FM / ideep

/usr/local/lib64/python3.6/site-packages/caffe2/python/ideep/

/usr/local/lib64/python3.6/site-packages/caffe2/python/ideep

Name	Size	Mode	Actions
__pycache__/	-	0755	rm
adam_op_test.py	3212	0644	edit dl rm
blobs_queue_db_test.py	4045	0644	edit dl rm
channel_shuffle_op_test.py	1286	0644	edit dl rm
concat_split_op_test.py	5532	0644	edit dl rm
convfusion_op_test.py	31919	0644	edit dl rm
conv_op_test.py	5666	0644	edit dl rm
conv_transpose_test.py	2595	0644	edit dl rm
copy_op_test.py	3074	0644	edit dl rm
dropout_op_test.py	2075	0644	edit dl rm
elementwise_sum_op_test.py	6317	0644	edit dl rm
expanddims_squeeze_op_test.py	4285	0644	edit dl rm
fc_op_test.py	11745	0644	edit dl rm
leaky_relu_op_test.py	2854	0644	edit dl rm
LRN_op_test.py	1195	0644	edit dl rm
moment_sgd_op_test.py	1777	0644	edit dl rm
operator_fallback_op_test.py	3450	0644	edit dl rm
order_switch_op_test.py	2299	0644	edit dl rm
pool_op_test.py	4296	0644	edit dl rm
pre_convert_test.py	4115	0644	edit dl rm
relu_op_test.py	3705	0644	edit dl rm
reshape_op_test.py	5919	0644	edit dl rm
shape_op_test.py	2631	0644	edit dl rm
sigmoid_op_test.py	775	0644	edit dl rm
softmax_op_test.py	932	0644	edit dl rm
spatial_bn_op_test.py	5237	0644	edit dl rm
test_ideep_net.py	4094	0644	edit dl rm
transform_ideep_net.py	11683	0644	edit dl rm
transpose_op_test.py	1284	0644	edit dl rm
weightedsum_op_test.py	1560	0644	edit dl rm
__init__.py	0	0644	edit dl rm

Edit: /usr/local/lib64/python3.6/site-packages/caffe2/python/ideep/spatial_bn_op_test.py (5237B)

@unittest.skipIf(not workspace.C.use_mkldnn, "No MKLDNN support.")
class TestSpatialBN(hu.HypothesisTestCase):
    @given(size=st.integers(7, 10),
           input_channels=st.integers(7, 10),
           batch_size=st.integers(1, 3),
           seed=st.integers(0, 65535),
           order=st.sampled_from(["NCHW"]),
           epsilon=st.floats(min_value=1e-5, max_value=1e-2),
           inplace=st.sampled_from([True, False]),
           **mu.gcs)
    @settings(deadline=1000)
    def test_spatialbn_test_mode(
            self, size, input_channels, batch_size, seed, order, epsilon,
            inplace, gc, dc):
        op = core.CreateOperator(
            "SpatialBN",
            ["X", "scale", "bias", "mean", "var"],
            ["X" if inplace else "Y"],
            order=order,
            is_test=True,
            epsilon=epsilon
        )

def reference_spatialbn_test(X, scale, bias, mean, var):
            if order == "NCHW":
                scale = scale[np.newaxis, :, np.newaxis, np.newaxis]
                bias = bias[np.newaxis, :, np.newaxis, np.newaxis]
                mean = mean[np.newaxis, :, np.newaxis, np.newaxis]
                var = var[np.newaxis, :, np.newaxis, np.newaxis]
            return ((X - mean) / np.sqrt(var + epsilon) * scale + bias,)

np.random.seed(1701)
        scale = np.random.rand(input_channels).astype(np.float32) + 0.5
        bias = np.random.rand(input_channels).astype(np.float32) - 0.5
        mean = np.random.randn(input_channels).astype(np.float32)
        var = np.random.rand(input_channels).astype(np.float32) + 0.5
        X = np.random.rand(
            batch_size, input_channels, size, size).astype(np.float32) - 0.5

if order == "NHWC":
            X = X.swapaxes(1, 2).swapaxes(2, 3)

self.assertDeviceChecks(dc, op, [X, scale, bias, mean, var], [0])

@given(size=st.integers(7, 10),
           input_channels=st.integers(7, 10),
           batch_size=st.integers(1, 3),
           seed=st.integers(0, 65535),
           order=st.sampled_from(["NCHW"]),
           epsilon=st.floats(1e-5, 1e-2),
           inplace=st.sampled_from([True, False]),
           **mu.gcs)
    def test_spatialbn_train_mode(
            self, size, input_channels, batch_size, seed, order, epsilon,
            inplace, gc, dc):
        print("dc0: {}, dc1: {}".format(dc[0], dc[1]))
        op = core.CreateOperator(
            "SpatialBN",
            ["X", "scale", "bias", "running_mean", "running_var"],
            ["X" if inplace else "Y",
            "running_mean", "running_var", "saved_mean", "saved_var"],
            order=order,
            is_test=False,
            epsilon=epsilon,
        )
        np.random.seed(1701)
        scale = np.random.rand(input_channels).astype(np.float32) + 0.5
        bias = np.random.rand(input_channels).astype(np.float32) - 0.5
        running_mean = np.random.randn(input_channels).astype(np.float32)
        running_var = np.random.rand(input_channels).astype(np.float32) + 0.5
        X = np.random.rand(
            batch_size, input_channels, size, size).astype(np.float32) - 0.5

if order == "NHWC":
            X = X.swapaxes(1, 2).swapaxes(2, 3)

# TODO: It looks like IDEEP spatial_bn op outputs save_var (output[4])
        # as the reciprocal of CPU op's output. Need to check back and add
        # output[4] for comparison
        self.assertDeviceChecks(dc, op, [X, scale, bias, running_mean, running_var],
            [0, 1, 2, 3])

@given(size=st.integers(7, 10),
           input_channels=st.integers(1, 10),
           batch_size=st.integers(1, 3),
           seed=st.integers(0, 65535),
           order=st.sampled_from(["NCHW"]),
           epsilon=st.floats(min_value=1e-5, max_value=1e-2),
           **mu.gcs)
    @settings(deadline=None, max_examples=50)
    def test_spatialbn_train_mode_gradient_check(
            self, size, input_channels, batch_size, seed, order, epsilon,
            gc, dc):
        op = core.CreateOperator(
            "SpatialBN",
            ["X", "scale", "bias", "mean", "var"],
            ["Y", "mean", "var", "saved_mean", "saved_var"],
            order=order,
            is_test=False,
            epsilon=epsilon,
        )
        np.random.seed(seed)
        scale = np.random.rand(input_channels).astype(np.float32) + 0.5
        bias = np.random.rand(input_channels).astype(np.float32) - 0.5
        mean = np.random.randn(input_channels).astype(np.float32)
        var = np.random.rand(input_channels).astype(np.float32) + 0.5
        X = np.random.rand(
            batch_size, input_channels, size, size).astype(np.float32) - 0.5
        if order == "NHWC":
            X = X.swapaxes(1, 2).swapaxes(2, 3)

for input_to_check in [0, 1, 2]:  # dX, dScale, dBias
            self.assertGradientChecks(gc, op, [X, scale, bias, mean, var],
                                      input_to_check, [0])

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

from hypothesis import given, settings import hypothesis.strategies as st import numpy as np import unittest from caffe2.python import core, workspace import caffe2.python.hypothesis_test_util as hu import caffe2.python.ideep_test_util as mu @unittest.skipIf(not workspace.C.use_mkldnn, "No MKLDNN support.") class TestSpatialBN(hu.HypothesisTestCase): @given(size=st.integers(7, 10), input_channels=st.integers(7, 10), batch_size=st.integers(1, 3), seed=st.integers(0, 65535), order=st.sampled_from(["NCHW"]), epsilon=st.floats(min_value=1e-5, max_value=1e-2), inplace=st.sampled_from([True, False]), **mu.gcs) @settings(deadline=1000) def test_spatialbn_test_mode( self, size, input_channels, batch_size, seed, order, epsilon, inplace, gc, dc): op = core.CreateOperator( "SpatialBN", ["X", "scale", "bias", "mean", "var"], ["X" if inplace else "Y"], order=order, is_test=True, epsilon=epsilon ) def reference_spatialbn_test(X, scale, bias, mean, var): if order == "NCHW": scale = scale[np.newaxis, :, np.newaxis, np.newaxis] bias = bias[np.newaxis, :, np.newaxis, np.newaxis] mean = mean[np.newaxis, :, np.newaxis, np.newaxis] var = var[np.newaxis, :, np.newaxis, np.newaxis] return ((X - mean) / np.sqrt(var + epsilon) * scale + bias,) np.random.seed(1701) scale = np.random.rand(input_channels).astype(np.float32) + 0.5 bias = np.random.rand(input_channels).astype(np.float32) - 0.5 mean = np.random.randn(input_channels).astype(np.float32) var = np.random.rand(input_channels).astype(np.float32) + 0.5 X = np.random.rand( batch_size, input_channels, size, size).astype(np.float32) - 0.5 if order == "NHWC": X = X.swapaxes(1, 2).swapaxes(2, 3) self.assertDeviceChecks(dc, op, [X, scale, bias, mean, var], [0]) @given(size=st.integers(7, 10), input_channels=st.integers(7, 10), batch_size=st.integers(1, 3), seed=st.integers(0, 65535), order=st.sampled_from(["NCHW"]), epsilon=st.floats(1e-5, 1e-2), inplace=st.sampled_from([True, False]), **mu.gcs) def test_spatialbn_train_mode( self, size, input_channels, batch_size, seed, order, epsilon, inplace, gc, dc): print("dc0: {}, dc1: {}".format(dc[0], dc[1])) op = core.CreateOperator( "SpatialBN", ["X", "scale", "bias", "running_mean", "running_var"], ["X" if inplace else "Y", "running_mean", "running_var", "saved_mean", "saved_var"], order=order, is_test=False, epsilon=epsilon, ) np.random.seed(1701) scale = np.random.rand(input_channels).astype(np.float32) + 0.5 bias = np.random.rand(input_channels).astype(np.float32) - 0.5 running_mean = np.random.randn(input_channels).astype(np.float32) running_var = np.random.rand(input_channels).astype(np.float32) + 0.5 X = np.random.rand( batch_size, input_channels, size, size).astype(np.float32) - 0.5 if order == "NHWC": X = X.swapaxes(1, 2).swapaxes(2, 3) # TODO: It looks like IDEEP spatial_bn op outputs save_var (output[4]) # as the reciprocal of CPU op's output. Need to check back and add # output[4] for comparison self.assertDeviceChecks(dc, op, [X, scale, bias, running_mean, running_var], [0, 1, 2, 3]) @given(size=st.integers(7, 10), input_channels=st.integers(1, 10), batch_size=st.integers(1, 3), seed=st.integers(0, 65535), order=st.sampled_from(["NCHW"]), epsilon=st.floats(min_value=1e-5, max_value=1e-2), **mu.gcs) @settings(deadline=None, max_examples=50) def test_spatialbn_train_mode_gradient_check( self, size, input_channels, batch_size, seed, order, epsilon, gc, dc): op = core.CreateOperator( "SpatialBN", ["X", "scale", "bias", "mean", "var"], ["Y", "mean", "var", "saved_mean", "saved_var"], order=order, is_test=False, epsilon=epsilon, ) np.random.seed(seed) scale = np.random.rand(input_channels).astype(np.float32) + 0.5 bias = np.random.rand(input_channels).astype(np.float32) - 0.5 mean = np.random.randn(input_channels).astype(np.float32) var = np.random.rand(input_channels).astype(np.float32) + 0.5 X = np.random.rand( batch_size, input_channels, size, size).astype(np.float32) - 0.5 if order == "NHWC": X = X.swapaxes(1, 2).swapaxes(2, 3) for input_to_check in [0, 1, 2]: # dX, dScale, dBias self.assertGradientChecks(gc, op, [X, scale, bias, mean, var], input_to_check, [0]) if __name__ == "__main__": unittest.main()