juusopi/grocery-fi-ner

Finnish Grocery NER + Text Classification Pipeline

spaCy pipeline for Finnish grocery text: named entity recognition (NER) and text classification (textcat) in a single model backed by TurkuNLP/bert-base-finnish-cased-v1.

Both tasks share one transformer encoder (transformer → ner → textcat).

Labels

NER: NOTE, PRODUCT, QUANTITY, UNIT

Textcat categories:

Performance (dev set)

NER

Metric	Score
F1	0.999
Precision	0.999
Recall	0.999

Per-label:

Label	Precision	Recall	F1
QUANTITY	1.000	1.000	1.000
UNIT	1.000	1.000	1.000
PRODUCT	0.998	0.998	0.998
NOTE	0.998	0.999	0.999

Text Classification

Metric	Score
Macro AUC	0.000

Per-category F1:

Category	F1

Usage

import spacy

nlp = spacy.load("juusopi/grocery-fi-ner")
doc = nlp("500 g omenaa")

# NER
for ent in doc.ents:
    print(ent.text, ent.label_)

# Text classification
best = max(doc.cats, key=doc.cats.get)
print("Category:", best)

Model details

• Base model: TurkuNLP/bert-base-finnish-cased-v1
• spaCy version: >=3.8.11,<3.9.0
• Pipeline version: 0.0.0