Yeah, sorry about that. 💚 There's been a lot of UI drift lately and we really need to train another iteration of the model soon.

Yeah that's what I'm thinking

There's been a lot of UI drift lately so it's probably time to retrain the model, I've noticed it's been really bad at labelling Bluesky screenshots as Twitter recently.

I'm juggling a half dozen deadlines right now but I'll try to do that sometime in the next while, please bear with me.

Update, it should be working better now, I've switched the queue to prioritise newer posts. 🤞🏼

It's still limping along! I need to retrain the model I think, there's been enough UI drift that it's starting to get really bad at missing screenshots. It's always a huge amount of work to do that so it admittedly might take awhile.

We don't have anything to do with that user's lists, sorry. We simply label screenshots, that's it.

How weird, it doesn't look much like Bluesky!

Thanks for tagging, I'll add that to the false positive training set, hopefully I can improve that in the future.