ruby-dnn 0.13.2 → 0.13.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d6410781490b2e9ce5ca1370bcca4905d4476bce2d49db64dda5f6b4a933f89f
-  data.tar.gz: e7ab72ce5b64a85c116ecbf81798aa6abd6f3356ca1a5e8512b2f0e6090b01c7
+  metadata.gz: cd3d1988c50c875fa619ab6c525fd604c8f264731c965f31725dd80ca8b17d72
+  data.tar.gz: 7149096f7222c5a4d42bfa69121a99c5df97ca054ae90de7f7c7fc9a294e7c53
 SHA512:
-  metadata.gz: a08c6f35faccf3b18e61f04535fbf67ac6c961f57a52a75c82be573269cb1d0dd5e8d2d9d215a3c311919c84ceb82c94c0c78db04de2b872fa98078e475596e7
-  data.tar.gz: 221da0acff79523f23e881d5a53d511348162840bd79d27221d2ec65e0c1c508efc2090b79646711856c20cd4e407aa8a8d08a51650587083a2af1284cc92abf
+  metadata.gz: 4ba83b8274994a9b90be17231791d1935de015dc2963ec4cc47f98f12782c8f4ac342aa220492c4e43d4a5d7780dae2011ba1b6fc967a23538b22318890bbb5a
+  data.tar.gz: 5374ea0e296ba510ce13d602daa4ffbc31757800ba9a83003f4795f195e4917511935d1bd7279ba8ce7204e78e73c7d2f918789452f329177fbd80eee981df3c

data/Rakefile

@@ -8,15 +8,23 @@ Rake::TestTask.new(:test) do |t|
   t.test_files = FileList["test/*_test.rb"]
 end
 
-task :build_dataset_loader do
+task :build_cifar_loader do
   sh "cd ext/cifar_loader; ruby extconf.rb; make"
 end
 
-task :build_image_io do
+task :build_rb_stb_image do
   sh "cd ext/rb_stb_image; ruby extconf.rb; make"
 end
 
-task :default => [:test, :build_dataset_loader, :build_image_io]
+task :clean_cifar_loader do
+  sh "cd ext/cifar_loader; make clean; unlink Makefile"
+end
+
+task :clean_rb_stb_image do
+  sh "cd ext/rb_stb_image; make clean; unlink Makefile"
+end
+
+task :default => [:test, :build_cifar_loader, :build_rb_stb_image]
 
 task :doc do
   src_list = Dir["lib/dnn/core/*.rb"]

data/lib/dnn/core/embedding.rb

@@ -28,7 +28,7 @@ module DNN
 
       def build
         @built = true
-        @weight = Param.new(Xumo::SFloat.new(@input_length), 0)
+        @weight = Param.new(Xumo::SFloat.new(@input_length), Xumo::SFloat[0])
         @weight_initializer.init_param(self, @weight)
         @weight_regularizer.param = @weight if @weight_regularizer
       end

data/lib/dnn/core/layers.rb

@@ -168,8 +168,8 @@ module DNN
         @bias_initializer = bias_initializer
         @weight_regularizer = weight_regularizer
         @bias_regularizer = bias_regularizer
-        @weight = Param.new(nil, 0)
-        @bias = use_bias ? Param.new(nil, 0) : nil
+        @weight = Param.new(nil, Xumo::SFloat[0])
+        @bias = use_bias ? Param.new(nil, Xumo::SFloat[0]) : nil
       end
 
       def regularizers

data/lib/dnn/core/losses.rb

@@ -17,7 +17,7 @@ module DNN
         end
         loss_value = forward(y, t)
         loss_value += regularizers_forward(layers) if layers
-        loss_value
+        loss_value.is_a?(Float) ? Xumo::SFloat[loss_value] : loss_value
       end
 
       def forward(y, t)

data/lib/dnn/core/models.rb

@@ -82,12 +82,14 @@ module DNN
       # @param [Numo::SFloat] x Input training data.
       # @param [Numo::SFloat] y Output training data.
       # @param [Integer] epochs Number of training.
+      # @param [Integer] initial_epoch Initial epoch.
       # @param [Integer] batch_size Batch size used for one training.
       # @param [Array | NilClass] test If you to test the model for every 1 epoch,
       #                                specify [x_test, y_test]. Don't test to the model, specify nil.
       # @param [Boolean] verbose Set true to display the log. If false is set, the log is not displayed.
       def train(x, y, epochs,
                 batch_size: 1,
+                initial_epoch: 1,
                 test: nil,
                 verbose: true)
         raise DNN_Error.new("The model is not optimizer setup complete.") unless @optimizer
@@ -95,7 +97,7 @@ module DNN
         check_xy_type(x, y)
         iter = Iterator.new(x, y)
         num_train_datas = x.is_a?(Array) ? x[0].shape[0] : x.shape[0]
-        (1..epochs).each do |epoch|
+        (initial_epoch..epochs).each do |epoch|
           call_callbacks(:before_epoch, epoch)
           puts "【 epoch #{epoch}/#{epochs} 】" if verbose
           iter.foreach(batch_size) do |x_batch, y_batch, index|
@@ -136,6 +138,7 @@ module DNN
       # Setup the model before use this method.
       # @param [Numo::SFloat] x Input training data.
       # @param [Numo::SFloat] y Output training data.
+      # @param [Integer] batch_size Batch size used for one test.
       # @return [Float | Numo::SFloat] Return loss value in the form of Float or Numo::SFloat.
       def train_on_batch(x, y)
         raise DNN_Error.new("The model is not optimizer setup complete.") unless @optimizer
@@ -146,7 +149,7 @@ module DNN
         loss_value = @loss_func.loss(x, y, layers)
         dy = @loss_func.backward(x, y)
         backward(dy)
-        @optimizer.update(layers.uniq)
+        @optimizer.update(layers)
         @loss_func.regularizers_backward(layers)
         call_callbacks(:after_train_on_batch, loss_value)
         loss_value
@@ -155,6 +158,7 @@ module DNN
       # Evaluate model and get accuracy of test data.
       # @param [Numo::SFloat] x Input test data.
       # @param [Numo::SFloat] y Output test data.
+      # @param [Integer] batch_size Batch size used for one test.
       # @return [Array] Returns the test data accuracy and mean loss in the form [accuracy, mean_loss].
       def accuracy(x, y, batch_size: 100)
         check_xy_type(x, y)
@@ -186,18 +190,21 @@ module DNN
         [correct, loss_value]
       end
 
-      private def evaluate(y, t)
-        if y.shape[1..-1] == [1]
+      # Implement the process to evaluate this model.
+      # @param [Numo::SFloat] x Input test data.
+      # @param [Numo::SFloat] y Output test data.
+      private def evaluate(x, y)
+        if x.shape[1..-1] == [1]
           correct = 0
-          y.shape[0].times do |i|
+          x.shape[0].times do |i|
             if @loss_func.is_a?(Losses::SigmoidCrossEntropy)
-              correct += 1 if (y[i, 0] < 0 && t[i, 0] < 0.5) || (y[i, 0] >= 0 && t[i, 0] >= 0.5)
+              correct += 1 if (x[i, 0] < 0 && y[i, 0] < 0.5) || (x[i, 0] >= 0 && y[i, 0] >= 0.5)
             else
-              correct += 1 if (y[i, 0] < 0 && t[i, 0] < 0) || (y[i, 0] >= 0 && t[i, 0] >= 0)
+              correct += 1 if (x[i, 0] < 0 && y[i, 0] < 0) || (x[i, 0] >= 0 && y[i, 0] >= 0)
             end
           end
         else
-          correct = y.max_index(axis: 1).eq(t.max_index(axis: 1)).count
+          correct = x.max_index(axis: 1).eq(y.max_index(axis: 1)).count
         end
         correct
       end
@@ -271,7 +278,7 @@ module DNN
           end
         end
         get_layers.(@last_link)
-        @layers_cache = layers
+        @layers_cache = layers.uniq
       end
 
       # Get the all has param layers.
@@ -316,8 +323,8 @@ module DNN
       end
 
       def naming
-        layers.uniq.each do |layer|
-          id = layers.uniq.select { |l| l.is_a?(layer.class) }.index(layer)
+        layers.each do |layer|
+          id = layers.select { |l| l.is_a?(layer.class) }.index(layer)
           class_name = layer.class.name.split("::").last
           layer.name = "#{class_name}_#{id}".to_sym unless layer.name
           if layer.is_a?(Layers::HasParamLayer)

data/lib/dnn/core/normalizations.rb

@@ -22,8 +22,8 @@ module DNN
 
       def build(input_shape)
         super
-        @gamma = Param.new(Xumo::SFloat.ones(*output_shape), 0)
-        @beta = Param.new(Xumo::SFloat.zeros(*output_shape), 0)
+        @gamma = Param.new(Xumo::SFloat.ones(*output_shape), Xumo::SFloat[0])
+        @beta = Param.new(Xumo::SFloat.zeros(*output_shape), Xumo::SFloat[0])
         @running_mean = Param.new(Xumo::SFloat.zeros(*output_shape))
         @running_var = Param.new(Xumo::SFloat.zeros(*output_shape))
       end

data/lib/dnn/core/optimizers.rb

@@ -38,7 +38,7 @@ module DNN
         clip_grads(target_params) if @clip_norm
         update_params(target_params)
         target_params.each do |param|
-          param.grad = Xumo::SFloat.zeros(*param.data.shape)
+          param.grad = Xumo::SFloat[0]
         end
       end
 
@@ -58,7 +58,7 @@ module DNN
       end
 
       private def clip_grads(params)
-        norm = Math.sqrt(params.reduce(0) { |sum, param| sum + (param.grad == 0 ? 0 : (param.grad ** 2).sum) })
+        norm = Math.sqrt(params.reduce(0) { |total, param| total + (param.grad ** 2).sum })
         return if norm <= @clip_norm
         rate = @clip_norm / (norm + 1e-7)
         params.each do |param|

data/lib/dnn/core/rnn_layers.rb

@@ -33,7 +33,7 @@ module DNN
         @return_sequences = return_sequences
         @layers = []
         @hidden = Param.new
-        @recurrent_weight = Param.new(nil, 0)
+        @recurrent_weight = Param.new(nil, Xumo::SFloat[0])
         @recurrent_weight_initializer = recurrent_weight_initializer
         @recurrent_weight_regularizer = recurrent_weight_regularizer
       end

data/lib/dnn/core/savers.rb

@@ -21,7 +21,7 @@ module DNN
       end
 
       def set_all_params_data(params_data)
-        all_params = @model.has_param_layers.uniq.map { |layer|
+        all_params = @model.has_param_layers.map { |layer|
           layer.get_params.values
         }.flatten
         all_params.each do |param|
@@ -91,7 +91,7 @@ module DNN
       end
 
       def get_all_params_data
-        all_params = @model.has_param_layers.uniq.map { |layer|
+        all_params = @model.has_param_layers.map { |layer|
           layer.get_params.values
         }.flatten
         all_params.to_h { |param| [param.name, param.data] }

data/lib/dnn/version.rb

@@ -1,3 +1,3 @@
 module DNN
-  VERSION = "0.13.2"
+  VERSION = "0.13.3"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ruby-dnn
 version: !ruby/object:Gem::Version
-  version: 0.13.2
+  version: 0.13.3
 platform: ruby
 authors:
 - unagiootoro
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2019-09-28 00:00:00.000000000 Z
+date: 2019-10-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: numo-narray