chattronics

README

Chattronics

How to run

There are 2 main ways to run chattronics: real and mocked.

Given that to run using the real GPT model you require an API key (which will be used to charge you), it's best to use the mock model for testing purposes.

Mocked

Just run make run-mock and follow the instructions in the terminal.

Real GPT Model

First, setup you API key using a environment variable file. Create a file named .env in root and with the exact following format

OPENAI_API_KEY=your-api-key-here

Then, just run make run-real model=gpt-4 for GPT-4 or make run-real for GPT-3.5-Turbo.

To set the temperature just add a temperatur=0.2 flag to the run make command. The default value is 0.1.

Run results

Each execution will create a folder inside the runs folder. If this directory doesn't exist for you, it'll be created after executing.

Each run folder will have the following format

runs
└─ Mar08_14-57-30
    ├── diagram.svg
    ├── circuit files
    └── logs
        ├── console.logs
        ├── reviewer.json
        └── architect.json

Diagrams, circuits and other output files are stored in the specific run folder and all logs are stored in logs. The console.logs file is basically the stdout of the application, i.e., Debug, Info, Warn, Error and Fatal logs will be sent here. The json logs are the interactions with GPT in structured form. The name of these json files are referring to the persona of the model in question.