Model for predicting how much a specific, nonstandard unit (meaning units that aren't household units, such as "cup" or "kilo") of an ingredient weighs. Has been trained on USDA FoodData Central data. Do numpy.exp1m on the resulting logit to get the predicted weight in grams.
Prompt format:
FOOD: {food}
UNIT: {unit}
COMMENTS: {comments}
where food is a food description in the USDA FoodData Central format, unit is the parsed nonstandard unit and comments is either an array of comments in the form of [comment1, comment2] or none.
Citation: U.S. Department of Agriculture, Agricultural Research Service, Beltsville Human Nutrition Research Center. FoodData Central. [Internet]. [cited 2026]. Available from https://fdc.nal.usda.gov/.
- Downloads last month
- 26
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for Forkulous/tf-forkulous-units
Base model
microsoft/deberta-v3-base