• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1# Copyright 2019 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.nn as nn
18from mindspore import Tensor, context
19from mindspore.common.api import _cell_graph_executor
20from mindspore.ops import operations as P
21from ....train_step_wrap import train_step_with_loss_warp
22
23
24class DenseMutMulNet(nn.Cell):
25    def __init__(self):
26        super(DenseMutMulNet, self).__init__()
27        self.fc1 = nn.Dense(128, 768, activation='relu')
28        self.fc2 = nn.Dense(128, 768, activation='relu')
29        self.fc3 = nn.Dense(128, 768, activation='relu')
30        self.fc4 = nn.Dense(768, 768, activation='relu')
31        self.relu4 = nn.ReLU()
32        self.relu5 = nn.ReLU()
33        self.transpose = P.Transpose()
34        self.matmul1 = P.MatMul()
35        self.matmul2 = P.MatMul()
36
37    def construct(self, x):
38        q = self.fc1(x)
39        k = self.fc2(x)
40        v = self.fc3(x)
41        k = self.transpose(k, (1, 0))
42        c = self.relu4(self.matmul1(q, k))
43        s = self.relu5(self.matmul2(c, v))
44        s = self.fc4(s)
45        return s
46
47
48def test_dmnet_train_step():
49    context.reset_auto_parallel_context()
50    input_ = Tensor(np.ones([32, 128]).astype(np.float32) * 0.01)
51    label = Tensor(np.zeros([32, 768]).astype(np.float32))
52    net = DenseMutMulNet()
53    net = train_step_with_loss_warp(DenseMutMulNet())
54    net.set_train()
55    _cell_graph_executor.compile(net, input_, label)
56