From ca1b87c147246197b873e0d6bf2cc7341b31d4bb Mon Sep 17 00:00:00 2001 From: Franck Dary <franck.dary@lis-lab.fr> Date: Wed, 4 Dec 2019 16:30:50 +0100 Subject: [PATCH] Increased dict size --- UD_any/tokeparser/tokeparser.dicts | 18 +++++++++--------- UD_any/tokeparser_incremental/tokeparser.dicts | 18 +++++++++--------- UD_any/tokeparser_sequential/tokeparser.dicts | 18 +++++++++--------- 3 files changed, 27 insertions(+), 27 deletions(-) diff --git a/UD_any/tokeparser/tokeparser.dicts b/UD_any/tokeparser/tokeparser.dicts index 7067316..7378953 100644 --- a/UD_any/tokeparser/tokeparser.dicts +++ b/UD_any/tokeparser/tokeparser.dicts @@ -12,7 +12,7 @@ Tokenizer_form.f 80 Embeddings _ 200000 Tokenizer_lemma 80 Embeddings _ 20 Tokenizer_letters 80 Embeddings _ 5000 Tokenizer_labels 18 Embeddings _ 200 -Tokenizer_morpho 22 Embeddings _ 2000 +Tokenizer_morpho 22 Embeddings _ 8000 # TAGGER Tagger_actions 18 Embeddings _ 200 Tagger_bool 16 Embeddings _ 20 @@ -25,7 +25,7 @@ Tagger_form.f 80 Embeddings _ 200000 Tagger_lemma 80 Embeddings _ 20 Tagger_letters 80 Embeddings _ 5000 Tagger_labels 18 Embeddings _ 200 -Tagger_morpho 22 Embeddings _ 2000 +Tagger_morpho 22 Embeddings _ 8000 # MORPHO Morpho_actions 18 Embeddings _ 200 Morpho_bool 16 Embeddings _ 20 @@ -38,7 +38,7 @@ Morpho_form.f 80 Embeddings _ 200000 Morpho_lemma 80 Embeddings _ 20 Morpho_letters 80 Embeddings _ 5000 Morpho_labels 18 Embeddings _ 200 -Morpho_morpho 22 Embeddings _ 20000 +Morpho_morpho 22 Embeddings _ 80000 # LEMMATIZER_RULE Lemmatizer_Rules_actions 18 Embeddings _ 2000 Lemmatizer_Rules_bool 16 Embeddings _ 20 @@ -46,12 +46,12 @@ Lemmatizer_Rules_int 16 Embeddings _ 2000 Lemmatizer_Rules_eos 16 Embeddings _ 20 Lemmatizer_Rules_gov 16 Embeddings _ 20 Lemmatizer_Rules_pos 18 Embeddings _ 30 -Lemmatizer_Rules_form 80 Embeddings _ 100000 +Lemmatizer_Rules_form 80 Embeddings _ 300000 Lemmatizer_Rules_form.f 80 Embeddings _ 20 Lemmatizer_Rules_lemma 80 Embeddings _ 20 Lemmatizer_Rules_letters 80 Embeddings _ 5000 Lemmatizer_Rules_labels 18 Embeddings _ 20 -Lemmatizer_Rules_morpho 22 Embeddings _ 2000 +Lemmatizer_Rules_morpho 22 Embeddings _ 8000 # LEMMATIZER_CASE Lemmatizer_Case_actions 18 Embeddings _ 2000 Lemmatizer_Case_bool 16 Embeddings _ 20 @@ -59,12 +59,12 @@ Lemmatizer_Case_int 16 Embeddings _ 2000 Lemmatizer_Case_eos 16 Embeddings _ 20 Lemmatizer_Case_gov 16 Embeddings _ 20 Lemmatizer_Case_pos 18 Embeddings _ 30 -Lemmatizer_Case_form 80 Embeddings _ 100000 +Lemmatizer_Case_form 80 Embeddings _ 300000 Lemmatizer_Case_form.f 80 Embeddings _ 20 Lemmatizer_Case_lemma 80 Embeddings _ 20 Lemmatizer_Case_letters 80 Embeddings _ 5000 Lemmatizer_Case_labels 18 Embeddings _ 20 -Lemmatizer_Case_morpho 22 Embeddings _ 2000 +Lemmatizer_Case_morpho 22 Embeddings _ 8000 # PARSER Parser_actions 18 Embeddings _ 200 Parser_bool 16 Embeddings _ 20 @@ -77,7 +77,7 @@ Parser_form.f 80 Embeddings _ 200000 Parser_lemma 80 Embeddings _ 20 Parser_letters 80 Embeddings _ 5000 Parser_labels 18 Embeddings _ 200 -Parser_morpho 22 Embeddings _ 2000 +Parser_morpho 22 Embeddings _ 8000 # SEGMENTER Segmenter_actions 18 Embeddings _ 200 Segmenter_bool 16 Embeddings _ 20 @@ -90,4 +90,4 @@ Segmenter_form.f 80 Embeddings _ 200000 Segmenter_lemma 80 Embeddings _ 20 Segmenter_letters 80 Embeddings _ 5000 Segmenter_labels 18 Embeddings _ 200 -Segmenter_morpho 22 Embeddings _ 2000 +Segmenter_morpho 22 Embeddings _ 8000 diff --git a/UD_any/tokeparser_incremental/tokeparser.dicts b/UD_any/tokeparser_incremental/tokeparser.dicts index 7067316..7378953 100644 --- a/UD_any/tokeparser_incremental/tokeparser.dicts +++ b/UD_any/tokeparser_incremental/tokeparser.dicts @@ -12,7 +12,7 @@ Tokenizer_form.f 80 Embeddings _ 200000 Tokenizer_lemma 80 Embeddings _ 20 Tokenizer_letters 80 Embeddings _ 5000 Tokenizer_labels 18 Embeddings _ 200 -Tokenizer_morpho 22 Embeddings _ 2000 +Tokenizer_morpho 22 Embeddings _ 8000 # TAGGER Tagger_actions 18 Embeddings _ 200 Tagger_bool 16 Embeddings _ 20 @@ -25,7 +25,7 @@ Tagger_form.f 80 Embeddings _ 200000 Tagger_lemma 80 Embeddings _ 20 Tagger_letters 80 Embeddings _ 5000 Tagger_labels 18 Embeddings _ 200 -Tagger_morpho 22 Embeddings _ 2000 +Tagger_morpho 22 Embeddings _ 8000 # MORPHO Morpho_actions 18 Embeddings _ 200 Morpho_bool 16 Embeddings _ 20 @@ -38,7 +38,7 @@ Morpho_form.f 80 Embeddings _ 200000 Morpho_lemma 80 Embeddings _ 20 Morpho_letters 80 Embeddings _ 5000 Morpho_labels 18 Embeddings _ 200 -Morpho_morpho 22 Embeddings _ 20000 +Morpho_morpho 22 Embeddings _ 80000 # LEMMATIZER_RULE Lemmatizer_Rules_actions 18 Embeddings _ 2000 Lemmatizer_Rules_bool 16 Embeddings _ 20 @@ -46,12 +46,12 @@ Lemmatizer_Rules_int 16 Embeddings _ 2000 Lemmatizer_Rules_eos 16 Embeddings _ 20 Lemmatizer_Rules_gov 16 Embeddings _ 20 Lemmatizer_Rules_pos 18 Embeddings _ 30 -Lemmatizer_Rules_form 80 Embeddings _ 100000 +Lemmatizer_Rules_form 80 Embeddings _ 300000 Lemmatizer_Rules_form.f 80 Embeddings _ 20 Lemmatizer_Rules_lemma 80 Embeddings _ 20 Lemmatizer_Rules_letters 80 Embeddings _ 5000 Lemmatizer_Rules_labels 18 Embeddings _ 20 -Lemmatizer_Rules_morpho 22 Embeddings _ 2000 +Lemmatizer_Rules_morpho 22 Embeddings _ 8000 # LEMMATIZER_CASE Lemmatizer_Case_actions 18 Embeddings _ 2000 Lemmatizer_Case_bool 16 Embeddings _ 20 @@ -59,12 +59,12 @@ Lemmatizer_Case_int 16 Embeddings _ 2000 Lemmatizer_Case_eos 16 Embeddings _ 20 Lemmatizer_Case_gov 16 Embeddings _ 20 Lemmatizer_Case_pos 18 Embeddings _ 30 -Lemmatizer_Case_form 80 Embeddings _ 100000 +Lemmatizer_Case_form 80 Embeddings _ 300000 Lemmatizer_Case_form.f 80 Embeddings _ 20 Lemmatizer_Case_lemma 80 Embeddings _ 20 Lemmatizer_Case_letters 80 Embeddings _ 5000 Lemmatizer_Case_labels 18 Embeddings _ 20 -Lemmatizer_Case_morpho 22 Embeddings _ 2000 +Lemmatizer_Case_morpho 22 Embeddings _ 8000 # PARSER Parser_actions 18 Embeddings _ 200 Parser_bool 16 Embeddings _ 20 @@ -77,7 +77,7 @@ Parser_form.f 80 Embeddings _ 200000 Parser_lemma 80 Embeddings _ 20 Parser_letters 80 Embeddings _ 5000 Parser_labels 18 Embeddings _ 200 -Parser_morpho 22 Embeddings _ 2000 +Parser_morpho 22 Embeddings _ 8000 # SEGMENTER Segmenter_actions 18 Embeddings _ 200 Segmenter_bool 16 Embeddings _ 20 @@ -90,4 +90,4 @@ Segmenter_form.f 80 Embeddings _ 200000 Segmenter_lemma 80 Embeddings _ 20 Segmenter_letters 80 Embeddings _ 5000 Segmenter_labels 18 Embeddings _ 200 -Segmenter_morpho 22 Embeddings _ 2000 +Segmenter_morpho 22 Embeddings _ 8000 diff --git a/UD_any/tokeparser_sequential/tokeparser.dicts b/UD_any/tokeparser_sequential/tokeparser.dicts index 7067316..7378953 100644 --- a/UD_any/tokeparser_sequential/tokeparser.dicts +++ b/UD_any/tokeparser_sequential/tokeparser.dicts @@ -12,7 +12,7 @@ Tokenizer_form.f 80 Embeddings _ 200000 Tokenizer_lemma 80 Embeddings _ 20 Tokenizer_letters 80 Embeddings _ 5000 Tokenizer_labels 18 Embeddings _ 200 -Tokenizer_morpho 22 Embeddings _ 2000 +Tokenizer_morpho 22 Embeddings _ 8000 # TAGGER Tagger_actions 18 Embeddings _ 200 Tagger_bool 16 Embeddings _ 20 @@ -25,7 +25,7 @@ Tagger_form.f 80 Embeddings _ 200000 Tagger_lemma 80 Embeddings _ 20 Tagger_letters 80 Embeddings _ 5000 Tagger_labels 18 Embeddings _ 200 -Tagger_morpho 22 Embeddings _ 2000 +Tagger_morpho 22 Embeddings _ 8000 # MORPHO Morpho_actions 18 Embeddings _ 200 Morpho_bool 16 Embeddings _ 20 @@ -38,7 +38,7 @@ Morpho_form.f 80 Embeddings _ 200000 Morpho_lemma 80 Embeddings _ 20 Morpho_letters 80 Embeddings _ 5000 Morpho_labels 18 Embeddings _ 200 -Morpho_morpho 22 Embeddings _ 20000 +Morpho_morpho 22 Embeddings _ 80000 # LEMMATIZER_RULE Lemmatizer_Rules_actions 18 Embeddings _ 2000 Lemmatizer_Rules_bool 16 Embeddings _ 20 @@ -46,12 +46,12 @@ Lemmatizer_Rules_int 16 Embeddings _ 2000 Lemmatizer_Rules_eos 16 Embeddings _ 20 Lemmatizer_Rules_gov 16 Embeddings _ 20 Lemmatizer_Rules_pos 18 Embeddings _ 30 -Lemmatizer_Rules_form 80 Embeddings _ 100000 +Lemmatizer_Rules_form 80 Embeddings _ 300000 Lemmatizer_Rules_form.f 80 Embeddings _ 20 Lemmatizer_Rules_lemma 80 Embeddings _ 20 Lemmatizer_Rules_letters 80 Embeddings _ 5000 Lemmatizer_Rules_labels 18 Embeddings _ 20 -Lemmatizer_Rules_morpho 22 Embeddings _ 2000 +Lemmatizer_Rules_morpho 22 Embeddings _ 8000 # LEMMATIZER_CASE Lemmatizer_Case_actions 18 Embeddings _ 2000 Lemmatizer_Case_bool 16 Embeddings _ 20 @@ -59,12 +59,12 @@ Lemmatizer_Case_int 16 Embeddings _ 2000 Lemmatizer_Case_eos 16 Embeddings _ 20 Lemmatizer_Case_gov 16 Embeddings _ 20 Lemmatizer_Case_pos 18 Embeddings _ 30 -Lemmatizer_Case_form 80 Embeddings _ 100000 +Lemmatizer_Case_form 80 Embeddings _ 300000 Lemmatizer_Case_form.f 80 Embeddings _ 20 Lemmatizer_Case_lemma 80 Embeddings _ 20 Lemmatizer_Case_letters 80 Embeddings _ 5000 Lemmatizer_Case_labels 18 Embeddings _ 20 -Lemmatizer_Case_morpho 22 Embeddings _ 2000 +Lemmatizer_Case_morpho 22 Embeddings _ 8000 # PARSER Parser_actions 18 Embeddings _ 200 Parser_bool 16 Embeddings _ 20 @@ -77,7 +77,7 @@ Parser_form.f 80 Embeddings _ 200000 Parser_lemma 80 Embeddings _ 20 Parser_letters 80 Embeddings _ 5000 Parser_labels 18 Embeddings _ 200 -Parser_morpho 22 Embeddings _ 2000 +Parser_morpho 22 Embeddings _ 8000 # SEGMENTER Segmenter_actions 18 Embeddings _ 200 Segmenter_bool 16 Embeddings _ 20 @@ -90,4 +90,4 @@ Segmenter_form.f 80 Embeddings _ 200000 Segmenter_lemma 80 Embeddings _ 20 Segmenter_letters 80 Embeddings _ 5000 Segmenter_labels 18 Embeddings _ 200 -Segmenter_morpho 22 Embeddings _ 2000 +Segmenter_morpho 22 Embeddings _ 8000 -- GitLab