【论文精读】HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in Hugging Face

HuggingGPT: SoLVing AI Tasks with ChatGPT and its Friends in Hugging FaCe

• 前言
• Abstract
• 1 Introduction
• 2 Related Works
• 3 HuggingGPT
• • 3.1 Task Planning
  • • Specification-based Instruction
    • Demonstration-based Parsing
    • 3.2 Model Selection
    • • In-context Task-model Assignment
      • 3.3 Task Execution
      • • Resource Dependency
        • 3.4 RespoNSE Generation
        • 4 Experiments
        • • 4.1 Settings
          • 4.2 Qualitative Results
          • 4.3 Quantitative Evaluation
          • 5 Limitations
          • 6 Conclusion
          • 阅读总结

            前言

            又是一篇利用LLM进行规划的工作，中稿于NeurIPS 2023，它充分结合了专家模型解决特定任务的能力和LLMs的理解推理能力，工作新颖但效果一般，可以说提供了一种新的利用LLM解决通用任务的范式，在思想上值得借鉴。