informers 1.0.3 → 1.1.1

data/lib/informers/utils/math.rb

@@ -1,5 +1,75 @@
 module Informers
   module Utils
+    def self.interpolate_data(input, in_shape, out_shape, mode = "bilinear", align_corners = false)
+      in_channels, in_height, in_width = in_shape
+      out_height, out_width = out_shape
+
+      # TODO use mode and align_corners
+
+      # Output image dimensions
+      x_scale = out_width / in_width.to_f
+      y_scale = out_height / in_height.to_f
+
+      # Output image
+      out_img = Array.new(out_height * out_width * in_channels)
+
+      # Pre-calculate strides
+      in_stride = in_height * in_width
+      out_stride = out_height * out_width
+
+      out_height.times do |i|
+        out_width.times do |j|
+          # Calculate output offset
+          out_offset = i * out_width + j
+
+          # Calculate input pixel coordinates
+          x = (j + 0.5) / x_scale - 0.5
+          y = (i + 0.5) / y_scale - 0.5
+
+          # Calculate the four nearest input pixels
+          # We also check if the input pixel coordinates are within the image bounds
+          x1 = x.floor
+          y1 = y.floor
+          x2 = [x1 + 1, in_width - 1].min
+          y2 = [y1 + 1, in_height - 1].min
+
+          x1 = [x1, 0].max
+          y1 = [y1, 0].max
+
+          # Calculate the fractional distances between the input pixel and the four nearest pixels
+          s = x - x1
+          t = y - y1
+
+          # Perform bilinear interpolation
+          w1 = (1 - s) * (1 - t)
+          w2 = s * (1 - t)
+          w3 = (1 - s) * t
+          w4 = s * t
+
+          # Calculate the four nearest input pixel indices
+          y_stride = y1 * in_width
+          x_stride = y2 * in_width
+          idx1 = y_stride + x1
+          idx2 = y_stride + x2
+          idx3 = x_stride + x1
+          idx4 = x_stride + x2
+
+          in_channels.times do |k|
+            # Calculate channel offset
+            c_offset = k * in_stride
+
+            out_img[k * out_stride + out_offset] =
+              w1 * input[c_offset + idx1] +
+              w2 * input[c_offset + idx2] +
+              w3 * input[c_offset + idx3] +
+              w4 * input[c_offset + idx4]
+          end
+        end
+      end
+
+      out_img
+    end
+
     def self.softmax(arr)
       # Compute the maximum value in the array
       max_val = arr.max
@@ -17,6 +87,9 @@ module Informers
     end
 
     def self.sigmoid(arr)
+      if arr[0].is_a?(Array)
+        return arr.map { |a| sigmoid(a) }
+      end
       arr.map { |v| 1 / (1 + Math.exp(-v)) }
     end
 

data/lib/informers/utils/tensor.rb

@@ -22,5 +22,51 @@ module Informers
         row.map { |v| v / norm }
       end
     end
+
+    def self.stack(tensors, dim = 0)
+      tensors
+    end
+
+    def self.ones_like(tensor)
+      if tensor[0].is_a?(Array)
+        return tensor.map { |v| ones_like(v) }
+      end
+      tensor.map { |_| 1 }
+    end
+
+    def self.dims(tensor)
+      dims = []
+      while tensor.is_a?(Array)
+        dims << tensor.size
+        tensor = tensor[0]
+      end
+      dims
+    end
+
+    def self.interpolate(input, shape, mode = "bilinear", align_corners = false)
+      out_height, out_width = shape
+
+      # Input image dimensions
+      in_channels = dims(input)[-3] || 1
+      in_height = dims(input)[-2]
+      in_width = dims(input)[-1]
+
+      output = interpolate_data(
+        input.flatten,
+        [in_channels, in_height, in_width],
+        [out_height, out_width],
+        mode,
+        align_corners
+      )
+      reshape(output, [in_channels, out_height, out_width])
+    end
+
+    def self.reshape(arr, dims)
+      arr = arr.flatten
+      dims[1..-1].reverse_each do |dim|
+        arr = arr.each_slice(dim)
+      end
+      arr.to_a
+    end
   end
 end

data/lib/informers/version.rb

@@ -1,3 +1,3 @@
 module Informers
-  VERSION = "1.0.3"
+  VERSION = "1.1.1"
 end

data/lib/informers.rb

@@ -6,18 +6,24 @@ require "tokenizers"
 require "io/console"
 require "json"
 require "open-uri"
+require "open3"
 require "stringio"
 require "uri"
 
 # modules
+require_relative "informers/utils/audio"
 require_relative "informers/utils/core"
+require_relative "informers/utils/generation"
+require_relative "informers/utils/ffmpeg"
 require_relative "informers/utils/hub"
+require_relative "informers/utils/image"
 require_relative "informers/utils/math"
 require_relative "informers/utils/tensor"
 require_relative "informers/configs"
 require_relative "informers/env"
 require_relative "informers/model"
 require_relative "informers/models"
+require_relative "informers/processors"
 require_relative "informers/tokenizers"
 require_relative "informers/version"
 require_relative "informers/pipelines"

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: informers
 version: !ruby/object:Gem::Version
-  version: 1.0.3
+  version: 1.1.1
 platform: ruby
 authors:
 - Andrew Kane
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-08-29 00:00:00.000000000 Z
+date: 2024-10-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: onnxruntime
@@ -30,14 +30,14 @@ dependencies:
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 0.5.2
+        version: 0.5.3
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 0.5.2
+        version: 0.5.3
 description:
 email: andrew@ankane.org
 executables: []
@@ -53,9 +53,14 @@ files:
 - lib/informers/model.rb
 - lib/informers/models.rb
 - lib/informers/pipelines.rb
+- lib/informers/processors.rb
 - lib/informers/tokenizers.rb
+- lib/informers/utils/audio.rb
 - lib/informers/utils/core.rb
+- lib/informers/utils/ffmpeg.rb
+- lib/informers/utils/generation.rb
 - lib/informers/utils/hub.rb
+- lib/informers/utils/image.rb
 - lib/informers/utils/math.rb
 - lib/informers/utils/tensor.rb
 - lib/informers/version.rb
@@ -78,7 +83,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.5.11
+rubygems_version: 3.5.16
 signing_key:
 specification_version: 4
 summary: Fast transformer inference for Ruby