Merge branch 'master' into dev

README.md

@@ -312,6 +312,13 @@ unsupport_args:可选，Paddle API不支持的参数功能，通过该字段配
 paddle_default_kwargs :可选，当 paddle 参数更多 或者 参数默认值不一致 时，可以通过该配置，设置参数默认值。
 ```
 
+**需要注意的是**，如果一个API是别名API(alias API), 比如 `torch.nn.modules.GroupNorm` 和 `torch.nn.GroupNorm` 是同一个API，只是按照模块路径采用了不同的调用方式，那么就无需编写相关 Matcher，只需在 paconvert/api_alias_mapping.json 中增加如下该 API 的配置即可：
+```bash
+{
+  "torch.nn.modules.GroupNorm": "torch.nn.GroupNorm"
+}
+```
+
 对于一个待开发API，首先依据步骤1的映射关系，确定其属于哪种分类情况。
 
 对于以下映射关系的分类，都可以通过框架封装好的通用转换器：`GenericMatcher` 来处理：

paconvert/api_mapping.json

@@ -510,8 +510,8 @@
     ]
   },
   "torch.Tensor.coalesce": {
-    "Matcher": "GenericMatcher",
-    "paddle_api": "paddle.Tensor.coalesce"
+    "Matcher": "TensorFunc2PaddleFunc",
+    "paddle_api": "paddle.sparse.coalesce"
   },
   "torch.Tensor.conj": {
     "Matcher": "TensorUnchangeMatcher"
@@ -632,8 +632,11 @@
     ]
   },
   "torch.Tensor.diagflat": {
-    "Matcher": "GenericMatcher",
-    "paddle_api": "paddle.Tensor.diagflat"
+    "Matcher": "TensorFunc2PaddleFunc",
+    "paddle_api": "paddle.diagflat",
+    "args_list": [
+      "offset"
+    ]
   },
   "torch.Tensor.diagonal": {
     "Matcher": "GenericMatcher",
@@ -2002,8 +2005,11 @@
   },
   "torch.Tensor.slice_scatter": {},
   "torch.Tensor.slogdet": {
-    "Matcher": "GenericMatcher",
-    "paddle_api": "paddle.Tensor.slogdet"
+    "Matcher": "TensorSLogDetMatcher",
+    "paddle_api": "paddle.linalg.slogdet",
+    "args_list": [
+      "out"
+    ]
   },
   "torch.Tensor.smm": {},
   "torch.Tensor.softmax": {
@@ -2210,13 +2216,19 @@
   "torch.Tensor.transpose_": {},
   "torch.Tensor.triangular_solve": {},
   "torch.Tensor.tril": {
-    "Matcher": "GenericMatcher",
-    "paddle_api": "paddle.Tensor.tril"
+    "Matcher": "TensorFunc2PaddleFunc",
+    "paddle_api": "paddle.tril",
+    "args_list": [
+      "diagonal"
+    ]
   },
   "torch.Tensor.tril_": {},
   "torch.Tensor.triu": {
-    "Matcher": "GenericMatcher",
-    "paddle_api": "paddle.Tensor.triu"
+    "Matcher": "TensorFunc2PaddleFunc",
+    "paddle_api": "paddle.triu",
+    "args_list": [
+      "diagonal"
+    ]
   },
   "torch.Tensor.triu_": {},
   "torch.Tensor.true_divide": {

paconvert/api_matcher.py

@@ -3642,3 +3642,27 @@ def get_paddle_nodes(self, args, kwargs):
         code = "{}({})".format(self.get_paddle_api(), code)
         node = ast.parse(code.strip("\n")).body
         return node
+
+
+class TensorSLogDetMatcher(BaseMatcher):
+    def generate_code(self, kwargs):
+        out_v = kwargs.pop("out") if "out" in kwargs else None
+
+        if out_v:
+            API_TEMPLATE = textwrap.dedent(
+                """
+                res = paddle.linalg.slogdet({})
+                paddle.assign(res[0], {}[0]), paddle.assign(res[1], {}[1])
+                """
+            )
+            code = API_TEMPLATE.format(self.paddleClass, out_v, out_v)
+        else:
+            API_TEMPLATE = textwrap.dedent(
+                """
+                res = paddle.linalg.slogdet({})
+                res[0], res[1]
+                """
+            )
+            code = API_TEMPLATE.format(self.paddleClass)
+
+        return code

paconvert/transformer/basic_transformer.py

@@ -458,6 +458,7 @@ def trans_class_method(self, node, torch_api):
                         ast.Attribute,
                         ast.Subscript,
                         ast.BinOp,
+                        ast.Tuple,
                     ),
                 ):
                     self.insert_multi_node(node_list[0:-1])

tests/test_NLLLoss.py

@@ -0,0 +1,169 @@
+# Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import textwrap
+
+from apibase import APIBase
+
+obj = APIBase("torch.nn.NLLLoss")
+
+
+def test_case_1():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss()
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_2():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss(weight=weight)
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_3():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss(weight=weight, size_average=True)
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_4():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss(weight=weight, size_average=False)
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_5():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss(weight=weight, size_average=True, ignore_index=1)
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_6():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss(weight=weight, reduce=True)
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_7():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss(weight=weight, reduce=False)
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_8():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss(weight=weight, reduction='none')
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_9():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss(weight=weight, reduction='mean')
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_10():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        import torch.nn as nn
+        m = torch.tensor([[0.2, 0.1, 0.7], [0.2, 0.1, 0.7]])
+        weight = torch.tensor([3.,1.,1.])
+        target = torch.tensor([1,2])
+        loss = nn.NLLLoss(weight=weight, reduction='sum')
+        result = loss(m, target)
+        """
+    )
+    obj.run(pytorch_code, ["result"])

tests/test_Tensor_bernoulli_.py

@@ -0,0 +1,52 @@
+# Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import textwrap
+
+from apibase import APIBase
+
+obj = APIBase("torch.Tensor.bernoulli_")
+
+
+def test_case_1():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        src = torch.tensor([1., 2., 3., 4., 5., 6.])
+        result = src.bernoulli_(0.5)
+        """
+    )
+    obj.run(pytorch_code, ["result"], check_value=False)
+
+
+def test_case_2():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        src = torch.tensor([1., 2., 3., 4., 5., 6.])
+        result = src.bernoulli_(0.5, generator=None)
+        """
+    )
+    obj.run(pytorch_code, ["result"], check_value=False)
+
+
+def test_case_3():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        src = torch.tensor([1., 2., 3., 4., 5., 6.])
+        result = src.bernoulli_(p=0.5, generator=None)
+        """
+    )
+    obj.run(pytorch_code, ["result"], check_value=False)

tests/test_Tensor_bfloat16.py

@@ -0,0 +1,41 @@
+# Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import textwrap
+
+from apibase import APIBase
+
+obj = APIBase("torch.Tensor.bfloat16")
+
+
+def test_case_1():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        src = torch.tensor([1., 2., 3., 4., 5., 6.])
+        result = src.bfloat16().float()
+        """
+    )
+    obj.run(pytorch_code, ["result"])
+
+
+def test_case_2():
+    pytorch_code = textwrap.dedent(
+        """
+        import torch
+        src = torch.tensor([1., 2., 3., 4., 5., 6.])
+        result = src.bfloat16(memory_format=torch.preserve_format).float()
+        """
+    )
+    obj.run(pytorch_code, ["result"])