comfyui_LLM_party

1625
Updated 15 days ago
View on GitHub →See Common Issues →

Available Nodes

Repository Overview

load_openai_ebd

ComfyUI Node Documentation: load_openai_ebd

Overview

The load_openai_ebd node is part of the ComfyUI LLM Party suite designed to facilitate the integration of OpenAI word vector search capabilities within ComfyUI workflows. This node leverages OpenAI's embedding models to perform text similarity searches, helping users query and retrieve relevant pieces of information from their datasets or documents based on the inputted question or query.

Functionality

What This Node Does

The load_openai_ebd node specializes in using OpenAI's embedding models to encode text and perform similarity searches. It is particularly useful for querying large knowledge bases or text documents to find sections relevant to a particular question. The node can either work with locally stored data or directly with input text strings, splitting them into manageable chunks to facilitate efficient searching.

Inputs

The node requires several inputs to function efficiently:

Required Inputs

  1. model_name (STRING): Specifies the name of the embedding model to be used. Default is "text-embedding-3-small".
  2. question (STRING): The input query or question for which relevant information needs to be fetched.
  3. is_enable (BOOLEAN): A toggle to enable or disable the node operation. Default is True, meaning the node is active.
  4. k (INT): Determines the number of top similar chunks to retrieve from the knowledge base. Default is 5.
  5. chunk_size (INT): The size of each text chunk, in characters, for splitting the document. Default is 200.
  6. chunk_overlap (INT): The number of overlapping characters between consecutive chunks. Default is 50.

Optional Inputs

  1. file_content (STRING): Text content that the node will process into chunks and search through. This input is forcefully required if base path is not provided.
  2. base_path (STRING): Path to an existing database on disk for FAISS (an embedded search platform). Default is an empty string.
  3. base_url (STRING): Base URL for the embedding service if using an API. Default is an empty string.
  4. api_key (STRING): API key for accessing OpenAI services. Default is an empty string.
  5. is_ollama (BOOLEAN): Indicates if the Ollama format should be used for hosting models locally. Default is False.

Outputs

The node generates the following output:

  • ebd_response (STRING): The result from the similarity search, which includes the most relevant text portions related to the input question derived from the provided knowledge base or text content.

Usage in ComfyUI Workflows

The load_openai_ebd node can be integrated into ComfyUI workflows to enhance the capabilities of AI-driven applications. Its primary usage scenarios include:

  • Knowledge Retrieval: Implemented in workflows where retrieving relevant information from large datasets is crucial, such as customer service bots or research assistants.
  • Text Enhancement: Used in content generation workflows to ensure the inclusion of contextually relevant information.
  • Interactive Applications: Integrated into interactive applications where real-time query responses are necessary.

Special Features and Considerations

  • Flexibility: The node provides flexibility by allowing both local and API-based model hosting, and different embedding models can be specified as needed.
  • Configurable Chunking: It allows the configuration of chunk size and overlap, ensuring efficient processing of various text lengths.
  • APIs and Local Models: Supports both OpenAI and Ollama (for local models), providing versatile deployment options.
  • Considerations for API Keys: Users need to ensure that appropriate API keys are configured either through the node interface or via environment configurations for seamless operation.

In conclusion, the load_openai_ebd node for ComfyUI offers powerful text processing and similarity search features that can be seamlessly integrated into diverse workflows, making it a valuable tool for applications that require intelligent information retrieval and analysis capabilities.