AirLLM запускает ИИ-модели с 70 млрд параметров без суперкомпьютера #ИИ #Нейросети #Python

What if you could run an EI model with billions of parameters without any supercomputer? This is now possible with an open source Python library called Air LM. Instead of loading a massive model at once, it loads it layer by layer from your hard drive, like reading a book page by page instead of

memorizing it all at once. Also [music] has a Flash Attention feature that keeps memory usage almost constant, even with long input data. With it, models like the Lama 337 billion parameters [music] can work right on your MacBook or gaming computer. For students, researchers, and indie developers, this is a complete game changer.