Model Serving · Data artifact

Model Metadata Descriptor

Data artifactModel ServingModelsarc:ModelMetadataDescriptor

A machine-readable description of a model's architecture, context length, tokenizer settings and build environment that serving runtimes use to configure backend engines.

Responsibility. Informs the serving runtime how to initialise engines for a specific model.

Also known as: Hugging Face config.json, Model metadata

Direct neighbourhood (hover for relationship types)

Relationships

configures structural

LLM Inference Service Ch4.5

is written by dependency

Engine Builder Ch4.5

Design guidance

MUST provide complete and accurate architecture, context length and tokenizer metadata for every served model.
SHOULD record the CUDA and engine library versions used to build compiled engines as part of model metadata.

Classification

Risks mitigated: Engine initialisation failure from missing or malformed model metadataEngine version mismatch across environments

Sources

Ch4.5: T. Nguyen, "NVIDIA NIM and Triton Inference Server," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 4.5. ISBN: 9798244538229.