Inspiration
Cooking is one of the most human things to do. Experimenting with flavor isn't something a machine can do. But what if it did?
What it does
Sally's Big Byte is a website that hosts recipes generated by a machine learning algorithm.
How we built it
The machine learning model was built using tensorflow in python, and trained with Nvidia's CUDA library. The website was written in python using flask.
Challenges we ran into
Training with CUDA and getting CUDA versions working was a huge challenge. Our model also ran into some issues with training quality.
Accomplishments that we're proud of
Learning tensorflow in one day and getting a model that actually worked was surprising for sure, but we're proud of how it came out.
What we learned
Text generation is not necessarily the best way to handle complex steps like recipes, this might have been a problem better solved by categorizing recipe steps and tokenizing the steps and ingredients rather than the raw text.
What's next for Sally's Big Byte
Who knows for sure? The next step would be fine tune training the model, or starting with a new model that accounts for consistency.
Log in or sign up for Devpost to join the conversation.