Ricky Loynd
Ricky Loynd
This would be a reasonable extension, but it's currently out-of-scope for our Microsoft team. Maybe a community member would be interested in pursuing it.
@balakreshnan Can you provide more details about which sample code you are running, and how you prepared the virtual environment?
@sonichi fyi
Yes, Voyager's approach is very interesting. My team (Padawan) looked into it at some depth. I created this overview: 
Voyager itself has many limitations. But its skill library, based on a vector DB, inspired my approach to our [TeachableAgent](https://github.com/microsoft/autogen/blob/main/notebook/agentchat_teachability.ipynb). And I plan to enhance `TeachableAgent` with the ability to...
I like your analysis of the problem! Robust learning of compositional skills has been a long-standing challenge for RL agents. Voyager shows how an LLM-based agent can learn a hierarchy...
> I don't know if this recursive teachable agent is really a separate thing from `TeachableAgent` itself Yes, that's a key question. We've just created new issues #538 and #540...
> For the short-term solution, isn't that just a matter of configuring a Retrieve agent with a custom text splitter that chunks Python files into function blocks? Then you could...
A new relevant paper: https://arxiv.org/abs/2311.07635