Skip to content

Latest commit

 

History

History
76 lines (54 loc) · 2.85 KB

ai-support.spec.md

File metadata and controls

76 lines (54 loc) · 2.85 KB
Raw

AI Support Spec

This document describes the structure of the ai-support.json file. The JSON file is utilized by an automation to update our documentation with the AI libraries/abstractions/gateways that we support.

Structure

The general structure of the JSON file is that of an object who's keys map to a descriptor for a supported AI thing. Each descriptor has a kind property that indicates what sort of AI thing is being described.

kind = "abstraction"

An abstraction is a module that provides a single common interface to many LLMs or gateways. An abstraction descriptor has the following fields:

  • title (string): A human readable name for the abstraction.
  • featuresPreamble (string): An optional block of text that should be added to the document prior to the features table.
  • providersPreamble (string): An optional block of text that should be added to the document prior to the providers table.
  • features (object[]): An array of feature entities.
  • providers (object[]): An array of provider entities.

kind = "gateway"

A gateway is a service that provides access to multiple large language models (LLMs) through a single API. A gateway descriptor has the following fields:

  • title (string): A human readable name for the gateway.
  • preamble (string): An optional block of text that should be added to the document prior to the models table.
  • footnote (string): An optional block of text that should be added to the document subsequent to the models table.
  • models (object[]): An array of model entities.

kind = "sdk"

A SDK is a module that provides an API that is specific to a single LLM. An SDK descriptor has the following fields:

  • title (string): A human readable name for the SDK.
  • featuresPreamble (string): An optional block of text that should be added to the document prior to the features table.
  • features (object[]): An array of feature entities.

Entities

Feature

Describes an LLM feature. It is an object with the following fields:

  • title (string): A human readable name for the feature.
  • supported (boolean): Indicates if our instrumentation supports the feature or not.

Model

Describes an LLM, the features it supports, and the features we instrument. It is an object with the following fields:

  • name (string): A human readable name for the LLM.
  • features (object[]): An array of feature entities.

Provider

Describes an LLM or gateway that is supported by an abstraction. It is an object with the following fields:

  • name (string): A human readable name for the LLM or gateway.
  • supported (boolean): Indicates if we instrument this provider within the context of the abstraction.
  • transitively (boolean): Indicates if we instrument this provider directly in the instrumentation for the abstraction (false), or if we rely on a transitive instrumentation (true).