Dataset · fastino
Multi-task classification dataset across 17 domains
fastino/fast-decisions
1,700 labeled English documents from the GLiNER2.5-Decide suite for evaluating classifiers.
What's inside
The validation (dev) split of the Fast Decisions dataset, used for GLiNER2.5-Decide models, contains 1,700 English rows with 100 examples per each of the 17 domains. Each row includes the document text, task name, candidate list, and ground truth answers for multiple classification heads (both single-label and multi-label).
What you can do with it
- Evaluating and benchmarking zero-shot and multi-label text classification models
- Debugging decision-making pipelines in production NLP systems
- Tasks
- Text classification
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 1.2K
- Likes
- 36
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("fastino/fast-decisions")