Text files

The toolkit provide an API for importing text documents to be used for enhancing LLM knowledge on particular topics. The SERMAS toolkit processes a list of document descriptors and creates the emdeddings of the content storing them in a vector database. The new information is then used by the LLM provider at inference time.

The documents can be added by:

• adding them to the Repository configuration: the documents are automatically imported during the application import phase
• calling the import API:
  Endpoint: POST api/dialogue/document
  Payload: a list of DialogueDocumentDto object in the form:

  {
    appId: string;
    documentId: string;
    content?: string;
    metadata?: DialogueDocumentMetadataDto;
  }