Excited to be the first to achieve large scale training on full stack AMD compute and networking. Our technical report dives into deep detail on how our AMD training stack works Now we've proved out AMD training, excited to scale up. There are now no bottlenecks except compute.