We'll be running highly quantized, somewhat distilled versions of something similar to Llama on our devices before long, and I don't think the RLHF part will ta

We'll be running highly quantized, somewhat distilled versions of something similar to Llama on our devices before long, and I don't think the RLHF part will take long to be replicated, the biggest block there is just data.