Expand description
§Model Deployment Card
The ModelDeploymentCard (MDC) is the primary model configuration structure that will be available to any component that needs to interact with the model or its dependent artifacts.
The ModelDeploymentCard contains LLM model deployment configuration information:
- Display name and service name for the model
- Model information (ModelInfoType)
- Tokenizer configuration (TokenizerKind)
- Prompt formatter settings (PromptFormatterArtifact)
- Various metadata like revision, publish time, etc.
Structs§
Enums§
- Generation
Config - Model
Deployment Card Builder Error - Error type for ModelDeploymentCardBuilder
- Model
Info Type - Prompt
Context Mixin - Prompt
Formatter Artifact - Supported types of prompt formatters.
- Tokenizer
Kind
Constants§
- ROOT_
PATH - Identify model deployment cards in the key-value store