Mauve 👁💜 @mauve@mastodon.mauve.moe

@jazzbox35 Install instructions for what? I usually run models from my own harnesses these days.

Holy shit the new #Phi3 #LLM from microsoft is 3.8B params but performs better than a bunch of 7B models I've tried. Now I just need to find a way to get it to do function calling.

https://www.theverge.com/2024/4/23/24137534/microsoft-phi-3-launch-small-ai-language-model

Programming is wizardry, Prompt engineering is demonology.

@Filene @atax1a yeah absolutely. I read the book "The Structure Of Magic" to understand how to work with it better. I wish I could use it to talk to computers more effectively but it's also just way less reliable than computer code :P

@julian @evan @silverpill here is a post I managed to get resolving which might help for comparison https://github.com/RangerMauve/staticpub.mauve.moe/blob/default/helloworld.html

@julian @evan @silverpill In general I have had to do this for most implementations to catch gotchas 😅

@julian @evan @silverpill in my tests I found masto could get annoyed if it cannot fully resolve the author of the post and all the extra details. Unsure if that's relevant here. Try checking your http server logs vs the mastodon source code.

@julian @evan @silverpill Cool mind sending a link to the post json-ld and maybe the source code? I'd love to test it against my implementations and see what's up.

It's a majorly missed opportunity that tools like huggingface don't use bittorrent for distribution by default.

@julian @evan @silverpill I found Link tag to suffice. Do you also have webfinger implemented?

The Crow was the original Joker 2019

@atax1a NGL I would prefer a spoken programming language at this point.

Happy weed day to those that celebrate. 🍃🍃