FoodExtract-1K: Text Samples for Fine-Tuning Food Extraction Models

A 1,000-sample subset of the FoodExtract-135k dataset designed for fine-tuning small language models. Created by mrdbourke and last updated on March 25, 2026, its purpose is to train models to extract structured food and drink data from text, replicating the capability of larger models. The fine-tuned models can then filter large text corpora for food-related content.

Use Cases

• Fine-tuning small language models for structured data extraction based on the described task.
• Filtering large text datasets for food and drink mentions based on the model's described purpose.
• Benchmarking the performance of small LLMs against larger models on a specific extraction task.
• Creating training data pipelines for domain-specific text classification focused on culinary items.

Strengths

• Designed for a specific, replicable task: enabling a small LLM to match a larger model's performance on food extraction.
• Serves as a smaller, more manageable 1k-sample version for initial experimentation before using the full 135k dataset.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the full dataset is referenced but the exact size of this sample is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality and format require manual inspection after download.

Provenance

Source

huggingface user mrdbourke

Collection Method

Likely created as a sample for model fine-tuning, but the exact gathering method is unspecified.

Freshness

Last updated 2026-03-25 05:54:46; freshness should be verified.