When will incorporate Cerebras inference

[Yvette]

I feel like the inference speed now is still quite long for code generation, is there any possibility of adopting Cerebras inference to speed up? But seems it only supports llama models now.