The unmodified concurrency client and model deployment template structure are derived from NVIDIA Dynamo v1.6.1, examples/gpu-sharing/.
Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
These files are licensed under Apache-2.0; the license is included in vendor/dynamo/LICENSE. The model, runtime versions, two-GPU placement, scheduler selection, and environment placeholders were adapted for the reported experiments and this reproduction package.