comfyui_LLM_party

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

Available Nodes

Repository Overview

advance_ebd_tool

Advanced Embedding Tool Node Documentation

Overview

The Advanced Embedding Tool node, a part of the ComfyUI LLM Party project, is designed to facilitate advanced text embedding functionalities within the ComfyUI environment. This node allows users to perform similarity searches in large textual datasets, functioning as a sophisticated knowledge retrieval tool. Leveraging the power of embeddings, it enables seamless integration of language models and knowledge bases, supporting interactions that respond to user queries by extracting pertinent information from uploaded files or pre-existing databases.

Features

  • Advanced Text Embedding: Utilize Hugging Face embeddings to convert and manage text data efficiently.
  • Knowledge Base Integration: Build and maintain textual knowledge bases that can be queried effectively.
  • Configurable Parameters: Offers flexibility with multiple adjustable parameters including device selection, chunk size, and more.
  • Interoperability: Designed to work smoothly with other nodes and existing ComfyUI workflows.

Inputs

Required Inputs

  • Path: A string indicating the path to the embedding model.
  • Enable Status (is_enable): An option to enable or disable the node's functionality. Choose between "enable" and "disable".
  • K: An integer specifying the number of similar text segments to retrieve.
  • Device: Determine the computational device, options include "auto", "cuda", "mps", and "cpu".
  • Chunk Size: An integer that sets the size of text chunks for processing.
  • Chunk Overlap: An integer defining the overlap between text chunks.
  • File Name: A string used to name the knowledge base being created or accessed.

Optional Inputs

  • File Content: A string input to provide textual data directly.
  • Base Path: A string representing the path where a pre-existing knowledge base file is located.
  • EBD Model: Pre-loaded embedding model to be used, alternatively specify via the path input.

Outputs

  • Tool: The node generates an output tool in JSON format comprising function descriptors. This tool can be utilized to query textual data in a knowledge base, retrieving relevant content based on user requests.

Usage in Workflows

The Advanced Embedding Tool node can be incorporated into ComfyUI workflows aimed at constructing complex language model interfaces. It is particularly useful where there is a need to:

  1. Create and manage a local knowledge base from textual data which is segmented into chunks.
  2. Enable similarity-based search queries on the processed data to return relevant text segments.
  3. Incorporate the node into broader language processing or AI assistant workflows for knowledge extraction and management.

Special Features and Considerations

  • Device Auto-Detection: The node can auto-detect the best available device for processing, offering support for GPU, MPS, or CPU, ensuring optimized performance based on hardware capabilities.
  • Flexibility with Model Integration: It allows specifying custom or pre-built embedding models, either through direct paths or by using a previously loaded model.
  • Localized Language Support: The node supports configurations for both Chinese and English, as defined in the ComfyUI project settings.

In conclusion, the Advanced Embedding Tool node is a powerful utility for embedding-driven knowledge operations within the ComfyUI platform, enabling facile integration and functionality extension for language model applications.