@ryanramage yeah feel free to follow the main repo: https://github.com/RangerMauve/mind-goblin
Gonna push my latest version in the next week or so.
@ryanramage All within a few seconds with as little power or ram usage possible.
@ryanramage I'm giving the agent a tool call to use LLaVA to "see". I'm going to allow it access to the screen, an image file, or the camera.
I want a flow along the lines of:
- "Hey, save this text as a reminder for later"
- tool("see", "extract the text from the image") => take a pic and run through llava
- tool('save', "{summarized text}", ["reminder"]) => save to local database for later
- response: "Saved!"
- "What was the last reminder you saved?"
- tool('load', 'reminder', {limit: 1})
@starshine @delta Oh my bad! Misunderstood the last part.
@starshine @delta is doing some cool stuff. I think grant funding has been useful for getting these protocols set up.
I also use Matrix on the daily and it fits the criteria.
My hot take is that implementing ActivtyPub and getting it to talk to different implementations is not all that hard. I have had to do stuff that turns my brain inside out and "Oh what json properties do I need in this HTTP API with lots of examples in existing implementations" is a cakewalk. It just takes some time and tinkering.
@ww Maybe? Is this something a user would use from a cli? It seems to be more for web services from what I'm seeing.
@happyborg I've been using it with https://continue.dev as a local copilot. I mostly use it for doing basic refactoring or looking up syntax. I think search in general is getting worse and with local models you get the chance to do it all offline. I specifically care about having a fine tuned model people can have within their browser to let them make p2p apps fully offline and on mesh networks.
This post by @maggie has some great ideas on how #LLM tech can help enable #LocalFirst applications for regular folks. I've been wanting to do something similar within @agregore some day with local LLMs helping people author p2p web apps.
@enby_of_the_apocalypse @futurebird Sorry bad article it was the first result I found that was vaguely related. I'll see about getting a better link on Monday after I'm back at my pooter.
I was so excited I forgot to add the video! This is a single 7.5W laser running at 30kHz sample rate, the tracking is essentially a travelling salesman algorithm. Oh and the laser control software is all Seb’s too. Go check out Seb’s work! Truly terrific stuff https://seblee.co #asteroids #laser #emfcamp
As a general rule, the state should over-supply services. There should be slack in the system almost all the time. Relaxed GPs with lots of time to talk to their patients. So many teachers that the main trouble is finding rooms for them all.
Not only does this see a better quality of service mostly, but it also cushions the system in the event of an unexpected shock.
If you have just enough professionals to deliver at 100%, you don’t have enough professionals.
Occult Enby that's making local-first software with peer to peer protocols, mesh networks, and the web.
Exploring what a local-first cyberspace might look like in my spare time.