LLM Based Agents: Comparison Overview between LlamaIndex AgentWorkFlow and Microsoft AutoGen

Google Colab notebook for a talk I gave today

I have been experimenting with writing agents using small local LLMs, with mixed results. Experiments that perform poorly with small models running locally with Ollama usually work much better with larger models like OpenAI O1, etc. For my talk today I mostly used gpt-o1-mini, and ended with a simp,e example using Groq to run a distilled Deepseek R1 model.

Every thing is in this notebook: My talk Colab Notebook