Can that thing run an open source trainable model?
And I really have no idea from which I speak. So I'm going to babble a bit and hopefully someone here can correct and point me in a direction.
I would like a totally local AI model run. I don't want it to reach out and touch anything on the web unless I specifically tell it to. I want to feed it information though, maybe text or email or Real-Time camera and audio input or point to specific websites and it should digest those.
How about a home-based security system that has cameras all over the place outside and can tell the difference between deer and raccoon movement and a human attempting to be sneaky? What would it take to have a real-time home security system to be able to do that?
I'm not that paranoid. I'm just reasonably paranoid. Let's face it, I occasionally piss people off.
Does such a thing exist? And is it runnable on the above mentioned box in a reasonable fashion, ie: a training period should never be more than a day or so. And yes, training is different from the inference phase. But I expect to swap back and forth until it's trained and then let it run.
And if not that box, what box?
And does anyone recommend a specific model?
I would like a totally local AI model run. I don't want it to reach out and touch anything on the web unless I specifically tell it to. I want to feed it information though, maybe text or email or Real-Time camera and audio input or point to specific websites and it should digest those.
How about a home-based security system that has cameras all over the place outside and can tell the difference between deer and raccoon movement and a human attempting to be sneaky? What would it take to have a real-time home security system to be able to do that?
I'm not that paranoid. I'm just reasonably paranoid. Let's face it, I occasionally piss people off.
Does such a thing exist? And is it runnable on the above mentioned box in a reasonable fashion, ie: a training period should never be more than a day or so. And yes, training is different from the inference phase. But I expect to swap back and forth until it's trained and then let it run.
And if not that box, what box?
And does anyone recommend a specific model?