Details: https://spacy.io/models/pl#pl_core_news_md
File checksum:
8ff32b1591d984d03901d8c134281d062eb24e229d8d3a57f46a68bfb69452a6
Polish multi-task CNN trained on the National Corpus of Polish and UD Polish PDB. Assigns word vectors, POS tags, lemmas, dependency parses and named entities. Word vectors trained using FastText CBOW on Wikipedia and OSCAR (Common Crawl).
| Feature | Description |
|---|---|
| Name | pl_core_news_md
|
| Version | 2.3.0
|
| spaCy | >=2.3.0,<2.4.0
|
| Model size | 76 MB |
| Pipeline | tagger, parser, ner
|
| Vectors | 500000 keys, 20000 unique vectors (300 dimensions) |
| Sources | National Corpus of Polish (Mirosław Bańko, Rafał L. Górski, Barbara Lewandowska-Tomaszczyk, Marek Łaziński, Piotr Pęzik, Adam Przepiórkowski) UD Polish SZ v2.3 (Wróblewska, Alina; Zeman, Daniel; Mašek, Jan; Rosa, Rudolf) Morfeusz 2 Lemmas from the Grammatical Dictionary of Polish (SGJP) (Marcin Woliński, Zbigniew Bronk, Włodzimierz Gruszczyński, Witold Kieraś, Zygmunt Saloni, Danuta Skowrońska, Robert Wołosz) OSCAR (Common Crawl) Wikipedia (20200301) |
| License | GPL
|
| Author | Explosion and Ryszard Tuora |
Label Scheme
| Component | Labels |
|---|---|
tagger
| ADJ, ADJA, ADJC, ADJP, ADV, AGLT, BEDZIE, BREV, BURK, COMP, CONJ, DEPR, FIN, GER, IMPS, IMPT, INF, INTERJ, INTERP, NUM, NUMCOL, PACT, PANT, PCON, PPAS, PPRON12, PPRON3, PRAET, PRED, PREP, QUB, SIEBIE, SUBST, WINIEN, XXX, _SP
|
parser
| ROOT, acl, advcl, advmod, amod, appos, aux, aux:pass, case, cc, ccomp, conj, cop, dep, det, det:numgov, expl:pv, iobj, mark, nmod, nsubj, nsubj:pass, nummod, obj, obl, obl:arg, punct, xcomp
|
ner
| date, geogName, orgName, persName, placeName, time
|
Accuracy
| Type | Score |
|---|---|
LAS
| 84.47 |
UAS
| 90.41 |
TOKEN_ACC
| 99.83 |
TAGS_ACC
| 98.28 |
ENTS_F
| 84.68 |
ENTS_P
| 84.65 |
ENTS_R
| 84.70 |
Installation
pip install spacy
python -m spacy download pl_core_news_md