From 178b1b14594cd0c0a2cefd41b5390616d45ecd6c Mon Sep 17 00:00:00 2001
From: feifei-111 <wuzhanfei@baidu.com>
Date: Mon, 29 Aug 2022 10:15:57 +0800
Subject: [PATCH] [API] fix bug in to_tensor which returns err dtype (#45418)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

* DLTP-55757 [Bug] 【Paddle3D】pointpillars在paddle=dev上export model报错

* DLTP-55757 [Bug] 【Paddle3D】pointpillars在paddle=dev上export model报错

* fix default dtype

* coverage

* del no need cases
---
 .../dygraph_to_static/test_to_tensor.py       | 32 +++++++++++++++++--
 python/paddle/tensor/creation.py              | 22 +++++++++----
 2 files changed, 44 insertions(+), 10 deletions(-)

diff --git a/python/paddle/fluid/tests/unittests/dygraph_to_static/test_to_tensor.py b/python/paddle/fluid/tests/unittests/dygraph_to_static/test_to_tensor.py
index 27935e9a2f..63fce642f8 100644
--- a/python/paddle/fluid/tests/unittests/dygraph_to_static/test_to_tensor.py
+++ b/python/paddle/fluid/tests/unittests/dygraph_to_static/test_to_tensor.py
@@ -33,7 +33,7 @@ def case0(x):
 
 def case1(x):
     paddle.set_default_dtype("float64")
-    a = paddle.to_tensor([1.0, 2.0, 3.0], stop_gradient=False)
+    a = paddle.to_tensor([1, 2, 3], stop_gradient=False, dtype='float32')
 
     return a
 
@@ -68,13 +68,27 @@ def case4(x):
         place = paddle.CUDAPlace(0)
     else:
         place = paddle.CPUPlace()
-    a = paddle.to_tensor([1.0], place=place, dtype="float64")
-    b = paddle.to_tensor([2], place=place, stop_gradient=False, dtype="int64")
+    a = paddle.to_tensor([1], place=place)
+    b = paddle.to_tensor([2.1], place=place, stop_gradient=False, dtype="int64")
     c = paddle.to_tensor([a, b, [1]], dtype="float32")
 
     return c
 
 
+def case5(x):
+    paddle.set_default_dtype("float64")
+    a = paddle.to_tensor([1, 2])
+
+    return a
+
+
+def case6(x):
+    na = numpy.array([1, 2], dtype='int32')
+    a = paddle.to_tensor(na)
+
+    return a
+
+
 class TestToTensorReturnVal(unittest.TestCase):
 
     def test_to_tensor_badreturn(self):
@@ -111,6 +125,18 @@ class TestToTensorReturnVal(unittest.TestCase):
         self.assertTrue(a.stop_gradient == b.stop_gradient)
         self.assertTrue(a.place._equals(b.place))
 
+        a = paddle.jit.to_static(case5)(x)
+        b = case5(x)
+        self.assertTrue(a.dtype == b.dtype)
+        self.assertTrue(a.stop_gradient == b.stop_gradient)
+        self.assertTrue(a.place._equals(b.place))
+
+        a = paddle.jit.to_static(case6)(x)
+        b = case6(x)
+        self.assertTrue(a.dtype == b.dtype)
+        self.assertTrue(a.stop_gradient == b.stop_gradient)
+        self.assertTrue(a.place._equals(b.place))
+
 
 class TestStatic(unittest.TestCase):
 
diff --git a/python/paddle/tensor/creation.py b/python/paddle/tensor/creation.py
index 272c7f2ed8..b6291a9248 100644
--- a/python/paddle/tensor/creation.py
+++ b/python/paddle/tensor/creation.py
@@ -354,21 +354,29 @@ def _to_tensor_static(data, dtype=None, stop_gradient=None):
     if isinstance(data, Variable) and (dtype is None or dtype == data.dtype):
         output = data
     else:
+
+        if not isinstance(data, np.ndarray):
+            if np.isscalar(data) and not isinstance(data, str):
+                data = np.array([data])
+            elif isinstance(data, (list, tuple)):
+                data = np.array(data)
+
+            if isinstance(data,
+                          np.ndarray) and not dtype and data.dtype != 'object':
+                if data.dtype in ['float16', 'float32', 'float64']:
+                    data = data.astype(paddle.get_default_dtype())
+                elif data.dtype in ['int32']:
+                    data = data.astype('int64')
+
         if dtype:
             target_dtype = dtype
-        elif hasattr(data, 'dtype'):
+        elif hasattr(data, 'dtype') and data.dtype != 'object':
             target_dtype = data.dtype
         else:
             target_dtype = paddle.get_default_dtype()
 
         target_dtype = convert_dtype(target_dtype)
 
-        if not isinstance(data, np.ndarray):
-            if np.isscalar(data) and not isinstance(data, str):
-                data = np.array([data])
-            elif isinstance(data, (list, tuple)):
-                data = np.array(data)
-
         if isinstance(data, np.ndarray) and len(data.shape) > 0 and any(
                 isinstance(x, Variable) for x in data):
             if not all(
-- 
GitLab