πŸ”₯
Langtorch
  • πŸ‘‹ Introduction
  • πŸ‘₯ Our Approach
  • πŸ€– LLM Service
    • OpenAi
    • CohereAI
    • Minimax
    • Huggingface
  • πŸ”ƒ Processor
    • Input and Output
      • Pre-built types
        • SingleText
        • MultiChatMessage
    • Integration
    • Customize your own processor?
  • πŸ”€ Loader
  • πŸ”§ Pre-processing & Post-processing
    • πŸ”„ Parser
  • πŸ“ Prompt Template
    • Prompt Template Class
    • Annotation Based Prompt Template
  • πŸ’ͺ Capability
    • Capability Unit
    • Capability Node
    • Capability Graph or Capability DAG
  • πŸ•΅οΈβ€β™‚οΈ Agent
  • 🧠 Memory
  • 🧩 Semantic Cache
  • πŸ’Ύ Vector Store
  • πŸ”Œ Dependency Injection
  • πŸš€ Langtorch Hub
  • πŸ“Š Dashboard
  • πŸ—ΊοΈ Roadmap
  • [δΈ­ζ–‡]Langtorch one pager
  • [Eng]Langtorch one pager
Powered by GitBook
On this page
  1. πŸ’ͺ Capability

Capability Graph or Capability DAG

Capability Graph = linking multiple Capability Units

A Capability Directed Acyclic Graph (Capability DAG) is a structure that connects these nodes in a specific order, ensuring there are no cycles (i.e., loops) in the graph. The nodes are arranged in a sequence where the output of one node becomes the input for the next node. By processing the nodes in the correct order, the Capability DAG can accomplish complex tasks by combining the individual functions of its nodes.

For example, suppose you have three nodes in a Capability DAG:

  1. Node A: Adds two numbers and returns the sum.

  2. Node B: Multiplies two numbers and returns the product.

  3. Node C: Subtracts two numbers and returns the difference.

By connecting these nodes in a specific order, the Capability DAG can perform a series of calculations. If Node A's output is connected to Node B's input and Node B's output is connected to Node C's input, the graph will add two numbers, multiply the result by another number, and then subtract a final number.

PreviousCapability NodeNextπŸ•΅οΈβ€β™‚οΈ Agent

Last updated 2 years ago