[Bug] AttributeError: 'UserProxyAgent' object has no attribute 'register_for_execution'

[luohao123]

The example code have errors

AttributeError: 'UserProxyAgent' object has no attribute 'register_for_execution'

https://github.com/microsoft/autogen/blob/main/notebook/agentchat_function_call_currency_calculator.ipynb

Please edit it avoid misleading new comer users.

[rickyloynd-microsoft]

The problem could be that you are using the public pip version pyautogen=0.2.2, which doesn't support the function decorators. You can clone the repo then install it with pip install -e . from the autogen dir.

[luohao123]

@rickyloynd-microsoft When will the latest master code on pypi?

[luohao123]

Also, when using default docker run python code, always have this error:

import requests ModuleNotFoundError: No module named 'requests'

I can not say it's LLM produces error, but requests is just system lib, why it runs error like this

[rickyloynd-microsoft]

@rickyloynd-microsoft When will the latest master code on pypi?

We don't have a date, but almost certainly within 2 weeks.

[rickyloynd-microsoft]

Also, when using default docker run python code, always have this error:

import requests ModuleNotFoundError: No module named 'requests'

I can not say it's LLM produces error, but requests is just system lib, why it runs error like this

Are you able to run the sample code here without docker?

[luohao123]

I don't know, but as a common sense, requests should can be used out of box in python3

[rickyloynd-microsoft]

To identify your problem, please start with the simple example here before attempting docker.

[nise]

If I run the simple example suggested by rickyloynd-microsoft I get this:

ImportError                               Traceback (most recent call last)
Cell In[8], [line 1](vscode-notebook-cell:?execution_count=8&line=1)
----> [1](vscode-notebook-cell:?execution_count=8&line=1) from autogen import ConversableAgent, UserProxyAgent, config_list_from_json
      [4](vscode-notebook-cell:?execution_count=8&line=4) def main():
      [5](vscode-notebook-cell:?execution_count=8&line=5)     # Load LLM inference endpoints from an env variable or a file
      [6](vscode-notebook-cell:?execution_count=8&line=6)     # See https://microsoft.github.io/autogen/docs/FAQ#set-your-api-endpoints
      [7](vscode-notebook-cell:?execution_count=8&line=7)     # and OAI_CONFIG_LIST_sample.
      [8](vscode-notebook-cell:?execution_count=8&line=8)     # For example, if you have created a OAI_CONFIG_LIST file in the current working directory, that file will be used.
      [9](vscode-notebook-cell:?execution_count=8&line=9)     config_list = config_list #config_list_from_json(env_or_file="OAI_CONFIG_LIST")

ImportError: cannot import name 'ConversableAgent' from 'autogen' (unknown location)

[nise]

So the code is still not available at pypi?

[thinkall]

@nise could you try pip install -U pyautogen==0.2.29?

[LucasTrg]

@luohao123

I don't know, but as a common sense, requests should can be used out of box in python3

The requests module has never been a part of the python3 standard library : https://pypi.org/project/requests/ You're probably mixing it up with urllib.request