Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Add ONNX opset 7 support. #434

Merged
merged 1 commit into from
May 15, 2019
Merged

Add ONNX opset 7 support. #434

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
57 changes: 29 additions & 28 deletions python/src/nnabla/utils/converter/onnx/exporter.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -147,7 +147,7 @@ class ParameterState:

# OnnxExporter class
class OnnxExporter:
def __init__(self, nnp, batch_size, opset="6"):
def __init__(self, nnp, batch_size, opset="7"):
self._nnp = nnp.protobuf
self._batch_size = batch_size
self._model_proto = None
Expand All @@ -165,7 +165,6 @@ def __init__(self, nnp, batch_size, opset="6"):

# opset_6 default op table
table_op_set_6 = {
# optype with same names
"Dropout": partial(self.Dropout, "6"),
"Softmax": "Softmax",
"BatchNormalization": partial(self.BatchNormalization, "6"),
Expand All @@ -181,7 +180,6 @@ def __init__(self, nnp, batch_size, opset="6"):
"Exp": "Exp",
"Identity": "Identity",
"Pad": "Pad",
# optype with different names
"ReLU": "Relu",
"PReLU": "PRelu",
"LeakyReLU": "LeakyRelu",
Expand Down Expand Up @@ -211,16 +209,13 @@ def __init__(self, nnp, batch_size, opset="6"):
"LogicalXor": partial(self.BinaryOperator, "Xor", "6"),
"Maximum2": "Max",
"Minimum2": "Min",
# optype that gets converted
"Affine": partial(self.Affine, '6'),
"MulScalar": partial(self.ElementWiseScalar, "Mul", "6"),
"MinimumScalar": "Clip",
"MaximumScalar": "Clip",
"AddScalar": partial(self.ElementWiseScalar, "Add", "6"),
"PowScalar": partial(self.ElementWiseScalar, "Pow", "6"),
"SumPooling": partial(self.SumPooling, "6"),
# optype that should get merged
# with other operators
"BroadcastTo": "",
"Split": self.Split,
"Stack": self.Stack,
Expand All @@ -235,31 +230,36 @@ def __init__(self, nnp, batch_size, opset="6"):
"BinarySigmoid": self.BinarySigmoid
}

table_op_set_7 = {
"Dropout": partial(self.Dropout, "7"),
"BatchNormalization": partial(self.BatchNormalization, "7"),
"Less": partial(self.BinaryOperator, "Less", "7"),
"Greater": partial(self.BinaryOperator, "Greater", "7"),
"Equal": partial(self.BinaryOperator, "Equal", "7"),
"Add2": partial(self.BinaryOperator, "Add", "7"),
"Mul2": partial(self.BinaryOperator, "Mul", "7"),
"Div2": partial(self.BinaryOperator, "Div", "7"),
"Pow2": partial(self.BinaryOperator, "Pow", "7"),
"Sub2": partial(self.BinaryOperator, "Sub", "7"),
"LogicalAnd": partial(self.BinaryOperator, "And", "7"),
"LogicalOr": partial(self.BinaryOperator, "Or", "7"),
"LogicalXor": partial(self.BinaryOperator, "Xor", "7"),
"Affine": partial(self.Affine, '7'),
"MulScalar": partial(self.ElementWiseScalar, "Mul", "7"),
"AddScalar": partial(self.ElementWiseScalar, "Add", "7"),
"PowScalar": partial(self.ElementWiseScalar, "Pow", "7"),
"Deconvolution": partial(self.Deconvolution, "7"),
"SumPooling": partial(self.SumPooling, "7"),
"Unpooling": self.Unpooling_7,
}
table_op_set_7 = dict(table_op_set_6, **table_op_set_7)

# opset_9 table
table_op_set_9 = {
"Dropout": partial(self.Dropout, "9"),
"Add2": partial(self.BinaryOperator, "Add", "9"),
"Mul2": partial(self.BinaryOperator, "Mul", "9"),
"Div2": partial(self.BinaryOperator, "Div", "9"),
"Pow2": partial(self.BinaryOperator, "Pow", "9"),
"Sub2": partial(self.BinaryOperator, "Sub", "9"),
"Affine": partial(self.Affine, '9'),
"Slice": self.Slice_9,
"Unpooling": self.Unpooling_9,
"SumPooling": partial(self.SumPooling, "9"),
"BatchNormalization": partial(self.BatchNormalization, "9"),
"Deconvolution": partial(self.Deconvolution, "9"),
"MulScalar": partial(self.ElementWiseScalar, "Mul", "9"),
"AddScalar": partial(self.ElementWiseScalar, "Add", "9"),
"PowScalar": partial(self.ElementWiseScalar, "Pow", "9"),
"LogicalAnd": partial(self.BinaryOperator, "And", "9"),
"LogicalOr": partial(self.BinaryOperator, "Or", "9"),
"LogicalXor": partial(self.BinaryOperator, "Xor", "9"),
"Less": partial(self.BinaryOperator, "Less", "9"),
"Greater": partial(self.BinaryOperator, "Greater", "9"),
"Equal": partial(self.BinaryOperator, "Equal", "9"),
}
table_op_set_9 = dict(table_op_set_6, **table_op_set_9)
table_op_set_9 = dict(table_op_set_7, **table_op_set_9)

# opset_ support for SNPE
table_op_set_9_x = {
Expand All @@ -269,6 +269,7 @@ def __init__(self, nnp, batch_size, opset="6"):

opver_impl_map = {
"6": table_op_set_6,
"7": table_op_set_7,
"9": table_op_set_9,
"9x": table_op_set_9_x
}
Expand Down Expand Up @@ -310,7 +311,7 @@ def Dropout(self, opset, func):
name=func.name
)
return [n]
elif opset == "9":
elif opset == "7":
n = onnx.helper.make_node(
'Dropout',
func.input,
Expand Down Expand Up @@ -351,7 +352,7 @@ def BinaryOperator(self, func_name, opset, func):
name=func.name
)
nl.append(n)
else: # opset >= 9
else: # opset >= 7
input0_shape_len = len(self._var_dict[func.input[0]].dim)
input1_shape_len = len(self._var_dict[second_input].dim)
unsqueeze_output = fork_name("broadcast_unsqueeze")
Expand Down
Loading