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# ============================================================================ 15""" 16Function: 17 test network 18Usage: 19 python test_network_main.py --net lenet --target Ascend 20""" 21import argparse 22 23import numpy as np 24from models.alexnet import AlexNet 25from models.lenet import LeNet 26from models.resnetv1_5 import resnet50 27 28import mindspore.context as context 29import mindspore.nn as nn 30from mindspore import Tensor 31from mindspore.nn import TrainOneStepCell, WithLossCell 32from mindspore.nn.optim import Momentum 33 34context.set_context(mode=context.GRAPH_MODE, device_target="Ascend") 35 36 37def train(net, data, label): 38 learning_rate = 0.01 39 momentum = 0.9 40 41 optimizer = Momentum(filter(lambda x: x.requires_grad, net.get_parameters()), learning_rate, momentum) 42 criterion = nn.SoftmaxCrossEntropyWithLogits(sparse=True) 43 net_with_criterion = WithLossCell(net, criterion) 44 train_network = TrainOneStepCell(net_with_criterion, optimizer) # optimizer 45 train_network.set_train() 46 res = train_network(data, label) 47 print(res) 48 assert res 49 50 51def test_resnet50(): 52 data = Tensor(np.ones([32, 3, 224, 224]).astype(np.float32) * 0.01) 53 label = Tensor(np.ones([32]).astype(np.int32)) 54 net = resnet50(32, 10) 55 train(net, data, label) 56 57 58def test_lenet(): 59 net = LeNet() 60 data = Tensor(np.ones([net.batch_size, 3, 32, 32]).astype(np.float32) * 0.01) 61 label = Tensor(np.ones([net.batch_size]).astype(np.int32)) 62 train(net, data, label) 63 64 65def test_alexnet(): 66 data = Tensor(np.ones([32, 3, 227, 227]).astype(np.float32) * 0.01) 67 label = Tensor(np.ones([32]).astype(np.int32)) 68 net = AlexNet() 69 train(net, data, label) 70 71 72parser = argparse.ArgumentParser(description='MindSpore Testing Network') 73parser.add_argument('--net', default='resnet50', type=str, help='net name') 74parser.add_argument('--device', default='Ascend', type=str, help='device target') 75if __name__ == "__main__": 76 args = parser.parse_args() 77 context.set_context(device_target=args.device) 78 if args.net == 'resnet50': 79 test_resnet50() 80 elif args.net == 'lenet': 81 test_lenet() 82 elif args.net == 'alexnet': 83 test_alexnet() 84 else: 85 print("Please add net name like --net lenet") 86