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 16import pytest 17 18import mindspore.context as context 19from mindspore import Tensor 20from mindspore.nn import Cell 21from mindspore.ops import composite as C 22from mindspore.ops.operations import Minimum 23 24context.set_context(mode=context.GRAPH_MODE, device_target="CPU") 25grad = C.GradOperation(get_all=True, sens_param=True) 26 27 28class MinNetMe(Cell): 29 def __init__(self): 30 super(MinNetMe, self).__init__() 31 self.min = Minimum() 32 33 def construct(self, inputA, inputB): 34 x = self.min(inputA, inputB) 35 return x 36 37 38class GradWrap(Cell): 39 def __init__(self, network): 40 super(GradWrap, self).__init__() 41 self.network = network 42 43 def construct(self, inputA, inputB, sens): 44 gout = grad(self.network)(inputA, inputB, sens) 45 return gout 46 47 48def gen_data(inputA_np, inputB_np, grad_=None): 49 inputA_me = inputA_np 50 if isinstance(inputA_np, np.ndarray): 51 inputA_me = Tensor(inputA_me) 52 53 inputB_me = inputB_np 54 if isinstance(inputB_np, np.ndarray): 55 inputB_me = Tensor(inputB_np) 56 57 if grad_ is None: 58 grad_ = Tensor(grad_) 59 60 net_me = GradWrap(MinNetMe()) 61 net_me.set_train() 62 output = net_me(inputA_me, inputB_me, Tensor(grad_)) 63 return output 64 65 66@pytest.mark.level1 67@pytest.mark.platform_x86_cpu 68@pytest.mark.env_onecard 69def test_min_tensor_grad_4d(): 70 inputA_np = np.random.randn(1, 3, 2, 2).astype(np.float32) 71 inputB_np = np.random.randn(1, 3, 2, 2).astype(np.float32) 72 grad_ = np.random.randn(1, 3, 2, 2).astype(np.float32) 73 output = gen_data(inputA_np, inputB_np, grad_) 74 print(output[0].asnumpy()) 75 print(output[1].asnumpy()) 76 77 78@pytest.mark.level0 79@pytest.mark.platform_x86_cpu 80@pytest.mark.env_onecard 81def test_min_tensor_grad_result(): 82 inputA = np.array([[[[0.659578], [0.49113268], [0.75909054], [0.71681815], [0.30421826]]], 83 [[[0.30322495], [0.02858258], [0.06398096], [0.09519596], [0.12498625]]], 84 [[[0.7347768], [0.166469], [0.328553], [0.54908437], [0.23673844]]]]).astype(np.float32) 85 inputB = np.array([[[[0.9154968, 0.29014662, 0.6492294, 0.39918253, 0.1648203, 0.00861965]], 86 [[0.996885, 0.24152198, 0.3601213, 0.51664376, 0.7933056, 0.84706444]], 87 [[0.75606346, 0.974512, 0.3939527, 0.69697475, 0.83400667, 0.6348955]], 88 [[0.68492866, 0.24609096, 0.4924665, 0.22500521, 0.38474053, 0.5586104]]]]).astype(np.float32) 89 grad_ = np.array([[[[0.42891738, 0.03434946, 0.06192983, 0.21216309, 0.37450036, 0.6619524], 90 [0.8583447, 0.5765161, 0.1468952, 0.9975385, 0.6908136, 0.4903796], 91 [0.68952006, 0.39336833, 0.9049695, 0.66886294, 0.2338471, 0.913618], 92 [0.0428149, 0.6243054, 0.8519898, 0.12088962, 0.9735885, 0.45661286], 93 [0.41563734, 0.41607043, 0.4754915, 0.32207987, 0.33823156, 0.47422352]], 94 95 [[0.64478457, 0.22430937, 0.7682554, 0.46082005, 0.8938723, 0.20490853], 96 [0.44393885, 0.08278944, 0.4734108, 0.5543551, 0.39428464, 0.44424313], 97 [0.12612297, 0.76566416, 0.71133816, 0.81280327, 0.20583127, 0.54058075], 98 [0.41341263, 0.48118508, 0.00401995, 0.37259838, 0.05435474, 0.5240658], 99 [0.4081956, 0.48718935, 0.9132831, 0.67969185, 0.0119757, 0.8328054]], 100 101 [[0.91695577, 0.95370644, 0.263782, 0.7477626, 0.6448147, 0.8080634], 102 [0.15576603, 0.9104615, 0.3778708, 0.6912833, 0.2092224, 0.67462957], 103 [0.7087075, 0.7888326, 0.4672294, 0.98221505, 0.25210258, 0.98920417], 104 [0.7466197, 0.22702982, 0.01991269, 0.6846591, 0.7515228, 0.5890395], 105 [0.04531088, 0.21740614, 0.8406235, 0.36480767, 0.37733936, 0.02914464]], 106 107 [[0.33069974, 0.5497569, 0.9896345, 0.4167176, 0.78057563, 0.04659131], 108 [0.7747768, 0.21427679, 0.29893255, 0.7706969, 0.9755185, 0.42388415], 109 [0.3910244, 0.39381978, 0.37065396, 0.15558061, 0.05012341, 0.15870963], 110 [0.17791101, 0.47219893, 0.13899496, 0.32323205, 0.3628809, 0.02580585], 111 [0.30274773, 0.62890774, 0.11024303, 0.6980051, 0.35346958, 0.062852]]], 112 113 [[[0.6925081, 0.74668753, 0.80145043, 0.06598313, 0.665123, 0.15073007], 114 [0.11784806, 0.6385372, 0.5228278, 0.5349848, 0.84671104, 0.8096436], 115 [0.09516156, 0.63298017, 0.52382874, 0.36734378, 0.66497755, 0.6019127], 116 [0.46438488, 0.0194377, 0.9388292, 0.7286089, 0.29178405, 0.11872514], 117 [0.22101837, 0.6164887, 0.6139798, 0.11711904, 0.6227745, 0.09701069]], 118 119 [[0.80480653, 0.90034056, 0.8633447, 0.97415197, 0.08309154, 0.8446033], 120 [0.9473769, 0.791024, 0.26339203, 0.01155075, 0.2673186, 0.7116369], 121 [0.9687511, 0.24281934, 0.37777108, 0.09802654, 0.2421312, 0.87095344], 122 [0.6311381, 0.23368953, 0.0998995, 0.4364419, 0.9187446, 0.5043872], 123 [0.35226053, 0.09357589, 0.41317305, 0.85930043, 0.16249318, 0.5478765]], 124 125 [[0.14338651, 0.24859418, 0.4246941, 0.73034066, 0.47172204, 0.8717199], 126 [0.05415315, 0.78556925, 0.99214983, 0.7415298, 0.673708, 0.87817156], 127 [0.616975, 0.42843062, 0.05179814, 0.1566958, 0.04536059, 0.70166487], 128 [0.15493333, 0.776598, 0.4361967, 0.40253627, 0.89210516, 0.8144414], 129 [0.04816005, 0.29696834, 0.4586605, 0.3419852, 0.5595613, 0.74093205]], 130 131 [[0.1388035, 0.9168704, 0.64287645, 0.83864623, 0.48026922, 0.78323376], 132 [0.12724937, 0.83034366, 0.42557436, 0.50578654, 0.25630295, 0.15349793], 133 [0.27256685, 0.04547984, 0.5385756, 0.39270344, 0.7661698, 0.23722854], 134 [0.24620503, 0.25431684, 0.71564585, 0.01161419, 0.846467, 0.7043044], 135 [0.63272387, 0.11857849, 0.3772076, 0.16758402, 0.46743023, 0.05919575]]], 136 137 [[[0.18827082, 0.8912264, 0.6841404, 0.74436826, 0.9582085, 0.1083683], 138 [0.60695344, 0.09742349, 0.25074378, 0.87940735, 0.21116392, 0.39418384], 139 [0.744686, 0.35679692, 0.01308284, 0.45166633, 0.68166, 0.8634658], 140 [0.7331758, 0.21113694, 0.3935488, 0.87934476, 0.70728546, 0.09309767], 141 [0.12128611, 0.93696386, 0.81177396, 0.85402405, 0.5827289, 0.9776509]], 142 143 [[0.54069614, 0.66651285, 0.10646132, 0.17342485, 0.88795924, 0.03551182], 144 [0.25531697, 0.87946486, 0.74267226, 0.89230734, 0.95171434, 0.94697934], 145 [0.3708397, 0.507355, 0.97099817, 0.4918163, 0.17212386, 0.5008048], 146 [0.62530744, 0.25210327, 0.73966664, 0.71555346, 0.82484317, 0.6094874], 147 [0.4589691, 0.1386695, 0.27448782, 0.20373994, 0.27805242, 0.23292768]], 148 149 [[0.7414099, 0.2270226, 0.90431255, 0.47035843, 0.9581062, 0.5359226], 150 [0.79603523, 0.45549425, 0.80858237, 0.7705133, 0.017761, 0.98001194], 151 [0.06013146, 0.99240226, 0.33515573, 0.04110833, 0.41470334, 0.7130743], 152 [0.5687417, 0.5788611, 0.00722461, 0.6603336, 0.3420471, 0.75181854], 153 [0.4699261, 0.51390815, 0.343182, 0.81498754, 0.8942413, 0.46532857]], 154 155 [[0.4589523, 0.5534698, 0.2825786, 0.8205943, 0.78258514, 0.43154418], 156 [0.27020997, 0.01667354, 0.60871965, 0.90670526, 0.3208025, 0.96995634], 157 [0.85337156, 0.9711295, 0.1381724, 0.53670496, 0.7347996, 0.73380876], 158 [0.6137464, 0.54751194, 0.9037335, 0.23134394, 0.61411524, 0.26583543], 159 [0.70770144, 0.01813207, 0.24718016, 0.70329237, 0.7062925, 0.14399007]]]]).astype(np.float32) 160 output = gen_data(inputA, inputB, grad_) 161 expect0 = np.array([[[[5.7664223], [6.9810176], [2.6029902], [2.7598205], [6.763105]]], 162 [[[10.065580], [12.077245], [9.3383940], [11.522709], [8.889048]]], 163 [[[3.5789766], [13.424448], [8.7327460], [6.9677467], [9.635764]]]], np.float32) 164 expect1 = np.array([[[[0., 4.2504573, 2.5030296, 3.623167, 6.417151, 7.2115746]], 165 [[0., 4.3674493, 2.8031523, 2.5352, 0., 0.]], 166 [[0.7087075, 0., 2.040332, 2.1372325, 0., 2.9222295]], 167 [[1.0278877, 5.247942, 2.6855955, 5.494814, 3.565799, 0.66265094]]]], np.float32) 168 error0 = np.ones(shape=expect0.shape) * 1.0e-5 169 error1 = np.ones(shape=expect1.shape) * 1.0e-5 170 assert np.all(np.abs(output[0].asnumpy() - expect0) < error0) 171 assert np.all(np.abs(output[1].asnumpy() - expect1) < error1) 172