[andrejguran]

Very excited about the AI stats: 7B model running 30 tokens/ second, 13B+ parameters running on device, first token in 2.2sec. It seems we'll have more powerful AI models running on devices soon.