OpenAI's nightmare: DeepSeek R1 on a Raspberry Pi 5 ;)

https://www.youtube.com/watch?v=o1sN1lB76EA

OpenAI's nightmare: Deepseek R1 on a Raspberry Pi

YouTube
@geerlingguy I tried this today, works amazing
@theearthisapringle @geerlingguy does the autonomous model poop its pants when you ask about Beijing in 1989?

@geerlingguy

This is great news. Its not quite there yet but the potential for an LLM layer to sit on top of other apps an data sets (like my car and its 1000 page manual) is within sight. We may in a couple of years start to see an LLM optional user interface running locally on basic hardware.

Ultimately I dont need a poet, I want something that works like the computer on #StarTrek Voyager

#AI #LLM

@geerlingguy I'm yet to watch the video but as someone mulling over the idea of running an LLM at home but trying not to purchase loads of hardware to do it, I'm excited by the continued development in this area. Currently I'm using Gemini through their OpenAI-compatable API.
@geerlingguy I somehow approve of this content - it costs 5m to train an ai now, watch that cost plummet but there will be a big mkt. Let this year play out to see how quickly things evolve - it may be a lot faster than most anticipate? #free for commercial use #llm
@geerlingguy Something like 20-ish, sharing all the costs.

@geerlingguy

Winnie wants his Dyson Sphere !
.Now.

@geerlingguy
Useless unreliable tripe running on a Pi is STILL useless unreliable tripe, but if it bursts the AI hype bubble, it has redeeming value after all.

@geerlingguy Oh the irony; OpenAI and Microsoft are claiming that DeepSeek took their data without permission to create their own, yet OpenAI took everyone's data without permission to train it's LLM's

You can't make this up 🤣🤣🤣

@geerlingguy
golden:
... this works ... somewhat 😂
... winnie the pooh of tech 😂