3 ms·
People running LLMs on CPU are generally running them integer quantized, so they use fewer bits per parameter.
by sbierwagen 3y ago
People running LLMs on CPU are generally running them integer quantized, so they use fewer bits per parameter.
loading thread…