Tudorx95/NER_Economic_Political

GLiNER Fine-tuned for Political & Economic NER

Fine-tuned version of urchade/gliner_small-v2.1 on a custom politico-economic NER dataset. Trained to recognize 11 entity types.

Entity types

POLITICIAN, POLITICAL_PARTY, POLITICAL_ORG, FINANCIAL_ORG, ECONOMIC_INDICATOR, POLICY, LEGISLATION, MARKET_EVENT, CURRENCY, TRADE_AGREEMENT, GPE

Performance

Test set: 2122 examples. Evaluation mode: ent_type (label match, ignoring exact boundaries).

Global (micro-averaged):

• Precision: 0.6811
• Recall: 0.9094
• F1: 0.7789

Per label:

Label	Precision	Recall	F1
POLITICIAN	0.603	0.932	0.732
POLITICAL_PARTY	0.750	0.964	0.843
POLITICAL_ORG	0.324	0.497	0.392
FINANCIAL_ORG	0.257	0.471	0.332
ECONOMIC_INDICATOR	0.294	1.000	0.455
POLICY	0.111	0.250	0.154
LEGISLATION	0.238	1.000	0.385
MARKET_EVENT	0.188	0.710	0.297
CURRENCY	0.094	0.400	0.153
TRADE_AGREEMENT	0.122	0.357	0.182
GPE	0.842	0.971	0.901

Usage

from gliner import GLiNER

model = GLiNER.from_pretrained("Tudorx95/NER_Economic_Political")
labels = ["POLITICIAN", "POLITICAL_PARTY", "POLITICAL_ORG", "FINANCIAL_ORG",
          "ECONOMIC_INDICATOR", "POLICY", "LEGISLATION", "MARKET_EVENT",
          "CURRENCY", "TRADE_AGREEMENT", "GPE"]

text = "The Federal Reserve raised rates after President Biden signed the new bill."
entities = model.predict_entities(text, labels, threshold=0.5)
for e in entities:
    print(e["text"], "->", e["label"])

Training details

• Base model: urchade/gliner_small-v2.1
• Training examples: 5747
• Validation examples: 1228
• Epochs: 10
• Batch size: 8
• Learning rate: 3e-06