torch-rb 0.2.6 → 0.3.3

data/lib/torch/native/parser.rb

@@ -83,6 +83,8 @@ module Torch
                 else
                   v.is_a?(Integer)
                 end
+              when "int?"
+                v.is_a?(Integer) || v.nil?
               when "float"
                 v.is_a?(Numeric)
               when /int\[.*\]/
@@ -126,9 +128,11 @@ module Torch
         end
 
         func = candidates.first
+        args = func.args.map { |a| final_values[a[:name]] }
+        args << TensorOptions.new.dtype(6) if func.tensor_options
         {
           name: func.cpp_name,
-          args: func.args.map { |a| final_values[a[:name]] }
+          args: args
         }
       end
     end

data/lib/torch/nn/conv2d.rb

@@ -18,7 +18,6 @@ module Torch
         F.conv2d(input, @weight, @bias, @stride, @padding, @dilation, @groups)
       end
 
-      # TODO add more parameters
       def extra_inspect
         s = String.new("%{in_channels}, %{out_channels}, kernel_size: %{kernel_size}, stride: %{stride}")
         s += ", padding: %{padding}" if @padding != [0] * @padding.size

data/lib/torch/nn/functional.rb

@@ -373,7 +373,8 @@ module Torch
             end
 
           # weight and input swapped
-          Torch.embedding_bag(weight, input, offsets, scale_grad_by_freq, mode_enum, sparse, per_sample_weights)
+          ret, _, _, _ = Torch.embedding_bag(weight, input, offsets, scale_grad_by_freq, mode_enum, sparse, per_sample_weights)
+          ret
         end
 
         # distance functions
@@ -426,6 +427,9 @@ module Torch
         end
 
         def mse_loss(input, target, reduction: "mean")
+          if target.size != input.size
+            warn "Using a target size (#{target.size}) that is different to the input size (#{input.size}). This will likely lead to incorrect results due to broadcasting. Please ensure they have the same size."
+          end
           NN.mse_loss(input, target, reduction)
         end
 

data/lib/torch/nn/module.rb

@@ -286,8 +286,11 @@ module Torch
         str % vars
       end
 
+      # used for format
+      # remove tensors for performance
+      # so we can skip call to inspect
       def dict
-        instance_variables.map { |k| [k[1..-1].to_sym, instance_variable_get(k)] }.to_h
+        instance_variables.reject { |k| instance_variable_get(k).is_a?(Tensor) }.map { |k| [k[1..-1].to_sym, instance_variable_get(k)] }.to_h
       end
     end
   end

data/lib/torch/optim/optimizer.rb

@@ -32,9 +32,11 @@ module Torch
       end
 
       def state_dict
+        raise NotImplementedYet
+
         pack_group = lambda do |group|
-          packed = group.select { |k, _| k != :params }.to_h
-          packed[:params] = group[:params].map { |p| p.object_id }
+          packed = group.select { |k, _| k != :params }.map { |k, v| [k.to_s, v] }.to_h
+          packed["params"] = group[:params].map { |p| p.object_id }
           packed
         end
 
@@ -42,8 +44,8 @@ module Torch
         packed_state = @state.map { |k, v| [k.is_a?(Tensor) ? k.object_id : k, v] }.to_h
 
         {
-          state: packed_state,
-          param_groups: param_groups
+          "state" => packed_state,
+          "param_groups" => param_groups
         }
       end
 

data/lib/torch/tensor.rb

@@ -1,6 +1,7 @@
 module Torch
   class Tensor
     include Comparable
+    include Enumerable
     include Inspector
 
     alias_method :requires_grad?, :requires_grad
@@ -25,6 +26,14 @@ module Torch
       inspect
     end
 
+    def each
+      return enum_for(:each) unless block_given?
+
+      size(0).times do |i|
+        yield self[i]
+      end
+    end
+
     # TODO make more performant
     def to_a
       arr = _flat_data
@@ -38,10 +47,15 @@ module Torch
       end
     end
 
-    # TODO support dtype
-    def to(device, non_blocking: false, copy: false)
+    def to(device = nil, dtype: nil, non_blocking: false, copy: false)
+      device ||= self.device
       device = Device.new(device) if device.is_a?(String)
-      _to(device, _dtype, non_blocking, copy)
+
+      dtype ||= self.dtype
+      enum = DTYPE_TO_ENUM[dtype]
+      raise Error, "Unknown type: #{dtype}" unless enum
+
+      _to(device, enum, non_blocking, copy)
     end
 
     def cpu
@@ -89,8 +103,9 @@ module Torch
       Torch.empty(0, dtype: dtype)
     end
 
-    def backward(gradient = nil)
-      _backward(gradient)
+    def backward(gradient = nil, retain_graph: nil, create_graph: false)
+      retain_graph = create_graph if retain_graph.nil?
+      _backward(gradient, retain_graph, create_graph)
     end
 
     # TODO read directly from memory
@@ -153,6 +168,18 @@ module Torch
       neg
     end
 
+    def &(other)
+      logical_and(other)
+    end
+
+    def |(other)
+      logical_or(other)
+    end
+
+    def ^(other)
+      logical_xor(other)
+    end
+
     # TODO better compare?
     def <=>(other)
       item <=> other
@@ -161,49 +188,15 @@ module Torch
     # based on python_variable_indexing.cpp and
     # https://pytorch.org/cppdocs/notes/tensor_indexing.html
     def [](*indexes)
-      result = self
-      dim = 0
-      indexes.each do |index|
-        if index.is_a?(Numeric)
-          result = result._select_int(dim, index)
-        elsif index.is_a?(Range)
-          finish = index.end
-          finish += 1 unless index.exclude_end?
-          result = result._slice_tensor(dim, index.begin, finish, 1)
-          dim += 1
-        elsif index.is_a?(Tensor)
-          result = result.index([index])
-        elsif index.nil?
-          result = result.unsqueeze(dim)
-          dim += 1
-        elsif index == true
-          result = result.unsqueeze(dim)
-          # TODO handle false
-        else
-          raise Error, "Unsupported index type: #{index.class.name}"
-        end
-      end
-      result
+      _index(tensor_indexes(indexes))
     end
 
     # based on python_variable_indexing.cpp and
     # https://pytorch.org/cppdocs/notes/tensor_indexing.html
-    def []=(index, value)
+    def []=(*indexes, value)
       raise ArgumentError, "Tensor does not support deleting items" if value.nil?
-
       value = Torch.tensor(value, dtype: dtype) unless value.is_a?(Tensor)
-
-      if index.is_a?(Numeric)
-        copy_to(_select_int(0, index), value)
-      elsif index.is_a?(Range)
-        finish = index.end
-        finish += 1 unless index.exclude_end?
-        copy_to(_slice_tensor(0, index.begin, finish, 1), value)
-      elsif index.is_a?(Tensor)
-        index_put!([index], value)
-      else
-        raise Error, "Unsupported index type: #{index.class.name}"
-      end
+      _index_put_custom(tensor_indexes(indexes), value)
     end
 
     # native functions that need manually defined
@@ -217,13 +210,13 @@ module Torch
       end
     end
 
-    # native functions overlap, so need to handle manually
+    # parser can't handle overlap, so need to handle manually
     def random!(*args)
       case args.size
       when 1
         _random__to(*args)
       when 2
-        _random__from_to(*args)
+        _random__from(*args)
       else
         _random_(*args)
       end
@@ -236,8 +229,29 @@ module Torch
 
     private
 
-    def copy_to(dst, src)
-      dst.copy!(src)
+    def tensor_indexes(indexes)
+      indexes.map do |index|
+        case index
+        when Integer
+          TensorIndex.integer(index)
+        when Range
+          finish = index.end
+          if finish == -1 && !index.exclude_end?
+            finish = nil
+          else
+            finish += 1 unless index.exclude_end?
+          end
+          TensorIndex.slice(index.begin, finish)
+        when Tensor
+          TensorIndex.tensor(index)
+        when nil
+          TensorIndex.none
+        when true, false
+          TensorIndex.boolean(index)
+        else
+          raise Error, "Unsupported index type: #{index.class.name}"
+        end
+      end
     end
   end
 end

data/lib/torch/utils/data.rb

@@ -0,0 +1,23 @@
+module Torch
+  module Utils
+    module Data
+      class << self
+        def random_split(dataset, lengths)
+          if lengths.sum != dataset.length
+            raise ArgumentError, "Sum of input lengths does not equal the length of the input dataset!"
+          end
+
+          indices = Torch.randperm(lengths.sum).to_a
+          _accumulate(lengths).zip(lengths).map { |offset, length| Subset.new(dataset, indices[(offset - length)...offset]) }
+        end
+
+        private
+
+        def _accumulate(iterable)
+          sum = 0
+          iterable.map { |x| sum += x }
+        end
+      end
+    end
+  end
+end

data/lib/torch/utils/data/data_loader.rb

@@ -6,10 +6,22 @@ module Torch
 
         attr_reader :dataset
 
-        def initialize(dataset, batch_size: 1, shuffle: false)
+        def initialize(dataset, batch_size: 1, shuffle: false, collate_fn: nil)
           @dataset = dataset
           @batch_size = batch_size
           @shuffle = shuffle
+
+          @batch_sampler = nil
+
+          if collate_fn.nil?
+            if auto_collation?
+              collate_fn = method(:default_collate)
+            else
+              collate_fn = method(:default_convert)
+            end
+          end
+
+          @collate_fn = collate_fn
         end
 
         def each
@@ -25,8 +37,8 @@ module Torch
             end
 
           indexes.each_slice(@batch_size) do |idx|
-            batch = idx.map { |i| @dataset[i] }
-            yield collate(batch)
+            # TODO improve performance
+            yield @collate_fn.call(idx.map { |i| @dataset[i] })
           end
         end
 
@@ -36,7 +48,7 @@ module Torch
 
         private
 
-        def collate(batch)
+        def default_convert(batch)
           elem = batch[0]
           case elem
           when Tensor
@@ -44,11 +56,15 @@ module Torch
           when Integer
             Torch.tensor(batch)
           when Array
-            batch.transpose.map { |v| collate(v) }
+            batch.transpose.map { |v| default_convert(v) }
           else
-            raise NotImpelmentYet
+            raise NotImplementedYet
           end
         end
+
+        def auto_collation?
+          !@batch_sampler.nil?
+        end
       end
     end
   end

data/lib/torch/utils/data/subset.rb

@@ -0,0 +1,25 @@
+module Torch
+  module Utils
+    module Data
+      class Subset < Dataset
+        def initialize(dataset, indices)
+          @dataset = dataset
+          @indices = indices
+        end
+
+        def [](idx)
+          @dataset[@indices[idx]]
+        end
+
+        def length
+          @indices.length
+        end
+        alias_method :size, :length
+
+        def to_a
+          @indices.map { |i| @dataset[i] }
+        end
+      end
+    end
+  end
+end

data/lib/torch/version.rb

@@ -1,3 +1,3 @@
 module Torch
-  VERSION = "0.2.6"
+  VERSION = "0.3.3"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: torch-rb
 version: !ruby/object:Gem::Version
-  version: 0.2.6
+  version: 0.3.3
 platform: ruby
 authors:
 - Andrew Kane
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-06-29 00:00:00.000000000 Z
+date: 2020-08-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rice
@@ -259,8 +259,10 @@ files:
 - lib/torch/optim/rprop.rb
 - lib/torch/optim/sgd.rb
 - lib/torch/tensor.rb
+- lib/torch/utils/data.rb
 - lib/torch/utils/data/data_loader.rb
 - lib/torch/utils/data/dataset.rb
+- lib/torch/utils/data/subset.rb
 - lib/torch/utils/data/tensor_dataset.rb
 - lib/torch/version.rb
 homepage: https://github.com/ankane/torch.rb