Agents

Last updated Nov 14, 2024 Edit Source

See:
AI/Generative AI/Foundation models
AI/Deep learning/Multimodal learning

# Resources

ChatGPT’s next level is Agent AI: Auto-GPT, BabyAGI, AgentGPT, Microsoft Jarvis & friends
Auto-GPT - An experimental open-source attempt to make GPT-4 fully autonomous
- https://significant-gravitas.github.io/Auto-GPT/
- https://autogpt.net/
AgentGPT
BabyAGI
JARVIS / HuggingGPT
Agents in AutoGen | AutoGen (microsoft.github.io)
- Agent - an entity that can act on behalf of human intent
- we can incorporate both very simple agents that can solve simple tasks with a single backend, but also we can have agents that are composed of multiple simpler agents
- When we use the word ‘multi-agent’ and ‘single-agent’, I think there are at least two different dimensions:
  - Interface
    - From the user’s point of view, do they interact with the system in a single interaction point or do they see explicitly multiple agents working and need to interact with multiple of them?
    - A single interaction point can make many applications’ user experience more straightforward
  - Architecture
    - Are there multiple agents underneath running at the backend?
    - The multi-agent design of the architecture is easier to maintain, understand and extend than a single agent system
    - It’s very important to recognize that the multi-agent architecture is a good way to build a single agent
- A particular system can have a single-agent interface and a multi-agent architecture
- Ways of combining models and agents:
  - Different conversation patterns or chats: sequential, group, constrained group, customized group, nested
  - Different prompting and reasoning techniques such as reflection, ReAct
  - Tool use and code execution
  - Planning and task decomposition
  - Retrieve augmentation
  - Integrating multiple models, APIs, modalities and memories
Stateflow - enhancing LLM task-solving through state-driven workflows
EcoAssistant - Using LLM Assistants More Accurately and Affordably | AutoGen (microsoft.github.io)
- Built on top of Autogen
- #PAPER EcoAssistant: Using LLM Assistant More Affordably and Accurately (arxiv.org)
- JieyuZ2/EcoAssistant: EcoAssistant: using LLM assistant more affordably and accurately (github.com)
Agent for text-to-SQL with automatic error correction - Hugging Face Open-Source AI Cookbook
- Agent for text-to-SQL with automatic error correction
- Uses ReactCodeAgent in transformers.agents
Natural Language Queries Across Databases with LLMs, AutoGen, and Langchain (PostgreSQL) | by Manojpn | Medium
- Autogen for SQL queries
Step by Step guide to develop AI Multi-Agent system using Microsoft Semantic Kernel and GPT-4o | by Akshay Kokane | Jun, 2024 | Medium
- Multi-agent system with semantic kernel and openai assistant. Using .NET
https://numbersstation.ai/text-to-sql-that-isnt/
- https://www.youtube.com/watch?v=xmzda44hUgk
Agent for text-to-SQL with automatic error correction - Hugging Face Open-Source AI Cookbook
Natural Language Queries Across Databases with LLMs, AutoGen, and Langchain (PostgreSQL) | by Manojpn | Medium

# Courses

# Code

# References

#PAPER HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in HuggingFace (Shen 2023)
#PAPER MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering
- #CODE https://github.com/openai/mle-bench
- MLE-bench is a benchmark for measuring how well AI agents perform at machine learning engineering
#PAPER MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation
- #CODE GitHub - snap-stanford/MLAgentBench
#WHITEPAPER AIDE: Human-Level Performance in Data Science Competitions
- #CODE https://github.com/WecoAI/aideml
#PAPER OpenHands: An Open Platform for AI Software Developers as Generalist Agents
- #CODE https://github.com/opendevin/opendevin

CarlosGG's Knowledge Garden 🪴

Agents

# Resources

# Courses

# Code

# References

Backlinks

Interactive Graph