So, basically they did all the things their users said would kill the site and it [checks notes] killed the site.

Wait, that’s not Björn.

I agree - I stopped following because of how toxic the politics got. Now I’m following it again.

Did not know that - it makes it so much worse. They didn’t even bother writing a good prompt!

Big +1 for LRCGET. It is extremely powerful and dead simple to use. I’m generally able to point it at a directory and press go. It will return lyrics about as accurately and reliably as any manual search, and it has the (experimental) ability to embed the lyrics as well.

I prefer embedded lyrics, since I know that they will be available and inseparable from the files regardless of what my library looks like in the future.

Ooh, ooh, Nick Cage as Charlie Kaufman writing and then stealing a National Treasure adaptation.

Would watch.

Okay, but how do you tell the difference between a Ponzi scheme and normal Hollywood movie financing? /s

Okay, “hodl gang”

I’m saving this post for three months because I’m sure it’ll age like wine.

Signal from the kitchen seems more likely than a Earth sized TNO

welp, I’m sure that will never be misused…

Not for long

It’s tough keeping your feet dry when you’re… strolling through a dewy meadow.

“Where are his glasses? He can’t see without his glasses!”

press lap and check it again in two years.

How did you do the white hairs?

It seems like if you can recreate the training, the model could be considered open.

Lots of open source software performs operations on proprietary data, or accesses proprietary data/endpoints. That doesn’t make the software itself compromised.

LLMs are perhaps a little bit different in that the output (weights) is the part of the software that is useful.

It seems pretty obvious that an open weight model is a prime vector for social engineering/disinformation. If you can’t see the training data or mechanism, you are putting an awful lot of trust in the source. Better make sure that source has your best interests in mind.

Never tried it. It looks like maybe the mobile app is not available from open source repos.