Clara is an AI chat bot that responds in rhymes and can generate poems! When a poem is generated, the bot will dynamically change the theme of this page to reflect the theme of the poem.
To run this app locally, you will need an OpenAI API key with chat completion permissions.
Once you have an API key, set the environment variable OPENAI_API_KEY to the value of the API key. You can do this any way you like, but this repo does support .env files in both .env and server/.env locations, so you could set it up like this if that's easier:
echo 'OPENAI_API_KEY=<paste-token-here>' > .envThis is a mono repo with a backend and frontend service.
You will need npm to install dependencies. In this directory, run npm install.
You will need cargo for the backend dependencies. That's it :)
Now that you have dependencies installed, run just this one command to boot up both the frontend and the backend: npm run dev
That's it! Now visit http://localhost:5173 in a browser! Have fun!
To test the frontend, run npm test in this directory. To test the backend, navigate to the server/ directory and run cargo test.
- Basic AI chat features, but Clara always speaks in rhymes!
- Clara, when prompted, will write a poem and change the theme of the chat bot webpage to match the poem's theme
- Themes can be saved across sessions! Just click the "Save Theme" button when you are not on the default theme, then refresh. The theme will still be there. This is saved using the browser's local storage feature.
The model runs slow when asking it for a poem, because of the size of the CSS file it needs to process to generate new styles. It's much faster if you just chat with it and don't ask for a poem. Still thinking of ways around the speed issue.
There is no image uploading feature.
Run docker compose up --build to see it run in the terminal, docker compose up --build -d to let it run in detached mode (so you can close your terminal while the app still runs). To shut it down once it's in detached mode, run docker compose down. If those commands fail, try docker-compose in place of docker compose, which is the way that some debian-based distros are set up. Once it's running, you can visit http://localhost:49152 in a browser to try it out.
If you don't like the default port, you can set the CLARA_PORT env var to your desired port on the host machine. This env var won't do anything if set within the container. So you can use it like this: env CLARA_PORT=5000 docker compose up if you want to run on port 5000.
In release mode only, the backend is configured to serve static files from dist, which is the directory that Vite builds the static files in by default. So all you need to do is run npm run prod. Vite will automatically build the static files, Cargo will compile the Rust backend in release mode, and the server will start running on port 49152. The server will serve the static frontend files. Visit http://localhost:49152 once it is running to try it out.