๐ AG2 v0.5.0 is here! Two super cool features:
1. ReasoningAgent: Your o1 alternative
- Tree of Thoughts with beam search
- Systematic multi-path reasoning
- Smart solution evaluation
2. Structured Output & GraphRAG
- Pydantic schemas for perfect JSON responses @OpenAI
- Graph-based knowledge retrieval @falkordb
- See demo: Trip planner with both features!
See it in action: https://t.co/KZpY6YF5Bt and https://t.co/7OI02eYC73
๐ฏโโ๏ธ Join +20k agent builders on Discord: https://t.co/yiHAG0yUPf
โญ Star AG2: https://t.co/GaEetk82tn
#AG2 #AgentOS #AI #OpenSource
Can a LLM solve a multi-step problem with only the most basic tools?
Try do the problem by hand first.
Full notebook:
https://t.co/a9OhRK6ndh
#lionagi#AI#AGI#LLM#OpenAI#GenAI
Define a Research Workflow for a RAG-powered agent ๐๐ค
This is a cool notebook by @quantoceanli showing how you can build an agentic workflow to do scientific research from ArXiv, Wikipedia, textbooks, and more.
Have it first fetch relevant abstracts from ArXiv, propose an idea, and lookup information through sources. Afterwards trace through intermediate responses in between tool lookups.
Built with @llama_index + LionAGI - check it out!
https://t.co/gpedEXHMps
Can a LLM solve a multi-step problem with only the most basic tools?
Try do the problem by hand first.
Full notebook:
https://t.co/a9OhRK6ndh
#lionagi#AI#AGI#LLM#OpenAI#GenAI
@MertLovesAI@llama_index yes, that way, the whole process can iteratively updated and truly 'auto' , but code interpreter doesn't do pip install, so can't do integration test.
RAG Assisted Auto Developer ๐๐งโ๐ป
Hereโs a neat cookbook by @quantoceanli to build a devbot that can 1) understand a codebase, 2) write additional code based on the codebase.
Itโs a nice mix of different tools: @llama_index to index an existing codebase, Autogen / OpenAI Code Interpreter to write/test code, and https://t.co/Y2mLO3pRVB as the orchestration layer to define the flow.
Check it out: https://t.co/7E5RIbSktl
Here's how you can easily build an ArXiv research assistant ๐งโ๐ฌ๐ค with @llama_index + LionAGI by @quantoceanli ๐
1) Define a @llama_index RAG pipeline as an agent tool
2) Plug tool into LionAGI, letting you compose custom agentic pipelines
Check it out: https://t.co/GaSf1ie59N
Community Showcase ๐: LionAGI, a simple but powerful agent framework by @quantoceanli designed for interacting with data ๐ ๏ธ
โ Easily compose your own workflow - can be sequential or a loop. Can include chain-of-thought and use of external tools.
โ Efficient data operations for reading, chunking, binning, writing, storing and managing data.
โ Supports concurrent calls, function calling, JSON mode with @OpenAI out of the box.
Best of all, you can easily combine with a @llama_index RAG pipeline to build automated AI assistants over your data. Check out how to build an ArXiv research assistant below! ๐งโ๐ฌ๐ค
ArXiv research assistant with LionAGI + LlamaIndex: https://t.co/fBpN5UTqJ9
LionAGI repo: https://t.co/9xE10vF0f2
LionAGI Docs: https://t.co/BAuP5wXq3l
Full credits to @quantoceanli. This is a great addition to the LLM developer community ๐