Hacker News new | past | comments | ask | show | jobs | submit
make3 | on: We're releasing the weights of Mistral Large 4
Training neural networks is requires extremely high speed interconnect between the humongous GPUs