Recipe generator using cohere and annoy, built for the Cohere AI Hackathon #3.
Install dependencies:
pip install -r requirements.txt
You'll need to make a .env file. Place it in the root directory. It should contain your api key as follows:
API-KEY=<your-api-key>
To run the application:
python main.py
You'll be prompted with
What ingredients do you have?
Enter the ingredients in lower case separated by semicolon-space, eg: tomatoes; onions; etc.;. The output will be 3 lines:
- The name of the generated recipe
- The additional ingredients (formatted as a list)
- The directions (formatted as a list)
An example file for the data can be found in ./test_recipes.csv. It is a csv file with 1000 recipes.
The util.py module contains a function load_recipes() that accepts a filepath and returns a pandas.DataFrame containing the columns
ingredients: List[str]steps: List[str]name: str
- Come up with a catchy name
- Suggest ingredients boolean: Whether to add ingredients or not
You can start application by typing streamlit run streamlit_app.py in the console.