1# 2# Copyright (C) 2018 The Android Open Source Project 3# 4# Licensed under the Apache License, Version 2.0 (the "License"); 5# you may not use this file except in compliance with the License. 6# You may obtain a copy of the License at 7# 8# http://www.apache.org/licenses/LICENSE-2.0 9# 10# Unless required by applicable law or agreed to in writing, software 11# distributed under the License is distributed on an "AS IS" BASIS, 12# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13# See the License for the specific language governing permissions and 14# limitations under the License. 15# 16 17batches = 2 18units = 16 19input_size = 8 20 21model = Model() 22 23input = Input("input", "TENSOR_FLOAT32", "{%d, %d}" % (batches, input_size)) 24weights = Input("weights", "TENSOR_FLOAT32", "{%d, %d}" % (units, input_size)) 25recurrent_weights = Input("recurrent_weights", "TENSOR_FLOAT32", "{%d, %d}" % (units, units)) 26bias = Input("bias", "TENSOR_FLOAT32", "{%d}" % (units)) 27hidden_state_in = Input("hidden_state_in", "TENSOR_FLOAT32", "{%d, %d}" % (batches, units)) 28 29activation_param = Int32Scalar("activation_param", 1) # Relu 30 31hidden_state_out = IgnoredOutput("hidden_state_out", "TENSOR_FLOAT32", "{%d, %d}" % (batches, units)) 32output = Output("output", "TENSOR_FLOAT32", "{%d, %d}" % (batches, units)) 33 34model = model.Operation("RNN", input, weights, recurrent_weights, bias, hidden_state_in, 35 activation_param).To([hidden_state_out, output]) 36model = model.RelaxedExecution(True) 37 38input0 = { 39 weights: [ 40 0.461459, 0.153381, 0.529743, -0.00371218, 0.676267, -0.211346, 41 0.317493, 0.969689, -0.343251, 0.186423, 0.398151, 0.152399, 42 0.448504, 0.317662, 0.523556, -0.323514, 0.480877, 0.333113, 43 -0.757714, -0.674487, -0.643585, 0.217766, -0.0251462, 0.79512, 44 -0.595574, -0.422444, 0.371572, -0.452178, -0.556069, -0.482188, 45 -0.685456, -0.727851, 0.841829, 0.551535, -0.232336, 0.729158, 46 -0.00294906, -0.69754, 0.766073, -0.178424, 0.369513, -0.423241, 47 0.548547, -0.0152023, -0.757482, -0.85491, 0.251331, -0.989183, 48 0.306261, -0.340716, 0.886103, -0.0726757, -0.723523, -0.784303, 49 0.0354295, 0.566564, -0.485469, -0.620498, 0.832546, 0.697884, 50 -0.279115, 0.294415, -0.584313, 0.548772, 0.0648819, 0.968726, 51 0.723834, -0.0080452, -0.350386, -0.272803, 0.115121, -0.412644, 52 -0.824713, -0.992843, -0.592904, -0.417893, 0.863791, -0.423461, 53 -0.147601, -0.770664, -0.479006, 0.654782, 0.587314, -0.639158, 54 0.816969, -0.337228, 0.659878, 0.73107, 0.754768, -0.337042, 55 0.0960841, 0.368357, 0.244191, -0.817703, -0.211223, 0.442012, 56 0.37225, -0.623598, -0.405423, 0.455101, 0.673656, -0.145345, 57 -0.511346, -0.901675, -0.81252, -0.127006, 0.809865, -0.721884, 58 0.636255, 0.868989, -0.347973, -0.10179, -0.777449, 0.917274, 59 0.819286, 0.206218, -0.00785118, 0.167141, 0.45872, 0.972934, 60 -0.276798, 0.837861, 0.747958, -0.0151566, -0.330057, -0.469077, 61 0.277308, 0.415818 62 ], 63 recurrent_weights: [ 64 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 65 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 66 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 67 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 68 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 69 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 70 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 71 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 72 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 73 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 74 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 75 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 76 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 77 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 78 0.1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 79 0.1 80 ], 81 bias: [ 82 0.065691948, -0.69055247, 0.1107955, -0.97084129, -0.23957068, 83 -0.23566568, -0.389184, 0.47481549, -0.4791103, 0.29931796, 84 0.10463274, 0.83918178, 0.37197268, 0.61957061, 0.3956964, 85 -0.37609905 86 ], 87} 88 89 90test_inputs = [ 91 0.23689353, 0.285385, 0.037029743, -0.19858193, -0.27569133, 92 0.43773448, 0.60379338, 0.35562468, -0.69424844, -0.93421471, 93 -0.87287879, 0.37144363, -0.62476718, 0.23791671, 0.40060222, 94 0.1356622, -0.99774903, -0.98858172, -0.38952237, -0.47685933, 95 0.31073618, 0.71511042, -0.63767755, -0.31729108, 0.33468103, 96 0.75801885, 0.30660987, -0.37354088, 0.77002847, -0.62747043, 97 -0.68572164, 0.0069220066, 0.65791464, 0.35130811, 0.80834007, 98 -0.61777675, -0.21095741, 0.41213346, 0.73784804, 0.094794154, 99 0.47791874, 0.86496925, -0.53376222, 0.85315156, 0.10288584, 100 0.86684, -0.011186242, 0.10513687, 0.87825835, 0.59929144, 101 0.62827742, 0.18899453, 0.31440187, 0.99059987, 0.87170351, 102 -0.35091716, 0.74861872, 0.17831337, 0.2755419, 0.51864719, 103 0.55084288, 0.58982027, -0.47443086, 0.20875752, -0.058871567, 104 -0.66609079, 0.59098077, 0.73017097, 0.74604273, 0.32882881, 105 -0.17503482, 0.22396147, 0.19379807, 0.29120302, 0.077113032, 106 -0.70331609, 0.15804303, -0.93407321, 0.40182066, 0.036301374, 107 0.66521823, 0.0300982, -0.7747041, -0.02038002, 0.020698071, 108 -0.90300065, 0.62870288, -0.23068321, 0.27531278, -0.095755219, 109 -0.712036, -0.17384434, -0.50593495, -0.18646687, -0.96508682, 110 0.43519354, 0.14744234, 0.62589407, 0.1653645, -0.10651493, 111 -0.045277178, 0.99032974, -0.88255352, -0.85147917, 0.28153265, 112 0.19455957, -0.55479527, -0.56042433, 0.26048636, 0.84702539, 113 0.47587705, -0.074295521, -0.12287641, 0.70117295, 0.90532446, 114 0.89782166, 0.79817224, 0.53402734, -0.33286154, 0.073485017, 115 -0.56172788, -0.044897556, 0.89964068, -0.067662835, 0.76863563, 116 0.93455386, -0.6324693, -0.083922029 117] 118 119golden_outputs = [ 120 0.496726, 0, 0.965996, 0, 0.0584254, 0, 121 0, 0.12315, 0, 0, 0.612266, 0.456601, 122 0, 0.52286, 1.16099, 0.0291232, 123 124 0, 0, 0.524901, 0, 0, 0, 125 0, 1.02116, 0, 1.35762, 0, 0.356909, 126 0.436415, 0.0355727, 0, 0, 127 128 0, 0, 0, 0.262335, 0, 0, 129 0, 1.33992, 0, 2.9739, 0, 0, 130 1.31914, 2.66147, 0, 0, 131 132 0.942568, 0, 0, 0, 0.025507, 0, 133 0, 0, 0.321429, 0.569141, 1.25274, 1.57719, 134 0.8158, 1.21805, 0.586239, 0.25427, 135 136 1.04436, 0, 0.630725, 0, 0.133801, 0.210693, 137 0.363026, 0, 0.533426, 0, 1.25926, 0.722707, 138 0, 1.22031, 1.30117, 0.495867, 139 140 0.222187, 0, 0.72725, 0, 0.767003, 0, 141 0, 0.147835, 0, 0, 0, 0.608758, 142 0.469394, 0.00720298, 0.927537, 0, 143 144 0.856974, 0.424257, 0, 0, 0.937329, 0, 145 0, 0, 0.476425, 0, 0.566017, 0.418462, 146 0.141911, 0.996214, 1.13063, 0, 147 148 0.967899, 0, 0, 0, 0.0831304, 0, 149 0, 1.00378, 0, 0, 0, 1.44818, 150 1.01768, 0.943891, 0.502745, 0, 151 152 0.940135, 0, 0, 0, 0, 0, 153 0, 2.13243, 0, 0.71208, 0.123918, 1.53907, 154 1.30225, 1.59644, 0.70222, 0, 155 156 0.804329, 0, 0.430576, 0, 0.505872, 0.509603, 157 0.343448, 0, 0.107756, 0.614544, 1.44549, 1.52311, 158 0.0454298, 0.300267, 0.562784, 0.395095, 159 160 0.228154, 0, 0.675323, 0, 1.70536, 0.766217, 161 0, 0, 0, 0.735363, 0.0759267, 1.91017, 162 0.941888, 0, 0, 0, 163 164 0, 0, 1.5909, 0, 0, 0, 165 0, 0.5755, 0, 0.184687, 0, 1.56296, 166 0.625285, 0, 0, 0, 167 168 0, 0, 0.0857888, 0, 0, 0, 169 0, 0.488383, 0.252786, 0, 0, 0, 170 1.02817, 1.85665, 0, 0, 171 172 0.00981836, 0, 1.06371, 0, 0, 0, 173 0, 0, 0, 0.290445, 0.316406, 0, 174 0.304161, 1.25079, 0.0707152, 0, 175 176 0.986264, 0.309201, 0, 0, 0, 0, 177 0, 1.64896, 0.346248, 0, 0.918175, 0.78884, 178 0.524981, 1.92076, 2.07013, 0.333244, 179 180 0.415153, 0.210318, 0, 0, 0, 0, 181 0, 2.02616, 0, 0.728256, 0.84183, 0.0907453, 182 0.628881, 3.58099, 1.49974, 0 183] 184 185input_sequence_size = int(len(test_inputs) / input_size / batches) 186 187# TODO: enable the other data points after fixing reference issues 188#for i in range(input_sequence_size): 189for i in range(1): 190 input_begin = i * input_size 191 input_end = input_begin + input_size 192 input0[input] = test_inputs[input_begin:input_end] 193 input0[input].extend(input0[input]) 194 input0[hidden_state_in] = [0 for x in range(batches * units)] 195 output0 = { 196 hidden_state_out: [0 for x in range(batches * units)], 197 } 198 golden_start = i * units 199 golden_end = golden_start + units 200 output0[output] = golden_outputs[golden_start:golden_end] 201 output0[output].extend(output0[output]) 202 Example((input0, output0)) 203