4 ms·
> at home with the right hardware Where the right hardware is 10x4090s even at 4 bits quantization. I'm hoping we'll see these models get smaller, but the GPT-
by lolinder 2y ago
> at home with the right hardware
Where the right hardware is 10x4090s even at 4 bits quantization. I'm hoping we'll see these models get smaller, but the GPT-4-competitive one isn't really accessible for home use yet.
Still amazing that it's available at all, of course!
- petercooper 2y agoIt's hardly cheap starting at about $10k of hardware, but another potential option appears to be using Exo to spread the model across a few MBPs or Mac Studios: https://x.com/exolabs_/status/1814913116704288870 https://x.com/exolabs_/status/1814913116704288870
- niutech 2y agoOr maybe using Distributed Llama? https://github.com/b4rtaz/distributed-llama https://github.com/b4rtaz/distributed-llama