Merge pull request #4350 from vilyaair/typecodes

Prefer data type objects over character codes in the NumPy

chainer/function.py

@@ -24,7 +24,7 @@ def no_backprop_mode():
     that calling :func:`~chainer.Variable.backward` on ``y`` has no effect on
     the gradients of ``x``.
 
-    >>> x = chainer.Variable(np.array([1,], 'f'))
+    >>> x = chainer.Variable(np.array([1,], np.float32))
     >>> with chainer.no_backprop_mode():
     ...     y = x + 1
     >>> y.backward()
@@ -53,7 +53,7 @@ def force_backprop_mode():
     :func:`~chainer.Variable.backward` on ``y`` will compute and accumulate the
     gradients of the variables in the graph, in this case only ``x``.
 
-    >>> x = chainer.Variable(np.array([1,], 'f'))
+    >>> x = chainer.Variable(np.array([1,], np.float32))
     >>> with chainer.no_backprop_mode():
     ...     with chainer.force_backprop_mode():
     ...         y = x + 1

chainer/function_hook.py

@@ -55,7 +55,7 @@ class FunctionHook(object):
         ...     return F.exp(self.l(x1))
         >>> model1 = Model()
         >>> model2 = Model()
-        >>> x = chainer.Variable(np.zeros((1, 10), 'f'))
+        >>> x = chainer.Variable(np.zeros((1, 10), np.float32))
         >>> with chainer.function_hooks.TimerHook() as m:
         ...   _ = model1(x)
         ...   y = model2(x)

chainer/function_hooks/debug_print.py

@@ -34,7 +34,7 @@ class PrintHook(function_hook.FunctionHook):
 
         >>> from chainer import function_hooks
         >>> l = L.Linear(10, 10)
-        >>> x = chainer.Variable(np.zeros((1, 10), 'f'))
+        >>> x = chainer.Variable(np.zeros((1, 10), np.float32))
         >>> with chainer.function_hooks.PrintHook():
         ...     y = l(x)
         ...     z = F.sum(y)

chainer/functions/activation/clipped_relu.py

@@ -93,7 +93,7 @@ def clipped_relu(x, z=20.0):
 
     .. admonition:: Example
 
-        >>> x = np.random.uniform(-100, 100, (10, 20)).astype('f')
+        >>> x = np.random.uniform(-100, 100, (10, 20)).astype(np.float32)
         >>> z = 10.0
         >>> np.any(x < 0)
         True

chainer/functions/activation/crelu.py

@@ -70,7 +70,7 @@ def crelu(x, axis=1):
 
     .. admonition:: Example
 
-        >>> x = np.array([[-1, 0], [2, -3]], 'f')
+        >>> x = np.array([[-1, 0], [2, -3]], np.float32)
         >>> x
         array([[-1.,  0.],
                [ 2., -3.]], dtype=float32)

chainer/functions/activation/elu.py

@@ -103,7 +103,7 @@ def elu(x, alpha=1.0):
 
     .. admonition:: Example
 
-        >>> x = np.array([[-1, 0], [2, -3]], 'f')
+        >>> x = np.array([[-1, 0], [2, -3]], np.float32)
         >>> x
         array([[-1.,  0.],
                [ 2., -3.]], dtype=float32)

chainer/functions/activation/leaky_relu.py

@@ -110,7 +110,7 @@ def leaky_relu(x, slope=0.2):
 
     .. admonition:: Example
 
-        >>> x = np.array([[-1, 0], [2, -3], [-2, 1]], 'f')
+        >>> x = np.array([[-1, 0], [2, -3], [-2, 1]], np.float32)
         >>> x
         array([[-1.,  0.],
                [ 2., -3.],

chainer/functions/activation/log_softmax.py

@@ -154,7 +154,7 @@ def log_softmax(x):
 
     .. admonition:: Example
 
-        >>> x = np.array([[0, 1, 2], [0, 2, 4]], 'f')
+        >>> x = np.array([[0, 1, 2], [0, 2, 4]], np.float32)
         >>> x
         array([[0., 1., 2.],
                [0., 2., 4.]], dtype=float32)

chainer/functions/activation/lstm.py

@@ -332,9 +332,9 @@ def lstm(c_prev, x):
         Most typical preparation of ``x`` is:
 
         >>> n_units = 100
-        >>> y = chainer.Variable(np.zeros((1, n_units), 'f'))
-        >>> h = chainer.Variable(np.zeros((1, n_units), 'f'))
-        >>> c = chainer.Variable(np.zeros((1, n_units), 'f'))
+        >>> y = chainer.Variable(np.zeros((1, n_units), np.float32))
+        >>> h = chainer.Variable(np.zeros((1, n_units), np.float32))
+        >>> c = chainer.Variable(np.zeros((1, n_units), np.float32))
         >>> model = chainer.Chain()
         >>> with model.init_scope():
         ...   model.w = L.Linear(n_units, 4 * n_units)

chainer/functions/activation/maxout.py

@@ -38,9 +38,9 @@ def maxout(x, pool_size, axis=1):
         combination with a Linear link.
 
         >>> in_size, out_size, pool_size = 10, 10, 10
-        >>> bias = np.arange(out_size * pool_size).astype('f')
+        >>> bias = np.arange(out_size * pool_size).astype(np.float32)
         >>> l = L.Linear(in_size, out_size * pool_size, initial_bias=bias)
-        >>> x = np.zeros((1, in_size), 'f')  # prepare data
+        >>> x = np.zeros((1, in_size), np.float32)  # prepare data
         >>> x = l(x)
         >>> y = F.maxout(x, pool_size)
         >>> x.shape

chainer/functions/activation/relu.py

@@ -165,7 +165,7 @@ def relu(x):
 
     .. admonition:: Example
 
-        >>> x = np.array([[-1, 0], [2, -3], [-2, 1]], 'f')
+        >>> x = np.array([[-1, 0], [2, -3], [-2, 1]], np.float32)
         >>> np.any(x < 0)
         True
         >>> y = F.relu(x)

chainer/functions/activation/sigmoid.py

@@ -109,7 +109,7 @@ def sigmoid(x):
 
         It maps the input values into the range of :math:`[0, 1]`.
 
-        >>> x = np.arange(-2, 3, 2).astype('f')
+        >>> x = np.arange(-2, 3, 2).astype(np.float32)
         >>> x
         array([-2.,  0.,  2.], dtype=float32)
         >>> F.sigmoid(x)

chainer/functions/activation/slstm.py

@@ -253,10 +253,10 @@ def slstm(c_prev1, c_prev2, x1, x2):
         Most typical preparation of ``x1``, ``x2`` is:
 
         >>> n_units = 100
-        >>> h1 = chainer.Variable(np.zeros((1, n_units), 'f'))
-        >>> h2 = chainer.Variable(np.zeros((1, n_units), 'f'))
-        >>> c1 = chainer.Variable(np.zeros((1, n_units), 'f'))
-        >>> c2 = chainer.Variable(np.zeros((1, n_units), 'f'))
+        >>> h1 = chainer.Variable(np.zeros((1, n_units), np.float32))
+        >>> h2 = chainer.Variable(np.zeros((1, n_units), np.float32))
+        >>> c1 = chainer.Variable(np.zeros((1, n_units), np.float32))
+        >>> c2 = chainer.Variable(np.zeros((1, n_units), np.float32))
         >>> model1 = chainer.Chain()
         >>> with model1.init_scope():
         ...   model1.w = L.Linear(n_units, 4 * n_units)

chainer/functions/activation/softmax.py

@@ -145,7 +145,7 @@ def softmax(x, axis=1):
 
     .. admonition:: Example
 
-        >>> x = np.array([[0, 1, 2], [0, 2, 4]], 'f')
+        >>> x = np.array([[0, 1, 2], [0, 2, 4]], np.float32)
         >>> x
         array([[0., 1., 2.],
                [0., 2., 4.]], dtype=float32)

chainer/functions/activation/softplus.py

@@ -109,7 +109,7 @@ def softplus(x, beta=1.0):
 
     .. admonition:: Example
 
-        >>> x = np.arange(-2, 3, 2).astype('f')
+        >>> x = np.arange(-2, 3, 2).astype(np.float32)
         >>> x
         array([-2.,  0.,  2.], dtype=float32)
         >>> F.softplus(x, beta=1.0).data

chainer/functions/activation/tanh.py

@@ -104,7 +104,7 @@ def tanh(x):
 
     .. admonition:: Example
 
-        >>> x = np.arange(-1, 4, 2).astype('f')
+        >>> x = np.arange(-1, 4, 2).astype(np.float32)
         >>> x
         array([-1.,  1.,  3.], dtype=float32)
         >>> F.tanh(x).data

chainer/functions/activation/tree_lstm.py

@@ -269,11 +269,11 @@ def tree_lstm(*inputs):
         ...   model.w = L.Linear(10, 5 * 10)
         ...   model.v1 = L.Linear(10, 5 * 10)
         ...   model.v2 = L.Linear(10, 5 * 10)
-        >>> y = np.random.uniform(-1, 1, (4, 10)).astype('f')
-        >>> h1 = np.random.uniform(-1, 1, (4, 10)).astype('f')
-        >>> h2 = np.random.uniform(-1, 1, (4, 10)).astype('f')
-        >>> c1 = np.random.uniform(-1, 1, (4, 10)).astype('f')
-        >>> c2 = np.random.uniform(-1, 1, (4, 10)).astype('f')
+        >>> y = np.random.uniform(-1, 1, (4, 10)).astype(np.float32)
+        >>> h1 = np.random.uniform(-1, 1, (4, 10)).astype(np.float32)
+        >>> h2 = np.random.uniform(-1, 1, (4, 10)).astype(np.float32)
+        >>> c1 = np.random.uniform(-1, 1, (4, 10)).astype(np.float32)
+        >>> c2 = np.random.uniform(-1, 1, (4, 10)).astype(np.float32)
         >>> x = model.w(y) + model.v1(h1) + model.v2(h2)
         >>> c, h = F.tree_lstm(c1, c2, x)
 

chainer/functions/array/broadcast.py

@@ -64,11 +64,11 @@ def broadcast(*args):
 
     .. admonition:: Example
 
-        >>> x = np.random.uniform(0, 1, (3, 2)).astype('f')
+        >>> x = np.random.uniform(0, 1, (3, 2)).astype(np.float32)
         >>> y = F.broadcast(x)
         >>> np.all(x == y.data)
         True
-        >>> z = np.random.uniform(0, 1, (3, 2)).astype('f')
+        >>> z = np.random.uniform(0, 1, (3, 2)).astype(np.float32)
         >>> y, w = F.broadcast(x, z)
         >>> np.all(x == y.data) & np.all(z == w.data)
         True

chainer/functions/array/depth2space.py

@@ -67,7 +67,7 @@ def depth2space(X, r):
 
     .. admonition:: Example
 
-        >>> X = np.arange(24).reshape(1, 4, 2, 3).astype('f')
+        >>> X = np.arange(24).reshape(1, 4, 2, 3).astype(np.float32)
         >>> X.shape
         (1, 4, 2, 3)
         >>> X

chainer/functions/array/permutate.py

@@ -107,12 +107,12 @@ def permutate(x, indices, axis=0, inv=False):
 
     .. admonition:: Example
 
-        >>> x = np.arange(6).reshape((3, 2)).astype('f')
+        >>> x = np.arange(6).reshape((3, 2)).astype(np.float32)
         >>> x
         array([[0., 1.],
                [2., 3.],
                [4., 5.]], dtype=float32)
-        >>> indices = np.array([2, 0, 1], 'i')
+        >>> indices = np.array([2, 0, 1], np.int32)
         >>> y = F.permutate(x, indices)
         >>> y.data
         array([[4., 5.],
@@ -123,7 +123,7 @@ def permutate(x, indices, axis=0, inv=False):
         array([[2., 3.],
                [4., 5.],
                [0., 1.]], dtype=float32)
-        >>> indices = np.array([1, 0], 'i')
+        >>> indices = np.array([1, 0], np.int32)
         >>> y = F.permutate(x, indices, axis=1)
         >>> y.data
         array([[1., 0.],

chainer/functions/array/select_item.py

@@ -104,8 +104,8 @@ def select_item(x, t):
 
     .. admonition:: Example
 
-        >>> x = np.array([[0, 1, 2], [3, 4, 5]], 'f')
-        >>> t = np.array([0, 2], 'i')
+        >>> x = np.array([[0, 1, 2], [3, 4, 5]], np.float32)
+        >>> t = np.array([0, 2], np.int32)
         >>> y = F.select_item(x, t)
         >>> y.shape
         (2,)

chainer/functions/array/separate.py

@@ -58,7 +58,7 @@ def separate(x, axis=0):
 
     .. admonition:: Example
 
-        >>> x = np.arange(6).reshape((2, 3)).astype('f')
+        >>> x = np.arange(6).reshape((2, 3)).astype(np.float32)
         >>> x
         array([[0., 1., 2.],
                [3., 4., 5.]], dtype=float32)

chainer/functions/array/space2depth.py

@@ -66,7 +66,7 @@ def space2depth(X, r):
 
     .. admonition:: Example
 
-        >>> X = np.arange(24).reshape(1, 1, 4, 6).astype('f')
+        >>> X = np.arange(24).reshape(1, 1, 4, 6).astype(np.float32)
         >>> X.shape
         (1, 1, 4, 6)
         >>> X

chainer/functions/array/squeeze.py

@@ -80,7 +80,7 @@ def squeeze(x, axis=None):
 
     .. admonition:: Example
 
-        >>> x = np.array([[[[0, 1, 2]]], [[[3, 4, 5]]]], 'f')
+        >>> x = np.array([[[[0, 1, 2]]], [[[3, 4, 5]]]], np.float32)
         >>> x.shape
         (2, 1, 1, 3)
         >>> y = F.squeeze(x)

chainer/functions/array/swapaxes.py

@@ -40,7 +40,7 @@ def swapaxes(x, axis1, axis2):
 
     .. admonition:: Example
 
-        >>> x = np.array([[[0, 1, 2], [3, 4, 5]]], 'f')
+        >>> x = np.array([[[0, 1, 2], [3, 4, 5]]], np.float32)
         >>> x.shape
         (1, 2, 3)
         >>> y = F.swapaxes(x, axis1=0, axis2=1)

chainer/functions/array/transpose.py

@@ -45,7 +45,7 @@ def transpose(x, axes=None):
 
     .. admonition:: Example
 
-        >>> x = np.array([[[0, 1, 2], [3, 4, 5]]], 'f')
+        >>> x = np.array([[[0, 1, 2], [3, 4, 5]]], np.float32)
         >>> x.shape
         (1, 2, 3)
         >>> y = F.transpose(x)  # reverse the dimensions

chainer/functions/array/where.py

@@ -73,8 +73,8 @@ def where(condition, x, y):
         >>> cond
         array([[ True, False],
                [False,  True]])
-        >>> x = np.array([[1, 2], [3, 4]], 'f')
-        >>> y = np.zeros((2, 2), 'f')
+        >>> x = np.array([[1, 2], [3, 4]], np.float32)
+        >>> y = np.zeros((2, 2), np.float32)
         >>> F.where(cond, x, y).data
         array([[1., 0.],
                [0., 4.]], dtype=float32)

chainer/functions/connection/convolution_2d.py

@@ -660,13 +660,14 @@ def convolution_2d(x, W, b=None, stride=1, pad=0, cover_all=False, group=1,
         >>> h_i, w_i = 30, 40
         >>> h_k, w_k = 10, 10
         >>> h_p, w_p = 5, 5
-        >>> x = np.random.uniform(0, 1, (n, c_i, h_i, w_i)).astype('f')
+        >>> x = np.random.uniform(0, 1, (n, c_i, h_i, w_i)).astype(np.float32)
         >>> x.shape
         (10, 3, 30, 40)
-        >>> W = np.random.uniform(0, 1, (c_o, c_i, h_k, w_k)).astype('f')
+        >>> W = np.random.uniform(0, 1, (c_o, c_i, h_k, w_k)).\
+astype(np.float32)
         >>> W.shape
         (1, 3, 10, 10)
-        >>> b = np.random.uniform(0, 1, (c_o,)).astype('f')
+        >>> b = np.random.uniform(0, 1, (c_o,)).astype(np.float32)
         >>> b.shape
         (1,)
         >>> s_y, s_x = 5, 7

chainer/functions/connection/convolution_nd.py

@@ -410,13 +410,15 @@ def convolution_nd(x, W, b=None, stride=1, pad=0, cover_all=False):
         >>> d1, d2, d3 = 30, 40, 50
         >>> k1, k2, k3 = 10, 10, 10
         >>> p1, p2, p3 = 5, 5, 5
-        >>> x = np.random.uniform(0, 1, (n, c_i, d1, d2, d3)).astype('f')
+        >>> x = np.random.uniform(0, 1, (n, c_i, d1, d2, d3)).\
+astype(np.float32)
         >>> x.shape
         (10, 3, 30, 40, 50)
-        >>> W = np.random.uniform(0, 1, (c_o, c_i, k1, k2, k3)).astype('f')
+        >>> W = np.random.uniform(0, 1, (c_o, c_i, k1, k2, k3)).\
+astype(np.float32)
         >>> W.shape
         (1, 3, 10, 10, 10)
-        >>> b = np.random.uniform(0, 1, (c_o)).astype('f')
+        >>> b = np.random.uniform(0, 1, (c_o)).astype(np.float32)
         >>> b.shape
         (1,)
         >>> s1, s2, s3 = 2, 4, 6

chainer/functions/connection/deconvolution_2d.py

@@ -462,13 +462,14 @@ def deconvolution_2d(x, W, b=None, stride=1, pad=0, outsize=None, group=1,
         >>> h_i, w_i = 5, 10
         >>> h_k, w_k = 10, 10
         >>> h_p, w_p = 5, 5
-        >>> x = np.random.uniform(0, 1, (n, c_i, h_i, w_i)).astype('f')
+        >>> x = np.random.uniform(0, 1, (n, c_i, h_i, w_i)).astype(np.float32)
         >>> x.shape
         (10, 1, 5, 10)
-        >>> W = np.random.uniform(0, 1, (c_i, c_o, h_k, w_k)).astype('f')
+        >>> W = np.random.uniform(0, 1, (c_i, c_o, h_k, w_k)).\
+astype(np.float32)
         >>> W.shape
         (1, 3, 10, 10)
-        >>> b = np.random.uniform(0, 1, c_o).astype('f')
+        >>> b = np.random.uniform(0, 1, c_o).astype(np.float32)
         >>> b.shape
         (3,)
         >>> s_y, s_x = 5, 5

chainer/functions/connection/deconvolution_nd.py

@@ -385,13 +385,15 @@ def deconvolution_nd(x, W, b=None, stride=1, pad=0, outsize=None):
         >>> d1, d2, d3 = 5, 10, 15
         >>> k1, k2, k3 = 10, 10, 10
         >>> p1, p2, p3 = 5, 5, 5
-        >>> x = np.random.uniform(0, 1, (n, c_i, d1, d2, d3)).astype('f')
+        >>> x = np.random.uniform(0, 1, (n, c_i, d1, d2, d3)).\
+astype(np.float32)
         >>> x.shape
         (10, 3, 5, 10, 15)
-        >>> W = np.random.uniform(0, 1, (c_i, c_o, k1, k2, k3)).astype('f')
+        >>> W = np.random.uniform(0, 1, (c_i, c_o, k1, k2, k3)).\
+astype(np.float32)
         >>> W.shape
         (3, 1, 10, 10, 10)
-        >>> b = np.random.uniform(0, 1, (c_o)).astype('f')
+        >>> b = np.random.uniform(0, 1, (c_o)).astype(np.float32)
         >>> b.shape
         (1,)
         >>> s1, s2, s3 = 2, 4, 6
@@ -412,13 +414,15 @@ def deconvolution_nd(x, W, b=None, stride=1, pad=0, outsize=None):
         >>> d1, d2, d3 = 5, 10, 15
         >>> k1, k2, k3 = 10, 10, 10
         >>> p1, p2, p3 = 5, 5, 5
-        >>> x = np.random.uniform(0, 1, (n, c_i, d1, d2, d3)).astype('f')
+        >>> x = np.random.uniform(0, 1, (n, c_i, d1, d2, d3)).\
+astype(np.float32)
         >>> x.shape
         (10, 3, 5, 10, 15)
-        >>> W = np.random.uniform(0, 1, (c_i, c_o, k1, k2, k3)).astype('f')
+        >>> W = np.random.uniform(0, 1, (c_i, c_o, k1, k2, k3)).\
+astype(np.float32)
         >>> W.shape
         (3, 1, 10, 10, 10)
-        >>> b = np.random.uniform(0, 1, (c_o)).astype('f')
+        >>> b = np.random.uniform(0, 1, (c_o)).astype(np.float32)
         >>> b.shape
         (1,)
         >>> s1, s2, s3 = 2, 4, 6