From e4b3b72ebfd2b02df35bdcfddae6e507632a0d2c Mon Sep 17 00:00:00 2001
From: geekun <youjiangkun@huawei.com>
Date: Tue, 23 Jun 2020 16:29:22 +0800
Subject: [PATCH] fix infer value bug

---
 mindspore/common/tensor.py           | 8 ++++++++
 mindspore/ops/operations/math_ops.py | 6 ++++--
 2 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/mindspore/common/tensor.py b/mindspore/common/tensor.py
index 0a631b954fd..ddd7cbfabc0 100644
--- a/mindspore/common/tensor.py
+++ b/mindspore/common/tensor.py
@@ -22,6 +22,10 @@ from . import dtype as mstype
 from ._register_for_tensor import tensor_operator_registry
 
 __all__ = ['Tensor', 'MetaTensor']
+np_types = (np.int8, np.int16, np.int32, np.int64,
+            np.uint8, np.uint16, np.uint32, np.uint64, np.float16,
+            np.float32, np.float64, np.bool_)
+
 
 
 class Tensor(Tensor_):
@@ -54,6 +58,10 @@ class Tensor(Tensor_):
     """
 
     def __init__(self, input_data, dtype=None):
+        # If input data is numpy number, convert it to np array
+        if isinstance(input_data, np_types):
+            input_data = np.array(input_data)
+
         # If input_data is tuple/list/numpy.ndarray, it's support in check_type method.
         check_type('tensor input_data', input_data, (Tensor_, float, int))
         if dtype is not None:
diff --git a/mindspore/ops/operations/math_ops.py b/mindspore/ops/operations/math_ops.py
index d992d078ace..d5a83f1208b 100644
--- a/mindspore/ops/operations/math_ops.py
+++ b/mindspore/ops/operations/math_ops.py
@@ -830,7 +830,8 @@ class Neg(PrimitiveWithInfer):
     def infer_value(self, input_x):
         if input_x is not None:
             input_x = input_x.asnumpy()
-            return Tensor(-input_x)
+            out = np.array(-input_x, input_x.dtype)
+            return Tensor(out)
 
         return None
 
@@ -1609,7 +1610,8 @@ class Div(_MathBinaryOp):
         if x is not None and y is not None:
             x = x.asnumpy()
             y = y.asnumpy()
-            return Tensor(x / y)
+            out = np.array(x / y, x.dtype)
+            return Tensor(out)
         return None