gliner_datause

Fine-tune of urchade/gliner_large-v2.1 for data-use mention extraction (dataset / survey / census / registry mentions in economics research papers).

Labels

• NAMED_DATA — a proper name, title, or acronym of a specific data source
• DESCRIPTIVE_DATA — a source described in words but not named
• VAGUE_DATA — generic data wording with no identifiable source

Training

• base model: urchade/gliner_large-v2.1
• dataset: rafmacalaba/data-use-mentions (gliner config)
• epochs: 5
• learning rate: 5e-06
• batch size: 16
• precision: bf16

Evaluation (holdout)

thr	tp	fp	fn	precision	recall	f0.5	f1
0.10	7147	5212	199	0.5783	0.9729	0.6293	0.7254
0.20	7082	3928	264	0.6432	0.9641	0.6891	0.7716
0.30	7017	3234	329	0.6845	0.9552	0.7256	0.7975
0.40	6901	2549	445	0.7303	0.9394	0.7643	0.8217
0.50	6704	1874	642	0.7815	0.9126	0.8046	0.8420
0.60	6212	1196	1134	0.8386	0.8456	0.8400	0.8421
0.70	4923	645	2423	0.8842	0.6702	0.8311	0.7624

Best F0.5: 0.8400 (thr=0.6) Best F1: 0.8421 (thr=0.6)