-
-
Save mberman84/c5d77723d2848e890f41a7a226f36eec to your computer and use it in GitHub Desktop.
| # Must have conda installed | |
| # It costs approximately $0.2 (in GPT-4 API fees) to generate one example with analysis and design, and around $2.0 for a full project. | |
| conda create -n metagpt python=3.11.4 | |
| conda activate metagpt | |
| npm --version # to check you have npm installed | |
| # optional: install node if you don't have it | |
| npm install -g @mermaid-js/mermaid-cli | |
| git clone https://github.com/geekan/metagpt | |
| cd metagpt | |
| python -m pip install -r requirements.txt | |
| python setup.py install | |
| # create openai API key | |
| # insert API key into config/config.yaml | |
| # make sure to set the model correctly depending on which model you have access to | |
| python startup.py "Write a cli snake game based on pygame" # optional --code_review True (better code, costs more) | |
| # if you get ModuleNotFoundError: No module named 'bs4’, run "python -m pip install bs4" | |
| python startup.py "Write a cli snake game based on pygame" # do this again if you got the ModuleNotFoundError |
how to use this with GPT-3.5 ?
OPENAI_API_MODEL: "gpt-3.5-turbo"
with gpt-3.5-turbo, I see MAX_TOKENS is set to 1500 in config.yaml, but I am getting this error
This model's maximum context length is 4097 tokens. However, you requested 4238 tokens (2738 in the messages, 1500 in the completion). Please reduce the length of the messages or completion.
It failed to generate any code for me and would break each time I got to the code generation step. The requirements worked fine. After I commented out all the mermaid_to_file commands in design_api.py, then it finally generated code.
For line #7 the command to install nodejs within conda from the commandline is:
conda install -c conda-forge nodejs
Will this generate ReactJS webapps?
I am only seeing python results even after requesting ReactJS in the prompt.
Will this generate ReactJS webapps? I am only seeing python results even after requesting ReactJS in the prompt.
The prompt for the programmer persona is set to only output python. You will have to tweak it on your end to use anything different.
Will this generate ReactJS webapps? I am only seeing python results even after requesting ReactJS in the prompt.
The prompt for the programmer persona is set to only output python. You will have to tweak it on your end to use anything different.
Any idea how to do this?
TypeError: '>' not supported between instances of 'float' and 'str'
with gpt-3.5-turbo, I see MAX_TOKENS is set to 1500 in config.yaml, but I am getting this error
This model's maximum context length is 4097 tokens. However, you requested 4238 tokens (2738 in the messages, 1500 in the completion). Please reduce the length of the messages or completion.
The same situation was with me when I used gpt-3.5-turbo. When I changed it with gpt 4.0, it worked and created a program exactly like the one in Berman's video.
how to use this with GPT-3.5 ?