Ingrediate is an open-source application for generating custom recipes based on validated, user inputted ingredients.
Simply type in the contents of your fridge/pantry, choose a meal type you would like to make, select a recipe generation model, and generate! Ingrediate supports persistent data through user authentication with Firebase, and PostgreSQL.
- For use of Open Source LLM's within the application, download and install Ollama, and make sure you have both mistral:7b and llama2 downloaded locally.
- Ingrediate can also generate custom recipes with the paid OpenAI API service, you must have an account and your own personal API key to use. This will only apply if you want to use the text-davinci-003 model in the application.
- You must also have a PostgreSQL database instance with a connection string to be able to persist favorited recipes to the database.
- Finally, for user authentication you should set up a Firebase instance with a Web API key.
-
Fork and clone this repo.
-
cd
to the root directory runnpm install
-
Create a .env in server following the .env.example.
# Do not share your OpenAI API key with anyone! It should remain a secret.
OPENAI_API_KEY=
PORT=
PG_URI=
NODE_ENV=
FIREBASE_API_KEY=
-
Paste in your OpenAI API & Firebase API key for user auth, PG_URI connection string, and port to run your server.
-
Run
npm run dev
, ollama will launch, the development server and client will run concurrently, and Ingrediate will open in your web browser!
Developed By | Github | |
---|---|---|
Johanna Cameron |