• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1# Copyright 2020 Huawei Technologies Co., Ltd
2#
3# Licensed under the Apache License, Version 2.0 (the "License");
4# you may not use this file except in compliance with the License.
5# You may obtain a copy of the License at
6#
7# http://www.apache.org/licenses/LICENSE-2.0
8#
9# Unless required by applicable law or agreed to in writing, software
10# distributed under the License is distributed on an "AS IS" BASIS,
11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12# See the License for the specific language governing permissions and
13# limitations under the License.
14# ============================================================================
15import numpy as np
16
17import mindspore as ms
18from mindspore import context, Tensor, Parameter
19from mindspore.common.api import _cell_graph_executor
20from mindspore.nn import Cell
21from mindspore.ops import operations as P
22
23
24class Net(Cell):
25    def __init__(self, mul_weight, strategy1=None, strategy2=None):
26        super().__init__()
27        self.mul = P.Mul().shard(strategy1)
28        self.neg = P.Neg().shard(strategy2)
29        self.mul_weight = Parameter(mul_weight, "w1")
30
31    def construct(self, x, b):
32        out = self.mul(x, self.mul_weight)
33        out = self.neg(out)
34        return out
35
36
37class EvalNet(Cell):
38    def __init__(self, network, strategy2=None):
39        super().__init__()
40        self.network = network
41        self.relu = P.ReLU().shard(strategy2)
42
43    def construct(self, x, b):
44        out = self.network(x, b)
45        out1 = self.relu(out)
46        return out, out1
47
48
49_x = Tensor(np.ones([64, 64]), dtype=ms.float32)
50_w1 = Tensor(np.ones([64, 64]), dtype=ms.float32)
51_b = Tensor(np.ones([64, 64]), dtype=ms.float32)
52
53
54def test_train_and_eval():
55    context.set_context(mode=0)
56    context.set_auto_parallel_context(parallel_mode="semi_auto_parallel", device_num=16)
57    strategy1 = ((4, 4), (4, 4))
58    strategy2 = ((4, 4),)
59    net = Net(_w1, strategy1, strategy2)
60    eval_net = EvalNet(net, strategy2=strategy2)
61    net.set_auto_parallel()
62    net.set_train()
63    _cell_graph_executor.compile(net, _x, _b, phase='train', auto_parallel_mode=True)
64
65    eval_net.set_train(mode=False)
66    eval_net.set_auto_parallel()
67    _cell_graph_executor.compile(eval_net, _x, _b, phase='eval', auto_parallel_mode=True)
68
69    context.reset_auto_parallel_context()
70
71def test_train_and_eval_auto():
72    context.set_context(mode=0)
73    context.set_auto_parallel_context(parallel_mode="auto_parallel", device_num=16)
74    strategy1 = ((4, 4), (4, 4))
75    strategy2 = ((4, 4),)
76    net = Net(_w1, strategy1, strategy2)
77    eval_net = EvalNet(net, strategy2=strategy2)
78    net.set_auto_parallel()
79    net.set_train()
80    _cell_graph_executor.compile(net, _x, _b, phase='train', auto_parallel_mode=True)
81
82    eval_net.set_train(mode=False)
83    eval_net.set_auto_parallel()
84    _cell_graph_executor.compile(eval_net, _x, _b, phase='eval', auto_parallel_mode=True)
85
86    context.reset_auto_parallel_context()
87