Published February 6, 2022 | Version v1
Dataset Open

UDPipe Models for Morphologically enhanced Universal Dependencies for Korean (morphUD)

Creators

  • 1. UBC

Description

UDPipe 1 (https://ufal.mff.cuni.cz/udpipe/1) models for Morphologically enhanced Universal Dependencies for Korean (morphUD) including ko.vectors.xz files for morphUD and wordUD. 

Embedding files are available at https://doi.org/10.5281/zenodo.5869738

udpipe --train --tokenizer="dimension=64;epochs=100;segment_size=50;initialization_range=0.2;batch_size=50;learning_rate=0.005;learning_rate_final=0.001;dropout=0.3;early_stopping=1" --tagger="models=2;templates_1=tagger;guesser_suffix_rules_1=8;guesser_enrich_dictionary_1=6;guesser_prefixes_max_1=0;use_lemma_1=0;use_xpostag_1=1;use_feats_1=1;provide_lemma_1=0;provide_xpostag_1=1;provide_feats_1=1;prune_features_1=0;templates_2=lemmatizer;guesser_suffix_rules_2=8;guesser_enrich_dictionary_2=6;guesser_prefixes_max_2=4;use_lemma_2=1;use_xpostag_2=0;use_feats_2=0;provide_lemma_2=1;provide_xpostag_2=0;provide_feats_2=0;prune_features_2=1" --parser="iterations=30;embedding_upostag=20;embedding_feats=20;embedding_xpostag=0;embedding_form=300;embedding_form_file={wordUD|morphUD}.txt.embeding.vec;embedding_lemma=0;embedding_deprel=20;learning_rate=0.01;learning_rate_final=0.001;l2=0.5;hidden_layer=200;batch_size=10;transition_system=projective;transition_oracle=dynamic;structured_interval=10" --heldout={UD_Korean-GSD|UD_Korean-Kaist}/ko_{gsd|kaist}-ud-dev.conllu {UD_Korean-GSD|UD_Korean-Kaist}/ko_{gsd|kaist}-ud-{wordUD|morphUD}.model {UD_Korean-GSD|UD_Korean-Kaist}/ko_{gsd|kaist}-ud-train.conllu

Files

Files (2.5 GB)

Name Size Download all
md5:2afab432805d64f7f67fc852d217a1b5
349.1 MB Download
md5:cab5aa082ebd80eace91680208106229
862.3 MB Download
md5:3ec7f9ac3dee2387c016f2e48d6f664a
361.5 MB Download
md5:05135dc0965966dd369379777a3a767c
918.4 MB Download