From 225428d180f8ac93dc10c5b13934e2d5c8a7a888 Mon Sep 17 00:00:00 2001
From: Franck Dary <franck.dary@lis-lab.fr>
Date: Thu, 10 Dec 2020 14:22:09 +0100
Subject: [PATCH] Added tokeparser machines with one classifier per task
---
.../results_tokeparser_multi_10_12_20.txt | 65 +++++++++
.../tokeparser_base_multi/machine.rm | 127 +++++++++++++++++
.../tokeparser_incr_multi/machine.rm | 127 +++++++++++++++++
.../templates/tokeparser_seq_multi/machine.rm | 131 ++++++++++++++++++
4 files changed, 450 insertions(+)
create mode 100644 UD_any/results/results_tokeparser_multi_10_12_20.txt
create mode 100644 UD_any/templates/tokeparser_base_multi/machine.rm
create mode 100644 UD_any/templates/tokeparser_incr_multi/machine.rm
create mode 100644 UD_any/templates/tokeparser_seq_multi/machine.rm
diff --git a/UD_any/results/results_tokeparser_multi_10_12_20.txt b/UD_any/results/results_tokeparser_multi_10_12_20.txt
new file mode 100644
index 0000000..8cce207
--- /dev/null
+++ b/UD_any/results/results_tokeparser_multi_10_12_20.txt
@@ -0,0 +1,65 @@
+Corpus Metric F1.score Model
+--------------------------------------------------------------------------------
+UD_English-EWT LAS 73.62[±0.06]% tokeparser_incr_multi
+UD_English-EWT LAS 74.01% tokeparser_base_multi
+UD_English-EWT LAS 76.35[±0.11]% tokeparser_seq_multi
+
+UD_English-EWT Lemmas 96.97% tokeparser_seq_multi
+UD_English-EWT Lemmas 97.03[±0.06]% tokeparser_incr_multi
+UD_English-EWT Lemmas 97.03[±0.10]% tokeparser_base_multi
+
+UD_English-EWT Sentences 69.57[±1.01]% tokeparser_seq_multi
+UD_English-EWT Sentences 73.29[±0.41]% tokeparser_incr_multi
+UD_English-EWT Sentences 73.94[±0.69]% tokeparser_base_multi
+
+UD_English-EWT Tokens 98.43[±0.01]% tokeparser_seq_multi
+UD_English-EWT Tokens 98.44[±0.02]% tokeparser_incr_multi
+UD_English-EWT Tokens 98.47[±0.09]% tokeparser_base_multi
+
+UD_English-EWT UAS 76.93[±0.09]% tokeparser_incr_multi
+UD_English-EWT UAS 77.34[±0.00]% tokeparser_base_multi
+UD_English-EWT UAS 79.66[±0.10]% tokeparser_seq_multi
+
+UD_English-EWT UFeats 95.41[±0.04]% tokeparser_base_multi
+UD_English-EWT UFeats 95.41[±0.10]% tokeparser_incr_multi
+UD_English-EWT UFeats 95.52[±0.04]% tokeparser_seq_multi
+
+UD_English-EWT UPOS 93.92[±0.12]% tokeparser_base_multi
+UD_English-EWT UPOS 93.97[±0.16]% tokeparser_incr_multi
+UD_English-EWT UPOS 94.32[±0.01]% tokeparser_seq_multi
+
+UD_English-EWT Words 98.43[±0.01]% tokeparser_seq_multi
+UD_English-EWT Words 98.44[±0.02]% tokeparser_incr_multi
+UD_English-EWT Words 98.47[±0.09]% tokeparser_base_multi
+--------------------------------------------------------------------------------
+UD_French-GSD LAS 83.59[±0.34]% tokeparser_incr_multi
+UD_French-GSD LAS 84.29[±0.10]% tokeparser_base_multi
+UD_French-GSD LAS 85.34[±0.10]% tokeparser_seq_multi
+
+UD_French-GSD Lemmas 97.61% tokeparser_seq_multi
+UD_French-GSD Lemmas 97.61[±0.06]% tokeparser_incr_multi
+UD_French-GSD Lemmas 97.62[±0.00]% tokeparser_base_multi
+
+UD_French-GSD Sentences 91.17[±1.03]% tokeparser_base_multi
+UD_French-GSD Sentences 92.41[±0.52]% tokeparser_seq_multi
+UD_French-GSD Sentences 92.84[±0.48]% tokeparser_incr_multi
+
+UD_French-GSD Tokens 99.52[±0.00]% tokeparser_seq_multi
+UD_French-GSD Tokens 99.52[±0.04]% tokeparser_base_multi
+UD_French-GSD Tokens 99.54[±0.06]% tokeparser_incr_multi
+
+UD_French-GSD UAS 86.44[±0.34]% tokeparser_incr_multi
+UD_French-GSD UAS 87.00[±0.01]% tokeparser_base_multi
+UD_French-GSD UAS 87.90[±0.04]% tokeparser_seq_multi
+
+UD_French-GSD UFeats 95.47[±0.12]% tokeparser_incr_multi
+UD_French-GSD UFeats 95.56[±0.01]% tokeparser_base_multi
+UD_French-GSD UFeats 95.82[±0.03]% tokeparser_seq_multi
+
+UD_French-GSD UPOS 96.81[±0.06]% tokeparser_base_multi
+UD_French-GSD UPOS 96.84[±0.02]% tokeparser_incr_multi
+UD_French-GSD UPOS 96.95[±0.03]% tokeparser_seq_multi
+
+UD_French-GSD Words 99.25[±0.02]% tokeparser_seq_multi
+UD_French-GSD Words 99.25[±0.04]% tokeparser_base_multi
+UD_French-GSD Words 99.30[±0.08]% tokeparser_incr_multi
diff --git a/UD_any/templates/tokeparser_base_multi/machine.rm b/UD_any/templates/tokeparser_base_multi/machine.rm
new file mode 100644
index 0000000..142c979
--- /dev/null
+++ b/UD_any/templates/tokeparser_base_multi/machine.rm
@@ -0,0 +1,127 @@
+Name : Tokenizer, Tagger, Morpho, Lemmatizer, Parser and Segmenter Machine with one classifier for each state
+Classifier : tokenizer
+{
+ Transitions : {tokenizer,data/tokenizer.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0}
+ Context : Targets{b.-3 b.-2 b.-1} Columns{ID} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ SplitTrans : LSTM{1 1 0.0 1} In{64} Out{64}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : tagger
+{
+ Transitions : {tagger,data/tagger.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{ID UPOS} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : morpho
+{
+ Transitions : {morpho,data/morpho_whole.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{ID UPOS FEATS} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : lemmatizer
+{
+ Transitions : {lemmatizer_rules,data/lemmatizer_rules.ts lemmatizer_case,data/lemmatizer_case.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{ID UPOS FEATS} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : parser
+{
+ Transitions : {parser,data/parser_eager_rel_strict.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : segmenter
+{
+ Transitions : {segmenter,data/segmenter.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{EOS ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Splitwords : data/splitwords.ts
+Predictions : ID FORM UPOS FEATS LEMMA HEAD DEPREL EOS
+Strategy
+{
+ Block : End{cannotMove}
+ tokenizer tagger ENDWORD 0
+ tokenizer tagger SPLIT 0
+ tokenizer tokenizer * 0
+ tagger morpho * 0
+ morpho lemmatizer_rules * 0
+ lemmatizer_rules lemmatizer_case * 0
+ lemmatizer_case parser * 0
+ parser segmenter eager_SHIFT 0
+ parser segmenter eager_RIGHT_rel 0
+ parser parser * 0
+ segmenter tokenizer * 1
+}
diff --git a/UD_any/templates/tokeparser_incr_multi/machine.rm b/UD_any/templates/tokeparser_incr_multi/machine.rm
new file mode 100644
index 0000000..2f5f928
--- /dev/null
+++ b/UD_any/templates/tokeparser_incr_multi/machine.rm
@@ -0,0 +1,127 @@
+Name : Tokenizer, Tagger, Morpho, Lemmatizer, Parser and Segmenter Machine with one classifier for each state
+Classifier : tokenizer
+{
+ Transitions : {tokenizer,data/tokenizer.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{EOS ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ SplitTrans : LSTM{1 1 0.0 1} In{64} Out{64}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : tagger
+{
+ Transitions : {tagger,data/tagger.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{EOS ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : morpho
+{
+ Transitions : {morpho,data/morpho_whole.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{EOS ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : lemmatizer
+{
+ Transitions : {lemmatizer_rules,data/lemmatizer_rules.ts lemmatizer_case,data/lemmatizer_case.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{EOS ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : parser
+{
+ Transitions : {parser,data/parser_eager_rel_strict.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{EOS ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : segmenter
+{
+ Transitions : {segmenter,data/segmenter.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{EOS ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Splitwords : data/splitwords.ts
+Predictions : ID FORM UPOS FEATS LEMMA HEAD DEPREL EOS
+Strategy
+{
+ Block : End{cannotMove}
+ tokenizer tagger ENDWORD 0
+ tokenizer tagger SPLIT 0
+ tokenizer tokenizer * 0
+ tagger morpho * 0
+ morpho lemmatizer_rules * 0
+ lemmatizer_rules lemmatizer_case * 0
+ lemmatizer_case parser * 0
+ parser segmenter eager_SHIFT 0
+ parser segmenter eager_RIGHT_rel 0
+ parser parser * 0
+ segmenter tokenizer * 1
+}
diff --git a/UD_any/templates/tokeparser_seq_multi/machine.rm b/UD_any/templates/tokeparser_seq_multi/machine.rm
new file mode 100644
index 0000000..13f6e47
--- /dev/null
+++ b/UD_any/templates/tokeparser_seq_multi/machine.rm
@@ -0,0 +1,131 @@
+Name : Tokenizer, Tagger, Morpho, Lemmatizer, Parser and Segmenter Machine with one classifier for each state
+Classifier : tokenizer
+{
+ Transitions : {tokenizer,data/tokenizer.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 0} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0}
+ Context : Targets{b.-3 b.-2 b.-1} Columns{ID} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ SplitTrans : LSTM{1 1 0.0 1} In{64} Out{64}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : tagger
+{
+ Transitions : {tagger,data/tagger.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 10} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2} Columns{ID UPOS} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : morpho
+{
+ Transitions : {morpho,data/morpho_whole.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 10} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2} Columns{ID UPOS FEATS} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : lemmatizer
+{
+ Transitions : {lemmatizer_rules,data/lemmatizer_rules.ts lemmatizer_case,data/lemmatizer_case.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 10} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2} Columns{ID UPOS FEATS} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : parser
+{
+ Transitions : {parser,data/parser_eager_rel_strict.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 10} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{EOS ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Classifier : segmenter
+{
+ Transitions : {segmenter,data/segmenter.ts}
+ LossMultiplier : {}
+ Network type : Modular
+ Contextual : Window{-10 10} Columns{FORM} LSTM{1 1 0 1} In{64} Out{128} w2v{FORM,data/FORM.w2v} Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1}
+ Context : Targets{b.-3 b.-2 b.-1 b.0 b.1 b.2 s.0 s.1 s.2 b.0.0 s.0.0 s.0.-1 s.1.0 s.1.-1 s.2.0 s.2.-1} Columns{EOS ID UPOS FEATS DEPREL} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{prefix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ Focused : Column{suffix3:FORM} NbElem{3} Buffer{0} Stack{} LSTM{1 1 0 1} In{64} Out{64} w2v{}
+ RawInput : Left{5} Right{10} LSTM{1 1 0.0 1} In{32} Out{32}
+ History : NbElem{10} LSTM{1 1 0 1} In{32} Out{32}
+ InputDropout : 0.3
+ MLP : {1600 0.3 1600 0.3}
+ End
+ Optimizer : Adagrad {0.01 0.000001 0 0.0000000001}
+ Type : classification
+ Loss : crossentropy
+}
+Splitwords : data/splitwords.ts
+Predictions : ID FORM UPOS FEATS LEMMA HEAD DEPREL EOS
+Strategy
+{
+ Block : End{cannotMove}
+ tokenizer tokenizer ENDWORD 1
+ tokenizer tokenizer SPLIT 1
+ tokenizer tokenizer * 0
+ Block : End{cannotMove}
+ tagger tagger * 1
+ Block : End{cannotMove}
+ morpho morpho * 1
+ Block : End{cannotMove}
+ lemmatizer_rules lemmatizer_case * 0
+ lemmatizer_case lemmatizer_rules * 1
+ Block : End{cannotMove}
+ parser segmenter eager_SHIFT 0
+ parser segmenter eager_RIGHT_rel 0
+ parser parser * 0
+ segmenter parser * 1
+}
--
GitLab