A ai.meta.com Quantized Llama models with increased speed and a reduced memory footprint 508 points egnehots 2 years ago 122 comments
