Nebius and PyTorch built a high-performance DeepSeek V3 training recipe on TorchTitan, running on 256 Nebius B200 GPUs with Soperator. This post dives deeper into why GPU-initiated RDMA with NVSHMEM and DeepEP make such an impact in improving performance for large MoE models, and walks through the full IBGDA and DeepEP setup you can run on Nebius today.