Fix AttributeError: module 'numpy' has no attribute 'bool'. (#21165)

docs/python_docs/python/tutorials/packages/np/cheat-sheet.md

@@ -108,7 +108,7 @@ npx.load('my_arrays')
 # np.int64    # Signed 64-bit integer types
 # np.float32  # Standard double-precision floating point
 # np.complex  # Complex numbers represented by 128 floats
-# np.bool     # Boolean type storing TRUE and FALSE values
+# bool        # Boolean type storing TRUE and FALSE values (formerly `np.bool`)
 # np.object   # Python object type
 # np.string_  # Fixed-length string type
 # np.unicode_ # Fixed-length unicode type

python/mxnet/ndarray/ndarray.py

@@ -3078,7 +3078,7 @@ def indexing_key_expand_implicit_axes(key, shape):
     (slice(None, 2, None), None, 0, slice(None, None, None))
     >>> bool_array = np.array([[True, False, True, False],
                                [False, True, False, True],
-                               [True, False, True, False]], dtype=np.bool)
+                               [True, False, True, False]], dtype=bool)
     >>> indexing_key_expand_implicit_axes(np.s_[bool_array, None, 0:2], shape)
     (array([0, 0, 1, 1, 2, 2], dtype=int64), array([0, 2, 1, 3, 0, 2], dtype=int64), None, slice(None, 2, None))
     """
@@ -3109,7 +3109,7 @@ def indexing_key_expand_implicit_axes(key, shape):
                 )
             ell_idx = i
         else:
-            # convert primitive type boolean value to mx.np.bool type
+            # convert primitive type boolean value to bool (formerly mx.np.bool) type
             # otherwise will be treated as 1/0
             if isinstance(idx, bool):
                 idx = array(idx, dtype=np.bool_)

python/mxnet/ndarray/numpy/_op.py

@@ -391,7 +391,7 @@ def full(shape, fill_value, dtype=None, order='C', device=None, out=None):  # py
         device = str(device)
     if isinstance(fill_value, bool):
         fill_value = int(fill_value)
-        dtype = _np.bool if dtype is None else dtype
+        dtype = bool if dtype is None else dtype
     elif isinstance(fill_value, numeric_types):
         if dtype is None or dtype is float:
             dtype = dtype_from_number(fill_value)
@@ -526,7 +526,7 @@ def empty_like(prototype, dtype=None, order='C', subok=False, shape=None): # pyl
     dtype_list = {_np.float16: 'float16', _np.float32: 'float32', _np.float64: 'float64',
                   float: 'float64', _np.int8: 'int8', _np.int16: 'int16', _np.int32: 'int32',
                   _np.int64: 'int64', int:'int64', _np.uint8: 'uint8', _np.uint16: 'uint16',
-                  _np.uint32: 'uint32', _np.uint64: 'uint64', _np.bool: 'bool',
+                  _np.uint32: 'uint32', _np.uint64: 'uint64', bool: 'bool',
                   _np.bool_: 'bool_', bool: 'bool', None: 'None'}
     if order != 'C':
         raise NotImplementedError("Only support C-order at this moment")
@@ -8969,7 +8969,7 @@ def isposinf(x, out=None, **kwargs):
     >>> np.isposinf(np.array([-np.inf, 0., np.inf]))
     array([False, False,  True])
     >>> x = np.array([-np.inf, 0., np.inf])
-    >>> y = np.array([True, True, True], dtype=np.bool)
+    >>> y = np.array([True, True, True], dtype=bool)
     >>> np.isposinf(x, y)
     array([False, False,  True])
     >>> y
@@ -9015,7 +9015,7 @@ def isneginf(x, out=None, **kwargs):
     >>> np.isneginf(np.array([-np.inf, 0., np.inf]))
     array([ True, False, False])
     >>> x = np.array([-np.inf, 0., np.inf])
-    >>> y = np.array([True, True, True], dtype=np.bool)
+    >>> y = np.array([True, True, True], dtype=bool)
     >>> np.isneginf(x, y)
     array([ True, False, False])
     >>> y
@@ -9070,7 +9070,7 @@ def isfinite(x, out=None, **kwargs):
     >>> np.isfinite(np.array([np.log(-1.),1.,np.log(0)]))
     array([False,  True, False])
     >>> x = np.array([-np.inf, 0., np.inf])
-    >>> y = np.array([True, True, True], dtype=np.bool)
+    >>> y = np.array([True, True, True], dtype=bool)
     >>> np.isfinite(x, y)
     array([False,  True, False])
     >>> y

python/mxnet/numpy/multiarray.py

@@ -192,7 +192,7 @@ def _as_mx_np_array(object, device=None, zero_copy=False):
         return from_numpy(object, zero_copy and object.flags['C_CONTIGUOUS'])
     elif isinstance(object, (integer_types, numeric_types)):
         return object
-    elif isinstance(object, (_np.bool_, _np.bool)):
+    elif isinstance(object, (_np.bool_, bool)):
         return array(object, dtype=_np.bool_, device=device)
     elif isinstance(object, (list, tuple)):
         tmp = [_as_mx_np_array(arr, device=device, zero_copy=zero_copy) for arr in object]
@@ -782,7 +782,7 @@ def __getitem__(self, key):
         ndim = self.ndim  # pylint: disable=redefined-outer-name
         shape = self.shape  # pylint: disable=redefined-outer-name
         if isinstance(key, bool): # otherwise will be treated as 0 and 1
-            key = array(key, dtype=_np.bool, device=self.device)
+            key = array(key, dtype=bool, device=self.device)
         if isinstance(key, list):
             try:
                 new_key = _np.array(key)
@@ -966,13 +966,13 @@ def __setitem__(self, key, value):
         if isinstance(value, NDArray) and not isinstance(value, ndarray):
             raise TypeError('Cannot assign mx.nd.NDArray to mxnet.numpy.ndarray')
         if isinstance(key, bool): # otherwise will be treated as 0 and 1
-            key = array(key, dtype=_np.bool)
+            key = array(key, dtype=bool)
 
         # Handle single boolean assign of matching dimensionality and size first for higher speed
         # If the boolean array is mixed with other idices, it is instead expanded into (multiple)
         # integer array indices and will be handled by advanced assign.
         # Come before the check self.dim == 0 as it also handle the 0-dim case.
-        if isinstance(key, ndarray) and key.dtype == _np.bool:
+        if isinstance(key, ndarray) and key.dtype == bool:
             return self._set_np_boolean_indexing(key, value)
 
         # handle basic and advanced indexing
@@ -12123,7 +12123,7 @@ def isposinf(x, out=None, **kwargs):
     >>> np.isposinf(np.array([-np.inf, 0., np.inf]))
     array([False, False,  True])
     >>> x = np.array([-np.inf, 0., np.inf])
-    >>> y = np.array([True, True, True], dtype=np.bool)
+    >>> y = np.array([True, True, True], dtype=bool)
     >>> np.isposinf(x, y)
     array([False, False,  True])
     >>> y
@@ -12169,7 +12169,7 @@ def isneginf(x, out=None, **kwargs):
     >>> np.isneginf(np.array([-np.inf, 0., np.inf]))
     array([ True, False, False])
     >>> x = np.array([-np.inf, 0., np.inf])
-    >>> y = np.array([True, True, True], dtype=np.bool)
+    >>> y = np.array([True, True, True], dtype=bool)
     >>> np.isneginf(x, y)
     array([ True, False, False])
     >>> y
@@ -12224,7 +12224,7 @@ def isfinite(x, out=None, **kwargs):
     >>> np.isfinite(np.array([np.log(-1.),1.,np.log(0)]))
     array([False,  True, False])
     >>> x = np.array([-np.inf, 0., np.inf])
-    >>> y = np.array([True, True, True], dtype=np.bool)
+    >>> y = np.array([True, True, True], dtype=bool)
     >>> np.isfinite(x, y)
     array([False,  True, False])
     >>> y

python/mxnet/numpy/utils.py

@@ -38,7 +38,7 @@
 int8 = onp.dtype(onp.int8)
 int64 = onp.dtype(onp.int64)
 bool_ = onp.dtype(onp.bool_)
-bool = onp.dtype(onp.bool)
+bool = onp.dtype(bool)
 int16 = onp.dtype(onp.int16)
 uint16 = onp.dtype(onp.uint16)
 uint32 = onp.dtype(onp.uint32)

python/mxnet/symbol/numpy/_symbol.py

@@ -1334,7 +1334,7 @@ def full(shape, fill_value, dtype=None, order='C', ctx=None, out=None):  # pylin
         return ret
     if isinstance(fill_value, bool):
         fill_value = int(fill_value)
-        dtype = _np.bool if dtype is None else dtype
+        dtype = bool if dtype is None else dtype
     return _npi.full(shape=shape, value=fill_value, ctx=ctx, dtype=dtype, out=out)
 
 

python/mxnet/test_utils.py

@@ -77,7 +77,7 @@ def default_rtols():
     return {np.dtype(np.float16): 1e-2,
             np.dtype(np.float32): 1e-4,
             np.dtype(np.float64): 1e-5,
-            np.dtype(np.bool): 0,
+            np.dtype(bool): 0,
             np.dtype(np.int8): 0,
             np.dtype(np.uint8): 0,
             np.dtype(np.int32): 0,
@@ -90,7 +90,7 @@ def default_atols():
     return {np.dtype(np.float16): 1e-1,
             np.dtype(np.float32): 1e-3,
             np.dtype(np.float64): 1e-20,
-            np.dtype(np.bool): 0,
+            np.dtype(bool): 0,
             np.dtype(np.int8): 0,
             np.dtype(np.uint8): 0,
             np.dtype(np.int32): 0,

tests/python/dnnl/subgraphs/test_matmul_subgraph.py

@@ -51,7 +51,7 @@ def forward(self, x, mask):
       value = mx.np.reshape(value, (-2, -2, self._num_heads, -1))
       scores = mx.npx.batch_dot(mx.np.swapaxes(query, 1, 2), mx.np.swapaxes(key, 1, 2),
                                 transpose_b=True)
-      mask = mx.np.expand_dims(mask, axis=1).astype(np.bool)
+      mask = mx.np.expand_dims(mask, axis=1).astype(bool)
       attn_weights = mx.npx.masked_softmax(scores, mask=mask, axis=-1, temperature=self._scale)
       attn_weights = mx.npx.dropout(attn_weights, p=0.1)
       context_vec = mx.npx.batch_dot(attn_weights,

tests/python/gpu/test_fusion.py

@@ -289,7 +289,7 @@ def forward(self, valid_length):
 
     foo = Foo()
     foo.hybridize(static_alloc=True)
-    out = foo(mx.np.ones((10,), ctx=mx.gpu(), dtype=np.bool))
+    out = foo(mx.np.ones((10,), ctx=mx.gpu(), dtype=bool))
     mx.npx.waitall()
 
 @use_np

tests/python/unittest/test_numpy_interoperability.py

@@ -1497,8 +1497,8 @@ def _add_workload_gcd():
 
 
 def _add_workload_bitwise_or():
-    OpArgMngr.add_workload('bitwise_or', np.array([False, False, True, True], dtype=np.bool),
-                           np.array([False, True, False, True], dtype=np.bool))
+    OpArgMngr.add_workload('bitwise_or', np.array([False, False, True, True], dtype=bool),
+                           np.array([False, True, False, True], dtype=bool))
     for dtype in [np.int8, np.int32, np.int64]:
         zeros = np.array([0], dtype=dtype)
         ones = np.array([-1], dtype=dtype)
@@ -1509,8 +1509,8 @@ def _add_workload_bitwise_or():
 
 
 def _add_workload_bitwise_and():
-    OpArgMngr.add_workload('bitwise_and', np.array([False, False, True, True], dtype=np.bool),
-                           np.array([False, True, False, True], dtype=np.bool))
+    OpArgMngr.add_workload('bitwise_and', np.array([False, False, True, True], dtype=bool),
+                           np.array([False, True, False, True], dtype=bool))
     for dtype in [np.int8, np.int32, np.int64]:
         zeros = np.array([0], dtype=dtype)
         ones = np.array([-1], dtype=dtype)
@@ -1521,8 +1521,8 @@ def _add_workload_bitwise_and():
 
 
 def _add_workload_bitwise_xor():
-    OpArgMngr.add_workload('bitwise_xor', np.array([False, False, True, True], dtype=np.bool),
-                           np.array([False, True, False, True], dtype=np.bool))
+    OpArgMngr.add_workload('bitwise_xor', np.array([False, False, True, True], dtype=bool),
+                           np.array([False, True, False, True], dtype=bool))
     for dtype in [np.int8, np.int32, np.int64]:
         zeros = np.array([0], dtype=dtype)
         ones = np.array([-1], dtype=dtype)
@@ -1834,11 +1834,11 @@ def _add_workload_floor(array_pool):
 def _add_workload_logical_not(array_pool):
     OpArgMngr.add_workload('logical_not', np.ones(10, dtype=np.int32))
     OpArgMngr.add_workload('logical_not', array_pool['4x1'])
-    OpArgMngr.add_workload('logical_not', np.array([True, False, True, False], dtype=np.bool))
+    OpArgMngr.add_workload('logical_not', np.array([True, False, True, False], dtype=bool))
 
 
 def _add_workload_bitwise_not():
-    OpArgMngr.add_workload('bitwise_not', np.array([True, False, True, False], dtype=np.bool))
+    OpArgMngr.add_workload('bitwise_not', np.array([True, False, True, False], dtype=bool))
     for dtype in [np.int8, np.int32, np.int64]:
         zeros = np.array([0], dtype=dtype)
         ones = np.array([-1], dtype=dtype)
@@ -1847,7 +1847,7 @@ def _add_workload_bitwise_not():
 
 
 def _add_workload_invert():
-    OpArgMngr.add_workload('invert', np.array([True, False, True, False], dtype=np.bool))
+    OpArgMngr.add_workload('invert', np.array([True, False, True, False], dtype=bool))
     for dtype in [np.int8, np.int32, np.int64]:
         zeros = np.array([0], dtype=dtype)
         ones = np.array([-1], dtype=dtype)
@@ -2045,19 +2045,19 @@ def _add_workload_less_equal(array_pool):
 
 def _add_workload_logical_and(array_pool):
     OpArgMngr.add_workload('logical_and', np.array([0, 1, 2, 4, 2], dtype=np.float32), np.array([-2, 5, 1, 4, 3], dtype=np.float32))
-    OpArgMngr.add_workload('logical_and', np.array([False, False, True, True], dtype=np.bool),
-                           np.array([False, True, False, True], dtype=np.bool))
+    OpArgMngr.add_workload('logical_and', np.array([False, False, True, True], dtype=bool),
+                           np.array([False, True, False, True], dtype=bool))
 
 def _add_workload_logical_or(array_pool):
-    OpArgMngr.add_workload('logical_or', np.array([0, 1, 2, 4, 2], dtype=np.bool), np.array([-2, 5, 1, 4, 3], dtype=np.bool))
-    OpArgMngr.add_workload('logical_or', np.array([False, False, True, True], dtype=np.bool),
-                           np.array([False, True, False, True], dtype=np.bool))
+    OpArgMngr.add_workload('logical_or', np.array([0, 1, 2, 4, 2], dtype=bool), np.array([-2, 5, 1, 4, 3], dtype=bool))
+    OpArgMngr.add_workload('logical_or', np.array([False, False, True, True], dtype=bool),
+                           np.array([False, True, False, True], dtype=bool))
 
 
 def _add_workload_logical_xor(array_pool):
     OpArgMngr.add_workload('logical_xor', np.array([0, 1, 2, 4, 2], dtype=np.float32), np.array([-2, 5, 1, 4, 3], dtype=np.float32))
-    OpArgMngr.add_workload('logical_xor', np.array([False, False, True, True], dtype=np.bool),
-                           np.array([False, True, False, True], dtype=np.bool))
+    OpArgMngr.add_workload('logical_xor', np.array([False, False, True, True], dtype=bool),
+                           np.array([False, True, False, True], dtype=bool))
 
 
 def _add_workload_where():
@@ -2910,7 +2910,7 @@ def _add_workload_select():
     condlist = np.array([[ True,  True,  True, False, False,
                            False, False, False, False, False],
                          [ False, False, False, False, False,
-                           False,  True,  True,  True, True]], dtype=np.bool)
+                           False,  True,  True,  True, True]], dtype=bool)
     choicelist = np.array([[ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9],
                            [ 0,  1,  4,  9, 16, 25, 36, 49, 64, 81]])
     OpArgMngr.add_workload('select', condlist, choicelist)

tests/python/unittest/test_numpy_ndarray.py

@@ -43,7 +43,7 @@ def test_np_empty():
         (np.float32, np.float32),
         (np.float64, np.float64),
         (np.bool_, np.bool_),
-        (np.bool, np.bool_),
+        (bool, np.bool_),
         ('int8', np.int8),
         ('int32', np.int32),
         ('float16', np.float16),
@@ -80,7 +80,7 @@ def test_np_empty():
 
 @use_np
 def test_np_array_creation():
-    dtypes = [_np.int8, _np.int32, _np.float16, _np.float32, _np.float64, _np.bool, _np.bool_,
+    dtypes = [_np.int8, _np.int32, _np.float16, _np.float32, _np.float64, bool, _np.bool_,
               'int8', 'int32', 'float16', 'float32', 'float64', 'bool', None]
     objects = [
         [],
@@ -154,7 +154,7 @@ def check_zero_array_creation(shape, dtype):
                 assert type(y[1]) == np.ndarray
 
     for shape in shapes:
-        for dtype in [_np.bool, bool, _np.bool, 'bool']:
+        for dtype in [bool, 'bool']:
             check_zero_array_creation(shape, dtype)
 
 
@@ -207,7 +207,7 @@ def check_ones_array_creation(shape, dtype):
                 assert type(y[1]) == np.ndarray
 
     for shape in shapes:
-        for dtype in [_np.bool, bool, _np.bool, 'bool']:
+        for dtype in [bool, 'bool']:
             check_ones_array_creation(shape, dtype)
 
 
@@ -575,7 +575,7 @@ def check_astype_equal(itype, otype, copy, expect_zero_copy=False, hybridize=Fal
             assert id(mx_ret) == id(mx_data)
             assert id(np_ret) == id(np_data)
 
-    dtypes = [np.int8, np.uint8, np.int32, np.float16, np.float32, np.float64, np.bool, np.bool_,
+    dtypes = [np.int8, np.uint8, np.int32, np.float16, np.float32, np.float64, bool, np.bool_,
               'int8', 'uint8', 'int32', 'float16', 'float32', 'float64', 'bool']
 
     for itype, otype in itertools.product(dtypes, dtypes):
@@ -1218,7 +1218,7 @@ def test_boolean_index_tuple():
                   [[4, 5],
                    [6, 7]]], dtype=np.int32)
     b = np.array([[False,True],
-                  [True,False]],dtype=np.bool)
+                  [True,False]],dtype=bool)
     _np_a = a.asnumpy()
     _np_b = b.asnumpy()
     assert same(a[:, b].asnumpy(), _np_a[:, _np_b])
@@ -1252,7 +1252,7 @@ def test_boolean_index_assign():
     # test boolean indexing assign
     shape = (3, 2, 3)
     mx_data = np.random.uniform(size=shape)
-    mx_mask = np.array([[False,True], [True,False], [True,False]],dtype=np.bool)
+    mx_mask = np.array([[False,True], [True,False], [True,False]],dtype=bool)
     np_data = mx_data.asnumpy()
     np_mask = mx_mask.asnumpy()
 
@@ -1271,7 +1271,7 @@ def test_boolean_index_assign():
     mx_data[mx_mask, :] = 3
     assert_almost_equal(mx_data.asnumpy(), np_data, rtol=1e-3, atol=1e-5, use_broadcast=False)
 
-    mx_mask = np.array([[False,True, True],[False, True,False]],dtype=np.bool)
+    mx_mask = np.array([[False,True, True],[False, True,False]],dtype=bool)
     np_mask = mx_mask.asnumpy()
 
     np_data[0, np_mask] = 5
@@ -1310,7 +1310,7 @@ def test_boolean_index_autograd():
 
 @use_np
 def test_np_get_dtype():
-    dtypes = [_np.int8, _np.int32, _np.float16, _np.float32, _np.float64, _np.bool, _np.bool_,
+    dtypes = [_np.int8, _np.int32, _np.float16, _np.float32, _np.float64, bool, _np.bool_,
               'int8', 'int32', 'float16', 'float32', 'float64', 'bool', None]
     objects = [
         [],