2 ms·

Obviously cpu

• endorphine 3y ago
  Not obvious to me. I thought these AI things required vast amount of GPU power. Or is that for the training part?
  • ekianjo 3y ago
    Inference works fine on CPUs while a bit slow but fine for a chatbot
• vunderba 3y ago
  No. Ollama takes advantage of Metal on Mac M1.