Straico Official icon

Straico Official

Get data from Straico API

Actions24

RAG → Prompt Completion

AI-generated

Overview

This node operation sends a prompt to a specified RAG (Retrieval-Augmented Generation) resource to generate a completion based on the prompt and various search parameters. It is useful for scenarios where you want to leverage a RAG model to generate responses or completions informed by retrieved data chunks. For example, you can use it to query a knowledge base or document repository with a natural language prompt and get a contextually relevant generated response.

Use Case Examples

  1. Sending a customer support query prompt to a RAG model to get a detailed answer based on company documentation.
  2. Using a RAG model to complete a text prompt with information retrieved from a large dataset or knowledge base.

Properties

Name Meaning
RAG ID The unique identifier of the RAG resource to which the prompt will be sent.
Prompt The text prompt to send to the RAG model for completion.
Model The model identifier to use for generating the completion.
Search Type The type of search algorithm to use for retrieving relevant data chunks (e.g., similarity, mmr, similarity_score_threshold).
K The number of top results to consider in the search.
Fetch_k The number of results to fetch initially before filtering.
Lambda_mult A multiplier parameter used in the search algorithm to balance relevance and diversity.
Score_threshold The threshold score to filter search results.

Output

JSON

  • id - The unique identifier of the prompt completion response.
  • message - The generated completion message from the RAG model.
  • model - The model used to generate the completion.
  • usage - Usage statistics or metadata related to the completion.
  • created_at - Timestamp when the completion was created.

Dependencies

  • An API key credential for Straico API authentication

Troubleshooting

  • Ensure the RAG ID provided is valid and exists in the Straico system to avoid 404 errors.
  • Verify that the model specified is supported and correctly spelled to prevent model-related errors.
  • Check that all required parameters (RAG ID, prompt, model) are provided to avoid validation errors.
  • If the response is empty or unexpected, verify the search parameters (search type, k, fetch_k, lambda_mult, score_threshold) for correctness and adjust as needed.
  • Network or authentication errors may occur if the Straico API key credential is missing or invalid; ensure proper credential configuration.

Links

Discussion