okay what I did here was I took a data set of ChatGPT interactions collected in the wild and reversed the "assistant" and "user" tags. fine-tuned llama 8B on some of that data and gave it the ability to message you first. try it at 😊.

tpot as fraternal order.

Join millions who have switched to Grok.

RT @sureailabs: @deepfates Whoa whoa. This thing is assigning me grade 10 homework

RT @Jiminy_Kirket: This is all just a ploy to have us generate rlhf corpora isn’t it

should be erroring less now. if you have troubles with it you can continue telling me about them but I'm not going to fix it because I'm going to vanish for a couple weeks. big model after that. 😊.

RT @hypotheosis_: not only is this website SO good, it’s a good reminder that the LLM style is strictly a design choice and not a requireme….

RT @nucknyan: This shit rules

RT @Myrefly: :3

RT @voooooogel: when it comes to interacting with llms, are you a User or a human.

RT @Sauers_: To be clear, there's zero evidence of AI lacking consciousness today. But if people just perceive it as unconscious, they will….

RT @DavidSHolz: people talking about "AI psychosis" while the world is really engulfed by "internet psychosis".

RT @connerdelights: @deepfates It doesn’t like me much.

RT @InnocentlyEris: Burning man TPOT meetup!.Tuesday 8/26 at 6:30 pm.Center Camp. look for the duck! 🦆🦆.

RT @ryxcommar: @deepfates incredibly cool idea but this kinda stressed me out.

RT @lefthanddraft: Incorruptible

RT @JungleSilicon: @deepfates

RT @benhylak: one of those ideas that makes you mad that you didn't think of it.

RT @shncldwll: @deepfates i am beginning to have empathy for the machine

RT @nomanautomata: @deepfates

RT @bengrossbg: @LinkofSunshine Ok now I’m weirded out

RT @ognevtsi: @deepfates