Change np_compat to np_shape

python/mxnet/gluon/block.py

@@ -543,7 +543,7 @@ def __call__(self, *args):
 
         for hook in self._forward_hooks.values():
             hook(self, args, out)
-        if _mx_np.is_np_compat():
+        if _mx_np.is_np_shape():
             _check_all_np_ndarrays(_flatten(out, "output")[0])
         return out
 

python/mxnet/gluon/parameter.py

@@ -31,7 +31,7 @@
 from ..context import Context, cpu
 from .. import autograd
 from .utils import _indent, _brief_print_list, shape_is_known
-from ..util import is_np_compat
+from ..util import is_np_shape
 
 # pylint: disable= invalid-name
 tensor_types = (symbol.Symbol, ndarray.NDArray)
@@ -163,7 +163,7 @@ def shape(self, new_shape):
         if self._shape is None:
             self._shape = new_shape
             return
-        unknown_dim_size = -1 if is_np_compat() else 0
+        unknown_dim_size = -1 if is_np_shape() else 0
         assert len(self._shape) == len(new_shape) and \
             all(j in (unknown_dim_size, i) for i, j in zip(new_shape, self._shape)), \
             "Expected shape %s is incompatible with given shape %s."%(
@@ -282,7 +282,7 @@ def _finish_deferred_init(self):
                 initializer.create(default_init)(
                     initializer.InitDesc(self.name, {'__init__': init}), data)
                 # TODO(junwu): use np random operators when available
-                if is_np_compat():
+                if is_np_shape():
                     data = data.as_np_ndarray()  # convert to np.ndarray
 
             self._init_impl(data, ctx)
@@ -309,7 +309,7 @@ def _init_grad(self):
         self._grad = [ndarray.zeros(shape=i.shape, dtype=i.dtype, ctx=i.context,
                                     stype=self._grad_stype) for i in self._data]
         # TODO(junwu): use np.zeros
-        if is_np_compat():
+        if is_np_shape():
             self._grad = [arr.as_np_ndarray() for arr in self._grad]
 
         autograd.mark_variables(self._check_and_get(self._data, list),
@@ -558,7 +558,7 @@ def var(self):
             self._var = symbol.var(self.name, shape=self.shape, dtype=self.dtype,
                                    lr_mult=self.lr_mult, wd_mult=self.wd_mult,
                                    init=self.init, stype=self._stype)
-            if is_np_compat():
+            if is_np_shape():
                 self._var = self._var.as_np_ndarray()
         return self._var
 

python/mxnet/gluon/utils.py

@@ -38,7 +38,7 @@ class requests_failed_to_import(object):
 import numpy as np
 
 from .. import ndarray
-from ..util import is_np_compat
+from ..util import is_np_shape
 
 
 def split_data(data, num_slice, batch_axis=0, even_split=True):
@@ -453,7 +453,7 @@ def shape_is_known(shape):
     """Check whether a shape is completely known w/ or w/o np semantics."""
     if shape is None:
         return False
-    unknown_dim_size = -1 if is_np_compat() else 0
+    unknown_dim_size = -1 if is_np_shape() else 0
     if len(shape) == 0:
         return unknown_dim_size == -1
     for dim_size in shape:

python/mxnet/ndarray/numpy/_op.py

@@ -20,7 +20,7 @@
 from __future__ import absolute_import
 import numpy as _np
 from ...base import numeric_types
-from ...util import _sanity_check_params, use_np_compat, set_module
+from ...util import _sanity_check_params, set_module
 from ...context import current_context
 from . import _internal as _npi
 
@@ -90,7 +90,6 @@ def ones(shape, dtype=None, **kwargs):
 
 
 #pylint: disable= too-many-arguments, no-member, protected-access
-@use_np_compat
 def _ufunc_helper(lhs, rhs, fn_array, fn_scalar, lfn_scalar, rfn_scalar=None, out=None):
     """ Helper function for element-wise operation.
     The function will perform numpy-like broadcasting if needed and call different functions.

python/mxnet/ndarray/register.py

@@ -25,7 +25,7 @@
 from ..ndarray_doc import _build_doc
 
 from ..base import mx_uint, check_call, _LIB, py_str, _init_op_module, _Null, _is_np_op  # pylint: disable=unused-import
-from ..util import use_np_compat  # pylint: disable=unused-import
+from ..util import use_np_shape  # pylint: disable=unused-import
 
 
 def _verify_all_np_ndarrays(op_name, func_name, args, out):
@@ -176,7 +176,7 @@ def _generate_ndarray_function_code(handle, op_name, func_name, signature_only=F
     if is_np_op:
         doc_str_idx = 2
         code.append("""
-@use_np_compat""")
+@use_np_shape""")
     if arr_name:
         code.append("""
 def %s(*%s, **kwargs):"""%(func_name, arr_name))

python/mxnet/numpy/__init__.py

@@ -26,6 +26,6 @@
 from . import _op
 from . import _register
 from ._op import *  # pylint: disable=wildcard-import
-from ..util import use_np_compat, set_np_compat, np_compat, is_np_compat
+from ..util import use_np_shape, set_np_shape, np_shape, is_np_shape
 
 __all__ = []

python/mxnet/numpy/multiarray.py

@@ -30,7 +30,7 @@
 from . import _op as _mx_np_op
 from ..base import check_call, _LIB, NDArrayHandle
 from ..base import mx_real_t, c_array_buf, mx_uint, numeric_types
-from ..util import _sanity_check_params, set_module, use_np_compat
+from ..util import _sanity_check_params, set_module, use_np_shape
 from ..context import current_context
 from ..ndarray import numpy as _mx_nd_np
 from ..ndarray.numpy import _internal as _npi
@@ -75,7 +75,7 @@ def _np_ndarray_cls(handle, writable=True, stype=0):
 
 
 @set_module('mxnet.numpy')  # pylint: disable=invalid-name
-@use_np_compat
+@use_np_shape
 class ndarray(NDArray):
     """An array object represents a multidimensional, homogeneous array of fixed-size items.
     An associated data-type object describes the format of each element in the array
@@ -1140,7 +1140,6 @@ def ndim(self):
         return len(self.shape)
 
     @property
-    @use_np_compat
     def size(self):
         """Number of elements in the array."""
         return super(ndarray, self).size
@@ -1150,7 +1149,6 @@ def tostype(self, stype):
 
 
 @set_module('mxnet.numpy')
-@use_np_compat
 def empty(shape, dtype=None, **kwargs):
     """Return a new array of given shape and type, without initializing entries.
 
@@ -1183,7 +1181,7 @@ def empty(shape, dtype=None, **kwargs):
 
 
 @set_module('mxnet.numpy')
-@use_np_compat
+@use_np_shape
 def array(object, dtype=None, **kwargs):
     """
     Create an array.

tests/python/unittest/test_numpy_gluon.py

@@ -44,7 +44,7 @@ def __init__(self):
         def hybrid_forward(self, F, x, w):
             return F.dot(x, w)
 
-    @np.use_np_compat
+    @np.use_np_shape
     class TestBlock2(gluon.HybridBlock):
         def __init__(self):
             super(TestBlock2, self).__init__()
@@ -62,7 +62,7 @@ def hybrid_forward(self, F, x, w):
 
 
 def test_optimizer_with_np_ndarrays():
-    @np.use_np_compat
+    @np.use_np_shape
     class LinearRegression(gluon.HybridBlock):
         def __init__(self, num_input_dim=-1, num_hidden_dim=100, num_output_dim=10):
             super(LinearRegression, self).__init__()
@@ -78,7 +78,7 @@ def hybrid_forward(self, F, x, w1, w2):
             y_pred = h_relu.dot(w2)  # equivalent to F.np.dot(h_relu, w2)
             return y_pred
 
-    @np.use_np_compat
+    @np.use_np_shape
     class TotalLoss(gluon.HybridBlock):
         def hybrid_forward(self, F, pred, label):
             return ((pred - label) ** 2).sum()  # equivalent to F.np.sum(F.np.square(pred - label))

tests/python/unittest/test_numpy_ndarray.py

@@ -47,7 +47,7 @@ def test_array_creation():
 @with_seed()
 def test_zeros():
     # test np.zeros in Gluon
-    @np.use_np_compat
+    @np.use_np_shape
     class TestZeros(HybridBlock):
         def __init__(self, shape, dtype=None):
             super(TestZeros, self).__init__()
@@ -57,13 +57,13 @@ def __init__(self, shape, dtype=None):
         def hybrid_forward(self, F, x, *args, **kwargs):
             return x + F.np.zeros(shape, dtype)
 
-    @np.use_np_compat
+    @np.use_np_shape
     class TestZerosOutputType(HybridBlock):
         def hybrid_forward(self, F, x, *args, **kwargs):
             return x, F.np.zeros(shape=())
 
     # test np.zeros in imperative
-    @np.use_np_compat
+    @np.use_np_shape
     def check_zero_array_creation(shape, dtype):
         np_out = _np.zeros(shape=shape, dtype=dtype)
         mx_out = np.zeros(shape=shape, dtype=dtype)
@@ -97,7 +97,7 @@ def check_zero_array_creation(shape, dtype):
 @with_seed()
 def test_ones():
     # test np.ones in Gluon
-    @np.use_np_compat
+    @np.use_np_shape
     class TestOnes(HybridBlock):
         def __init__(self, shape, dtype=None):
             super(TestOnes, self).__init__()
@@ -107,13 +107,13 @@ def __init__(self, shape, dtype=None):
         def hybrid_forward(self, F, x, *args, **kwargs):
             return x * F.np.ones(shape, dtype)
 
-    @np.use_np_compat
+    @np.use_np_shape
     class TestOnesOutputType(HybridBlock):
         def hybrid_forward(self, F, x, *args, **kwargs):
             return x, F.np.ones(shape=())
 
     # test np.ones in imperative
-    @np.use_np_compat
+    @np.use_np_shape
     def check_ones_array_creation(shape, dtype):
         np_out = _np.ones(shape=shape, dtype=dtype)
         mx_out = np.ones(shape=shape, dtype=dtype)
@@ -156,7 +156,7 @@ def test_ndarray_binary_element_wise_ops():
     def get_np_ret(x1, x2, op):
         return np_op_map[op](x1, x2)
 
-    @np.use_np_compat
+    @np.use_np_shape
     class TestBinaryElementWiseOp(HybridBlock):
         def __init__(self, op, scalar=None, reverse=False):
             super(TestBinaryElementWiseOp, self).__init__()
@@ -219,7 +219,7 @@ def hybrid_forward(self, F, x, *args):
                 print(self._op)
                 assert False
 
-    @np.use_np_compat
+    @np.use_np_shape
     def check_binary_op_result(shape1, shape2, op, dtype=None):
         if shape1 is None:
             mx_input1 = abs(_np.random.uniform()) + 1
@@ -289,7 +289,7 @@ def check_binary_op_result(shape1, shape2, op, dtype=None):
 
 @with_seed()
 def test_hybrid_block_multiple_outputs():
-    @np.use_np_compat
+    @np.use_np_shape
     class TestAllNumpyOutputs(HybridBlock):
         def hybrid_forward(self, F, x, *args, **kwargs):
             return F.npe.relu(x), F.np.sum(x)
@@ -309,7 +309,7 @@ def hybrid_forward(self, F, x, *args, **kwargs):
             assert type(out1) is expected_out_type
             assert type(out2) is expected_out_type
 
-    @np.use_np_compat
+    @np.use_np_shape
     class TestMixedTypeOutputsFailure(HybridBlock):
         def hybrid_forward(self, F, x, *args, **kwargs):
             return F.relu(x.as_classic_ndarray()), F.np.sum(x)

tests/python/unittest/test_numpy_op.py

@@ -88,7 +88,7 @@ def is_int(dtype):
 
 
 @with_seed()
-@np.use_np_compat
+@np.use_np_shape
 def test_np_dot():
     shapes = [
         ((3, 0), (0, 4)),
@@ -132,7 +132,7 @@ def test_np_dot():
 
 @with_seed()
 def test_np_mean():
-    @np.use_np_compat
+    @np.use_np_shape
     class TestMean(HybridBlock):
         def __init__(self, axis=None, dtype=None, keepdims=False):
             super(TestMean, self).__init__()
@@ -194,7 +194,7 @@ def is_int(dtype):
 
 
 @with_seed()
-@np.use_np_compat
+@np.use_np_shape
 def test_np_transpose():
     # TODO(junwu): Add more test cases
     data = mx.sym.var('a').as_np_ndarray()
@@ -224,7 +224,7 @@ def test_np_transpose():
 
 
 @with_seed()
-@np.use_np_compat
+@np.use_np_shape
 def test_relu():
     # TODO(junwu): Add more test cases
     data = mx.sym.var('data').as_np_ndarray()
@@ -240,7 +240,7 @@ def test_relu():
 
 
 @with_seed()
-@np.use_np_compat
+@np.use_np_shape
 def test_sigmoid():
     # TODO(junwu): Add more test cases
     data = mx.sym.var('data').as_np_ndarray()
@@ -256,7 +256,7 @@ def test_sigmoid():
 
 
 @with_seed()
-@np.use_np_compat
+@np.use_np_shape
 def test_np_reshape():
     # TODO(junwu): Add more test cases
     data = mx.sym.var('a').as_np_ndarray()
@@ -272,7 +272,7 @@ def test_np_reshape():
 
 
 @with_seed()
-@np.use_np_compat
+@np.use_np_shape
 def test_np_maximum():
     # TODO(junwu): Add more test cases
     x1, x2 = mx.sym.var('x1').as_np_ndarray(), mx.sym.var('x2').as_np_ndarray()
@@ -293,7 +293,7 @@ def check_maximum(x1, x2):
 
 
 @with_seed()
-@np.use_np_compat
+@np.use_np_shape
 def test_np_minimum():
     # TODO(junwu): Add more test cases
     x1, x2 = mx.sym.var('x1').as_np_ndarray(), mx.sym.var('x2').as_np_ndarray()