Distributed SaveLoad implementation for semi-auto strategy (#59659)

* exclude xpu

* demo of running dygraph distributed save load

* support save cross mesh state_dict

* polish

* fix compute overlap bug

* test save load in dp_mp unittest

* fix get local file bug and test

* delete useless files, and rename var

* polish

* format codes

* test use_dist

* fix test

* info to debug

* fix test

* fix

* fix coverage ci

* fix docstring codes

* rename and codestyle

* get rid of use_dist argument

* fix copyright

* polish doc

* polish

* polish

* use tmp file path

Author: pangengzheng

python/paddle/distributed/__init__.py

@@ -103,6 +103,9 @@
 
 from . import rpc  # noqa: F401
 
+from .checkpoint.save_state_dict import save_state_dict
+from .checkpoint.load_state_dict import load_state_dict
+
 __all__ = [
     "io",
     "spawn",
@@ -157,5 +160,7 @@
     "Shard",
     "Replicate",
     "Partial",
+    "save_state_dict",
+    "load_state_dict",
     "shard_optimizer",
 ]

python/paddle/distributed/checkpoint/__init__.py

@@ -0,0 +1,21 @@
+# 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.
+
+from .save_state_dict import save_state_dict
+from .load_state_dict import load_state_dict
+
+__all__ = [
+    "save_state_dict",
+    "load_state_dict",
+]