gliner2_datause_smoke

Fine-tune of fastino/gliner2-large-v1 (GLiNER2) 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: fastino/gliner2-large-v1
• dataset: rafmacalaba/data-use-mentions (gliner2 config)
• epochs: 1
• encoder LR: 1e-05
• task LR: 0.0005
• batch size: 16
• precision: bf16

Evaluation (holdout, label-agnostic)

thr	tp	fp	fn	precision	recall	f0.5	f1
0.10	59	103	32	0.3642	0.6484	0.3992	0.4664
0.20	46	61	45	0.4299	0.5055	0.4432	0.4646
0.30	40	41	51	0.4938	0.4396	0.4819	0.4651
0.40	29	29	62	0.5000	0.3187	0.4489	0.3893
0.50	21	18	70	0.5385	0.2308	0.4251	0.3231
0.60	14	13	77	0.5185	0.1538	0.3518	0.2373
0.70	9	2	82	0.8182	0.0989	0.3333	0.1765

Best F0.5: 0.4819 (thr=0.3) Best F1: 0.4664 (thr=0.1)