gliner2_datause

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: 5
• encoder LR: 1e-05
• task LR: 0.0005
• batch size: 8
• precision: bf16

Evaluation (holdout, label-agnostic)

thr	tp	fp	fn	precision	recall	f0.5	f1
0.10	7037	3193	310	0.6879	0.9578	0.7290	0.8007
0.20	6915	2509	432	0.7338	0.9412	0.7676	0.8246
0.30	6789	2047	558	0.7683	0.9241	0.7951	0.8390
0.40	6576	1700	771	0.7946	0.8951	0.8128	0.8418
0.50	6300	1339	1047	0.8247	0.8575	0.8311	0.8408
0.60	5855	1001	1492	0.8540	0.7969	0.8419	0.8245
0.70	5072	630	2275	0.8895	0.6903	0.8410	0.7774

Best F0.5: 0.8419 (thr=0.6) Best F1: 0.8418 (thr=0.4)