Skip to main content

Setting up config.json

The configuration config.json file structure for the system is organized as follows:

  • ApiKey: Your unique API key for authenticating requests to the service's API.

CompletionsSettings

  • Url: The endpoint URL for text completions (e.g., https://api.openai.com/v1/chat/completions).
  • Model: The default language model, such as "gpt-4-1106-preview".
  • Temperature: Controls the randomness in responses, with 0.0 being the most deterministic.
  • SystemInstructions: Instructions defining the AI's response behavior and style.

EmbeddingsSettings

  • Url: Endpoint URL for obtaining embeddings (e.g., https://api.openai.com/v1/embeddings).
  • Model: Specifies the model for embeddings, like "text-embedding-3-small".

TranscriptionsSettings

  • Url: The URL for audio transcription services (e.g., https://api.openai.com/v1/audio/transcriptions).
  • Model: The model used for transcription, such as "whisper-1".

SpeechSettings

  • Url: Endpoint for speech generation services (e.g., https://api.openai.com/v1/audio/speech).
  • Voice: Specifies the voice to be used, for instance, "nova".

RagSettings

  • Chunksize: The size of text chunks for processing.
  • ChunkOverlap: The overlap size between chunks.
  • ChunkOverrun: Extra size allowed for a chunk.
  • ChunksToUse: The number of chunks to be used in processing.

VisionSettings

  • url: Endpoint for vision-related tasks (e.g., https://api.openai.com/v1/chat/completions).
  • detail: Level of detail required, such as "high".
  • model: The specific model for vision tasks, like "gpt-4-vision-preview".

ImageGenerationSettings

  • url: URL for image generation services (e.g., https://api.openai.com/v1/images/generations).
  • style: The style of the generated images, e.g., "vivid".
  • size: Specifies the size of generated images, such as "1024x1024".
  • quality: The quality of generated images, like "hd".
  • model: The model used for image generation, for instance, "dall-e-3".