openfoodfacts/ingredient-detection

This ingredient detection model was trained on the ingredient detection dataset v1.1 using this training code.

Training was tracked on Wandb.

This release provides the following assets:

Training-related assets:

• predictions directory: predictions on train and test dataset of the model, in: HTML format: easier to view JSONL format: either the raw or the aggregated (post-processed) version
• the HuggingFace serialized model, in the root directory

Serving assets:

• onnx.tar.gz: the model exported to ONNX format

Versions

1.1

New version based on v1.1 of the dataset. “organic”/”issu de l’agriculture biologique” suffixes are now considered as part of the ingredient list.s

1.0

First version based on v1.0 of the dataset.