b45e1a25a5
* sequences to ti transformations, support for seq_lengths input, update reference implemetations, add new tests * fix python api, update sequences to ti transformation * fix sequences to ti transformation * Update sequences to TI transformation: fix reverse sequence support * update single layer tests, fix TI reference impl, fix Sequences to TI transformations * ngraph code style * fix build * fix ngraph python api * resolver review comments, refactoring * Resolve review remarks * delete xfail
1671 lines
52 KiB
Python
1671 lines
52 KiB
Python
# ******************************************************************************
|
|
# Copyright 2017-2020 Intel Corporation
|
|
#
|
|
# 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
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# 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.
|
|
# ******************************************************************************
|
|
import numpy as np
|
|
import pytest
|
|
from _pyngraph import PartialShape
|
|
|
|
import ngraph as ng
|
|
import ngraph.opset1 as ng_opset1
|
|
from ngraph.impl import Type
|
|
|
|
from tests import skip_segfault
|
|
|
|
np_types = [np.float32, np.int32]
|
|
integral_np_types = [
|
|
np.int8,
|
|
np.int16,
|
|
np.int32,
|
|
np.int64,
|
|
np.uint8,
|
|
np.uint16,
|
|
np.uint32,
|
|
np.uint64,
|
|
]
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", np_types)
|
|
def test_binary_convolution(dtype):
|
|
strides = np.array([1, 1])
|
|
pads_begin = np.array([0, 0])
|
|
pads_end = np.array([0, 0])
|
|
dilations = np.array([1, 1])
|
|
mode = "xnor-popcount"
|
|
pad_value = 0.0
|
|
|
|
input0_shape = [1, 1, 9, 9]
|
|
input1_shape = [1, 1, 3, 3]
|
|
expected_shape = [1, 1, 7, 7]
|
|
|
|
parameter_input0 = ng.parameter(input0_shape, name="Input0", dtype=dtype)
|
|
parameter_input1 = ng.parameter(input1_shape, name="Input1", dtype=dtype)
|
|
|
|
node = ng.binary_convolution(
|
|
parameter_input0, parameter_input1, strides, pads_begin, pads_end, dilations, mode, pad_value,
|
|
)
|
|
|
|
assert node.get_type_name() == "BinaryConvolution"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == expected_shape
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", np_types)
|
|
def test_ctc_greedy_decoder(dtype):
|
|
input0_shape = [20, 8, 128]
|
|
input1_shape = [20, 8]
|
|
expected_shape = [8, 20, 1, 1]
|
|
|
|
parameter_input0 = ng.parameter(input0_shape, name="Input0", dtype=dtype)
|
|
parameter_input1 = ng.parameter(input1_shape, name="Input1", dtype=dtype)
|
|
|
|
node = ng.ctc_greedy_decoder(parameter_input0, parameter_input1)
|
|
|
|
assert node.get_type_name() == "CTCGreedyDecoder"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == expected_shape
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", np_types)
|
|
def test_deformable_convolution(dtype):
|
|
strides = np.array([1, 1])
|
|
pads_begin = np.array([0, 0])
|
|
pads_end = np.array([0, 0])
|
|
dilations = np.array([1, 1])
|
|
|
|
input0_shape = [1, 1, 9, 9]
|
|
input1_shape = [1, 1, 9, 9]
|
|
input2_shape = [1, 1, 3, 3]
|
|
expected_shape = [1, 1, 7, 7]
|
|
|
|
parameter_input0 = ng.parameter(input0_shape, name="Input0", dtype=dtype)
|
|
parameter_input1 = ng.parameter(input1_shape, name="Input1", dtype=dtype)
|
|
parameter_input2 = ng.parameter(input2_shape, name="Input2", dtype=dtype)
|
|
|
|
node = ng.deformable_convolution(
|
|
parameter_input0, parameter_input1, parameter_input2, strides, pads_begin, pads_end, dilations,
|
|
)
|
|
|
|
assert node.get_type_name() == "DeformableConvolution"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == expected_shape
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", np_types)
|
|
def test_deformable_psroi_pooling(dtype):
|
|
output_dim = 8
|
|
spatial_scale = 0.0625
|
|
group_size = 7
|
|
mode = "bilinear_deformable"
|
|
spatial_bins_x = 4
|
|
spatial_bins_y = 4
|
|
trans_std = 0.1
|
|
part_size = 7
|
|
|
|
input0_shape = [1, 392, 38, 63]
|
|
input1_shape = [300, 5]
|
|
input2_shape = [300, 2, 7, 7]
|
|
expected_shape = [300, 8, 7, 7]
|
|
|
|
parameter_input0 = ng.parameter(input0_shape, name="Input0", dtype=dtype)
|
|
parameter_input1 = ng.parameter(input1_shape, name="Input1", dtype=dtype)
|
|
parameter_input2 = ng.parameter(input2_shape, name="Input2", dtype=dtype)
|
|
|
|
node = ng.deformable_psroi_pooling(
|
|
parameter_input0,
|
|
parameter_input1,
|
|
output_dim,
|
|
spatial_scale,
|
|
group_size,
|
|
mode,
|
|
spatial_bins_x,
|
|
spatial_bins_y,
|
|
trans_std,
|
|
part_size,
|
|
offsets=parameter_input2,
|
|
)
|
|
|
|
assert node.get_type_name() == "DeformablePSROIPooling"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == expected_shape
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", np_types)
|
|
def test_floor_mod(dtype):
|
|
input0_shape = [8, 1, 6, 1]
|
|
input1_shape = [7, 1, 5]
|
|
expected_shape = [8, 7, 6, 5]
|
|
|
|
parameter_input0 = ng.parameter(input0_shape, name="Input0", dtype=dtype)
|
|
parameter_input1 = ng.parameter(input1_shape, name="Input1", dtype=dtype)
|
|
|
|
node = ng.floor_mod(parameter_input0, parameter_input1)
|
|
|
|
assert node.get_type_name() == "FloorMod"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == expected_shape
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", np_types)
|
|
def test_gather_tree(dtype):
|
|
input0_shape = [100, 1, 10]
|
|
input1_shape = [100, 1, 10]
|
|
input2_shape = [1]
|
|
input3_shape = []
|
|
expected_shape = [100, 1, 10]
|
|
|
|
parameter_input0 = ng.parameter(input0_shape, name="Input0", dtype=dtype)
|
|
parameter_input1 = ng.parameter(input1_shape, name="Input1", dtype=dtype)
|
|
parameter_input2 = ng.parameter(input2_shape, name="Input2", dtype=dtype)
|
|
parameter_input3 = ng.parameter(input3_shape, name="Input3", dtype=dtype)
|
|
|
|
node = ng.gather_tree(parameter_input0, parameter_input1, parameter_input2, parameter_input3)
|
|
|
|
assert node.get_type_name() == "GatherTree"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == expected_shape
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_lstm_cell_operator(dtype):
|
|
batch_size = 1
|
|
input_size = 16
|
|
hidden_size = 128
|
|
|
|
X_shape = [batch_size, input_size]
|
|
H_t_shape = [batch_size, hidden_size]
|
|
C_t_shape = [batch_size, hidden_size]
|
|
W_shape = [4 * hidden_size, input_size]
|
|
R_shape = [4 * hidden_size, hidden_size]
|
|
B_shape = [4 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_C_t = ng.parameter(C_t_shape, name="C_t", dtype=dtype)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
expected_shape = [1, 128]
|
|
|
|
node_default = ng.lstm_cell(
|
|
parameter_X, parameter_H_t, parameter_C_t, parameter_W, parameter_R, parameter_B, hidden_size,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "LSTMCell"
|
|
assert node_default.get_output_size() == 2
|
|
assert list(node_default.get_output_shape(0)) == expected_shape
|
|
assert list(node_default.get_output_shape(1)) == expected_shape
|
|
|
|
activations = ["tanh", "Sigmoid", "RELU"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 0.5
|
|
|
|
node_param = ng.lstm_cell(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "LSTMCell"
|
|
assert node_param.get_output_size() == 2
|
|
assert list(node_param.get_output_shape(0)) == expected_shape
|
|
assert list(node_param.get_output_shape(1)) == expected_shape
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_lstm_cell_operator_opset1(dtype):
|
|
batch_size = 1
|
|
input_size = 16
|
|
hidden_size = 128
|
|
|
|
X_shape = [batch_size, input_size]
|
|
H_t_shape = [batch_size, hidden_size]
|
|
C_t_shape = [batch_size, hidden_size]
|
|
W_shape = [4 * hidden_size, input_size]
|
|
R_shape = [4 * hidden_size, hidden_size]
|
|
B_shape = [4 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_C_t = ng.parameter(C_t_shape, name="C_t", dtype=dtype)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
expected_shape = [1, 128]
|
|
|
|
node_default = ng_opset1.lstm_cell(
|
|
parameter_X, parameter_H_t, parameter_C_t, parameter_W, parameter_R, parameter_B, hidden_size,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "LSTMCell"
|
|
assert node_default.get_output_size() == 2
|
|
assert list(node_default.get_output_shape(0)) == expected_shape
|
|
assert list(node_default.get_output_shape(1)) == expected_shape
|
|
|
|
activations = ["tanh", "Sigmoid", "RELU"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 0.5
|
|
|
|
node_param = ng_opset1.lstm_cell(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "LSTMCell"
|
|
assert node_param.get_output_size() == 2
|
|
assert list(node_param.get_output_shape(0)) == expected_shape
|
|
assert list(node_param.get_output_shape(1)) == expected_shape
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_lstm_sequence_operator_bidirectional_opset1(dtype):
|
|
batch_size = 1
|
|
input_size = 16
|
|
hidden_size = 128
|
|
num_directions = 2
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
C_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, 4 * hidden_size, input_size]
|
|
R_shape = [num_directions, 4 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_C_t = ng.parameter(C_t_shape, name="C_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "BIDIRECTIONAL"
|
|
node = ng_opset1.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node.get_type_name() == "LSTMSequence"
|
|
assert node.get_output_size() == 3
|
|
|
|
activations = ["RELU", "tanh", "Sigmoid"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 1.22
|
|
|
|
node_param = ng_opset1.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "LSTMSequence"
|
|
assert node_param.get_output_size() == 3
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_lstm_sequence_operator_reverse_opset1(dtype):
|
|
batch_size = 2
|
|
input_size = 4
|
|
hidden_size = 3
|
|
num_directions = 1
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
C_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, 4 * hidden_size, input_size]
|
|
R_shape = [num_directions, 4 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_C_t = ng.parameter(C_t_shape, name="C_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "REVERSE"
|
|
|
|
node_default = ng_opset1.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "LSTMSequence"
|
|
assert node_default.get_output_size() == 3
|
|
|
|
activations = ["RELU", "tanh", "Sigmoid"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 1.22
|
|
|
|
node_param = ng_opset1.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "LSTMSequence"
|
|
assert node_param.get_output_size() == 3
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_lstm_sequence_operator_forward_opset1(dtype):
|
|
batch_size = 2
|
|
input_size = 4
|
|
hidden_size = 3
|
|
num_directions = 1
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
C_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, 4 * hidden_size, input_size]
|
|
R_shape = [num_directions, 4 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_C_t = ng.parameter(C_t_shape, name="C_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "forward"
|
|
|
|
node_default = ng_opset1.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "LSTMSequence"
|
|
assert node_default.get_output_size() == 3
|
|
|
|
activations = ["RELU", "tanh", "Sigmoid"]
|
|
activation_alpha = [2.0]
|
|
activation_beta = [1.0]
|
|
clip = 0.5
|
|
|
|
node = ng_opset1.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node.get_type_name() == "LSTMSequence"
|
|
assert node.get_output_size() == 3
|
|
|
|
|
|
def test_gru_cell_operator():
|
|
batch_size = 1
|
|
input_size = 16
|
|
hidden_size = 128
|
|
|
|
X_shape = [batch_size, input_size]
|
|
H_t_shape = [batch_size, hidden_size]
|
|
W_shape = [3 * hidden_size, input_size]
|
|
R_shape = [3 * hidden_size, hidden_size]
|
|
B_shape = [3 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=np.float32)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=np.float32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=np.float32)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=np.float32)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=np.float32)
|
|
|
|
expected_shape = [1, 128]
|
|
|
|
node_default = ng.gru_cell(parameter_X, parameter_H_t, parameter_W, parameter_R, parameter_B, hidden_size)
|
|
|
|
assert node_default.get_type_name() == "GRUCell"
|
|
assert node_default.get_output_size() == 1
|
|
assert list(node_default.get_output_shape(0)) == expected_shape
|
|
|
|
activations = ["tanh", "relu"]
|
|
activations_alpha = [1.0, 2.0]
|
|
activations_beta = [1.0, 2.0]
|
|
clip = 0.5
|
|
linear_before_reset = True
|
|
|
|
# If *linear_before_reset* is set True, then B tensor shape must be [4 * hidden_size]
|
|
B_shape = [4 * hidden_size]
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=np.float32)
|
|
|
|
node_param = ng.gru_cell(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
activations,
|
|
activations_alpha,
|
|
activations_beta,
|
|
clip,
|
|
linear_before_reset,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "GRUCell"
|
|
assert node_param.get_output_size() == 1
|
|
assert list(node_param.get_output_shape(0)) == expected_shape
|
|
|
|
|
|
def test_gru_sequence():
|
|
batch_size = 2
|
|
input_size = 16
|
|
hidden_size = 32
|
|
seq_len = 8
|
|
seq_lengths = [seq_len] * batch_size
|
|
num_directions = 1
|
|
direction = "FORWARD"
|
|
|
|
X_shape = [batch_size, seq_len, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
W_shape = [num_directions, 3 * hidden_size, input_size]
|
|
R_shape = [num_directions, 3 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 3 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=np.float32)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=np.float32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=np.float32)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=np.float32)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=np.float32)
|
|
|
|
expected_shape_y = [batch_size, num_directions, seq_len, hidden_size]
|
|
expected_shape_h = [batch_size, num_directions, hidden_size]
|
|
|
|
node_default = ng.gru_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
seq_lengths,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "GRUSequence"
|
|
assert node_default.get_output_size() == 2
|
|
assert list(node_default.get_output_shape(0)) == expected_shape_y
|
|
assert list(node_default.get_output_shape(1)) == expected_shape_h
|
|
|
|
activations = ["tanh", "relu"]
|
|
activations_alpha = [1.0, 2.0]
|
|
activations_beta = [1.0, 2.0]
|
|
clip = 0.5
|
|
linear_before_reset = True
|
|
|
|
# If *linear_before_reset* is set True, then B tensor shape must be [4 * hidden_size]
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=np.float32)
|
|
|
|
node_param = ng.gru_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
seq_lengths,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activations_alpha,
|
|
activations_beta,
|
|
clip,
|
|
linear_before_reset,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "GRUSequence"
|
|
assert node_param.get_output_size() == 2
|
|
assert list(node_param.get_output_shape(0)) == expected_shape_y
|
|
assert list(node_param.get_output_shape(1)) == expected_shape_h
|
|
|
|
|
|
def test_rnn_sequence():
|
|
batch_size = 2
|
|
input_size = 16
|
|
hidden_size = 32
|
|
seq_len = 8
|
|
seq_lengths = [seq_len] * batch_size
|
|
num_directions = 1
|
|
direction = "FORWARD"
|
|
|
|
X_shape = [batch_size, seq_len, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
W_shape = [num_directions, hidden_size, input_size]
|
|
R_shape = [num_directions, hidden_size, hidden_size]
|
|
B_shape = [num_directions, hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=np.float32)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=np.float32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=np.float32)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=np.float32)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=np.float32)
|
|
|
|
expected_shape_y = [batch_size, num_directions, seq_len, hidden_size]
|
|
expected_shape_h = [batch_size, num_directions, hidden_size]
|
|
|
|
node_default = ng.rnn_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
seq_lengths,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "RNNSequence"
|
|
assert node_default.get_output_size() == 2
|
|
assert list(node_default.get_output_shape(0)) == expected_shape_y
|
|
assert list(node_default.get_output_shape(1)) == expected_shape_h
|
|
|
|
activations = ["relu"]
|
|
activations_alpha = [2.0]
|
|
activations_beta = [1.0]
|
|
clip = 0.5
|
|
|
|
node_param = ng.rnn_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
seq_lengths,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activations_alpha,
|
|
activations_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "RNNSequence"
|
|
assert node_param.get_output_size() == 2
|
|
assert list(node_param.get_output_shape(0)) == expected_shape_y
|
|
assert list(node_param.get_output_shape(1)) == expected_shape_h
|
|
|
|
|
|
@skip_segfault
|
|
def test_loop():
|
|
trip_count = 8
|
|
condition = True
|
|
|
|
node_default = ng.loop(trip_count, condition)
|
|
|
|
assert node_default.get_type_name() == "Loop"
|
|
|
|
|
|
def test_roi_pooling():
|
|
inputs = ng.parameter([2, 3, 4, 5], dtype=np.float32)
|
|
coords = ng.parameter([150, 5], dtype=np.float32)
|
|
node = ng.roi_pooling(inputs, coords, [6, 6], 0.0625, "Max")
|
|
|
|
assert node.get_type_name() == "ROIPooling"
|
|
assert node.get_output_size() == [6, 6]
|
|
assert list(node.get_output_shape(0)) == [150, 3, 6, 6]
|
|
assert node.get_output_element_type(0) == Type.f32
|
|
|
|
|
|
def test_psroi_pooling():
|
|
inputs = ng.parameter([1, 3, 4, 5], dtype=np.float32)
|
|
coords = ng.parameter([150, 5], dtype=np.float32)
|
|
node = ng.psroi_pooling(inputs, coords, 2, 6, 0.0625, 0, 0, "Avg")
|
|
|
|
assert node.get_type_name() == "PSROIPooling"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [150, 2, 6, 6]
|
|
assert node.get_output_element_type(0) == Type.f32
|
|
|
|
|
|
def test_convert_like():
|
|
parameter_data = ng.parameter([1, 2, 3, 4], name="data", dtype=np.float32)
|
|
like = ng.constant(1, dtype=np.int8)
|
|
|
|
node = ng.convert_like(parameter_data, like)
|
|
|
|
assert node.get_type_name() == "ConvertLike"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [1, 2, 3, 4]
|
|
assert node.get_output_element_type(0) == Type.i8
|
|
|
|
|
|
def test_bucketize():
|
|
data = ng.parameter([4, 3, 2, 1], name="data", dtype=np.float32)
|
|
buckets = ng.parameter([5], name="buckets", dtype=np.int64)
|
|
|
|
node = ng.bucketize(data, buckets, "i32")
|
|
|
|
assert node.get_type_name() == "Bucketize"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [4, 3, 2, 1]
|
|
assert node.get_output_element_type(0) == Type.i32
|
|
|
|
|
|
def test_region_yolo():
|
|
data = ng.parameter([1, 125, 13, 13], name="input", dtype=np.float32)
|
|
num_coords = 4
|
|
num_classes = 80
|
|
num_regions = 1
|
|
mask = [6, 7, 8]
|
|
axis = 0
|
|
end_axis = 3
|
|
do_softmax = False
|
|
|
|
node = ng.region_yolo(data, num_coords, num_classes, num_regions, do_softmax, mask, axis, end_axis)
|
|
|
|
assert node.get_type_name() == "RegionYolo"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [1, (80 + 4 + 1) * 3, 13, 13]
|
|
assert node.get_output_element_type(0) == Type.f32
|
|
|
|
|
|
def test_reorg_yolo():
|
|
data = ng.parameter([2, 24, 34, 62], name="input", dtype=np.int32)
|
|
stride = [2]
|
|
|
|
node = ng.reorg_yolo(data, stride)
|
|
|
|
assert node.get_type_name() == "ReorgYolo"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [2, 96, 17, 31]
|
|
assert node.get_output_element_type(0) == Type.i32
|
|
|
|
|
|
def test_embedding_bag_offsets_sum_1():
|
|
emb_table = ng.parameter([5, 2], name="emb_table", dtype=np.float32)
|
|
indices = ng.parameter([4], name="indices", dtype=np.int64)
|
|
offsets = ng.parameter([3], name="offsets", dtype=np.int64)
|
|
default_index = ng.parameter([], name="default_index", dtype=np.int64)
|
|
|
|
node = ng.embedding_bag_offsets_sum(emb_table, indices, offsets, default_index)
|
|
|
|
assert node.get_type_name() == "EmbeddingBagOffsetsSum"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [3, 2]
|
|
assert node.get_output_element_type(0) == Type.f32
|
|
|
|
|
|
def test_embedding_segments_sum_all_inputs():
|
|
emb_table = ng.parameter([5, 2], name="emb_table", dtype=np.float32)
|
|
indices = ng.parameter([4], name="indices", dtype=np.int64)
|
|
segment_ids = ng.parameter([4], name="segment_ids", dtype=np.int64)
|
|
num_segments = ng.parameter([], name="num_segments", dtype=np.int64)
|
|
default_index = ng.parameter([], name="default_index", dtype=np.int64)
|
|
per_sample_weights = ng.parameter([4], name="per_sample_weights", dtype=np.float32)
|
|
|
|
node = ng.embedding_segments_sum(
|
|
emb_table, indices, segment_ids, num_segments, default_index, per_sample_weights
|
|
)
|
|
|
|
assert node.get_type_name() == "EmbeddingSegmentsSum"
|
|
assert node.get_output_size() == 1
|
|
assert node.get_output_partial_shape(0).same_scheme(PartialShape([-1, 2]))
|
|
assert node.get_output_element_type(0) == Type.f32
|
|
|
|
|
|
def test_embedding_segments_sum_with_some_opt_inputs():
|
|
emb_table = ng.parameter([5, 2], name="emb_table", dtype=np.float32)
|
|
indices = ng.parameter([4], name="indices", dtype=np.int64)
|
|
segment_ids = ng.parameter([4], name="segment_ids", dtype=np.int64)
|
|
num_segments = ng.parameter([], name="num_segments", dtype=np.int64)
|
|
|
|
# only 1 out of 3 optional inputs
|
|
node = ng.embedding_segments_sum(emb_table, indices, segment_ids, num_segments)
|
|
|
|
assert node.get_type_name() == "EmbeddingSegmentsSum"
|
|
assert node.get_output_size() == 1
|
|
assert node.get_output_partial_shape(0).same_scheme(PartialShape([-1, 2]))
|
|
assert node.get_output_element_type(0) == Type.f32
|
|
|
|
|
|
def test_embedding_bag_packed_sum():
|
|
emb_table = ng.parameter([5, 2], name="emb_table", dtype=np.float32)
|
|
indices = ng.parameter([3, 3], name="indices", dtype=np.int64)
|
|
per_sample_weights = ng.parameter([3, 3], name="per_sample_weights", dtype=np.float32)
|
|
|
|
# only 1 out of 3 optional inputs
|
|
node = ng.embedding_bag_packed_sum(emb_table, indices, per_sample_weights)
|
|
|
|
assert node.get_type_name() == "EmbeddingBagPackedSum"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [3, 2]
|
|
assert node.get_output_element_type(0) == Type.f32
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", integral_np_types)
|
|
def test_interpolate(dtype):
|
|
image_shape = [1, 3, 1024, 1024]
|
|
output_shape = [64, 64]
|
|
attributes = {
|
|
"axes": [2, 3],
|
|
"mode": "cubic",
|
|
"pads_begin": np.array([2, 2], dtype=dtype),
|
|
}
|
|
|
|
image_node = ng.parameter(image_shape, dtype, name="Image")
|
|
|
|
node = ng.interpolate(image_node, output_shape, attributes)
|
|
expected_shape = [1, 3, 64, 64]
|
|
|
|
assert node.get_type_name() == "Interpolate"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == expected_shape
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
"int_dtype, fp_dtype",
|
|
[
|
|
(np.int8, np.float32),
|
|
(np.int16, np.float32),
|
|
(np.int32, np.float32),
|
|
(np.int64, np.float32),
|
|
(np.uint8, np.float32),
|
|
(np.uint16, np.float32),
|
|
(np.uint32, np.float32),
|
|
(np.uint64, np.float32),
|
|
(np.int32, np.float16),
|
|
(np.int32, np.float64),
|
|
],
|
|
)
|
|
def test_prior_box(int_dtype, fp_dtype):
|
|
image_shape = np.array([64, 64], dtype=int_dtype)
|
|
attributes = {
|
|
"offset": fp_dtype(0),
|
|
"min_size": np.array([2, 3], dtype=fp_dtype),
|
|
"aspect_ratio": np.array([1.5, 2.0, 2.5], dtype=fp_dtype),
|
|
}
|
|
|
|
layer_shape = ng.constant(np.array([32, 32], dtype=int_dtype), int_dtype)
|
|
|
|
node = ng.prior_box(layer_shape, image_shape, attributes)
|
|
|
|
assert node.get_type_name() == "PriorBox"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [2, 20480]
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
"int_dtype, fp_dtype",
|
|
[
|
|
(np.int8, np.float32),
|
|
(np.int16, np.float32),
|
|
(np.int32, np.float32),
|
|
(np.int64, np.float32),
|
|
(np.uint8, np.float32),
|
|
(np.uint16, np.float32),
|
|
(np.uint32, np.float32),
|
|
(np.uint64, np.float32),
|
|
(np.int32, np.float16),
|
|
(np.int32, np.float64),
|
|
],
|
|
)
|
|
def test_prior_box_clustered(int_dtype, fp_dtype):
|
|
image_size = np.array([64, 64], dtype=int_dtype)
|
|
attributes = {
|
|
"offset": fp_dtype(0.5),
|
|
"widths": np.array([4.0, 2.0, 3.2], dtype=fp_dtype),
|
|
"heights": np.array([1.0, 2.0, 1.0], dtype=fp_dtype),
|
|
}
|
|
|
|
output_size = ng.constant(np.array([19, 19], dtype=int_dtype), int_dtype)
|
|
|
|
node = ng.prior_box_clustered(output_size, image_size, attributes)
|
|
|
|
assert node.get_type_name() == "PriorBoxClustered"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [2, 4332]
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
"int_dtype, fp_dtype",
|
|
[
|
|
(np.int8, np.float32),
|
|
(np.int16, np.float32),
|
|
(np.int32, np.float32),
|
|
(np.int64, np.float32),
|
|
(np.uint8, np.float32),
|
|
(np.uint16, np.float32),
|
|
(np.uint32, np.float32),
|
|
(np.uint64, np.float32),
|
|
(np.int32, np.float16),
|
|
(np.int32, np.float64),
|
|
],
|
|
)
|
|
def test_detection_output(int_dtype, fp_dtype):
|
|
attributes = {
|
|
"num_classes": int_dtype(85),
|
|
"keep_top_k": np.array([64], dtype=int_dtype),
|
|
"nms_threshold": fp_dtype(0.645),
|
|
}
|
|
|
|
box_logits = ng.parameter([4, 1, 5, 5], fp_dtype, "box_logits")
|
|
class_preds = ng.parameter([2, 1, 4, 5], fp_dtype, "class_preds")
|
|
proposals = ng.parameter([2, 1, 4, 5], fp_dtype, "proposals")
|
|
aux_class_preds = ng.parameter([2, 1, 4, 5], fp_dtype, "aux_class_preds")
|
|
aux_box_preds = ng.parameter([2, 1, 4, 5], fp_dtype, "aux_box_preds")
|
|
|
|
node = ng.detection_output(box_logits, class_preds, proposals, attributes, aux_class_preds, aux_box_preds)
|
|
|
|
assert node.get_type_name() == "DetectionOutput"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [1, 1, 256, 7]
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
"int_dtype, fp_dtype",
|
|
[
|
|
(np.uint8, np.float32),
|
|
(np.uint16, np.float32),
|
|
(np.uint32, np.float32),
|
|
(np.uint64, np.float32),
|
|
(np.uint32, np.float16),
|
|
(np.uint32, np.float64),
|
|
],
|
|
)
|
|
def test_proposal(int_dtype, fp_dtype):
|
|
attributes = {
|
|
"base_size": int_dtype(1),
|
|
"pre_nms_topn": int_dtype(20),
|
|
"post_nms_topn": int_dtype(64),
|
|
"nms_thresh": fp_dtype(0.34),
|
|
"feat_stride": int_dtype(16),
|
|
"min_size": int_dtype(32),
|
|
"ratio": np.array([0.1, 1.5, 2.0, 2.5], dtype=fp_dtype),
|
|
"scale": np.array([2, 3, 3, 4], dtype=fp_dtype),
|
|
}
|
|
batch_size = 7
|
|
|
|
class_probs = ng.parameter([batch_size, 12, 34, 62], fp_dtype, "class_probs")
|
|
bbox_deltas = ng.parameter([batch_size, 24, 34, 62], fp_dtype, "bbox_deltas")
|
|
image_shape = ng.parameter([3], fp_dtype, "image_shape")
|
|
node = ng.proposal(class_probs, bbox_deltas, image_shape, attributes)
|
|
|
|
assert node.get_type_name() == "Proposal"
|
|
assert node.get_output_size() == 2
|
|
assert list(node.get_output_shape(0)) == [batch_size * attributes["post_nms_topn"], 5]
|
|
|
|
|
|
def test_tensor_iterator():
|
|
from ngraph.utils.tensor_iterator_types import (
|
|
GraphBody,
|
|
TensorIteratorSliceInputDesc,
|
|
TensorIteratorMergedInputDesc,
|
|
TensorIteratorInvariantInputDesc,
|
|
TensorIteratorBodyOutputDesc,
|
|
TensorIteratorConcatOutputDesc,
|
|
)
|
|
|
|
# Body parameters
|
|
body_timestep = ng.parameter([], np.int32, "timestep")
|
|
body_data_in = ng.parameter([1, 2, 2], np.float32, "body_in")
|
|
body_prev_cma = ng.parameter([2, 2], np.float32, "body_prev_cma")
|
|
body_const_one = ng.parameter([], np.int32, "body_const_one")
|
|
|
|
# CMA = cumulative moving average
|
|
prev_cum_sum = ng.multiply(ng.convert(body_timestep, "f32"), body_prev_cma)
|
|
curr_cum_sum = ng.add(prev_cum_sum, ng.squeeze(body_data_in, [0]))
|
|
elem_cnt = ng.add(body_const_one, body_timestep)
|
|
curr_cma = ng.divide(curr_cum_sum, ng.convert(elem_cnt, "f32"))
|
|
cma_hist = ng.unsqueeze(curr_cma, [0])
|
|
|
|
# TI inputs
|
|
data = ng.parameter([16, 2, 2], np.float32, "data")
|
|
# Iterations count
|
|
zero = ng.constant(0, dtype=np.int32)
|
|
one = ng.constant(1, dtype=np.int32)
|
|
initial_cma = ng.constant(np.zeros([2, 2], dtype=np.float32), dtype=np.float32)
|
|
iter_cnt = ng.range(zero, np.int32(16), np.int32(1))
|
|
ti_inputs = [iter_cnt, data, initial_cma, one]
|
|
|
|
graph_body = GraphBody([body_timestep, body_data_in, body_prev_cma, body_const_one], [curr_cma, cma_hist])
|
|
ti_slice_input_desc = [
|
|
# timestep
|
|
# input_idx, body_param_idx, start, stride, part_size, end, axis
|
|
TensorIteratorSliceInputDesc(0, 0, 0, 1, 1, -1, 0),
|
|
# data
|
|
TensorIteratorSliceInputDesc(1, 1, 0, 1, 1, -1, 0),
|
|
]
|
|
ti_merged_input_desc = [
|
|
# body prev/curr_cma
|
|
TensorIteratorMergedInputDesc(2, 2, 0),
|
|
]
|
|
ti_invariant_input_desc = [
|
|
# body const one
|
|
TensorIteratorInvariantInputDesc(3, 3),
|
|
]
|
|
|
|
# TI outputs
|
|
ti_body_output_desc = [
|
|
# final average
|
|
TensorIteratorBodyOutputDesc(0, 0, -1),
|
|
]
|
|
ti_concat_output_desc = [
|
|
# history of cma
|
|
TensorIteratorConcatOutputDesc(1, 1, 0, 1, 1, -1, 0),
|
|
]
|
|
|
|
node = ng.tensor_iterator(
|
|
ti_inputs,
|
|
graph_body,
|
|
ti_slice_input_desc,
|
|
ti_merged_input_desc,
|
|
ti_invariant_input_desc,
|
|
ti_body_output_desc,
|
|
ti_concat_output_desc,
|
|
)
|
|
|
|
assert node.get_type_name() == "TensorIterator"
|
|
assert node.get_output_size() == 2
|
|
# final average
|
|
assert list(node.get_output_shape(0)) == [2, 2]
|
|
# cma history
|
|
assert list(node.get_output_shape(1)) == [16, 2, 2]
|
|
|
|
|
|
def test_read_value():
|
|
init_value = ng.parameter([2, 2], name="init_value", dtype=np.int32)
|
|
|
|
node = ng.read_value(init_value, "var_id_667")
|
|
|
|
assert node.get_type_name() == "ReadValue"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [2, 2]
|
|
assert node.get_output_element_type(0) == Type.i32
|
|
|
|
|
|
def test_assign():
|
|
input_data = ng.parameter([5, 7], name="input_data", dtype=np.int32)
|
|
rv = ng.read_value(input_data, "var_id_667")
|
|
node = ng.assign(rv, "var_id_667")
|
|
|
|
assert node.get_type_name() == "Assign"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [5, 7]
|
|
assert node.get_output_element_type(0) == Type.i32
|
|
|
|
|
|
def test_extract_image_patches():
|
|
image = ng.parameter([64, 3, 10, 10], name="image", dtype=np.int32)
|
|
sizes = [3, 3]
|
|
strides = [5, 5]
|
|
rates = [1, 1]
|
|
padding = "VALID"
|
|
node = ng.extract_image_patches(image, sizes, strides, rates, padding)
|
|
|
|
assert node.get_type_name() == "ExtractImagePatches"
|
|
assert node.get_output_size() == 1
|
|
assert list(node.get_output_shape(0)) == [64, 27, 2, 2]
|
|
assert node.get_output_element_type(0) == Type.i32
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_lstm_sequence_operator_bidirectional(dtype):
|
|
batch_size = 1
|
|
input_size = 16
|
|
hidden_size = 128
|
|
num_directions = 2
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
C_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, 4 * hidden_size, input_size]
|
|
R_shape = [num_directions, 4 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_C_t = ng.parameter(C_t_shape, name="C_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "BIDIRECTIONAL"
|
|
node = ng.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node.get_type_name() == "LSTMSequence"
|
|
assert node.get_output_size() == 3
|
|
|
|
activations = ["RELU", "tanh", "Sigmoid"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 1.22
|
|
|
|
node_param = ng.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "LSTMSequence"
|
|
assert node_param.get_output_size() == 3
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_lstm_sequence_operator_reverse(dtype):
|
|
batch_size = 2
|
|
input_size = 4
|
|
hidden_size = 3
|
|
num_directions = 1
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
C_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, 4 * hidden_size, input_size]
|
|
R_shape = [num_directions, 4 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_C_t = ng.parameter(C_t_shape, name="C_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "REVERSE"
|
|
|
|
node_default = ng.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "LSTMSequence"
|
|
assert node_default.get_output_size() == 3
|
|
|
|
activations = ["RELU", "tanh", "Sigmoid"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 1.22
|
|
|
|
node_param = ng.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "LSTMSequence"
|
|
assert node_param.get_output_size() == 3
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_lstm_sequence_operator_forward(dtype):
|
|
batch_size = 2
|
|
input_size = 4
|
|
hidden_size = 3
|
|
num_directions = 1
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
C_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, 4 * hidden_size, input_size]
|
|
R_shape = [num_directions, 4 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_C_t = ng.parameter(C_t_shape, name="C_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "forward"
|
|
|
|
node_default = ng.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "LSTMSequence"
|
|
assert node_default.get_output_size() == 3
|
|
|
|
activations = ["RELU", "tanh", "Sigmoid"]
|
|
activation_alpha = [2.0]
|
|
activation_beta = [1.0]
|
|
clip = 0.5
|
|
|
|
node = ng.lstm_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_C_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node.get_type_name() == "LSTMSequence"
|
|
assert node.get_output_size() == 3
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_gru_sequence_operator_bidirectional(dtype):
|
|
batch_size = 1
|
|
input_size = 16
|
|
hidden_size = 128
|
|
num_directions = 2
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, 3 * hidden_size, input_size]
|
|
R_shape = [num_directions, 3 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 3 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "BIDIRECTIONAL"
|
|
node = ng.gru_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node.get_type_name() == "GRUSequence"
|
|
assert node.get_output_size() == 2
|
|
|
|
activations = ["RELU", "tanh"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 1.22
|
|
linear_before_reset = True
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
node_param = ng.gru_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
linear_before_reset
|
|
)
|
|
|
|
assert node_param.get_type_name() == "GRUSequence"
|
|
assert node_param.get_output_size() == 2
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_gru_sequence_operator_reverse(dtype):
|
|
batch_size = 2
|
|
input_size = 4
|
|
hidden_size = 3
|
|
num_directions = 1
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, 3 * hidden_size, input_size]
|
|
R_shape = [num_directions, 3 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 3 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "REVERSE"
|
|
|
|
node_default = ng.gru_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "GRUSequence"
|
|
assert node_default.get_output_size() == 2
|
|
|
|
activations = ["RELU", "tanh"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 1.22
|
|
linear_before_reset = True
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
node_param = ng.gru_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
linear_before_reset
|
|
)
|
|
|
|
assert node_param.get_type_name() == "GRUSequence"
|
|
assert node_param.get_output_size() == 2
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_gru_sequence_operator_forward(dtype):
|
|
batch_size = 2
|
|
input_size = 4
|
|
hidden_size = 3
|
|
num_directions = 1
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, 3 * hidden_size, input_size]
|
|
R_shape = [num_directions, 3 * hidden_size, hidden_size]
|
|
B_shape = [num_directions, 3 * hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "forward"
|
|
|
|
node_default = ng.gru_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "GRUSequence"
|
|
assert node_default.get_output_size() == 2
|
|
|
|
activations = ["RELU", "tanh"]
|
|
activation_alpha = [2.0]
|
|
activation_beta = [1.0]
|
|
clip = 0.5
|
|
linear_before_reset = True
|
|
B_shape = [num_directions, 4 * hidden_size]
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
node = ng.gru_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
linear_before_reset
|
|
)
|
|
|
|
assert node.get_type_name() == "GRUSequence"
|
|
assert node.get_output_size() == 2
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_rnn_sequence_operator_bidirectional(dtype):
|
|
batch_size = 1
|
|
input_size = 16
|
|
hidden_size = 128
|
|
num_directions = 2
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, hidden_size, input_size]
|
|
R_shape = [num_directions, hidden_size, hidden_size]
|
|
B_shape = [num_directions, hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "BIDIRECTIONAL"
|
|
node = ng.rnn_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node.get_type_name() == "RNNSequence"
|
|
assert node.get_output_size() == 2
|
|
|
|
activations = ["RELU", "tanh"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 1.22
|
|
|
|
node_param = ng.rnn_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "RNNSequence"
|
|
assert node_param.get_output_size() == 2
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_rnn_sequence_operator_reverse(dtype):
|
|
batch_size = 2
|
|
input_size = 4
|
|
hidden_size = 3
|
|
num_directions = 1
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, hidden_size, input_size]
|
|
R_shape = [num_directions, hidden_size, hidden_size]
|
|
B_shape = [num_directions, hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "REVERSE"
|
|
|
|
node_default = ng.rnn_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "RNNSequence"
|
|
assert node_default.get_output_size() == 2
|
|
|
|
activations = ["RELU", "tanh"]
|
|
activation_alpha = [1.0, 2.0, 3.0]
|
|
activation_beta = [3.0, 2.0, 1.0]
|
|
clip = 1.22
|
|
|
|
node_param = ng.rnn_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node_param.get_type_name() == "RNNSequence"
|
|
assert node_param.get_output_size() == 2
|
|
|
|
|
|
@pytest.mark.parametrize("dtype", [np.float32, np.float64])
|
|
def test_rnn_sequence_operator_forward(dtype):
|
|
batch_size = 2
|
|
input_size = 4
|
|
hidden_size = 3
|
|
num_directions = 1
|
|
seq_length = 2
|
|
|
|
X_shape = [batch_size, seq_length, input_size]
|
|
H_t_shape = [batch_size, num_directions, hidden_size]
|
|
seq_len_shape = [batch_size]
|
|
W_shape = [num_directions, hidden_size, input_size]
|
|
R_shape = [num_directions, hidden_size, hidden_size]
|
|
B_shape = [num_directions, hidden_size]
|
|
|
|
parameter_X = ng.parameter(X_shape, name="X", dtype=dtype)
|
|
parameter_H_t = ng.parameter(H_t_shape, name="H_t", dtype=dtype)
|
|
parameter_seq_len = ng.parameter(seq_len_shape, name="seq_len", dtype=np.int32)
|
|
parameter_W = ng.parameter(W_shape, name="W", dtype=dtype)
|
|
parameter_R = ng.parameter(R_shape, name="R", dtype=dtype)
|
|
parameter_B = ng.parameter(B_shape, name="B", dtype=dtype)
|
|
|
|
direction = "forward"
|
|
|
|
node_default = ng.rnn_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
)
|
|
|
|
assert node_default.get_type_name() == "RNNSequence"
|
|
assert node_default.get_output_size() == 2
|
|
|
|
activations = ["RELU", "tanh"]
|
|
activation_alpha = [2.0]
|
|
activation_beta = [1.0]
|
|
clip = 0.5
|
|
|
|
node = ng.rnn_sequence(
|
|
parameter_X,
|
|
parameter_H_t,
|
|
parameter_seq_len,
|
|
parameter_W,
|
|
parameter_R,
|
|
parameter_B,
|
|
hidden_size,
|
|
direction,
|
|
activations,
|
|
activation_alpha,
|
|
activation_beta,
|
|
clip,
|
|
)
|
|
|
|
assert node.get_type_name() == "RNNSequence"
|
|
assert node.get_output_size() == 2
|