Skip to main content

Data Model

RouteConfig​

Fields:

  • organization: The organizational unit or company associated with this route.
  • owner: The individual or entity responsible for this route.
  • rate_limit: Maximum number of requests allowed for this route in a given time period.
  • retries: Number of times the request should be retried in case of failures.
  • archive: Boolean indicating whether route data should be archived.
  • retention: Data retention period.
  • budget: Monetary or computational budget allocated for the operations of this route.
  • dlp: DLP configuration.

budget​

Fields:

  • enabled: Enable/Disable Data Loss Prevention(DLP).
  • annual: Annual budget limit.
  • monthly: Monthly budget limit.
  • weekly: Weekly budget limit.
  • currency: Currency for the budget.

dlp​

Fields:

  • enabled: Enable/Disable Data Loss Prevention(DLP).
  • strategy: Data Loss Prevention strategy inspect or de-identify.

model​

Fields:

  • name: The name of the LLM model.
  • provider: Entity or service providing this LLM model.
  • suffix: Additional descriptive or versioning information for the LLM model.
  • weight: Weight of the model.

Route​

Fields:

  • name: The unique identifier for this route.
  • type: The type chat, completions, embeddings of this route.
  • enabled: Enable/Disable this route.
  • models: Associated models for processing or handling requests on this route.
  • config: Configuration settings, like rate limits, archive, budget associated with this route.

Routes​

Fields:

  • routes: List of route objects.

ProviderConfig​

Fields:

  • api_base: Base URL of the API.
  • api_type: Type of the API.
  • api_version: Version of the API.
  • deployment_name: Name of the deployment.
  • organization: Name of the organization.

Provider​

Fields:

  • name: The unique identifier for this provider.
  • type: Type of the Provider.
  • enabled: Whether the provider is enabled.
  • vault_enabled: Whether the secrets vault is enabled.
  • config: Configuration for the provider.

Providers​

Fields:

  • providers: List of provider objects.

TemplateConfig​

Fields:

  • infoTypes: List of infoTypes.
  • transformation: Transformation to be used.
  • notify: Whether to notify.
  • reject: Whether to reject.
  • likelihood: Indicate how likely it is that a piece of data matches infoTypes.
  • routePrompt: Prompt to be used for the route.

infoType​

Fields:

  • name: Name of the infoType.
  • description: Description of the infoType.
  • regex: Regex of the infoType.
  • category: Category of the infoType.

Transformation​

Fields:

  • method: Method of the transformation Mask, Redact, Replace, etc.

Template​

Fields:

  • name: The unique identifier for this template.
  • description: Description for this template.
  • type: Type of the Template.
  • enabled: Whether the template is enabled.
  • config: Configuration for the template.

Templates​

Fields:

  • templates: List of template objects.

Secret​

Fields:

  • api_key: Key of the Secret.
  • api_key_secret_name: Name of the Secret.
  • api_key_secret_key: API Key of the Secret.
  • api_key_secret_key_javelin: Virtual API Key of the Secret.
  • provider_name: Provider Name of the Secret.
  • query_param_key: Query Param Key of the Secret.
  • header_key: Header Key of the Secret.
  • group: Group of the Secret.
  • enabled: Whether the secret is enabled.

Secrets​

Fields:

  • secrets: List of secret objects.

Usage​

Fields:

  • completion_tokens: Number of tokens used in the completion.
  • prompt_tokens: Number of tokens used in the prompt.
  • total_tokens: Total number of tokens used.

Message​

Fields:

  • role: Role or purpose of the message, e.g., 'error', 'info', 'warning'.
  • content: The actual content or body of the message.

Choice​

Fields:

  • finish_reason: Reason for the completion finish.
  • index: Index of the choice.
  • message: Message details.

QueryResponse​

Fields:

  • choices: List of choice.
  • created: Creation timestamp.
  • id: Unique identifier of the response.
  • model: Model associated with the response.
  • object: Object type like chat.completion.
  • system_fingerprint: System fingerprint if available.
  • usage: Metrics or data about the computational resources used.
  • javelin: Processing data & metrics.