Cerebras launches processor for AI theoretically equivalent to 62 Nvidia H100s

And like the previous two wafer-scale processors, virtually no one will use them.

But can I use it just by going `model.to("cerebras")`?