AiGen
  • Introduction to AiGen
  • AiGen
    • Jarvis
      • Controllers
        • Sample function
        • Backtracking Function
        • Forking Function
      • Execution Process
      • Token Generation and Controller Intervention
    • Marvin
      • Execution Process
      • Examples
        • Contextual Sentiment Analysis
        • Adaptive Learning Content Creation
        • Multilingual Customer Support Automation
  • Economic Framework
  • Token specifics
  • Additional links
    • Support
    • Whitepaper
  • AI Dictionary
Powered by GitBook
On this page

Introduction to AiGen

AiGen is a coordination layer for different components of the AI tech stack, with a foundational controller and scorer layer. Jarvis, the controller layer, incorporates custom logic during token-by-token decoding. These controllers offer flexibility and versatility in guiding LLMs' behavior, allowing for dynamic editing of prompts and generated text.

The platform offers plug-and-play capabilities, enabling LLMs and compute providers such as Bittensor's subnets, Hugging Face Transformers, Aakash Network, and others to effortlessly integrate. Lastly, Marvin, the scorer layer, accepts an instruction and response pair, and computes a relevance score ranging from 1 to 100, reflecting the predicted accuracy of the response relative to the given instruction.

AiGen provides a cohesive platform for collaboration among various AI modules. With its versatile controller and scorer layers, AiGen empowers developers and users to enhance the quality and efficiency of AI models while facilitating interoperability with leading LLMs and compute providers.

NextAiGen

Last updated 1 year ago