How to use code interpreter to save gpt generated documents such as csv files to local disk

[scortLi]

I have a requirement to generate an excel table according to some known data and save the table to the local disk, I can not save to the local disk, it is always saved in the directory（ /mnt/data/XXX.csv）, I guess it should be the directory of the virtual environment. I have an idea to execute the code generated by the code interpreter locally, but I don't know how to execute it or save the file. Is there any other simpler solution?

[qingyun-wu]

A simple solution I can see is to not use docker (set use_docker to False in the code_execution_config argument ). In that case, the program saves generated files to a dir on your local machine. You can either use the default work_dir or specify a specific local directory (via the work_dir field of the code_execution_config argument) to save the generated file to the local disk.

Check code_execution_config in this SDK doc: https://microsoft.github.io/autogen/docs/reference/agentchat/conversable_agent

"work_dir (Optional, str): The working directory for the code execution. If None, a default working directory will be used. The default working directory is the "extensions" directory under "path_to_autogen"."

[scortLi]

A simple solution I can see is to not use docker (set use_docker to False in the code_execution_config argument ). In that case, the program saves generated files to a dir on your local machine. You can either use the default work_dir or specify a specific local directory (via the work_dir field of the code_execution_config argument) to save the generated file to the local disk.

Check code_execution_config in this SDK doc: https://microsoft.github.io/autogen/docs/reference/agentchat/conversable_agent

"work_dir (Optional, str): The working directory for the code execution. If None, a default working directory will be used. The default working directory is the "extensions" directory under "path_to_autogen"."
If I call code interpreter in openai's assistant, how do I guarantee to run python code locally and save files such as scv locally

[gagb]

@lihui123456 can you share a snippet of your code? Which agent class are you using?

[scortLi]

@gagb here is my code:
llm_config = {
"config_list": config_list,
"assistant_id": assistant_id,
"tools": [
{
"type": "code_interpreter"
}
],
}
gpt_assistant = GPTAssistantAgent(
name="assistant",
instructions="你是一个代码解释器助手，如果需要使用python代码请用代码来解决问题",
llm_config=llm_config)

user_proxy = UserProxyAgent(
name="user_proxy",
is_termination_msg=lambda msg: "TERMINATE" in msg["content"],
code_execution_config={
"work_dir": "coding",
"use_docker": False, # set to True or image name like "python:3" to use docker
},
human_input_mode="NEVER"
)
user_proxy.initiate_chat(
gpt_assistant,
message="随机生成一个csv文件,并保存在本地",
)

If I execute the above code, the path of the generated CSV file is /mnt/data/XXX.csv, as if in a virtual environment, how can I use the file generated by gpt's assistant to locally or download the file generated in the virtual environment? thank you.

[gagb]

That's what I thought. This is about GPTAssistantAgent -- this was on the roadmap (#602) but never addressed. I just created an issue for it: #916

[gagb]

Closing in lieu of #916