based browser if you'd like.
I submitted feedback in the hopes that the logs were also transmitted.
Let me know if I can help troubleshoot this issue.
Ok please check again, I hope to have fixed the issues!
if anyone is stable, please post your stack (linux + firefox, etc)
Im checking out this new project. The website looks just like all my tailwinds, shadCN vibe coded projects haha
reminds me back when bootstraps was all the crazy, every website looked the same
tailwindcss + shadcn is a pretty standardized an functional layout, that's what the site was intended to be functional ๐
That did fix the issue! Thanks for fixing it so quick.
Some more feedback if you don't mind. The Core Assessments don't reset after submission which can cause a form error since if the previous selection is not changed.
Please check again, should be fixed!
<@&1053371578529878106> Hey dataset builders! Thanks to the kindness of tensorpool, every annotator that annotates 2 songs (and listens to each of them at least 75%) gets a 20$ tensorpool voucher to use for ML training or inference! ๐