def _convert_repeat_vector(converter: KerasConverter, k_op: "keras.layers.RepeatVector"):
x = converter.get_variable(converter.get_input_tensor(k_op)[0])
assert x.order == OrderNC, f"[KerasConverter] Currently only OrderNC is supported for input variable order of " \
f"keras.layers.RepeatVector: x.order={x.order}"
N = x.shape_dict[Axis.N]
n = k_op.n
C = x.shape_dict[Axis.C]
# TODO: Implement more efficient version
# ex) x.shape=(N=2, C=3), n=2
#
# x(N, C) * w(C, n*C) = y(N, n*C) = y(N, n, C)
# -----------------------------------------------------------------------------
# [1, 2, 3] [1, 0, 0, 1, 0, 0] [1, 2, 3, 1, 2, 3] [[1, 2, 3], [1, 2, 3]]
# [4, 5, 6] * [0, 1, 0, 0, 1, 0] = [4, 5, 6, 4, 5, 6] = [[4, 5, 6], [4, 5, 6]]
# [0, 0, 1, 0, 0, 1]
#
w = ConstantVariable(np.tile(np.eye(C), (1, n)), OrderCN)
y, = Linear(None)(x, w)
y = y.reshape([N, n, C], OrderNTC)
converter.set_variable(converter.get_output_tensor(k_op)[0], y)
def __call__(self, inputs: List[Variable]) -> Tuple[Variable]:
# noinspection PyUnresolvedReferences
linear_opr = Linear(generate_unique_name(self.cfunc.label))
x = inputs[0]
w = inputs[1]
if x.ndim == 4 and w.ndim == 2:
# wを4次元に拡張 (NC -> NCHW)
x_shape_dict = x.shape_dict
w_shape_dict = w.shape_dict
assert x_shape_dict[Axis.C] * x_shape_dict[Axis.H] * x_shape_dict[Axis.W] == w_shape_dict[Axis.C]
assert w.order is OrderNC
w.order = OrderNCHW
w_new_shape = [w_shape_dict[Axis.N], x_shape_dict[Axis.C], x_shape_dict[Axis.H],
x_shape_dict[Axis.W]]
w.shape = w_new_shape
w.data = w.data.reshape(w_new_shape)
opr_out, = linear_opr(inputs[0], inputs[1])
if len(inputs) == 3:
# biasあり
# noinspection PyUnresolvedReferences
bias_opr = AxiswiseBias(generate_unique_name(self.cfunc.label), axis=Axis.C)
self.hidden_vars.append(opr_out)
opr_out, = bias_opr(opr_out, inputs[2])
return opr_out,
def convert_layer_dense(self, layer_config: Dict[str, object], inputs: List[Variable]) -> List[Variable]:
assert len(inputs) == 1
input = inputs[0]
name: str = layer_config["name"]
weight_array = self.weights[f"{name}/{name}/kernel:0"].value
weight_var = ConstantVariable(weight_array, OrderCN) # shape: (in, out)
linear_opr = Linear(name)
y, = linear_opr(input, weight_var)
if layer_config["use_bias"]:
bias_array = self.weights[f"{name}/{name}/bias:0"].value
bias_var = ConstantVariable(bias_array, OrderC)
bias_opr = AxiswiseBias(name + "_bias", Axis.C)
y, = bias_opr(y, bias_var)
act_opr: Operator = None
activation_type: str = layer_config["activation"]
if activation_type == "relu":
act_opr = Relu(name + "_activation")
elif activation_type == "softmax":
warn("omitting softmax activation")
else:
raise NotImplementedError(f"Unknown activation {activation_type}")
if act_opr is not None:
y, = act_opr(y)
return [y]
def _convert_linear_function(
converter: ChainerConverter,
c_op: "chainer.functions.connection.linear.LinearFunction"):
x = converter.get_variable(c_op.inputs[0])
w = converter.get_variable(c_op.inputs[1]) # type: ConstantVariable
x2, = Reshape(None,
in_order=x.order,
out_order=OrderNC,
out_shape=[x.shape[0], mul(x.shape[1:])])(x)
w2, = ReinterpretAxis(None, in_order=w.order, out_order=OrderNC)(w)
w2, = Transpose(None)(w2)
w2.change_order(OrderCN)
y, = Linear(None)(x2, w2)
y, = ReinterpretAxis(None,
in_order=y.order,
out_order=Order([x.order.axes[0],
w.order.axes[0]]))(y)
if len(c_op.inputs) == 3:
# with bias
b = converter.get_variable(c_op.inputs[2])
check_broadcast_constraints(y, b)
y = y + b
converter.set_variable(c_op.outputs[0](), y)
def _convert_linear_function(
converter: ChainerConverter,
c_op: "chainer.functions.connection.linear.LinearFunction"):
linear_opr = Linear(None)
x = converter.get_variable(c_op.inputs[0])
w = converter.get_variable(c_op.inputs[1])
if x.ndim == 4 and w.ndim == 2:
# wを4次元に拡張 (NC -> NCHW)
x_shape_dict = x.shape_dict
w_shape_dict = w.shape_dict
assert x_shape_dict[Axis.C] * x_shape_dict[Axis.H] * x_shape_dict[
Axis.W] == w_shape_dict[Axis.C]
assert w.order is OrderNC
w.order = OrderNCHW
w_new_shape = [
w_shape_dict[Axis.N], x_shape_dict[Axis.C], x_shape_dict[Axis.H],
x_shape_dict[Axis.W]
]
w.shape = w_new_shape
w.data = w.data.reshape(w_new_shape)
y, = linear_opr(x, w)
if len(c_op.inputs) == 3:
# with bias
bias_opr = AxiswiseBias(None, axis=Axis.C)
bias = converter.get_variable(c_op.inputs[2])
y, = bias_opr(y, bias)
converter.set_variable(c_op.outputs[0](), y)
def _convert_mul(converter: ChainerConverter, c_op: "chainer.functions.math.basic_math.Mul"):
x1 = converter.get_variable(c_op.inputs[0])
x2 = converter.get_variable(c_op.inputs[1])
x1 = x1.reinterpret_axes(OrderNC)
x2 = x2.reinterpret_axes(OrderCN)
y, = Linear(None)(x1, x2)
converter.set_variable(c_op.outputs[0](), y)
def _convert_mul(converter: ChainerConverter, c_op: chainer.functions.math.basic_math.Mul):
x1 = converter.get_variable(c_op.inputs[0])
x2 = converter.get_variable(c_op.inputs[1])
x1, = ReinterpretAxis(None, x1.order, OrderNC)(x1)
x2, = ReinterpretAxis(None, x2.order, OrderCN)(x2)
y, = Linear(None)(x1, x2)
converter.set_variable(c_op.outputs[0](), y)
def _convert_linear_function(
converter: ChainerConverter,
c_op: "chainer.functions.connection.linear.LinearFunction"):
x = converter.get_variable(c_op.inputs[0])
w = converter.get_variable(c_op.inputs[1]) # type: ConstantVariable
x2 = x.reshape([x.shape[0], mul(x.shape[1:])], OrderNC)
w2 = w.reinterpret_axes(OrderNC)
y, = Linear(None)(x2, w2)
y = y.reinterpret_axes(Order([x.order.axes[0], w.order.axes[0]]))
if len(c_op.inputs) == 3:
# with bias
b = converter.get_variable(c_op.inputs[2])
check_broadcast_constraints(y, b)
y = y + b
converter.set_variable(c_op.outputs[0](), y)
def _convert_dense(converter: KerasConverter, k_op: "keras.layers.Dense"):
x = converter.get_variable(converter.get_input_tensor(k_op)[0])
w = converter.convert_to_constant_variable(k_op.kernel, OrderCN)
y, = Linear(None)(x, w)
if k_op.use_bias:
b = converter.convert_to_constant_variable(k_op.bias, OrderC)
y = y + b
y = do_activation(k_op.activation, y)
converter.set_variable(converter.get_output_tensor(k_op)[0], y)
def test_internal_softmax():
linear1 = Linear('linear')
softmax1 = Softmax('softmax', axis=Axis.C)
linear2 = Linear('linear')
softmax2 = Softmax('softmax', axis=Axis.C)
x = Variable([4, 5], OrderNC)
w1 = Variable([4, 5], OrderNC)
w2 = Variable([3, 4], OrderNC)
h, = linear1(x, w1)
h, = softmax1(h)
h, = linear2(h, w2)
y, = softmax2(h)
graph = Graph([x], [y])
graph, _ = RemoveLastSoftmax().optimize(graph)
ops = listup_operators(graph)
assert len(ops) == 3 and isinstance(ops[0], Linear) and isinstance(ops[1], Softmax) and isinstance(ops[2], Linear)
assert len(graph.outputs) == 1 and ops[2].outputs["y"] == graph.outputs[0]
def matmul_handler(converter: TensorFlowConverter, tf_op: "tf.Operation"):
a = converter.get_variable(tf_op.inputs[0])
b = converter.get_variable(tf_op.inputs[1])
transposed_a = tf_op.get_attr("transpose_a")
transposed_b = tf_op.get_attr("transpose_b")
if a.ndim > 2 or b.ndim > 2:
raise NotImplementedError(
"[TensorFlowConverter] Currently, MatMul is supported only 2D * 2D case."
)
c_axes = []
if transposed_a:
c_axes.append(a.order.axes[-1])
if a.order != OrderCN:
a = a.reinterpret_axes(OrderCN)
else:
c_axes.append(a.order.axes[-2])
if a.order != OrderNC:
a = a.reinterpret_axes(OrderNC)
if transposed_b:
c_axes.append(Axis())
if b.order != OrderNC:
b = b.reinterpret_axes(OrderNC)
else:
c_axes.append(Axis())
if b.order != OrderCN:
b = b.reinterpret_axes(OrderCN)
c_normalized, = Linear(None)(a, b)
c = c_normalized.reinterpret_axes(Order(c_axes))
converter.set_variable(tf_op.outputs[0], c)
def matmul_handler(converter: TensorFlowConverter, tf_op: "tf.Operation"):
a = converter.get_variable(tf_op.inputs[0])
b = converter.get_variable(tf_op.inputs[1])
transposed_a = tf_op.get_attr("transpose_a")
transposed_b = tf_op.get_attr("transpose_b")
if a.ndim > 2 or b.ndim > 2:
raise NotImplementedError(
"[TensorFlowConverter] Currently, MatMul is supported only 2D * 2D case."
)
c_axes = []
if transposed_a:
c_axes.append(a.order.axes[-1])
a_axis_K = a.order.axes[-2]
if a.order != OrderCN:
a, = ReinterpretAxis(None, in_order=a.order, out_order=OrderCN)(a)
else:
c_axes.append(a.order.axes[-2])
a_axis_K = a.order.axes[-1]
if a.order != OrderNC:
a, = ReinterpretAxis(None, in_order=a.order, out_order=OrderNC)(a)
if transposed_b:
b_axis_K = b.order.axes[-1]
c_axes.append(AxisVar())
if b.order != OrderNC:
b, = ReinterpretAxis(None, in_order=b.order, out_order=OrderNC)(b)
else:
c_axes.append(AxisVar())
if b.order != OrderCN:
b, = ReinterpretAxis(None, in_order=b.order, out_order=OrderCN)(b)
b_axis_K = b.order.axes[-2]
if flags.AGGRESSIVE_ORDER_INFERENCE:
# Assumption: 2 inner multiplied axes are same.
unify(a_axis_K, b_axis_K)
c_normalized, = Linear(None)(a, b)
c, = ReinterpretAxis(None,
in_order=c_normalized.order,
out_order=Order(c_axes))(c_normalized)
converter.set_variable(tf_op.outputs[0], c)
def generate_graph_model2(caption_net, hidden_num):
# inputs
var_input_img = Variable([1, 1, hidden_num], OrderNTC)
var_input_word = Variable([1, 1], OrderNT)
var_switch_img = Variable([1, 1, hidden_num], OrderNTC)
var_switch_word = Variable([1, 1, hidden_num], OrderNTC)
var_last_h = Variable([1, hidden_num], OrderNC)
var_last_c = Variable([1, hidden_num], OrderNC)
# prepare for lstm
var_emb_word, = Embedding(None)(var_input_word,
ConstantVariable(
caption_net.word_vec.W.data,
OrderCN)) # OrderNTC
var_lstm_input = (var_emb_word * var_switch_word) + \
(var_input_img * var_switch_img)
# lstm
lstm_opr = LSTM(None,
use_bias=True,
return_sequences=False,
activation="tanh",
recurrent_activation="sigmoid",
use_initial_h=True,
use_initial_c=True)
w_input = _convert_lstm_to_webdnn_order(caption_net.lstm.upward.W.data.T)
w_hidden = _convert_lstm_to_webdnn_order(caption_net.lstm.lateral.W.data.T)
b = _convert_lstm_to_webdnn_order(
caption_net.lstm.upward.b.data[None, :])[0]
var_lstm_h, var_lstm_c = lstm_opr(
x=var_lstm_input,
w_input=ConstantVariable(w_input, OrderCN),
w_hidden=ConstantVariable(w_hidden, OrderCN),
b=ConstantVariable(b, OrderC),
initial_h=var_last_h,
initial_c=var_last_c)
# word probability
var_word_score, = Linear(None)(var_lstm_h,
ConstantVariable(
caption_net.out_word.W.data.T, OrderCN))
var_word_score_biased, = AxiswiseBias(None, axis=Axis.C)(
var_word_score, ConstantVariable(caption_net.out_word.b.data, OrderC))
var_word_prob, = Softmax(None, axis=Axis.C)(var_word_score_biased)
return Graph([
var_input_img, var_input_word, var_switch_img, var_switch_word,
var_last_h, var_last_c
], [var_word_prob, var_lstm_h, var_lstm_c])
def test_NHWC_HWCN():
vx = np.random.rand(2, 3, 4, 5)
vw = np.random.rand(3, 4, 5, 2)
vy = np.tensordot(vx, vw, ((1, 2, 3), (0, 1, 2)))
x = Variable(vx.shape, order=OrderNHWC)
w = ConstantVariable(vw, order=OrderHWCN)
y, = Linear(None)(x, w)
generate_kernel_test_case(description=f"Linear: NHWC*HWCN",
backend=["fallback", "webassembly", "webgpu"],
graph=Graph([x], [y]),
inputs={x: vx},
expected={y: vy},
raise_skip=False)
def test_NC_CN():
vx = np.random.rand(3, 4)
vw = np.random.rand(4, 5)
vy = np.dot(vx, vw)
x = Variable(vx.shape, order=OrderNC)
w = ConstantVariable(vw, order=OrderCN)
y, = Linear(None)(x, w)
generate_kernel_test_case(description=f"Linear: NC*CN",
backend=["fallback", "webassembly", "webgpu"],
graph=Graph([x], [y]),
inputs={x: vx},
expected={y: vy},
raise_skip=False)
def test_single_softmax():
linear = Linear('linear')
softmax = Softmax('softmax', axis=Axis.C)
x = Variable([4, 5], OrderNC)
w = Variable([4, 5], OrderNC)
h, = linear(x, w)
y, = softmax(h)
graph = Graph([x], [y])
graph, _ = RemoveLastSoftmax().optimize(graph)
ops = listup_operators(graph)
assert len(ops) == 1 and isinstance(ops[0], Linear)
assert len(graph.outputs) == 1 and ops[0].outputs["y"] == graph.outputs[0]
def test_every_order():
orders = [OrderNC, OrderCN, OrderNHWC, OrderHWNC, OrderHWCN, OrderNCHW, OrderCNHW, OrderCHWN]
for order1, order2 in itertools.product(orders, orders): # type: Order, Order
if not order1.check_same_axes(order2):
continue
default_order = {
2: OrderNC,
4: OrderNHWC
}
op = Linear("op")
x1 = Variable(np.arange(order1.ndim) + 1, default_order[order2.ndim])
x2 = Variable(np.arange(order2.ndim) + 1, default_order[order2.ndim])
x1.change_order(order1)
x2.change_order(order2)
y, = op(x1, x2)
assert y.shape_dict[Axis.N] == x1.shape_dict[Axis.N]
assert y.shape_dict[Axis.C] == x2.shape_dict[Axis.N]
