Simple conversational command line GPT that you can run locally with OpenAI API to avoid web usage constraints.

Add interactive code interpreter similar to the web version GPT-4 with Code Interpreter from OpenAI that goes back-and-forth between code generation and GPT reasoning. This is different from LangChain Python Agent or codeinterpreter-api that only generate code one shot. Only focus on python interpreter with numeric usages for now. There are some examples in the examples folder.

Export OPENAI_API_KEY and run python gpt_dialogue.py or python code_interpreter.py. There are special commands like "save", "pretext", "reset" in the conversation that you can use for convenience.