adapt parallel interface change

chapter05/resnet/resnet_cifar.py

@@ -31,7 +31,8 @@ import mindspore.dataset.transforms.c_transforms as C2
 from mindspore import Tensor
 from mindspore.ops import operations as P
 from mindspore.nn.optim.momentum import Momentum
-from mindspore.train.model import Model, ParallelMode
+from mindspore.train.model import Model
+from mindspore.context import ParallelMode
 from mindspore import context
 from mindspore.train.callback import ModelCheckpoint, CheckpointConfig, LossMonitor
 from mindspore.train.serialization import load_checkpoint, load_param_into_net
@@ -113,7 +114,7 @@ def create_dataset(repeat_num=1, training=True):
 
 if __name__ == '__main__':
     if args_opt.mode == 'train' and args_opt.run_distribute:
-        context.set_auto_parallel_context(device_num=args_opt.device_num, parallel_mode=ParallelMode.DATA_PARALLEL, mirror_mean=True)
+        context.set_auto_parallel_context(device_num=args_opt.device_num, parallel_mode=ParallelMode.DATA_PARALLEL, gradients_mean=True)
         auto_parallel_context().set_all_reduce_fusion_split_indices([140])
         init()
 

chapter07/run_pretrain.py

@@ -24,7 +24,7 @@ import mindspore.communication.management as D
 import mindspore.common.dtype as mstype
 from mindspore import context
 from mindspore.train.model import Model
-from mindspore.train.parallel_utils import ParallelMode
+from mindspore.context import ParallelMode
 from mindspore.nn.wrap.loss_scale import DynamicLossScaleUpdateCell
 from mindspore.train.callback import ModelCheckpoint, CheckpointConfig, TimeMonitor
 from mindspore.train.serialization import load_checkpoint, load_param_into_net
@@ -77,7 +77,7 @@ def run_pretrain():
             ckpt_save_dir = args_opt.save_checkpoint_path + 'ckpt_' + str(rank) + '/'
 
         context.reset_auto_parallel_context()
-        context.set_auto_parallel_context(parallel_mode=ParallelMode.DATA_PARALLEL, mirror_mean=True,
+        context.set_auto_parallel_context(parallel_mode=ParallelMode.DATA_PARALLEL, gradients_mean=True,
                                           device_num=device_num)
         from mindspore.parallel._auto_parallel_context import auto_parallel_context
         if bert_net_cfg.num_hidden_layers == 12:

chapter07/src/bert_for_pre_training.py

@@ -24,7 +24,7 @@ from mindspore.common.tensor import Tensor
 from mindspore.common.parameter import Parameter, ParameterTuple
 from mindspore.common import dtype as mstype
 from mindspore.nn.wrap.grad_reducer import DistributedGradReducer
-from mindspore.train.parallel_utils import ParallelMode
+from mindspore.context import ParallelMode
 from mindspore.communication.management import get_group_size
 from mindspore import context
 from mindspore.ops import _selected_ops
@@ -280,7 +280,7 @@ class BertTrainOneStepCell(nn.Cell):
             self.reducer_flag = True
         self.grad_reducer = None
         if self.reducer_flag:
-            mean = context.get_auto_parallel_context("mirror_mean")
+            mean = context.get_auto_parallel_context("gradients_mean")
             degree = get_group_size()
             self.grad_reducer = DistributedGradReducer(optimizer.parameters, mean, degree)