The application is composed of two parts - the api
and the ui
.
Switch to the api
directory. All commands below are to be run there.
uv venv
source .venv/bin/activate
uv pip install -r requirements.txt
This should get all the required packages set up. Now, let's set the environment.
Head to Google AI Studio to get a Gemini API Key.
And grab a Weights and Biases key from https://wandb.ai/authorize.
Now, set the following in the .env
file:
GEMINI_API_KEY="YOUR_KEY_HERE"
GOOGLE_GENAI_USE_VERTEXAI=false
You should now be able to run the API endpoint locally by:
uvicorn main:app --reload
This should bring up the API at http://localhost:8000
.
Now, switch to the ui
directory from the project root in a new terminal tab/window.
pnpm i
This should be enough to install all the requirements. Now we can run it with
pnpm dev
That should do the trick for launching the UI on your local browser at http://localhost:3000
.