fleet_base.py 34.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
#   Copyright (c) 2020 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 __future__ import print_function
16
import copy
17
import warnings
18
import paddle
19
from paddle.fluid.framework import dygraph_only
20
from paddle.fluid import compiler
21
from .role_maker import UserDefinedRoleMaker, PaddleCloudRoleMaker, RoleMakerBase
22
from .strategy_compiler import StrategyCompiler
23
from .distributed_strategy import DistributedStrategy
24 25
from .meta_optimizer_factory import MetaOptimizerFactory
from .runtime_factory import RuntimeFactory
26
from paddle.fluid.wrapped_decorator import wrap_decorator
27
from paddle.fluid.dygraph import parallel_helper
28

29

30 31 32 33 34 35 36 37 38 39 40 41
def _inited_runtime_handler_(func):
    def __impl__(*args, **kwargs):
        cls = args[0]

        if cls._runtime_handle is None:
            raise ValueError("Fleet can not find suitable runtime handler")

        return func(*args, **kwargs)

    return __impl__


42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57
def _is_non_distributed_check_(func):
    def __impl__(*args, **kwargs):
        cls = args[0]

        if cls._role_maker is not None and cls._role_maker._is_non_distributed(
        ) is True:
            warnings.warn(
                "%s() function doesn't work when use non_distributed fleet." %
                (func.__name__))
            return

        return func(*args, **kwargs)

    return __impl__


58
inited_runtime_handler = wrap_decorator(_inited_runtime_handler_)
59
is_non_distributed_check = wrap_decorator(_is_non_distributed_check_)
60 61


62 63 64
class Fleet(object):
    """
    Unified API for distributed training of PaddlePaddle
65
    Please reference the https://github.com/PaddlePaddle/FleetX for details
66 67 68 69 70


    Returns:
        Fleet: A Fleet instance

71
    Example for collective training:
72 73
        .. code-block:: python

74
            import paddle.distributed.fleet as fleet
75 76 77

            fleet.init(is_collective=True)

78 79 80
            strategy = fleet.DistributedStrategy()
            optimizer = paddle.optimizer.SGD(learning_rate=0.001)
            optimizer = fleet.distributed_optimizer(optimizer, strategy=strategy)
81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96

            # do distributed training


    Example for parameter server training:

        .. code-block:: python

            import paddle.distributed.fleet as fleet

            fleet.init()

            strategy = fleet.DistributedStrategy()
            optimizer = paddle.optimizer.SGD(learning_rate=0.001)
            optimizer = fleet.distributed_optimizer(optimizer, strategy=strategy)

97 98
            if fleet.is_first_worker():
                print("this is first worker")
99

100 101
            print("current node index: {}".format(fleet.worker_index()))
            print("total number of worker num: {}".format(fleet.worker_num()))
102

103 104 105
            if fleet.is_worker():
                print("this is worker")
            print("worker endpoints: {}".format(fleet.worker_endpoints(to_string=True)))
106

107 108
            print("server num: {}".format(fleet.server_num()))
            print("server endpoints: {}".format(fleet.server_endpoints(to_string=True)))
109

110 111 112
            if fleet.is_server():
                print("this is server")
            fleet.stop_worker()
113 114


115 116 117
    """

    def __init__(self):
118
        self._role_maker = None
119
        self.strategy_compiler = None
120
        self._is_collective = False
121
        self._runtime_handle = None
122

123 124 125 126
    def init(self, role_maker=None, is_collective=False):
        """
        Initialize role_maker in Fleet.

127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161
        This function is responsible for the distributed architecture
        what you want to run your code behind.

        Args:
            role_maker (RoleMakerBase, optional): A ``RoleMakerBase`` containing the configuration
                of environment variables related to distributed training.If you did not initialize 
                the rolemaker by yourself, it will be automatically initialized to PaddleRoleMaker.
                The default value is None.
            is_collective (Boolean, optional): A ``Boolean`` variable determines whether the program 
                runs on the CPU or GPU. False means set distributed training using CPU, and True means
                GPU.The default value is False.The default value is False.
        Returns:
            None

        Examples1:

            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()

        Examples2:

            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init(is_collective=True)

        Examples3:

            .. code-block:: python

                import paddle.distributed.fleet as fleet
                role = fleet.PaddleCloudRoleMaker
                fleet.init(role)
162

163
        """
164 165

        if role_maker is None:
166 167 168 169 170 171
            if isinstance(is_collective, bool):
                self._is_collective = is_collective
                self._role_maker = PaddleCloudRoleMaker(
                    is_collective=self._is_collective)
            else:
                raise ValueError(
172 173
                    "`is_collective` should be instance of `bool`, but got {}".
                    format(type(is_collective)))
174
        else:
175 176 177 178 179 180
            if isinstance(role_maker, RoleMakerBase):
                self._role_maker = role_maker
            else:
                raise ValueError(
                    "`role_maker` should be subclass of `RoleMakerBase`, but got {}".
                    format(type(role_maker)))
181
        self._role_maker._generate_role()
182

183 184 185
        import paddle.distributed.fleet as fleet
        fleet.util._set_role_maker(self._role_maker)

186
        self.strategy_compiler = StrategyCompiler()
187 188 189 190 191 192
        if paddle.fluid.framework.in_dygraph_mode():
            if parallel_helper._is_parallel_ctx_initialized():
                warnings.warn(
                    "The dygraph parallel environment has been initialized.")
            else:
                paddle.distributed.init_parallel_env()
193 194 195 196 197 198 199 200

    def is_first_worker(self):
        """
        Check whether the node is the first instance of worker.

        Returns:
            bool: True if this is the first node of worker,
                  False if not.
201

202 203 204 205 206 207 208 209
        Examples:

            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()
                fleet.is_first_worker()

210
        """
211
        return self._role_maker._is_first_worker()
212 213 214 215 216 217 218

    def worker_index(self):
        """
        Get current worker index.

        Returns:
            int: node id
219 220 221 222 223 224 225 226

        Examples:

            .. code-block:: python
                import paddle.distributed.fleet as fleet
                fleet.init()
                fleet.worker_index()

227
        """
228
        return self._role_maker._worker_index()
229 230 231 232 233 234 235

    def worker_num(self):
        """
        Get current total worker number.

        Returns:
            int: worker numbers
236

237 238 239 240 241 242 243
        Examples:
            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()
                fleet.worker_num()

244
        """
245
        return self._role_maker._worker_num()
246 247 248 249 250 251 252 253

    def is_worker(self):
        """
        Check whether the node is an instance of worker.

        Returns:
            bool: True if this is a node of worker,
                  False if not.
254 255 256 257 258 259 260 261

        Examples:
            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()
                fleet.is_worker()

262
        """
263
        return self._role_maker._is_worker()
264 265 266

    def worker_endpoints(self, to_string=False):
        """
267
        Get current worker endpoints, such as ["127.0.0.1:1001", "127.0.0.1:1002"].
268 269 270

        Returns:
            list/string: server endpoints
271 272 273 274 275 276 277 278

        Examples:
            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()
                fleet.worker_endpoints()

279 280
        """
        if to_string:
281
            return ",".join(self._role_maker._get_trainer_endpoints())
282
        else:
283
            return self._role_maker._get_trainer_endpoints()
284 285 286 287 288 289 290

    def server_num(self):
        """
        Get current total worker number.

        Returns:
            int: server number
291 292 293 294 295 296

        Examples:
            .. code-block:: python
            import paddle.distributed.fleet as fleet
            fleet.init()
            fleet.server_num()
297
        """
298
        return len(self._role_maker._get_pserver_endpoints())
299 300 301 302 303 304 305

    def server_index(self):
        """
        Get current server index.

        Returns:
            int: node id
306 307 308 309 310 311 312 313

        Examples:
            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()
                fleet.server_index()

314
        """
315
        return self._role_maker._server_index()
316 317 318 319 320 321 322

    def server_endpoints(self, to_string=False):
        """
        Get current server endpoints, such as ["127.0.0.1:1001", "127.0.0.1:1002"].

        Returns:
            list/string: server endpoints
323 324 325 326 327 328 329 330

        Examples:
            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()
                fleet.server_endpoints()

331
        """
332

333
        if to_string:
334
            return ",".join(self._role_maker._get_pserver_endpoints())
335
        else:
336
            return self._role_maker._get_pserver_endpoints()
337 338 339 340 341 342 343 344

    def is_server(self):
        """
        Check whether the node is an instance of server.

        Returns:
            bool: True if this is a node of server,
                  False if not.
345 346 347 348 349 350 351 352

        Examples:

            .. code-block:: python
                import paddle.distributed.fleet as fleet
                fleet.init()
                fleet.is_server()

353
        """
354
        return self._role_maker._is_server(
355
        ) or self._role_maker._is_heter_worker()
356 357 358

    def barrier_worker(self):
        """
359 360 361 362
        barrier all workers

        Returns:
            None
363
        """
364
        self._role_maker._barrier("worker")
365

366
    @is_non_distributed_check
367
    @inited_runtime_handler
368 369
    def init_worker(self):
        """
370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387
        initialize `Communicator` for parameter server training.


        Returns:
            None

        Examples:

            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()

                # build net
                # fleet.distributed_optimizer(...)

                fleet.init_worker()

388 389 390
        """
        self._runtime_handle._init_worker()

391
    @is_non_distributed_check
392
    @inited_runtime_handler
393
    def init_server(self, *args, **kwargs):
394
        """
395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413
        init_server executor to initialize startup program,
        if the `args` is not empty, it will run load_persistables for increment training.


        Returns:
            None

        Examples:

            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()

                # build net
                # fleet.distributed_optimizer(...)

                fleet.init_server()

414
        """
415
        self._runtime_handle._init_server(*args, **kwargs)
416

417
    @is_non_distributed_check
418
    @inited_runtime_handler
419 420
    def run_server(self):
        """
421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438
        run server will run pserver main program with executor.

        Returns:
            None

        Examples:

            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()

                # build net
                # fleet.distributed_optimizer(...)

                if fleet.is_server():
                    fleet.init_server()

439 440 441
        """
        self._runtime_handle._run_server()

442
    @is_non_distributed_check
443
    @inited_runtime_handler
444 445
    def stop_worker(self):
        """
446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
        stop `Communicator` and give training complete notice to parameter server.

        Returns:
            None

        Examples:

            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()

                # build net
                # fleet.distributed_optimizer(...)

                fleet.init_server()

463 464 465
        """
        self._runtime_handle._stop_worker()

466 467 468 469 470 471 472
    def save_inference_model(self,
                             executor,
                             dirname,
                             feeded_var_names,
                             target_vars,
                             main_program=None,
                             export_for_deployment=True):
473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492
        """
        save inference model for inference.

        Returns:
            None

        Examples:

            .. code-block:: python

                import paddle.distributed.fleet as fleet
                fleet.init()

                # build net
                # fleet.distributed_optimizer(...)

                fleet.init_server()

        """

493 494 495 496 497
        self._runtime_handle._save_inference_model(
            executor, dirname, feeded_var_names, target_vars, main_program,
            export_for_deployment)

    def save_persistables(self, executor, dirname, main_program=None):
498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537
        """

        saves all persistable variables from :code:`main_program` to
        the folder :code:`dirname`. You can refer to

        The :code:`dirname` is used to specify the folder where persistable variables
        are going to be saved. If you would like to save variables in separate
        files, set :code:`filename` None.

        Args:
            executor(Executor): The executor to run for saving persistable variables.
                                You can refer to :ref:`api_guide_executor_en` for
                                more details.

            dirname(str, optional): The saving directory path.
                                When you need to save the parameter to the memory, set it to None.
            main_program(Program, optional): The program whose persistbale variables will
                                             be saved. Default: None.


        Returns:
            None

        Examples:

            .. code-block:: text

                import paddle.distributed.fleet as fleet
                import paddle.fluid as fluid

                fleet.init()

                # build net
                # fleet.distributed_optimizer(...)

                exe = fluid.Executor(fluid.CPUPlace())
                fleet.save_persistables(exe, "dirname", fluid.default_main_program())

        """

538 539
        self._runtime_handle._save_persistables(executor, dirname, main_program)

540
    def distributed_optimizer(self, optimizer, strategy=None):
541
        """
542 543 544 545 546 547 548 549 550
        Optimizer for distributed training.

        For the distributed training, this method would rebuild a new instance of DistributedOptimizer.
        Which has basic Optimizer function and special features for distributed training.

        Args:
            optimizer(Optimizer): The executor to run for init server.
            strategy(DistributedStrategy): Extra properties for distributed optimizer.

551
        Returns:
552
            Fleet: instance of fleet.
553 554

        Examples:
555

556
            .. code-block:: python
557 558 559 560 561 562 563 564

                import paddle.distributed.fleet as fleet
                role = fleet.role_maker.PaddleCloudRoleMaker(is_collective=True)
                fleet.init(role)
                strategy = fleet.DistributedStrategy()
                optimizer = paddle.optimizer.SGD(learning_rate=0.001)
                optimizer = fleet.distributed_optimizer(optimizer, strategy=strategy)

565 566
        """
        self.user_defined_optimizer = optimizer
567 568 569
        if paddle.fluid.framework.in_dygraph_mode():
            return self

570 571
        if strategy == None:
            strategy = DistributedStrategy()
572
        self.user_defined_strategy = strategy
D
Dong Daxiang 已提交
573
        self.valid_strategy = None
574 575
        return self

576 577 578
    @dygraph_only
    def distributed_model(self, model):
        """
579 580 581 582 583 584 585
        Return distributed data parallel model (Only work in dygraph mode)

        Args:
            model (Layer): the user-defind model which inherits Layer.

        Returns:
            distributed data parallel model which inherits Layer.
586 587

        Examples:
588

589 590
            .. code-block:: python

591 592 593 594 595 596 597 598 599
                import paddle
                import paddle.nn as nn
                from paddle.distributed import fleet

                class LinearNet(nn.Layer):
                    def __init__(self):
                        super(LinearNet, self).__init__()
                        self._linear1 = nn.Linear(10, 10)
                        self._linear2 = nn.Linear(10, 1)
600

601 602
                    def forward(self, x):
                        return self._linear2(self._linear1(x))
603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634

                # 1. enable dynamic mode
                paddle.disable_static()

                # 2. initialize fleet environment
                fleet.init(is_collective=True)

                # 3. create layer & optimizer
                layer = LinearNet()
                loss_fn = nn.MSELoss()
                adam = paddle.optimizer.Adam(
                    learning_rate=0.001, parameters=layer.parameters())

                # 4. get data_parallel model using fleet
                adam = fleet.distributed_optimizer(adam)
                dp_layer = fleet.distributed_model(layer)

                # 5. run layer
                inputs = paddle.randn([10, 10], 'float32')
                outputs = dp_layer(inputs)
                labels = paddle.randn([10, 1], 'float32')
                loss = loss_fn(outputs, labels)

                print("loss:", loss.numpy())

                loss = dp_layer.scale_loss(loss)
                loss.backward()
                dp_layer.apply_collective_grads()

                adam.step()
                adam.clear_grad()

635

636 637 638 639 640 641 642 643 644
        """
        assert model is not None
        self.model = paddle.DataParallel(model)
        return self.model

    @dygraph_only
    def state_dict(self):
        """
        Get state dict information from optimizer.
645
        (Only work in dygraph mode)
646 647 648 649 650 651 652

        Returns: 
            state_dict(dict) : dict contains all the Tensor used by optimizer

        Examples:
            .. code-block:: python

653 654 655 656 657 658
                import numpy as np
                import paddle
                from paddle.distributed import fleet

                paddle.disable_static()
                fleet.init(is_collective=True)
659

660 661
                value = np.arange(26).reshape(2, 13).astype("float32")
                a = paddle.fluid.dygraph.to_variable(value)
662

663 664
                layer = paddle.nn.Linear(13, 5)
                adam = paddle.optimizer.Adam(learning_rate=0.01, parameters=layer.parameters())
665

666 667 668
                adam = fleet.distributed_optimizer(adam)
                dp_layer = fleet.distributed_model(layer)
                state_dict = adam.state_dict()
669 670 671 672 673 674 675 676
        """
        # imitate target optimizer retrieval
        return self.user_defined_optimizer.state_dict()

    @dygraph_only
    def set_state_dict(self, state_dict):
        """
        Load optimizer state dict.
677
        (Only work in dygraph mode)
678 679 680 681

        Args: 
            state_dict(dict) : Dict contains all the Tensor needed by optimizer

682 683
        Returns:
            None
684 685 686 687

        Examples:
            .. code-block:: python

688 689 690
                import numpy as np
                import paddle
                from paddle.distributed import fleet
691

692 693 694 695 696
                paddle.disable_static()
                fleet.init(is_collective=True)

                value = np.arange(26).reshape(2, 13).astype("float32")
                a = paddle.fluid.dygraph.to_variable(value)
697

698 699
                layer = paddle.nn.Linear(13, 5)
                adam = paddle.optimizer.Adam(learning_rate=0.01, parameters=layer.parameters())
700

701 702 703 704 705 706
                adam = fleet.distributed_optimizer(adam)
                dp_layer = fleet.distributed_model(layer)
                state_dict = adam.state_dict()
                paddle.framework.save(state_dict, "paddle_dy")
                para_state_dict, opti_state_dict = paddle.framework.load( "paddle_dy")
                adam.set_state_dict(opti_state_dict)
707 708 709 710 711 712 713 714
        """
        # imitate target optimizer retrieval
        return self.user_defined_optimizer.set_state_dict(state_dict)

    @dygraph_only
    def set_lr(self, value):
        """
        Set the value of the learning rate manually in the optimizer. 
715
        (Only work in dygraph mode)
716

717 718 719
        Args:
            value (float|Tensor): the value of learning rate

720 721
        Returns: 
            None 
722 723 724 725

        Examples:
            .. code-block:: python

726 727 728
                import numpy as np
                import paddle
                from paddle.distributed import fleet
729

730 731
                paddle.disable_static()
                fleet.init(is_collective=True)
732

733 734
                value = np.arange(26).reshape(2, 13).astype("float32")
                a = paddle.fluid.dygraph.to_variable(value)
735

736 737
                layer = paddle.nn.Linear(13, 5)
                adam = paddle.optimizer.Adam(learning_rate=0.01, parameters=layer.parameters())
738

739 740 741 742 743 744 745 746 747 748 749 750 751 752
                adam = fleet.distributed_optimizer(adam)
                dp_layer = fleet.distributed_model(layer)

                lr_list = [0.2, 0.3, 0.4, 0.5, 0.6]
                for i in range(5):
                    adam.set_lr(lr_list[i])
                    lr = adam.get_lr()
                    print("current lr is {}".format(lr))
                # Print:
                #    current lr is 0.2
                #    current lr is 0.3
                #    current lr is 0.4
                #    current lr is 0.5
                #    current lr is 0.6
753 754 755 756 757 758 759 760
        """
        # imitate target optimizer retrieval
        return self.user_defined_optimizer.set_lr(value)

    @dygraph_only
    def get_lr(self):
        """
        Get current step learning rate.
761
        (Only work in dygraph mode)
762 763 764 765 766 767 768

        Returns:
            float: The learning rate of the current step.

        Examples:
            .. code-block:: python

769 770 771 772 773 774
                import numpy as np
                import paddle
                from paddle.distributed import fleet

                paddle.disable_static()
                fleet.init(is_collective=True)
775

776 777
                value = np.arange(26).reshape(2, 13).astype("float32")
                a = paddle.fluid.dygraph.to_variable(value)
778

779 780
                layer = paddle.nn.Linear(13, 5)
                adam = paddle.optimizer.Adam(learning_rate=0.01, parameters=layer.parameters())
781

782 783
                adam = fleet.distributed_optimizer(adam)
                dp_layer = fleet.distributed_model(layer)
784

785 786
                lr = adam.get_lr()
                print(lr) # 0.01
787 788 789 790 791 792 793 794
        """
        # imitate target optimizer retrieval
        return self.user_defined_optimizer.get_lr()

    @dygraph_only
    def step(self):
        """
        Execute the optimizer once.
795
        (Only work in dygraph mode)
796

797 798
        Returns:
            None
799 800 801 802

        Examples:
            .. code-block:: python

803 804 805
                import paddle
                import paddle.nn as nn
                from paddle.distributed import fleet
806

807 808 809 810 811
                class LinearNet(nn.Layer):
                    def __init__(self):
                        super(LinearNet, self).__init__()
                        self._linear1 = nn.Linear(10, 10)
                        self._linear2 = nn.Linear(10, 1)
812

813 814
                    def forward(self, x):
                        return self._linear2(self._linear1(x))
815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854

                # 1. enable dynamic mode
                paddle.disable_static()

                # 2. initialize fleet environment
                fleet.init(is_collective=True)

                # 3. create layer & optimizer
                layer = LinearNet()
                loss_fn = nn.MSELoss()
                adam = paddle.optimizer.Adam(
                    learning_rate=0.001, parameters=layer.parameters())

                # 4. get data_parallel model using fleet
                adam = fleet.distributed_optimizer(adam)
                dp_layer = fleet.distributed_model(layer)

                # 5. run layer
                inputs = paddle.randn([10, 10], 'float32')
                outputs = dp_layer(inputs)
                labels = paddle.randn([10, 1], 'float32')
                loss = loss_fn(outputs, labels)

                print("loss:", loss.numpy())

                loss = dp_layer.scale_loss(loss)
                loss.backward()
                dp_layer.apply_collective_grads()

                adam.step()
                adam.clear_grad()


        """
        # imitate target optimizer retrieval
        return self.user_defined_optimizer.step()

    @dygraph_only
    def clear_grad(self):
        """
855 856
        Clear the gradients of all optimized parameters for model.
        (Only work in dygraph mode)
857

858 859
        Returns: 
            None
860 861 862 863

        Examples:
            .. code-block:: python

864 865 866
                import paddle
                import paddle.nn as nn
                from paddle.distributed import fleet
867

868 869 870 871 872
                class LinearNet(nn.Layer):
                    def __init__(self):
                        super(LinearNet, self).__init__()
                        self._linear1 = nn.Linear(10, 10)
                        self._linear2 = nn.Linear(10, 1)
873

874 875
                    def forward(self, x):
                        return self._linear2(self._linear1(x))
876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911

                # 1. enable dynamic mode
                paddle.disable_static()

                # 2. initialize fleet environment
                fleet.init(is_collective=True)

                # 3. create layer & optimizer
                layer = LinearNet()
                loss_fn = nn.MSELoss()
                adam = paddle.optimizer.Adam(
                    learning_rate=0.001, parameters=layer.parameters())

                # 4. get data_parallel model using fleet
                adam = fleet.distributed_optimizer(adam)
                dp_layer = fleet.distributed_model(layer)

                # 5. run layer
                inputs = paddle.randn([10, 10], 'float32')
                outputs = dp_layer(inputs)
                labels = paddle.randn([10, 1], 'float32')
                loss = loss_fn(outputs, labels)

                print("loss:", loss.numpy())

                loss = dp_layer.scale_loss(loss)
                loss.backward()
                dp_layer.apply_collective_grads()

                adam.step()
                adam.clear_grad()

        """
        # imitate target optimizer retrieval
        return self.user_defined_optimizer.clear_grad()

912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934
    def minimize(self,
                 loss,
                 startup_program=None,
                 parameter_list=None,
                 no_grad_set=None):
        """
        Add distributed operations to minimize ``loss`` by updating ``parameter_list``.

        Args:
            loss (Variable): A ``Variable`` containing the value to minimize.
            startup_program (Program, optional): :ref:`api_fluid_Program` for
                initializing parameters in ``parameter_list``. The default value
                is None, at this time :ref:`api_fluid_default_startup_program` will be used.
            parameter_list (Iterable, optional): Iterable of ``Variable`` or ``Variable.name`` to update
                to minimize ``loss``. The default value is None, at this time all parameters
                will be updated.
            no_grad_set (set, optional): Set of ``Variable``  or ``Variable.name`` that don't need
                to be updated. The default value is None.

        Returns:
            tuple: tuple (optimize_ops, params_grads), A list of operators appended
            by minimize and a list of (param, grad) variable pairs, param is
            ``Parameter``, grad is the gradient value corresponding to the parameter.
935 936
            The returned tuple can be passed to ``fetch_list`` in ``Executor.run()`` to
            indicate program pruning. If so, the program will be pruned by ``feed`` and
937 938 939
            ``fetch_list`` before run, see details in ``Executor``.

        Examples:
940
            .. code-block:: python
941

942 943
                import paddle
                import paddle.distributed.fleet as fleet
944

945 946 947 948 949 950 951 952 953 954 955 956
                fc_1 = paddle.fluid.layers.fc(input=input_x, size=hid_dim, act='tanh')
                fc_2 = paddle.fluid.layers.fc(input=fc_1, size=hid_dim, act='tanh')
                prediction = paddle.fluid.layers.fc(input=[fc_2], size=label_dim, act='softmax')
                cost = paddle.fluid.layers.cross_entropy(input=prediction, label=input_y)
                avg_cost = paddle.fluid.layers.mean(x=cost)

                role = fleet.role_maker.PaddleCloudRoleMaker(is_collective=True)
                fleet.init(role)
                strategy = fleet.DistributedStrategy()
                optimizer = paddle.optimizer.SGD(learning_rate=0.001)
                optimizer = fleet.distributed_optimizer(optimizer, strategy=strategy)
                optimizer.minimize(avg_cost)
957

958
                # for more examples, please reference https://github.com/PaddlePaddle/FleetX
959 960

        """
961 962 963 964 965
        if paddle.fluid.framework.in_dygraph_mode():
            # imitate target optimizer retrieval
            target_opt = self.user_defined_optimizer
            return target_opt.minimize(loss)

966
        context = {}
967 968
        # cache original feed forward program
        self.origin_main_program = loss.block.program
969 970
        context["origin_main_program"] = self.origin_main_program
        context["loss"] = loss
971 972
        if startup_program == None:
            self.origin_startup_program = \
973 974
                paddle.static.default_startup_program().clone(for_test=False)
            startup_program = paddle.static.default_startup_program()
975 976 977
        else:
            self.origin_startup_program = \
                startup_program.clone(for_test=False)
978

979 980
        context["origin_startup_program"] = startup_program
        context["role_maker"] = self._role_maker
981 982 983 984 985

        # compile time
        distributed_optimizer_list = \
            MetaOptimizerFactory()._get_valid_meta_optimizers(
                self.user_defined_optimizer)
D
Dong Daxiang 已提交
986

987 988 989 990 991 992 993 994 995 996
        context["user_defined_strategy"] = copy.copy(self.user_defined_strategy)

        # trigger the auto-parallel in very strict condition
        # strategy = DistributedStrategy()
        # strategy.auto = True
        # optimizer = paddle.optimizer.SGD(learning_rate=0.1)
        # optimizer = fleet.distributed_optimizer(optimizer, strategy)
        if self.user_defined_strategy._is_strict_auto():
            # turn on all the strategy for each optimizer
            for opt in distributed_optimizer_list:
997
                opt._enable_strategy(self.user_defined_strategy, context)
998

999 1000
        valid_optimizer_list = []
        valid_graph_optimizer_list = []
D
Dong Daxiang 已提交
1001
        can_not_apply_optimizer_list = []
1002 1003 1004 1005 1006 1007 1008
        # recall meta optimizers for ranking
        for opt in distributed_optimizer_list:
            opt._set_basic_info(loss, self._role_maker,
                                self.user_defined_optimizer,
                                self.user_defined_strategy)
            if opt._can_apply() and not opt._is_graph_out():
                valid_optimizer_list.append(opt)
D
Dong Daxiang 已提交
1009
            elif opt._can_apply() and opt._is_graph_out():
1010
                valid_graph_optimizer_list.append(opt)
D
Dong Daxiang 已提交
1011 1012
            else:
                can_not_apply_optimizer_list.append(opt)
1013
        # combine recalled meta optimizers to be a valid meta optimizer
D
Dong Daxiang 已提交
1014
        meta_optimizer, graph_optimizer = \
1015 1016 1017 1018
            self.strategy_compiler.generate_optimizer(
                loss, self._role_maker, self.user_defined_optimizer,
                self.user_defined_strategy, valid_optimizer_list,
                valid_graph_optimizer_list)
D
Dong Daxiang 已提交
1019

D
Dong Daxiang 已提交
1020 1021
        valid_strategy = self.strategy_compiler._get_valid_strategy(
            self.user_defined_strategy, can_not_apply_optimizer_list)
1022 1023 1024

        context["valid_strategy"] = valid_strategy

D
Dong Daxiang 已提交
1025
        self.valid_strategy = valid_strategy
1026
        self.valid_strategy._enable_env()
D
Dong Daxiang 已提交
1027

1028 1029
        optimize_ops = []
        params_grads = []
1030

1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044
        if self._role_maker._is_non_distributed() and not self._is_collective:
            if self._runtime_handle is None:
                self._runtime_handle = RuntimeFactory()._create_runtime(context)

            compiled_program = compiler.CompiledProgram(
                self.origin_main_program).with_data_parallel(
                    loss_name=loss.name, share_vars_from=None)
            loss.block.program._graph = compiled_program
            return self.user_defined_optimizer.minimize(
                loss,
                startup_program=startup_program,
                parameter_list=parameter_list,
                no_grad_set=no_grad_set)

1045 1046 1047 1048 1049 1050
        if meta_optimizer:
            optimize_ops, params_grads = meta_optimizer.minimize(
                loss,
                startup_program=startup_program,
                parameter_list=parameter_list,
                no_grad_set=no_grad_set)
1051

1052
            default_program = paddle.static.default_main_program()
1053 1054 1055 1056

            if id(default_program) != id(loss.block.program):
                paddle.fluid.framework.switch_main_program(loss.block.program)

1057 1058 1059 1060 1061 1062
        else:
            optimize_ops, params_grads = self.user_defined_optimizer.minimize(
                loss,
                startup_program=startup_program,
                parameter_list=parameter_list,
                no_grad_set=no_grad_set)
1063

1064 1065
        context["program_optimize_ops"] = optimize_ops
        context["program_params_grads"] = params_grads
1066

1067
        if graph_optimizer:
D
Dong Daxiang 已提交
1068
            optimize_ops, params_grads = graph_optimizer.minimize(
1069 1070 1071 1072 1073 1074 1075 1076
                loss,
                startup_program=startup_program,
                parameter_list=parameter_list,
                no_grad_set=no_grad_set)
            # since we do not encourage users to use graph operations
            # if a graph optimizer takes effect, mostly
            # optimizers_ops and params_grads are None
            # i.e. users can not modify current computation graph anymore
1077 1078 1079
            context["graph_optimize_ops"] = optimize_ops
            context["graph_optimize_grads"] = params_grads

1080
        if self._runtime_handle is None:
1081
            self._runtime_handle = RuntimeFactory()._create_runtime(context)
1082

1083 1084
        import paddle.distributed.fleet as fleet
        fleet.util._set_strategy(context["valid_strategy"])
1085 1086

        return optimize_ops, params_grads