test_adam_op.py 10.6 KB
Newer Older
1
#   Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
D
dzhwinter 已提交
2
#
D
dzhwinter 已提交
3 4 5
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
D
dzhwinter 已提交
6
#
D
dzhwinter 已提交
7
#     http://www.apache.org/licenses/LICENSE-2.0
D
dzhwinter 已提交
8
#
D
dzhwinter 已提交
9 10 11 12 13 14
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

15 16 17
import unittest
import numpy as np
from op_test import OpTest
T
typhoonzero 已提交
18 19
from paddle.v2.fluid import core
from paddle.v2.fluid.op import Operator
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51


class TestAdamOp1(OpTest):
    def setUp(self):
        '''Test Adam Op with supplied attributes
        '''
        self.op_type = "adam"
        param = np.random.uniform(-1, 1, (102, 105)).astype("float32")
        grad = np.random.uniform(-1, 1, (102, 105)).astype("float32")
        moment1 = np.random.uniform(-1, 1, (102, 105)).astype("float32")
        # The second moment is positive
        moment2 = np.random.random((102, 105)).astype("float32")

        learning_rate = 0.004
        beta1 = 0.78
        beta2 = 0.836
        epsilon = 1e-4
        beta1_pow = beta1**10
        beta2_pow = beta2**10

        self.inputs = {
            'Param': param,
            'Grad': grad,
            'Moment1': moment1,
            'Moment2': moment2,
            'LearningRate': np.array([learning_rate]).astype("float32"),
            'Beta1Pow': np.array([beta1_pow]).astype("float32"),
            'Beta2Pow': np.array([beta2_pow]).astype("float32")
        }

        self.attrs = {'epsilon': epsilon, 'beta1': beta1, 'beta2': beta2}

52 53
        param_out, moment1_out, \
            moment2_out = adam_step(self.inputs, self.attrs)
54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94

        self.outputs = {
            'Moment1Out': moment1_out,
            'Moment2Out': moment2_out,
            'ParamOut': param_out
        }

    def test_check_output(self):
        self.check_output()


class TestAdamOp2(OpTest):
    def setUp(self):
        '''Test Adam Op with supplied attributes
        '''
        self.op_type = "adam"
        param = np.random.uniform(-1, 1, (102, 105)).astype("float32")
        grad = np.random.uniform(-1, 1, (102, 105)).astype("float32")
        moment1 = np.random.uniform(-1, 1, (102, 105)).astype("float32")
        # The second moment is positive
        moment2 = np.random.random((102, 105)).astype("float32")

        learning_rate = 0.001
        beta1 = 0.9
        beta2 = 0.999
        epsilon = 1e-8
        beta1_pow = beta1**10
        beta2_pow = beta2**10

        self.inputs = {
            'Param': param,
            'Grad': grad,
            'Moment1': moment1,
            'Moment2': moment2,
            'LearningRate': np.array([learning_rate]).astype("float32"),
            'Beta1Pow': np.array([beta1_pow]).astype("float32"),
            'Beta2Pow': np.array([beta2_pow]).astype("float32")
        }

        attributes = {'epsilon': epsilon, 'beta1': beta1, 'beta2': beta2}

95 96
        param_out, moment1_out, \
            moment2_out = adam_step(self.inputs, attributes)
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141

        self.outputs = {
            'Moment1Out': moment1_out,
            'Moment2Out': moment2_out,
            'ParamOut': param_out
        }

    def test_check_output(self):
        self.check_output()


class TestAdamOpMultipleSteps(OpTest):
    def setUp(self):
        '''Test Adam Operator with supplied attributes
        '''
        self.op_type = "adam"
        self.num_steps = 10

        param = np.random.uniform(-1, 1, (102, 105)).astype("float32")
        grad = np.random.uniform(-1, 1, (102, 105)).astype("float32")
        moment1 = np.random.uniform(-1, 1, (102, 105)).astype("float32")
        # The second moment is positive
        moment2 = np.random.random((102, 105)).astype("float32")

        learning_rate = 0.001
        beta1 = 0.9
        beta2 = 0.999
        epsilon = 1e-8
        beta1_pow = beta1**10
        beta2_pow = beta2**10

        self.inputs = {
            'Param': param,
            'Grad': grad,
            'Moment1': moment1,
            'Moment2': moment2,
            'LearningRate': np.array([learning_rate]).astype("float32"),
            'Beta1Pow': np.array([beta1_pow]).astype("float32"),
            'Beta2Pow': np.array([beta2_pow]).astype("float32")
        }

        self.attrs = {'epsilon': epsilon, 'beta1': beta1, 'beta2': beta2}

    def test_check_output(self):
        for _ in range(self.num_steps):
142 143
            param_out, moment1_out, \
                moment2_out = adam_step(self.inputs, self.attrs)
144 145 146 147 148 149 150 151 152 153 154 155 156 157

            self.outputs = {
                'Moment1Out': moment1_out,
                'Moment2Out': moment2_out,
                'ParamOut': param_out
            }

            # Verify output for this step
            self.check_output()

            # Output of this step becomes input for next step
            self.inputs['Param'] = param_out
            self.inputs['Moment1'] = moment1_out
            self.inputs['Moment2'] = moment2_out
158 159 160 161

            # Update powers of Beta1 and Beta2 for next time step
            self.inputs['Beta1Pow'] *= self.attrs['beta1']
            self.inputs['Beta2Pow'] *= self.attrs['beta1']
162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189

            # Randomize gradient for next step
            self.inputs['Grad'] = np.random.uniform(
                -1, 1, (102, 105)).astype("float32")


def adam_step(inputs, attributes):
    '''
    Simulate one step of the adam optimizer
    :param inputs: dict of inputs
    :param attributes: dict of attributes
    :return tuple: tuple of output param, moment1, moment2,
    beta1 power accumulator and beta2 power accumulator
    '''
    param = inputs['Param']
    grad = inputs['Grad']
    moment1 = inputs['Moment1']
    moment2 = inputs['Moment2']
    lr = inputs['LearningRate']
    beta1_pow = inputs['Beta1Pow']
    beta2_pow = inputs['Beta2Pow']

    beta1 = attributes['beta1']
    beta2 = attributes['beta2']
    epsilon = attributes['epsilon']

    moment1_out = beta1 * moment1 + (1 - beta1) * grad
    moment2_out = beta2 * moment2 + (1 - beta2) * np.square(grad)
190
    lr_t = lr * np.sqrt(1 - beta2_pow) / (1 - beta1_pow)
191
    param_out = param - lr_t * (moment1_out / (np.sqrt(moment2_out) + epsilon))
192
    return param_out, moment1_out, moment2_out
193 194


T
wip  
typhoonzero 已提交
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214
def adam_step_sparse(inputs, attributes, height, rows, row_numel, np_grad):
    '''
    Simulate one step of the adam optimizer
    :param inputs: dict of inputs
    :param attributes: dict of attributes
    :return tuple: tuple of output param, moment1, moment2,
    beta1 power accumulator and beta2 power accumulator
    '''
    param = inputs['Param']
    # grad = inputs['Grad']
    moment1 = inputs['Moment1']
    moment2 = inputs['Moment2']
    lr = inputs['LearningRate']
    beta1_pow = inputs['Beta1Pow']
    beta2_pow = inputs['Beta2Pow']

    beta1 = attributes['beta1']
    beta2 = attributes['beta2']
    epsilon = attributes['epsilon']

T
typhoonzero 已提交
215 216 217
    moment1_out = np.zeros(shape=[height, row_numel])
    moment2_out = np.zeros(shape=[height, row_numel])
    param_out = np.zeros(shape=[height, row_numel])
T
wip  
typhoonzero 已提交
218 219 220 221 222 223 224

    for idx, row_id in enumerate(rows):
        moment1_out[row_id] = beta1 * moment1[row_id] + (1 - beta1
                                                         ) * np_grad[idx]
        moment2_out[row_id] = beta2 * moment2[row_id] + (
            1 - beta2) * np.square(np_grad[idx])
        lr_t = lr * np.sqrt(1 - beta2_pow) / (1 - beta1_pow)
T
typhoonzero 已提交
225 226
        param_out[row_id] = param[row_id] - lr_t * (moment1_out[row_id] / (
            np.sqrt(moment2_out[row_id]) + epsilon))
T
wip  
typhoonzero 已提交
227 228 229 230 231 232 233 234 235 236 237
    return param_out, moment1_out, moment2_out


class TestSparseAdamOp(unittest.TestCase):
    def setup(self, scope, place):
        beta1 = 0.78
        beta2 = 0.836
        epsilon = 1e-4

        height = 10
        rows = [0, 4, 7]
T
typhoonzero 已提交
238
        self.rows = rows
T
wip  
typhoonzero 已提交
239
        row_numel = 12
T
typhoonzero 已提交
240
        self.row_numel = row_numel
T
wip  
typhoonzero 已提交
241 242 243 244
        self.dense_inputs = {
            "Param": np.full((height, row_numel), 5.0).astype("float32"),
            "Moment1": np.full((height, row_numel), 5.0).astype("float32"),
            "Moment2": np.full((height, row_numel), 5.0).astype("float32"),
T
typhoonzero 已提交
245 246
            'Beta1Pow': np.array([beta1**10]).astype("float32"),
            'Beta2Pow': np.array([beta2**10]).astype("float32"),
T
wip  
typhoonzero 已提交
247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265
            "LearningRate": np.full((1), 2.0).astype("float32")
        }
        self.attrs = {'epsilon': epsilon, 'beta1': beta1, 'beta2': beta2}

        grad_selected_rows = scope.var('Grad').get_selected_rows()
        grad_selected_rows.set_height(height)
        grad_selected_rows.set_rows(rows)
        np_array = np.ones((len(rows), row_numel)).astype("float32")
        np_array[0, 0] = 2.0
        np_array[2, 8] = 4.0

        grad_tensor = grad_selected_rows.get_tensor()
        grad_tensor.set(np_array, place)

        self.sparse_inputs = ["Grad"]

        param_out, mom1, mom2 = adam_step_sparse(
            self.dense_inputs, self.attrs, height, rows, row_numel, np_array)
        self.outputs = {
T
typhoonzero 已提交
266
            "ParamOut": param_out,
T
wip  
typhoonzero 已提交
267 268 269 270 271 272 273 274 275 276 277 278 279 280 281
            "Moment1Out": mom1,
            "Moment2Out": mom2
        }

    def check_with_place(self, place):
        scope = core.Scope()
        self.setup(scope, place)

        op_args = dict()
        for key, np_array in self.dense_inputs.iteritems():
            var = scope.var(key).get_tensor()
            var.set(np_array, place)
            op_args[key] = key
        for s in self.sparse_inputs:
            op_args[s] = s
T
typhoonzero 已提交
282 283 284 285
        for s in self.outputs:
            var = scope.var(s).get_tensor()
            var.set(self.outputs[s], place)
            op_args[s] = s
T
wip  
typhoonzero 已提交
286 287 288 289
        for k in self.attrs:
            op_args[k] = self.attrs[k]

        # create and run sgd operator
T
typhoonzero 已提交
290 291
        adam_op = Operator("adam", **op_args)
        adam_op.run(scope, place)
T
wip  
typhoonzero 已提交
292 293 294 295

        for key, np_array in self.outputs.iteritems():
            out_var = scope.var(key).get_tensor()
            actual = np.array(out_var)
T
typhoonzero 已提交
296 297 298 299 300 301
            actual = actual.reshape([actual.size])
            np_array = np_array.reshape([np_array.size])
            for idx, row_id in enumerate(self.rows):
                j = 0
                while j < self.row_numel:
                    pos = row_id * self.row_numel + j
T
update  
typhoonzero 已提交
302 303
                    self.assertLess((actual[pos] - np_array[pos]) / actual[pos],
                                    0.00001)
T
typhoonzero 已提交
304
                    j += 1
T
wip  
typhoonzero 已提交
305 306 307

    def test_sparse_sgd(self):
        places = [core.CPUPlace()]
308
        if core.is_compiled_with_cuda():
T
wip  
typhoonzero 已提交
309 310 311 312 313
            places.append(core.CUDAPlace(0))
        for place in places:
            self.check_with_place(place)


314 315
if __name__ == "__main__":
    unittest.main()