Food Facts Extractor Fine-tuned from Mistral 7B Instruct

Introduction

This repository contains a fine-tuned version of the Mistral 7B Instruct model, specifically tailored for extracting food facts from text. The model has been fine-tuned on a private dataset to enhance its performance in extracting food-related information accurately.

Output Format

"*ingredient, [ingredient: subingredient, subingredient], ingredient !Containment phrase: ingredient, ingredient !*Absence phrase: ingredient #Message regarding text or from the text"

An asterisk (*) symbol preceding an ingredient signifies potential ambiguity or inconsistencies in the surrounding text. This marker highlights sections where the extracted information might contain multiple errors or illogical constructions. It serves as a flag, prompting a closer examination of these areas to verify the accuracy of the data.
Square brackets ([]) are used to enclose ingredients that possess subcomponents or supplementary details. This structured approach allows for the clear identification of each constituent within the ingredient, regardless of whether they are separated by colons or dashes. This consistent formatting facilitates the effortless interpretation and analysis of the data.
Following the extracted text, a prominent !Containment phrase: ingredient, ingredient tag is appended. This critical marker serves as a clear and immediate indicator, alerting users to the potential presence of hazardous substances within the listed ingredients. By drawing heightened awareness to such crucial information, this tag promotes the adoption of appropriate safety measures when handling or consuming the associated product. For example, !Содержит: орехи, молоко, злаки. An asterisk (*) symbol means the absence of an ingredient, for example: !*Без: ГМО.

License

This project is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0).

Supported Languages

English (en)
Russian (ru)