Word embeddings CLARIN.SI-embed.sr 1.0

Word embeddings CLARIN.SI-embed.sr 1.0

CLARIN.SI data & tools

Avtorji: Ljubešić, Nikola

Identifikator vnosa: http://hdl.handle.net/11356/1206

Datum objave: 2018-12-10

Vrsta: lexicalConceptualResource, text

Velikost: 1480566 entries

Jezik(i): Serbian

Opis: CLARIN.SI-embed.sr contains word embeddings induced from the srWaC web corpus. The embeddings are based on the skip-gram model of fastText trained on 554,606,544 tokens of running text for (1) 881,150 lowercased surface forms (e.g., "srbije") and (2) 599,416 lowercased lemmas with added part-of-speech information (e.g., "srbija#Np").

Izdajatelj: Jožef Stefan Institute

Ključne besede: word embeddings lemmatisation part-of-speech tagging

Zbirke: CLARIN.SI data & tools

Ta vnos je bil nadomeščen z novejšim.

http://hdl.handle.net/11356/1789

Prikaži polni zapis vnosa

Datoteke v tem vnosu

To je vnos

Publicly Available

z licenco:
Creative Commons - Attribution 4.0 International (CC BY 4.0)

Ime: embed.sr-lemma.ft.sg.bin
Velikost: 1.2 GB
Format: Neznano
Opis: Lemma embeddings, binary
MD5: 00cb8eb2c8a438178018bcff5b426f75

Prenesi datoteko

Ime: embed.sr-lemma.ft.sg.vec.gz
Velikost: 192.69 MB
Format: application/gzip
Opis: Lemma embeddings, text format
MD5: e8e69b35175e47bce8e8e7c7869a77cc

Prenesi datoteko

Ime: embed.sr-token.ft.sg.bin
Velikost: 1.42 GB
Format: Neznano
Opis: Token embeddings, binary
MD5: fbfb884ff7746c0187e185ecd3a4a7e1

Prenesi datoteko

Ime: embed.hr-token.ft.sg.vec.gz
Velikost: 561.41 MB
Format: application/gzip
Opis: Token embeddings, text format
MD5: 21b828b6287daa181f6eeb572a99948f

Prenesi datoteko