rbbt-dm 1.1.56 → 1.1.57

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8ab1295a1cb43602045e0c75226cc71e1b8eb8fbc7ce72f869f1636309745ad7
-  data.tar.gz: 0c6bd97a2e8c81deb4435531344fd5b145382ddafd06012e93d8695de89f3ff2
+  metadata.gz: 072d57536b223931947dfd71d5e359961c6bfe44da0870cd2dbab440aa91ba6c
+  data.tar.gz: 7c5d2a3fc627992231b621b6efe4d2160aa093b11c89f31f728ede5121d2edc9
 SHA512:
-  metadata.gz: 9f2b34158e345c703e60a94d7481e397add3651b0fc5fd7c80205b2c44ae9578e1ccac2b9d67e3679afa37f60d56fe3defed8fbd42c81765e1239b90e8bc06da
-  data.tar.gz: fc1d764b9a240b60bd1de30e22f534485f7319c0396b291e28b0fe1ad48acde7cdeb5de7448509550892cd17e743b5388348f1f40584233280b5ec42cc95940c
+  metadata.gz: 5216b3179bc4a809829b79bc4c447159e88444dc7943da3d0c3643b728e3eb5a6da7c08a8538fb88db1e463f95e55e104976ac81f010aeaf729a6cb1c4ca1374
+  data.tar.gz: 12baae07f04ca3770dfef81c3166385badb7944d25b7072364631e5f93b419e04c53d5b7157934b3d1522ef83e84acf0c4e2c5d39ad3604b7ef8f5f460d2f750

data/lib/rbbt/vector/model/spaCy.rb

@@ -4,13 +4,13 @@ require 'rbbt/nlp/spaCy'
 class SpaCyModel < VectorModel
   attr_accessor :config
 
-  def spacy(&block)
+  def self.spacy(&block)
     RbbtPython.run "spacy" do 
       RbbtPython.module_eval(&block)
     end
   end
   
-  def initialize(dir, config, lang = 'en_core_web_md')
+  def initialize(dir, config, categories = %w(positive negative), lang = 'en_core_web_md')
     @config = case
               when Path === config
                 config.read
@@ -33,17 +33,18 @@ class SpaCyModel < VectorModel
       tmpconfig = File.join(file, 'config')
       tmptrain = File.join(file, 'train.spacy')
       SpaCy.config(@config, tmpconfig)
-      spacy do
+      SpaCyModel.spacy do
         nlp = SpaCy.nlp(lang)
         docs = []
         RbbtPython.iterate nlp.pipe(texts.zip(labels), as_tuples: true), :bar => "Training documents into spacy format" do |doc,label|
-          if %w(1 true pos).include?(label.to_s.downcase) 
-            doc.cats["positive"] = 1
-            doc.cats["negative"] = 0
-          else
-            doc.cats["positive"] = 0
-            doc.cats["negative"] = 1
-          end
+          doc.cats[label] = 1
+          #if %w(1 true pos).include?(label.to_s.downcase) 
+          #  doc.cats["positive"] = 1
+          #  doc.cats["negative"] = 0
+          #else
+          #  doc.cats["positive"] = 0
+          #  doc.cats["negative"] = 1
+          #end
           docs << doc
         end
 
@@ -51,7 +52,7 @@ class SpaCyModel < VectorModel
         doc_bin.to_disk(tmptrain)
       end
 
-      gpu = Rbbt::Config.get('gpu_id', :spacy, :spacy_train)
+      gpu = Rbbt::Config.get('gpu_id', :spacy, :spacy_train, :default => 0)
       CMD.cmd_log(:spacy, "train #{tmpconfig} --output #{file} --paths.train #{tmptrain} --paths.dev #{tmptrain}",  "--gpu-id" => gpu)
     end
  
@@ -59,14 +60,15 @@ class SpaCyModel < VectorModel
       texts = features
 
       docs = []
-      spacy do
+      SpaCyModel.spacy do
         nlp = spacy.load("#{file}/model-best")
 
         Log::ProgressBar.with_bar texts.length, :desc => "Evaluating documents" do |bar|
           texts.collect do |text|
             cats = nlp.(text).cats
             bar.tick
-            cats['positive'] > cats['negative']  ? 1 : 0
+            cats.sort_by{|l,v| v.to_f }.last.first
+            #cats['positive'] > cats['negative']  ? 1 : 0
           end
         end
       end

data/lib/rbbt/vector/model.rb

@@ -320,6 +320,8 @@ cat(paste(label, sep="\\n", collapse="\\n"));
       acc = []
       labels.each do |good_label|
         values = VectorModel.f1_metrics(test, predicted, good_label)
+        tp, tn, fp, fn, pr, re, f1 = values
+        Log.debug "Partial CV #{good_label} - P:#{"%.3f" % pr} R:#{"%.3f" % re} F1:#{"%.3f" % f1} - #{[tp.to_s, tn.to_s, fp.to_s, fn.to_s] * " "}"
         acc << values
       end
       Misc.zip_fields(acc).collect{|s| Misc.mean(s)}
@@ -340,12 +342,21 @@ cat(paste(label, sep="\\n", collapse="\\n"));
     end
 
     begin
-      feature_folds = Misc.divide(@features, folds)
-      labels_folds = Misc.divide(@labels, folds)
+      if folds == 1
+        feature_folds = [@features]
+        labels_folds = [@labels]
+      else
+        feature_folds = Misc.divide(@features, folds)
+        labels_folds = Misc.divide(@labels, folds)
+      end
 
       folds.times do |fix|
 
-        rest = (0..(folds-1)).to_a - [fix]
+        if folds == 1
+          rest = [fix]
+        else
+          rest = (0..(folds-1)).to_a - [fix]
+        end
 
         test_set = feature_folds[fix]
         train_set = feature_folds.values_at(*rest).inject([]){|acc,e| acc += e; acc}
@@ -355,6 +366,7 @@ cat(paste(label, sep="\\n", collapse="\\n"));
 
         @features = train_set
         @labels = train_labels
+
         self.train
         predictions = self.eval_list test_set, false
 
@@ -362,6 +374,8 @@ cat(paste(label, sep="\\n", collapse="\\n"));
 
         different_labels = test_labels.uniq
 
+        Log.debug do "Accuracy Fold #{fix}: #{(100 * test_labels.zip(predictions).select{|t,p| t == p }.length.to_f / test_labels.length).round(2)}%"  end
+
         tp, tn, fp, fn, pr, re, f1 = VectorModel.f1_metrics(test_labels, predictions, good_label)
 
         if multiclass 
@@ -377,7 +391,7 @@ cat(paste(label, sep="\\n", collapse="\\n"));
       @features = orig_features
       @labels = orig_labels
     end
-    self.train
+    self.train unless folds == 1
     res
   end
 end

data/share/spaCy/cpu/textcat_multilabel_accuracy.conf

@@ -0,0 +1,86 @@
+# This is an auto-generated partial config. To use it with 'spacy train'
+# you can run spacy init fill-config to auto-fill all default settings:
+# python -m spacy init fill-config ./base_config.cfg ./config.cfg
+[paths]
+train = null
+dev = null
+
+[system]
+gpu_allocator = null
+
+[nlp]
+lang = "en"
+pipeline = ["tok2vec","textcat_multilabel"]
+batch_size = 1000
+
+[components]
+
+[components.tok2vec]
+factory = "tok2vec"
+
+[components.tok2vec.model]
+@architectures = "spacy.Tok2Vec.v2"
+
+[components.tok2vec.model.embed]
+@architectures = "spacy.MultiHashEmbed.v2"
+width = ${components.tok2vec.model.encode.width}
+attrs = ["ORTH", "SHAPE"]
+rows = [5000, 2500]
+include_static_vectors = true
+
+[components.tok2vec.model.encode]
+@architectures = "spacy.MaxoutWindowEncoder.v2"
+width = 256
+depth = 8
+window_size = 1
+maxout_pieces = 3
+
+[components.textcat_multilabel]
+factory = "textcat_multilabel"
+
+[components.textcat_multilabel.model]
+@architectures = "spacy.TextCatEnsemble.v2"
+nO = null
+
+[components.textcat_multilabel.model.tok2vec]
+@architectures = "spacy.Tok2VecListener.v1"
+width = ${components.tok2vec.model.encode.width}
+
+[components.textcat_multilabel.model.linear_model]
+@architectures = "spacy.TextCatBOW.v1"
+exclusive_classes = true
+ngram_size = 1
+no_output_layer = false
+
+[corpora]
+
+[corpora.train]
+@readers = "spacy.Corpus.v1"
+path = ${paths.train}
+max_length = 2000
+
+[corpora.dev]
+@readers = "spacy.Corpus.v1"
+path = ${paths.dev}
+max_length = 0
+
+[training]
+dev_corpus = "corpora.dev"
+train_corpus = "corpora.train"
+
+[training.optimizer]
+@optimizers = "Adam.v1"
+
+[training.batcher]
+@batchers = "spacy.batch_by_words.v1"
+discard_oversize = false
+tolerance = 0.2
+
+[training.batcher.size]
+@schedules = "compounding.v1"
+start = 100
+stop = 1000
+compound = 1.001
+
+[initialize]
+vectors = "en_core_web_lg"

data/share/spaCy/cpu/textcat_multilabel_efficiency.conf

@@ -0,0 +1,78 @@
+# This is an auto-generated partial config. To use it with 'spacy train'
+# you can run spacy init fill-config to auto-fill all default settings:
+# python -m spacy init fill-config ./base_config.cfg ./config.cfg
+[paths]
+train = null
+dev = null
+
+[system]
+gpu_allocator = null
+
+[nlp]
+lang = "en"
+pipeline = ["tok2vec","textcat_multilabel"]
+batch_size = 1000
+
+[components]
+
+[components.tok2vec]
+factory = "tok2vec"
+
+[components.tok2vec.model]
+@architectures = "spacy.Tok2Vec.v2"
+
+[components.tok2vec.model.embed]
+@architectures = "spacy.MultiHashEmbed.v2"
+width = ${components.tok2vec.model.encode.width}
+attrs = ["ORTH", "SHAPE"]
+rows = [5000, 2500]
+include_static_vectors = false
+
+[components.tok2vec.model.encode]
+@architectures = "spacy.MaxoutWindowEncoder.v2"
+width = 96
+depth = 4
+window_size = 1
+maxout_pieces = 3
+
+[components.textcat_multilabel]
+factory = "textcat_multilabel"
+
+[components.textcat_multilabel.model]
+@architectures = "spacy.TextCatBOW.v1"
+exclusive_classes = true
+ngram_size = 1
+no_output_layer = false
+
+[corpora]
+
+[corpora.train]
+@readers = "spacy.Corpus.v1"
+path = ${paths.train}
+max_length = 2000
+
+[corpora.dev]
+@readers = "spacy.Corpus.v1"
+path = ${paths.dev}
+max_length = 0
+
+[training]
+dev_corpus = "corpora.dev"
+train_corpus = "corpora.train"
+
+[training.optimizer]
+@optimizers = "Adam.v1"
+
+[training.batcher]
+@batchers = "spacy.batch_by_words.v1"
+discard_oversize = false
+tolerance = 0.2
+
+[training.batcher.size]
+@schedules = "compounding.v1"
+start = 100
+stop = 1000
+compound = 1.001
+
+[initialize]
+vectors = null

data/share/spaCy/gpu/textcat_multilabel_accuracy.conf

@@ -0,0 +1,84 @@
+# This is an auto-generated partial config. To use it with 'spacy train'
+# you can run spacy init fill-config to auto-fill all default settings:
+# python -m spacy init fill-config ./base_config.cfg ./config.cfg
+[paths]
+train = null
+dev = null
+
+[system]
+gpu_allocator = "pytorch"
+
+[nlp]
+lang = "en"
+pipeline = ["transformer","textcat_multilabel"]
+batch_size = 128
+
+[components]
+
+[components.transformer]
+factory = "transformer"
+
+[components.transformer.model]
+@architectures = "spacy-transformers.TransformerModel.v1"
+name = "emilyalsentzer/Bio_ClinicalBERT"
+tokenizer_config = {"use_fast": true}
+
+[components.transformer.model.get_spans]
+@span_getters = "spacy-transformers.strided_spans.v1"
+window = 128
+stride = 96
+
+[components.textcat_multilabel]
+factory = "textcat_multilabel"
+
+[components.textcat_multilabel.model]
+@architectures = "spacy.TextCatEnsemble.v2"
+nO = null
+
+[components.textcat_multilabel.model.tok2vec]
+@architectures = "spacy-transformers.TransformerListener.v1"
+grad_factor = 1.0
+
+[components.textcat_multilabel.model.tok2vec.pooling]
+@layers = "reduce_mean.v1"
+
+[components.textcat_multilabel.model.linear_model]
+@architectures = "spacy.TextCatBOW.v1"
+exclusive_classes = true
+ngram_size = 1
+no_output_layer = false
+
+[corpora]
+
+[corpora.train]
+@readers = "spacy.Corpus.v1"
+path = ${paths.train}
+max_length = 500
+
+[corpora.dev]
+@readers = "spacy.Corpus.v1"
+path = ${paths.dev}
+max_length = 0
+
+[training]
+accumulate_gradient = 3
+dev_corpus = "corpora.dev"
+train_corpus = "corpora.train"
+
+[training.optimizer]
+@optimizers = "Adam.v1"
+
+[training.optimizer.learn_rate]
+@schedules = "warmup_linear.v1"
+warmup_steps = 250
+total_steps = 20000
+initial_rate = 5e-5
+
+[training.batcher]
+@batchers = "spacy.batch_by_padded.v1"
+discard_oversize = true
+size = 2000
+buffer = 256
+
+[initialize]
+vectors = null

data/share/spaCy/gpu/textcat_multilabel_efficiency.conf

@@ -0,0 +1,73 @@
+# This is an auto-generated partial config. To use it with 'spacy train'
+# you can run spacy init fill-config to auto-fill all default settings:
+# python -m spacy init fill-config ./base_config.cfg ./config.cfg
+[paths]
+train = null
+dev = null
+
+[system]
+gpu_allocator = "pytorch"
+
+[nlp]
+lang = "en"
+pipeline = ["transformer","textcat_multilabel"]
+batch_size = 128
+
+[components]
+
+[components.transformer]
+factory = "transformer"
+
+[components.transformer.model]
+@architectures = "spacy-transformers.TransformerModel.v1"
+name = "roberta-base"
+tokenizer_config = {"use_fast": true}
+
+[components.transformer.model.get_spans]
+@span_getters = "spacy-transformers.strided_spans.v1"
+window = 128
+stride = 96
+
+[components.textcat_multilabel]
+factory = "textcat_multilabel"
+
+[components.textcat_multilabel.model]
+@architectures = "spacy.TextCatBOW.v1"
+exclusive_classes = true
+ngram_size = 1
+no_output_layer = false
+
+[corpora]
+
+[corpora.train]
+@readers = "spacy.Corpus.v1"
+path = ${paths.train}
+max_length = 500
+
+[corpora.dev]
+@readers = "spacy.Corpus.v1"
+path = ${paths.dev}
+max_length = 0
+
+[training]
+accumulate_gradient = 3
+dev_corpus = "corpora.dev"
+train_corpus = "corpora.train"
+
+[training.optimizer]
+@optimizers = "Adam.v1"
+
+[training.optimizer.learn_rate]
+@schedules = "warmup_linear.v1"
+warmup_steps = 250
+total_steps = 20000
+initial_rate = 5e-5
+
+[training.batcher]
+@batchers = "spacy.batch_by_padded.v1"
+discard_oversize = true
+size = 2000
+buffer = 256
+
+[initialize]
+vectors = null

data/test/rbbt/vector/model/test_spaCy.rb

@@ -23,18 +23,23 @@ class TestSpaCyModel < Test::Unit::TestCase
       good = tsv.select("Recommended IND" => '1')
       bad = tsv.select("Recommended IND" => '0')
 
-      gsize = 2000
-      bsize = 500
+      gsize = 200
+      bsize = 50
       good.keys[0..gsize-1].each do |text|
         next if text.nil? || text.empty?
-        model.add text, '1'
+        model.add text, 'good'
       end
 
       bad.keys[0..bsize-1].each do |text|
-        model.add text, '0'
+        model.add text, 'bad'
       end
 
-      model.cross_validation
+      model.cross_validation 1
+
+      model = VectorModel.new dir
+
+      assert Misc.counts(model.eval_list(good.keys[0..50]))['good'] > 40
+      assert Misc.counts(model.eval_list(bad.keys[0..50]))['bad'] > 40
     end
 
     def test_svm_spacy

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-dm
 version: !ruby/object:Gem::Version
-  version: 1.1.56
+  version: 1.1.57
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2021-12-20 00:00:00.000000000 Z
+date: 2022-05-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util
@@ -116,8 +116,12 @@ files:
 - share/R/heatmap.3.R
 - share/spaCy/cpu/textcat_accuracy.conf
 - share/spaCy/cpu/textcat_efficiency.conf
+- share/spaCy/cpu/textcat_multilabel_accuracy.conf
+- share/spaCy/cpu/textcat_multilabel_efficiency.conf
 - share/spaCy/gpu/textcat_accuracy.conf
 - share/spaCy/gpu/textcat_efficiency.conf
+- share/spaCy/gpu/textcat_multilabel_accuracy.conf
+- share/spaCy/gpu/textcat_multilabel_efficiency.conf
 - test/rbbt/matrix/test_barcode.rb
 - test/rbbt/network/test_paths.rb
 - test/rbbt/statistics/test_fdr.rb