return to table of content

Llama 3.1 405B now runs at 969 tokens/s on Cerebras Inference