[ipsum2]

I would like to see time to first inference for typical models (llama-7b first token, SDXL 1 step, etc)