def decorrelate_transformer_converter(scope, operator, container):
op = operator.raw_operator
opv = container.target_opset
out = operator.outputs
# We retrieve the unique input.
X = operator.inputs[0]
# In most case, computation happen in floats.
# But it might be with double. ONNX is very strict
# about types, every constant should have the same
# type as the input.
dtype = guess_numpy_type(X.type)
# We tell in ONNX language how to compute the unique output.
# op_version=opv tells which opset is requested
Y = OnnxMatMul(OnnxSub(X, op.mean_.astype(dtype), op_version=opv),
op.coef_.astype(dtype),
op_version=opv,
output_names=out[:1])
Y.add_to(scope, container)
def decorrelate_transformer_converter(scope, operator, container):
op = operator.raw_operator
opv = container.target_opset
out = operator.outputs
X = operator.inputs[0]
dtype = guess_numpy_type(X.type)
Y1 = OnnxMatMul(
OnnxSub(X, op.mean_.astype(dtype), op_version=opv),
op.coef_.astype(dtype),
op_version=opv, output_names=out[:1])
Y2 = OnnxGemm(X, op.coef_.astype(dtype),
(- op.mean_ @ op.coef_).astype(dtype),
op_version=opv, alpha=1., beta=1.,
output_names=out[1:2])
Y1.add_to(scope, container)
Y2.add_to(scope, container)
def live_decorrelate_transformer_converter(scope, operator, container):
# shortcuts
op = operator.raw_operator
opv = container.target_opset
out = operator.outputs
# We retrieve the unique input.
X = operator.inputs[0]
# We guess its type. If the operator ingests float (or double),
# it outputs float (or double).
proto_dtype = guess_proto_type(X.type)
dtype = guess_numpy_type(X.type)
# Lines in comment specify the numpy computation
# the ONNX code implements.
# mean_ = numpy.mean(X, axis=0, keepdims=True)
mean = OnnxReduceMean(X, axes=[0], keepdims=1, op_version=opv)
# This is trick I often use. The converter automatically
# chooses a name for every output. In big graph,
# it is difficult to know which operator is producing which output.
# This line just tells every node must prefix its ouputs with this string.
# It also applies to all inputs nodes unless this method
# was called for one of these nodes.
mean.set_onnx_name_prefix('mean')
# X2 = X - mean_
X2 = OnnxSub(X, mean, op_version=opv)
# V = X2.T @ X2 / X2.shape[0]
N = OnnxGatherElements(OnnxShape(X, op_version=opv),
numpy.array([0], dtype=numpy.int64),
op_version=opv)
Nf = OnnxCast(N, to=proto_dtype, op_version=opv)
# Every output involved in N and Nf is prefixed by 'N'.
Nf.set_onnx_name_prefix('N')
V = OnnxDiv(OnnxMatMul(OnnxTranspose(X2, op_version=opv),
X2,
op_version=opv),
Nf,
op_version=opv)
V.set_onnx_name_prefix('V1')
# V += numpy.identity(V.shape[0]) * self.alpha
V = OnnxAdd(V,
op.alpha * numpy.identity(op.nf_, dtype=dtype),
op_version=opv)
V.set_onnx_name_prefix('V2')
# L, P = numpy.linalg.eig(V)
LP = OnnxEig(V, eigv=True, op_version=opv)
LP.set_onnx_name_prefix('LP')
# Linv = L ** (-0.5)
# Notation LP[0] means OnnxPow is taking the first output
# of operator OnnxEig, LP[1] would mean the second one
# LP is not allowed as it is ambiguous
Linv = OnnxPow(LP[0], numpy.array([-0.5], dtype=dtype), op_version=opv)
Linv.set_onnx_name_prefix('Linv')
# diag = numpy.diag(Linv)
diag = OnnxMul(OnnxEyeLike(numpy.zeros((op.nf_, op.nf_),
dtype=numpy.int64),
k=0,
op_version=opv),
Linv,
op_version=opv)
diag.set_onnx_name_prefix('diag')
# root = P @ diag @ P.transpose()
trv = OnnxTranspose(LP[1], op_version=opv)
coef_left = OnnxMatMul(LP[1], diag, op_version=opv)
coef_left.set_onnx_name_prefix('coef_left')
coef = OnnxMatMul(coef_left, trv, op_version=opv)
coef.set_onnx_name_prefix('coef')
# Same part as before.
Y = OnnxMatMul(X2, coef, op_version=opv, output_names=out[:1])
Y.set_onnx_name_prefix('Y')
# The last line specifies the final output.
# Every node involved in the computation is added to the ONNX
# graph at this stage.
Y.add_to(scope, container)
def live_decorrelate_transformer_converter(scope, operator, container):
op = operator.raw_operator
opv = container.target_opset
out = operator.outputs
# We retrieve the unique input.
X = operator.inputs[0]
proto_dtype = guess_proto_type(X.type)
dtype = guess_numpy_type(X.type)
# new part
# mean_ = numpy.mean(X, axis=0, keepdims=True)
mean = OnnxReduceMean(X, axes=[0], keepdims=1, op_version=opv)
mean.set_onnx_name_prefix('mean')
# X2 = X - mean_
X2 = OnnxSub(X, mean, op_version=opv)
# V = X2.T @ X2 / X2.shape[0]
N = OnnxGatherElements(OnnxShape(X, op_version=opv),
numpy.array([0], dtype=numpy.int64),
op_version=opv)
Nf = OnnxCast(N, to=proto_dtype, op_version=opv)
Nf.set_onnx_name_prefix('N')
V = OnnxDiv(OnnxMatMul(OnnxTranspose(X2, op_version=opv),
X2,
op_version=opv),
Nf,
op_version=opv)
V.set_onnx_name_prefix('V1')
# V += numpy.identity(V.shape[0]) * self.alpha
V = OnnxAdd(V,
op.alpha * numpy.identity(op.nf_, dtype=dtype),
op_version=opv)
V.set_onnx_name_prefix('V2')
# L, P = numpy.linalg.eig(V)
LP = OnnxEig(V, eigv=True, op_version=opv)
LP.set_onnx_name_prefix('LP')
# Linv = L ** (-0.5)
Linv = OnnxPow(LP[0], numpy.array([-0.5], dtype=dtype), op_version=opv)
Linv.set_onnx_name_prefix('Linv')
# diag = numpy.diag(Linv)
diag = OnnxMul(OnnxEyeLike(numpy.array([op.nf_, op.nf_],
dtype=numpy.int64),
k=0,
op_version=opv),
Linv,
op_version=opv)
diag.set_onnx_name_prefix('diag')
# root = P @ diag @ P.transpose()
trv = OnnxTranspose(LP[1], op_version=opv)
coef_left = OnnxMatMul(LP[1], diag, op_version=opv)
coef_left.set_onnx_name_prefix('coef_left')
coef = OnnxMatMul(coef_left, trv, op_version=opv)
coef.set_onnx_name_prefix('coef')
# Same part as before.
Y = OnnxMatMul(X2, coef, op_version=opv, output_names=out[:1])
Y.set_onnx_name_prefix('Y')
Y.add_to(scope, container)