gradient scale (#33862)

Author: FeixLiu

paddle/fluid/framework/distributed_strategy.proto

@@ -119,6 +119,16 @@ message ExecutionStrategy {
   optional bool use_thread_barrier = 4 [ default = false ];
 }
 
+message GradientScaleConfig {
+  // Optional value ['avg', 'sum', 'customized']
+  // If avg, loss@grad will be divided by the number of devices,
+  // that is, the gradient will be accumulated and averaged among
+  // multiple devices.
+  // Else if sum, the gradient will accumulated among multiple
+  // devices.
+  optional string scale_strategy = 1 [ default = 'avg' ];
+}
+
 message AsyncConfig {
   optional int32 k_steps = 1 [ default = -1 ];
   optional int32 max_merge_var_num = 2 [ default = 1 ];
@@ -195,6 +205,7 @@ message DistributedStrategy {
   optional TensorParallelConfig tensor_parallel_configs = 113;
   optional BuildStrategy build_strategy = 201;
   optional ExecutionStrategy execution_strategy = 202;
+  optional GradientScaleConfig gradient_scale_configs = 203;
 }
 
 message DistributedJobInfo {

python/paddle/distributed/fleet/base/distributed_strategy.py

@@ -254,6 +254,28 @@ def build_strategy(self, strategy):
                 getattr(self.strategy.build_strategy,
                         f.name).extend(getattr(strategy, f.name))
 
+    @property
+    def gradient_scale_configs(self):
+        """
+        Set the strategy of gradient scale
+        Examples:
+
+          .. code-block:: python
+            import paddle.distributed.fleet as fleet
+            strategy = fleet.DistributedStrategy()
+            strategy.gradient_scale_configs = {'scale_strategy': 'avg'}
+
+        Note that, strategy must be in 'avg', 'sum' or 'customized'
+        """
+        return get_msg_dict(self.strategy.gradient_scale_configs)
+
+    @gradient_scale_configs.setter
+    @is_strict_auto
+    def gradient_scale_configs(self, config):
+        check_configs_key(self.strategy.gradient_scale_configs, config,
+                          'gradient_scale_configs')
+        assign_configs_value(self.strategy.gradient_scale_configs, config)
+
     @property
     def a_sync(self):
         """

python/paddle/distributed/fleet/meta_optimizers/graph_execution_optimizer.py

@@ -18,6 +18,7 @@
 from .meta_optimizer_base import MetaOptimizerBase
 from ..base.private_helper_function import wait_server_ready
 import logging
+from paddle.static import BuildStrategy
 
 __all__ = []
 
@@ -147,6 +148,17 @@ def _try_to_compile(self, startup_program, main_program, loss):
         local_build_strategy.nccl_comm_num = \
             dist_strategy.nccl_comm_num
 
+        gradient_scale_configs = self.user_defined_strategy.gradient_scale_configs
+        scale_strategys = {
+            'avg': BuildStrategy.GradientScaleStrategy.CoeffNumDevice,
+            'sum': BuildStrategy.GradientScaleStrategy.One,
+            'customized': BuildStrategy.GradientScaleStrategy.Customized,
+        }
+        assert gradient_scale_configs['scale_strategy'] in scale_strategys, \
+            "gradient_scale_configs.scale_strategy must be 'avg', 'sum' or 'customized'"
+        local_build_strategy.gradient_scale_strategy = \
+            scale_strategys[gradient_scale_configs['scale_strategy']]
+
         if self.user_defined_strategy.recompute == True:
             logging.warn(
                 "set enable_sequential_execution=True since you have enable the recompute strategy"

python/paddle/fluid/tests/unittests/CMakeLists.txt

@@ -107,6 +107,7 @@ if(((NOT WITH_ROCM) AND (NOT WITH_GPU)) OR WIN32)
     LIST(REMOVE_ITEM TEST_OPS test_collective_wait)
     LIST(REMOVE_ITEM TEST_OPS test_memcpy_op)
     LIST(REMOVE_ITEM TEST_OPS test_raw_program_optimizer)
+    LIST(REMOVE_ITEM TEST_OPS test_fleet_gradient_scale)
 endif()
 
 if(WIN32)

python/paddle/fluid/tests/unittests/test_fleet_gradient_scale.py

@@ -0,0 +1,73 @@
+#   Copyright (c) 2021 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
+
+import unittest
+
+import paddle
+import paddle.fluid as fluid
+import paddle.distributed.fleet as fleet
+import numpy as np
+import os
+
+
+class TestGradientScale(unittest.TestCase):
+    def setUp(self):
+        os.environ["PADDLE_TRAINER_ID"] = "0"
+        os.environ["PADDLE_TRAINER_ENDPOINTS"] = "127.0.0.1:36001"
+
+    def mlp(self, input_x, input_y, hid_dim=128, label_dim=2):
+        fc_1 = paddle.static.nn.fc(x=input_x, size=hid_dim, activation='tanh')
+        fc_2 = paddle.static.nn.fc(x=fc_1, size=hid_dim, activation='tanh')
+        prediction = paddle.static.nn.fc(x=[fc_2],
+                                         size=label_dim,
+                                         activation='softmax')
+        cost = paddle.nn.functional.cross_entropy(
+            input=prediction, label=input_y)
+        avg_cost = paddle.mean(x=cost)
+        return avg_cost
+
+    def gen_data(self):
+        return {
+            "x": np.random.random(size=(128, 32)).astype('float32'),
+            "y": np.random.randint(
+                2, size=(128, 1)).astype('int64')
+        }
+
+    def test_single_gpu(self):
+        paddle.enable_static()
+        fleet.init(is_collective=True)
+        main_program = paddle.static.Program()
+        startup_program = paddle.static.Program()
+        strategy = fleet.DistributedStrategy()
+        strategy.gradient_scale_configs = {'scale_strategy': 'sum'}
+        with fluid.program_guard(main_program, startup_program):
+            with fluid.unique_name.guard():
+                input_x = paddle.static.data(
+                    name="x", shape=[None, 32], dtype='float32')
+                input_y = paddle.static.data(
+                    name="y", shape=[None, 1], dtype='int64')
+                cost = self.mlp(input_x=input_x, input_y=input_y)
+                output_name = cost.name
+                optimizer = fleet.distributed_optimizer(fluid.optimizer.Adam(),
+                                                        strategy)
+                optimizer.minimize(cost)
+
+        final_strategy = fleet._final_strategy()
+        assert final_strategy.gradient_scale_configs['scale_strategy'] == 'sum'
+
+
+if __name__ == "__main__":
+    unittest.main()