[ipsum2]

You can run llama models on a personal computer, even though it was trained on >10,000 GPUs.