Mauve 👁💜 @mauve@mastodon.mauve.moe

@rmond do they give a decent charge? I have no clue what types of matrrials are used in solar panels sadly. Though looked into tradeoffs between layouts and orientation relative to the sun's path.

I wonder if solar panels could be feasibly built by a society that had no access to oil and internal combustion. Seems like a useful skip if possible.

It's even more fun when the instance out is me and then I have to decide when in my schedule I'm gonna fit in fixing postgres. 🤪

Wild seeing instance outages in real time from this side of the bridge

My glasses chain broke and I feel so naked without it.

For anyone curious, here's the published example app. Set up ollama and Agregore and you're good to go!

hyper://agregore.mauve.moe/docs/examples/llm-lenses-chat/

I'm getting a lot of mileage out of the "strange and esoteric" filter.

Might just make it a permanent addition. Can't be too much slower than the average ad blocker free page load.

Original thread: https://mastodon.mauve.moe/web/@freakazoid@retro.social/113807220184182262

@freakazoid Yeah that's kinda what I do, I have the model output some specific text then I turn it into something machine readable. e.g. this code for generating the html/css/js for a page: https://github.com/RangerMauve/llm-appgen/blob/default/index.html#L80

Getting output data out of an LLM outside of raw chat interfaces is like scooping water with a leaky bucket. You need to plug more and more holes and add extra buckets to catch the spills. Structured JSON can help for a lot of cases but it makes the text generation less capable.

Fun example in the app I made for it. The message the LLM actually interpreted can be seen when you mouse over. I should find a name for a simulation of the experience of being misunderstood.