Llama 8B at 1800 tokens per second on Cerebrasinference.cerebras.ai2 points·huevosabio··0 commentsOpen articleSaveView on HN