This is... pretty gross, actually.

I tried listening to the spotify version of Legends of Avantris on the way to work this week, and... nope.

It has background music for some fucking reason and it's hard to actually *listen* to what people are fucking saying.

I have no fucking idea why some #Podcast and #ActualPlay folks think this is a good idea

It's not 'immersive', it's fucking distracting.

'LLMs need data moderation, and this is usually done by people in third world countries, a-'

... and so does Reddit and YT and FB and Twitter and every other major service that you use, on a similar scale.

I understand that this might piss you off for LLMs, and I get that this might seem like whataboutery, but I think it's interesting how this is only being raised about LLMs and not about, you know, *gestures* everything else.

On one hand, I wanna do model kits.

On the other hand, I wanna do more dataset shit

On the third hand: headachey

This dataset began with a purpose, but honestly, it's just fun to work on, so it may outlive its original project.

This, uh, this isn't getting out of hand.

Nope.

In fairness to me, that's not all one, cohesive dataset.

It's uh. Four different ones?

But, uh, yeah, that's still a fucking lot of data.

Let's push a bit of #controversy with a #poll, shall we?

When is it ok to clone someone's voice for a project?

I'll be interested to see the results for this one.

Coincidentally, I read a definition of archaeology vs grave robbery that was... 'so long as they died 50 years ago'

That'd be the mid 70s for us, btw.