publish pytorch examples

Deepak-Rai-1027 · Nov 30, 2018 · 28a3fe6 · 28a3fe6
1 parent b0c9ca0
commit 28a3fe6
Show file tree

Hide file tree

Showing 6 changed files with 1,351 additions and 5 deletions.
diff --git a/README.md b/README.md
@@ -208,7 +208,7 @@ Suppose `x` shape was `(3, 4, 5)`, then `y` has shape ...
 - keras, tensorflow.layers, mxnet and gluon: `(3, 20)`
 - pytorch: no such function
 
-<!-- TODO examples for depth-to-space and pixel shuffle? transpose vs permute? -->
+<!-- TODO examples for depth-to-space and pixel shuffle? transpose vs permute? torch.repeat is numpy.tile -->
 
 ## Supported frameworks
 

diff --git a/docs/pytorch-examples.html b/docs/pytorch-examples.html
diff --git a/einops/_backends.py b/einops/_backends.py
@@ -84,6 +84,9 @@ def reduce(self, x, operation, axes):
     def stack_on_zeroth_dimension(self, tensors: list):
         raise NotImplementedError()
 
+    def tile(self, x, repeats):
+        raise NotImplementedError()
+
     def is_float_type(self, x):
         # some backends (torch) can't compute average for non-floating types.
         # Decided to drop average for all backends if type is not floating
@@ -137,6 +140,9 @@ def arange(self, start, stop):
     def stack_on_zeroth_dimension(self, tensors: list):
         return self.np.stack(tensors)
 
+    def tile(self, x, repeats):
+        return self.np.tile(x, repeats)
+
     def is_float_type(self, x):
         return x.dtype in ('float16', 'float32', 'float64', 'float128')
 
@@ -170,6 +176,9 @@ def arange(self, start, stop):
     def stack_on_zeroth_dimension(self, tensors: list):
         return self.mx.nd.stack(*tensors)
 
+    def tile(self, x, repeats):
+        return self.mx.nd.tile(x, repeats)
+
     def is_float_type(self, x):
         return 'float' in str(x.dtype)
 
@@ -225,6 +234,9 @@ def arange(self, start, stop):
     def stack_on_zeroth_dimension(self, tensors: list):
         return self.mx.symbol.stack(*tensors)
 
+    def tile(self, x, repeats):
+        return self.mx.symbol.tile(x, repeats)
+
     def is_float_type(self, x):
         return 'float' in str(x.infer_type()[1][0])
 
@@ -274,6 +286,9 @@ def transpose(self, x, axes):
     def stack_on_zeroth_dimension(self, tensors: list):
         return self.torch.stack(tensors)
 
+    def tile(self, x, repeats):
+        return x.repeat(repeats)
+
     def is_float_type(self, x):
         return x.dtype in [self.torch.float16, self.torch.float32, self.torch.float64]
 
@@ -304,6 +319,9 @@ def arange(self, start, stop):
     def stack_on_zeroth_dimension(self, tensors: list):
         return self.cupy.stack(tensors)
 
+    def tile(self, x, repeats):
+        return self.cupy.tile(x, repeats)
+
     def is_float_type(self, x):
         return x.dtype in ('float16', 'float32', 'float64', 'float128')
 
@@ -338,6 +356,9 @@ def reduce(self, x, operation, axes):
     def stack_on_zeroth_dimension(self, tensors: list):
         return self.chainer.functions.stack(tensors)
 
+    def tile(self, x, repeats):
+        return self.chainer.functions.tile(x, repeats)
+
     def is_float_type(self, x):
         return x.dtype in ('float16', 'float32', 'float64', 'float128')
 
@@ -394,6 +415,9 @@ def transpose(self, x, axes):
     def stack_on_zeroth_dimension(self, tensors: list):
         return self.tf.stack(tensors)
 
+    def tile(self, x, repeats):
+        return self.tf.tile(x, repeats)
+
     def is_float_type(self, x):
         return x.dtype in ('float16', 'float32', 'float64', 'float128')
 
@@ -436,6 +460,9 @@ def transpose(self, x, axes):
     def stack_on_zeroth_dimension(self, tensors: list):
         return self.K.stack(tensors)
 
+    def tile(self, x, repeats):
+        return self.K.tile(x, repeats)
+
     def is_float_type(self, x):
         return 'float' in self.K.dtype(x)
 

diff --git a/einops/einops.py b/einops/einops.py
@@ -448,7 +448,7 @@ def rearrange(tensor, pattern, **axes_lengths):
     :return: tensor of the same type as input. If possible, a view to the original tensor is returned.
 
     When composing axes, C-order enumeration used (consecutive elements have different last axis)
-    More examples and explanations can be found in the einops guide.
+    More source_examples and explanations can be found in the einops guide.
     """
     if isinstance(tensor, list):
         if len(tensor) == 0:

diff --git a/einops/layers/__init__.py b/einops/layers/__init__.py
@@ -12,7 +12,7 @@ class RearrangeMixin:
     :param pattern: str, rearrangement pattern
     :param axes_lengths: any additional specification of dimensions
 
-    See einops.rearrange for examples.
+    See einops.rearrange for source_examples.
     """
 
     def __init__(self, pattern, **axes_lengths):
@@ -50,7 +50,7 @@ class ReduceMixin:
     :param reduction: one of available reductions ('min', 'max', 'sum', 'mean', 'prod'), case-sensitive
     :param axes_lengths: any additional specification of dimensions
 
-    See einops.reduce for examples.
+    See einops.reduce for source_examples.
     """
 
     def __init__(self, pattern, reduction, **axes_lengths):

diff --git a/tests/test_ops.py b/tests/test_ops.py
@@ -382,7 +382,7 @@ def unet_like_1d(x, usual_convolution):
              ]
 
     for backend in imp_op_backends:
-        print('testing examples for ', backend.framework_name)
+        print('testing source_examples for ', backend.framework_name)
         for test in tests:
             x = numpy.arange(10 * 20 * 30 * 40).reshape([10, 20, 30, 40])
             result1 = test(x)
@@ -585,3 +585,40 @@ def new_way(input, num_classes, num_anchors, anchors, stride_h, stride_w):
                       stride_h=stride_h, stride_w=stride_w, anchors=anchors)
     result1 = result1.reshape(result2.shape)
     assert torch.allclose(result1, result2)
+
+
+def test_tiling_imperatives():
+    for backend in imp_op_backends:
+        print('Tiling tests for ', backend.framework_name)
+        input = numpy.arange(2 * 3 * 5, dtype='int64').reshape([2, 1, 3, 1, 5])
+        test_cases = [
+            (1, 1, 1, 1, 1),
+            (1, 2, 1, 3, 1),
+            (3, 1, 1, 4, 1),
+        ]
+        for repeats in test_cases:
+            expected = numpy.tile(input, repeats)
+            converted = backend.from_numpy(input)
+            repeated = backend.tile(converted, repeats)
+            result = backend.to_numpy(repeated)
+            assert numpy.array_equal(result, expected)
+
+
+def test_tiling_symbolic():
+    for backend in sym_op_backends:
+        print('Tiling tests for ', backend.framework_name)
+        input = numpy.arange(2 * 3 * 5, dtype='int64').reshape([2, 1, 3, 1, 5])
+        test_cases = [
+            (1, 1, 1, 1, 1),
+            (1, 2, 1, 3, 1),
+            (3, 1, 1, 4, 1),
+        ]
+        for repeats in test_cases:
+            expected = numpy.tile(input, repeats)
+            sym = backend.create_symbol(input.shape)
+            result = backend.eval_symbol(backend.tile(sym, repeats), [[sym, input]])
+            assert numpy.array_equal(result, expected)
+
+            sym = backend.create_symbol([None] * len(input.shape))
+            result = backend.eval_symbol(backend.tile(sym, repeats), [[sym, input]])
+            assert numpy.array_equal(result, expected)