🤖 AgentWiki

Welcome to the LLM Agents Wiki, a Wikipedia for understanding and leveraging Large Language Model Agents. Dive into the cutting-edge developments, explore various types and design patterns, and discover the libraries and tools that empower these intelligent systems to perform autonomously across diverse applications.

Large Language Model (LLM) Agents are AI systems that utilize large language models to perform tasks autonomously. By understanding natural language, reasoning through complex problems, and interacting with external tools and environments, LLM Agents represent a significant advancement in artificial intelligence. They are capable of planning, executing, and adapting their actions based on given objectives and feedback from their environment.

In an LLM-powered autonomous agent system, the LLM functions as the agent’s brain, complemented by several key components:

• Planning
  • Task Decomposition
  • Self-Reflection
• Memory
  • Types of Memory
  • Maximum Inner Product Search (MIPS)
• Tool Use

These components enable the agent to plan complex tasks, remember past interactions, and utilize external tools to extend their capabilities.

• Reasoning and Planning: LLM Agents analyze complex tasks, devise strategies, and plan sequences of actions to achieve specific goals.
• Tool Utilization: They interact with external tools, APIs, databases, and services to extend their capabilities beyond text generation.
• Memory and Context Management: By maintaining context over interactions, LLM Agents can reference previous information and maintain coherent long-term objectives.
• Natural Language Understanding: Advanced language comprehension allows LLM Agents to interpret and generate human-like text.
• Autonomy and Adaptability: LLM Agents operate independently, making decisions and adapting to new information or changes in their environment.

Planning involves breaking down complex tasks into manageable subgoals, devising strategies, and sequencing actions.

• Task Decomposition
  • Chain-of-Thought (CoT) Reasoning
  • Tree of Thoughts
  • LLM+P (LLM plus Planning)
• Self-Reflection
  • ReAct (Reasoning and Acting)
  • Reflexion
  • Chain of Hindsight (CoH)
  • Algorithm Distillation (AD)

Memory allows agents to retain and recall information over extended periods, enhancing their ability to maintain context and learn from past interactions.

• Types of Memory
  • Sensory Memory
  • Short-Term Memory (STM) or Working Memory
  • Long-Term Memory (LTM)
    • Explicit/Declarative Memory
    • Implicit/Procedural Memory
• Maximum Inner Product Search (MIPS)
  • Locality-Sensitive Hashing (LSH)
  • ANNOY (Approximate Nearest Neighbors Oh Yeah)
  • Hierarchical Navigable Small World (HNSW)
  • FAISS (Facebook AI Similarity Search)
  • ScaNN (Scalable Nearest Neighbors)

Tool use extends the agent's capabilities by allowing interaction with external tools and APIs.

• MRKL Systems
• Tool Augmented Language Models (TALM)
• Toolformer
• ChatGPT Plugins and OpenAI API Function Calling
• HuggingGPT
• API-Bank

• Chain-of-Thought (CoT) Reasoning
• ReAct (Reasoning and Acting)
• AutoGPT
• BabyAGI
• AgentGPT
• Plan-and-Execute Agents
• Conversational Agents
• Tool-Using Agents

• Prompt Chaining
• Reinforcement Learning from Human Feedback (RLHF)
• Agent Loop (Perception-Thought-Action Cycle)
• Context Window Management
• Tool Integration Patterns
• Memory Augmentation
• Modular Architecture

• LangChain
• LlamaIndex (GPT Index)
• Hugging Face Transformers
• OpenAI API
• Microsoft Guidance
• AutoGPT and BabyAGI Implementations
• Haystack

• Autonomous Task Execution
• Customer Support and Virtual Assistants
• Research Assistance
• Education and Tutoring
• Content Creation
• Software Development Assistance
• Data Retrieval and Processing

• ChemCrow
• Autonomous Scientific Research Agents

• Generative Agents (Park et al. 2023)

• AutoGPT
• GPT-Engineer

• Finite Context Length
• Long-Term Planning and Task Decomposition
• Reliability of Natural Language Interface

The field of LLM Agents is rapidly evolving, with significant advancements including:

• Enhanced Reasoning Abilities
• Tool Use Integration
• Memory and Retrieval Augmented Models
• Ethical and Safe AI Practices
• Open-Source Agent Frameworks

Embark on your journey with LLM Agents by exploring the following resources:

• Introduction to LLM Agents
• LangChain Documentation
• OpenAI API Reference
• AutoGPT GitHub Repository
• ReAct Framework
• Hugging Face Transformers

Stay updated with the latest trends, research, and developments in the field of LLM Agents by joining our community:

• Discussion Forums
• Contribute to Open-Source Projects
• Attend Workshops and Webinars

Explore. Learn. Innovate. Unlock the transformative potential of Large Language Model Agents and be at the forefront of the AI revolution.

nlp language-model agent steerability prompting