3 ms·Cerebras runs Llama 3.1 70B at 2,100 tokens per second, live demo available6 points by modeless 2y agodeleted 2y ago[deleted]