Deploying a 1.3B GPT-3 Model with NVIDIA NeMo Megatron