🍳 Structured FoodExtract with a Fine-Tuned Gemma 3 270M

Extract food and drink items from text with a fine-tuned SLM (Small Language Model) or more specifically a fine-tuned Gemma 3 270M.

Our model has been fine-tuned on the FoodExtract-1k dataset.

• Input (str): Raw text strings or image captions (e.g. "A photo of a dog sitting on a beach" or "A breakfast plate with bacon, eggs and toast")
• Output (str): Generated text with food/not_food classification as well as noun extracted food and drink items and various food tags.

For example:

• Input: "For breakfast I had eggs, bacon and toast and a glass of orange juice"
• Output:

food_or_drink: 1
tags: fi, di
foods: eggs, bacon, toast
drinks: orange juice