From 393d5037e77d12c90dccbd513b501071e5e2448c Mon Sep 17 00:00:00 2001
From: Franck Dary <franck.dary@lis-lab.fr>
Date: Wed, 29 Apr 2020 15:07:12 +0200
Subject: [PATCH] Updated some machines for modular network
---
UD_any/parser/machine.rm | 43 +++++++++++--------------------------
UD_any/tagger/machine.rm | 39 ++++++++-------------------------
UD_any/tagmorpho/machine.rm | 19 ++++++++++++++++
UD_any/tokenizer/machine.rm | 41 ++++++++++-------------------------
4 files changed, 52 insertions(+), 90 deletions(-)
create mode 100644 UD_any/tagmorpho/machine.rm
diff --git a/UD_any/parser/machine.rm b/UD_any/parser/machine.rm
index c5fb678..a6a12ac 100644
--- a/UD_any/parser/machine.rm
+++ b/UD_any/parser/machine.rm
@@ -1,36 +1,19 @@
Name : Parser Machine
Classifier : parser
{
- Transitions : {parser,data/parser.ts}
- Network type : LSTM
- Unknown value threshold : 1
- Buffer context : {-3 -2 -1 0 1 2}
- Stack context : {2 1 0}
- Columns : {FORM UPOS}
- Focused buffer : {-1 0}
- Focused stack : {2 1 0}
- Focused columns : {ID EOS FORM FEATS DEPREL}
- Max nb elements : {1 1 10 10 1}
- Raw input left window : -1
- Raw input right window : -1
- Embeddings size : 128
- MLP : {2048 0.3}
- Context LSTM size : 512
- Focused LSTM size : 256
- Rawinput LSTM size : 32
- Split trans LSTM size : 256
- Num layers : 3
- BiLSTM : true
- LSTM dropout : 0.1
- Total input dropout : 0.3
- Embeddings dropout : 0.3
- Dropout 2d : false
- Tree embedding columns : {DEPREL}
- Tree embedding buffer : {-1}
- Tree embedding stack : {0 1 2}
- Tree embedding nb : {6}
- Tree embedding size : 128
- Optimizer : Adam {0.0005 0.9 0.999 0.00000001 0.00001 true}
+ Transitions : {parser,data/parser.ts}
+ Network type : Modular
+ Context : Unk{1} Buffer{-3 -2 -1 0 1 2} Stack{2 1 0} Columns{FORM UPOS} GRU{1 1 0.0 1} In{128} Out{512}
+ Focused : Column{ID} NbElem{1} Buffer{0} Stack{0} GRU{1 1 0.0 1} In{128} Out{256}
+ Focused : Column{EOS} NbElem{1} Buffer{-1 0} Stack{} GRU{1 1 0.0 1} In{128} Out{256}
+ Focused : Column{FORM} NbElem{10} Buffer{-1 0} Stack{2 1 0} GRU{1 1 0.0 1} In{128} Out{256}
+ Focused : Column{FEATS} NbElem{10} Buffer{-1 0} Stack{2 1 0} GRU{1 1 0.0 1} In{128} Out{256}
+ Focused : Column{DEPREL} NbElem{1} Buffer{} Stack{2 1 0} GRU{1 1 0.0 1} In{128} Out{256}
+ DepthLayerTree : Columns{DEPREL} Buffer{-1} Stack{2 1 0} LayerSizes{6} GRU{1 1 0.0 1} In{128} Out{128}
+ InputDropout : 0.3
+ MLP : {2048 0.3}
+ End
+ Optimizer : Adam {0.0005 0.9 0.999 0.00000001 0.00001 true}
}
Predictions : HEAD DEPREL EOS
Strategy : sequential
diff --git a/UD_any/tagger/machine.rm b/UD_any/tagger/machine.rm
index cbfd0b7..29f91ee 100644
--- a/UD_any/tagger/machine.rm
+++ b/UD_any/tagger/machine.rm
@@ -1,36 +1,15 @@
Name : Tagger Machine
Classifier : tagger
{
- Transitions : {tagger,data/tagger.ts}
- Network type : LSTM
- Unknown value threshold : 1
- Buffer context : {-3 -2 -1 0 1 2}
- Stack context : {}
- Columns : {FORM UPOS}
- Focused buffer : {-1 0}
- Focused stack : {}
- Focused columns : {ID FORM}
- Max nb elements : {1 10}
- Raw input left window : -1
- Raw input right window : -1
- Embeddings size : 128
- MLP : {2048 0.3}
- Context LSTM size : 512
- Focused LSTM size : 256
- Rawinput LSTM size : 32
- Split trans LSTM size : 256
- Num layers : 3
- BiLSTM : true
- LSTM dropout : 0.1
- Total input dropout : 0.3
- Embeddings dropout : 0.3
- Dropout 2d : false
- Tree embedding columns : {}
- Tree embedding buffer : {}
- Tree embedding stack : {}
- Tree embedding nb : {}
- Tree embedding size : 0
- Optimizer : Adam {0.0005 0.9 0.999 0.00000001 0.00001 true}
+ Transitions : {tagger,data/tagger.ts}
+ Network type : Modular
+ Context : Unk{1} Buffer{-3 -2 -1 0 1 2} Stack{} Columns{FORM UPOS} GRU{1 1 0.0 1} In{128} Out{512}
+ Focused : Column{ID} NbElem{1} Buffer{-1 0} Stack{} GRU{1 1 0.0 1} In{128} Out{256}
+ Focused : Column{FORM} NbElem{10} Buffer{-1 0} Stack{} GRU{1 1 0.0 1} In{128} Out{256}
+ InputDropout : 0.3
+ MLP : {2048 0.3}
+ End
+ Optimizer : Adam {0.0005 0.9 0.999 0.00000001 0.00001 true}
}
Predictions : UPOS
Strategy : sequential
diff --git a/UD_any/tagmorpho/machine.rm b/UD_any/tagmorpho/machine.rm
new file mode 100644
index 0000000..7e07d6f
--- /dev/null
+++ b/UD_any/tagmorpho/machine.rm
@@ -0,0 +1,19 @@
+Name : Feats Machine
+Classifier : morpho
+{
+ Transitions : {tagger,data/tagger.ts morpho,data/morpho_parts.ts}
+ Network type : Modular
+ Context : Unk{1} Buffer{-3 -2 -1 0 1 2} Stack{} Columns{FORM UPOS} LSTM{1 1 0.0 1} In{128} Out{512}
+ Focused : Column{ID} NbElem{1} Buffer{-1 0} Stack{} LSTM{1 1 0.0 1} In{128} Out{256}
+ Focused : Column{FORM} NbElem{10} Buffer{-1 0} Stack{} LSTM{1 1 0.0 1} In{128} Out{256}
+ Focused : Column{FEATS} NbElem{10} Buffer{-1 0} Stack{} LSTM{1 1 0.0 1} In{128} Out{256}
+ InputDropout : 0.3
+ MLP : {2048 0.3}
+ End
+ Optimizer : Adam {0.0005 0.9 0.999 0.00000001 0.00001 true}
+}
+Predictions : UPOS FEATS
+Strategy : sequential
+ tagger morpho 1
+ morpho morpho NOTHING 1
+ morpho morpho 0
diff --git a/UD_any/tokenizer/machine.rm b/UD_any/tokenizer/machine.rm
index bc622d2..1b26358 100644
--- a/UD_any/tokenizer/machine.rm
+++ b/UD_any/tokenizer/machine.rm
@@ -1,36 +1,17 @@
Name : Tokenizer Machine
Classifier : tokenizer
{
- Transitions : {tokenizer,data/tokenizer.ts}
- Network type : LSTM
- Unknown value threshold : 1
- Buffer context : {-3 -2 -1}
- Stack context : {}
- Columns : {FORM}
- Focused buffer : {-1 0}
- Focused stack : {}
- Focused columns : {ID FORM}
- Max nb elements : {1 10}
- Raw input left window : 5
- Raw input right window : 5
- Embeddings size : 128
- MLP : {2048 0.3}
- Context LSTM size : 512
- Focused LSTM size : 256
- Rawinput LSTM size : 32
- Split trans LSTM size : 256
- Num layers : 3
- BiLSTM : true
- LSTM dropout : 0.1
- Total input dropout : 0.3
- Embeddings dropout : 0.3
- Dropout 2d : false
- Tree embedding columns : {}
- Tree embedding buffer : {}
- Tree embedding stack : {}
- Tree embedding nb : {}
- Tree embedding size : 0
- Optimizer : Adam {0.0005 0.9 0.999 0.00000001 0.00001 true}
+ Transitions : {tokenizer,data/tokenizer.ts}
+ Network type : Modular
+ Context : Unk{1} Buffer{-3 -2 -1} Stack{} Columns{FORM} GRU{1 1 0.0 1} In{128} Out{512}
+ Focused : Column{ID} NbElem{1} Buffer{-1 0} Stack{} GRU{1 1 0.0 1} In{128} Out{256}
+ Focused : Column{FORM} NbElem{10} Buffer{-1 0} Stack{} GRU{1 1 0.0 1} In{128} Out{256}
+ RawInput : Left{5} Right{5} GRU{1 1 0.0 1} In{32} Out{32}
+ SplitTrans : GRU{1 1 0.0 1} In{128} Out{128}
+ InputDropout : 0.3
+ MLP : {2048 0.3}
+ End
+ Optimizer : Adam {0.0005 0.9 0.999 0.00000001 0.00001 true}
}
Splitwords : data/splitwords.ts
Predictions : ID FORM EOS
--
GitLab