Lamini software comes preinstalled to run and train LLMs on parity with NVIDIA CUDA on H100.
Pics
ssh into a node: https://twitter.com/realSharonZhou/status/1748418515852525571
machine humming: https://twitter.com/realSharonZhou/status/1748530474367254699
chip itself: https://twitter.com/realSharonZhou/status/1752463299667697714