Enable Hybrid Training and Inference With DGX Cloud and OCI GPU Infrastructure (Presented by Oracle)
, Distinguished Cloud Architect, Oracle Company
We'll present a comprehensive overview of the integration and optimization of large language model (LLM) training using NVIDIA DGX Cloud on Oracle Cloud Infrastructure (OCI), coupled with efficient inference deployment on OCI's GPU infrastructure. We'll specifically focus on how the NVIDIA Triton MLOps pipeline plays a pivotal role in deploying trained models seamlessly from DGX Cloud to OCI GPUs, thereby enabling a robust and scalable hybrid AI infrastructure.