易用性提升69个API核对

paconvert/api_mapping.json

@@ -10505,8 +10505,7 @@
       "remove_duplicate"
     ],
     "kwargs_change": {
-      "recurse": "include_sublayers",
-      "remove_duplicate": ""
+      "recurse": "include_sublayers"
     },
     "min_input_args": 0
   },
@@ -10523,10 +10522,12 @@
       "prefix",
       "remove_duplicate"
     ],
-    "unsupport_args": [
-      "memo",
-      "remove_duplicate"
-    ],
+    "kwargs_change": {
+      "memo": "layers_set"
+    },
+    "paddle_default_kwargs": {
+      "include_self": "True"
+    },
     "min_input_args": 0
   },
   "torch.nn.Module.named_parameters": {
@@ -10539,8 +10540,7 @@
       "remove_duplicate"
     ],
     "kwargs_change": {
-      "recurse": "include_sublayers",
-      "remove_duplicate": ""
+      "recurse": "include_sublayers"
     }
   },
   "torch.nn.Module.parameters": {

tests/test_linalg_lu_factor_ex.py

@@ -25,7 +25,6 @@ def test_case_1():
         import torch
         x = torch.tensor([[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]], dtype=torch.float64)
         LU, pivots, info = torch.linalg.lu_factor_ex(x)
-        info = info.item()
         """
     )
     obj.run(pytorch_code, ["LU", "pivots", "info"])
@@ -37,7 +36,6 @@ def test_case_2():
         import torch
         x = torch.tensor([[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]], dtype=torch.float64)
         LU, pivots, info = torch.linalg.lu_factor_ex(A=x)
-        info = info.item()
         """
     )
     obj.run(pytorch_code, ["LU", "pivots", "info"])
@@ -49,7 +47,6 @@ def test_case_3():
         import torch
         x = torch.tensor([[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]], dtype=torch.float64)
         LU, pivots, info = torch.linalg.lu_factor_ex(pivot=True, A=x)
-        info = info.item()
         """
     )
     obj.run(pytorch_code, ["LU", "pivots", "info"])
@@ -62,7 +59,6 @@ def test_case_4():
         x = torch.tensor([[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]], dtype=torch.float64)
         out = (torch.tensor([], dtype=torch.float64), torch.tensor([], dtype=torch.int), torch.tensor([], dtype=torch.int))
         LU, pivots, info = torch.linalg.lu_factor_ex(x, pivot=True, check_errors=False, out=out)
-        info = info.item()
         """
     )
     obj.run(pytorch_code, ["LU", "pivots", "info"])
@@ -75,7 +71,17 @@ def test_case_5():
         x = torch.tensor([[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]], dtype=torch.float64)
         out = (torch.tensor([], dtype=torch.float64), torch.tensor([], dtype=torch.int), torch.tensor([], dtype=torch.int))
         LU, pivots, info = torch.linalg.lu_factor_ex(A=x, pivot=True, check_errors=True, out=out)
-        info = info.item()
+        """
+    )
+    obj.run(pytorch_code, ["LU", "pivots", "info"])
+
+
+def test_case_6():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        x = torch.tensor([[[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]]], dtype=torch.float64)
+        LU, pivots, info = torch.linalg.lu_factor_ex(x)
         """
     )
     obj.run(pytorch_code, ["LU", "pivots", "info"])

tests/test_linalg_solve.py

@@ -127,3 +127,52 @@ def test_case_9():
         """
     )
     obj.run(pytorch_code, ["result", "out"])
+
+
+def test_case_10():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        x = torch.tensor([[3.0, 1],[1, 2]])
+        y = torch.tensor([[9.0, 8, 3], [-4.0, 6, 2]])
+        out = torch.tensor([])
+        result = torch.linalg.solve(A=x, B=y, out=out, left=True)
+        """
+    )
+    obj.run(pytorch_code, ["result", "out"])
+
+
+def test_case_11():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        x = torch.tensor([[3.0, 1],[1, 2]])
+        y = torch.tensor([[9.0, 8, 3], [-4.0, 6, 2]])
+        result = torch.linalg.solve(x, y, left=True)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_12():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        x = torch.tensor([[3.0, 1],[1, 2]])
+        y = torch.tensor([[9.0, 8, 3], [-4.0, 6, 2]])
+        result = torch.linalg.solve(A=x, B=y, left=True)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_13():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        x = torch.tensor([[3.0, 1],[1, 2]])
+        y = torch.tensor([[9.0, 8, 3], [-4.0, 6, 2]])
+        result = torch.linalg.solve(B=y, left=True, A=x)
+        """
+    )
+    obj.run(pytorch_code, ["result"])

tests/test_nn_Module_named_modules.py

@@ -39,12 +39,7 @@ def test_case_1():
         result = name_list
         """
     )
-    obj.run(
-        pytorch_code,
-        ["result"],
-        unsupport=True,
-        reason="doesn't support param remove_duplicate",
-    )
+    obj.run(pytorch_code, ["result"])
 
 
 def test_case_2():
@@ -65,12 +60,7 @@ def test_case_2():
         result = name_list
         """
     )
-    obj.run(
-        pytorch_code,
-        ["result"],
-        unsupport=True,
-        reason="paddle only support to output remove duplicate version",
-    )
+    obj.run(pytorch_code, ["result"])
 
 
 def test_case_3():
@@ -92,9 +82,4 @@ def test_case_3():
         result = name_list
         """
     )
-    obj.run(
-        pytorch_code,
-        ["result"],
-        unsupport=True,
-        reason="paddle doesn't support the memo parameter",
-    )
+    obj.run(pytorch_code, ["result"])

tests/test_nn_Module_named_parameters.py

@@ -181,3 +181,99 @@ def forward(self, x):
         """
     )
     obj.run(pytorch_code, ["result"], check_value=False)
+
+
+def test_case_8():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        result = []
+        class TestForHook(nn.Module):
+            def __init__(self):
+                super().__init__()
+                self.register_parameter('w1', nn.Parameter(torch.randn(2, 3)))
+                self.register_parameter('w1', nn.Parameter(torch.randn(3, 3)))
+                self.register_parameter('w2', nn.Parameter(torch.randn(1, 3)))
+                self.register_parameter('w2', nn.Parameter(torch.randn(1, 2)))
+            def forward(self, x):
+                return x
+
+        a = TestForHook()
+        for a,b in a.named_parameters(remove_duplicate=False):
+            result.append(b)
+        """
+    )
+    obj.run(pytorch_code, ["result"], check_value=False)
+
+
+def test_case_9():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        result = []
+        class TestForHook(nn.Module):
+            def __init__(self):
+                super().__init__()
+                self.register_parameter('w1', nn.Parameter(torch.randn(2, 3)))
+                self.register_parameter('w1', nn.Parameter(torch.randn(3, 3)))
+                self.register_parameter('w2', nn.Parameter(torch.randn(1, 3)))
+                self.register_parameter('w2', nn.Parameter(torch.randn(1, 2)))
+            def forward(self, x):
+                return x
+
+        a = TestForHook()
+        for a,b in a.named_parameters(remove_duplicate=True):
+            result.append(b)
+        """
+    )
+    obj.run(pytorch_code, ["result"], check_value=False)
+
+
+def test_case_10():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        result = []
+        class TestForHook(nn.Module):
+            def __init__(self):
+                super().__init__()
+                self.register_parameter('w1', nn.Parameter(torch.randn(2, 3)))
+                self.register_parameter('w1', nn.Parameter(torch.randn(3, 3)))
+                self.register_parameter('w2', nn.Parameter(torch.randn(1, 3)))
+                self.register_parameter('w2', nn.Parameter(torch.randn(1, 2)))
+            def forward(self, x):
+                return x
+
+        a = TestForHook()
+        for a,b in a.named_parameters(prefix = "wfs", recurse = False, remove_duplicate = False):
+            result.append(b)
+        """
+    )
+    obj.run(pytorch_code, ["result"], check_value=False)
+
+
+def test_case_11():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        result = []
+        class TestForHook(nn.Module):
+            def __init__(self):
+                super().__init__()
+                self.register_parameter('w1', nn.Parameter(torch.randn(2, 3)))
+                self.register_parameter('w1', nn.Parameter(torch.randn(3, 3)))
+                self.register_parameter('w2', nn.Parameter(torch.randn(1, 3)))
+                self.register_parameter('w2', nn.Parameter(torch.randn(1, 2)))
+            def forward(self, x):
+                return x
+
+        a = TestForHook()
+        for a,b in a.named_parameters(remove_duplicate = False, recurse = True, prefix = "wfs"):
+            result.append(b)
+        """
+    )
+    obj.run(pytorch_code, ["result"], check_value=False)

tests/test_optim_Adadelta.py

@@ -76,9 +76,7 @@ def test_case_6():
     )
 
 
-# if `weight_decay` set as int rather than float,
-# paddle will raise TypeError: 'int' object is not callable
-def _test_case_7():
+def test_case_7():
     pytorch_code = textwrap.dedent(
         generate_optimizer_test_code(
             "torch.optim.Adadelta(params=conv.parameters(), weight_decay=0)"

tests/test_optim_Adagrad.py

@@ -70,9 +70,7 @@ def test_case_6():
     obj.run(pytorch_code, ["result"], rtol=1.0e-5)
 
 
-# if `weight_decay` set as int rather than float,
-# paddle will raise TypeError: 'int' object is not callable
-def _test_case_7():
+def test_case_7():
     pytorch_code = textwrap.dedent(
         generate_optimizer_test_code(
             "torch.optim.Adagrad(params=conv.parameters(), weight_decay=0)"

tests/test_optim_Adam.py

@@ -77,3 +77,21 @@ def test_case_7():
         )
     )
     obj.run(pytorch_code, ["result"], rtol=1.0e-5)
+
+
+def test_case_8():
+    pytorch_code = textwrap.dedent(
+        generate_optimizer_test_code(
+            "torch.optim.Adam(conv.parameters(), weight_decay=1)"
+        )
+    )
+    obj.run(pytorch_code, ["result"], rtol=1.0e-5)
+
+
+def test_case_9():
+    pytorch_code = textwrap.dedent(
+        generate_optimizer_test_code(
+            "torch.optim.Adam(params=conv.parameters(), lr=0.001, betas=(0.9, 0.999), eps=1e-08, weight_decay=0)"
+        )
+    )
+    obj.run(pytorch_code, ["result"], rtol=1.0e-5)

tests/test_optim_AdamW.py

@@ -70,3 +70,21 @@ def test_case_6():
         generate_optimizer_test_code("torch.optim.AdamW(conv.parameters())")
     )
     obj.run(pytorch_code, ["result"], rtol=1.0e-5)
+
+
+def test_case_7():
+    pytorch_code = textwrap.dedent(
+        generate_optimizer_test_code(
+            "torch.optim.AdamW(conv.parameters(), weight_decay=1)"
+        )
+    )
+    obj.run(pytorch_code, ["result"], rtol=1.0e-5)
+
+
+def test_case_8():
+    pytorch_code = textwrap.dedent(
+        generate_optimizer_test_code(
+            "torch.optim.AdamW(params=conv.parameters(), lr=0.001, betas=(0.9, 0.999), eps=1e-08, weight_decay=0)"
+        )
+    )
+    obj.run(pytorch_code, ["result"], rtol=1.0e-5)

tests/test_optim_Adamax.py

@@ -68,3 +68,21 @@ def test_case_6():
         generate_optimizer_test_code("torch.optim.Adamax(conv.parameters())")
     )
     obj.run(pytorch_code, ["result"], rtol=1.0e-2)
+
+
+def test_case_7():
+    pytorch_code = textwrap.dedent(
+        generate_optimizer_test_code(
+            "torch.optim.Adamax(conv.parameters(), weight_decay=0, eps=1e-7)"
+        )
+    )
+    obj.run(pytorch_code, ["result"], rtol=1.0e-2)
+
+
+def test_case_8():
+    pytorch_code = textwrap.dedent(
+        generate_optimizer_test_code(
+            "torch.optim.Adamax(conv.parameters(), weight_decay=1, eps=1e-7)"
+        )
+    )
+    obj.run(pytorch_code, ["result"], rtol=1.0e-2)

tests/test_optim_RMSprop.py

@@ -54,9 +54,7 @@ def test_case_4():
     obj.run(pytorch_code, ["result"], rtol=1.0e-2)
 
 
-# if `weight_decay` set as int rather than float,
-# paddle will raise TypeError: 'int' object is not callable
-def _test_case_5():
+def test_case_5():
     pytorch_code = textwrap.dedent(
         generate_optimizer_test_code(
             "torch.optim.RMSprop(conv.parameters(), weight_decay=0)"

tests/test_optim_SGD.py

@@ -43,9 +43,7 @@ def test_case_3():
     obj.run(pytorch_code, ["result"], rtol=1.0e-5)
 
 
-# if `weight_decay` set as int rather than float,
-# paddle will raise TypeError: 'int' object is not callable
-def _test_case_4():
+def test_case_4():
     pytorch_code = textwrap.dedent(
         generate_optimizer_test_code(
             "torch.optim.SGD(conv.parameters(), lr=0.1, weight_decay=0)"