• Adam Osewski's avatar
    Use GroupConvolutionBackpropData v1 in ONNX ConvTranspose. (#4158) · d80c8e42
    Adam Osewski authored
    * Use ConvolutionBackpropData in ONNX ConvTranspose operator.
    
    * Replace deprecated functions and update documentation.
    
    * Remove GroupConvolutionTranspose fused op.
    
    * Add more UT for GroupConvolutionBackpropData.
    
    - They're transformed from GroupConvolutionTranspose
    
    * Fix validate_and_infed types for ConvolutionBackpropData v1.
    
    * Add new constructor for GroupConvolutionBackpropData
    
    - Additionally do some code formatting.
    
    * Improve validation in GroupConvBackprop v0.
    
    * Update/fix upgrade & downgrade passes for
    (Group)ConvolutionBackpropData.
    
    * Add decomposition and validation for GroupConvolutionBackpropData:v1
    
    * Type properties UT for GroupConvolutionBackprop v1
    
    * Remove include of deleted header.
    
    * Add squeeze to opset1::builder.
    
    * Use opset1 squeeze.
    
    * Helper functions for padding and output spatial shape inference.
    
    These helper functions are specialized only for *ConvolutionBackpropData operators.
    
    * Update documentation.
    
    * Add more checks and remove deprecated assert.
    
    * Refactor GroupConvolutionBackpropData validation.
    
    - Use helper functions to infer padding and output shape.
    - Change get_output_shape to return only spatial features dimensions
    
    * Remove unnecessary code from donwgrade pass.
    
    The GroupConvolutionBackpropData node has already computed all those required
    information.
    
    * Call Node API. Small review suggestions.
    
    * Work only on spatial dimensions inside helper functions.
    
    * Enhance documentation for ConvolutionBackpropData
    
    * Refactoring.
    
    * Reuse values computed in downgraded operator.
    
    * Check for correct filters input shape.
    
    * Add additional check on auto pad value.
    
    * Remove unused variables.
    
    * Rename helper function.
    
    * Add more checks.
    
    * Override is dynamic and add more restrictive check.
    
    * Fix merge artifact and force unsigned type for argument.
    
    * Review comments: renaming.
    
    * Change Convolution + bias pattern.
    
    Change from bias broadcasting to appropriate convolution output shape
    into reshape into (1, C, 1, 1) shape, which sould work better.
    
    * Add few more test cases for ONNX ConvTranspose op.
    
    * Skip one UT failing on PlaidML backend due to accuracy.
    d80c8e42
Name
Last commit
Last update
..
abc.in.cpp Loading commit data...
abs.in.cpp Loading commit data...
acos.in.cpp Loading commit data...
add.in.cpp Loading commit data...
aliased_output.in.cpp Loading commit data...
all.in.cpp Loading commit data...
any.in.cpp Loading commit data...
api.in.cpp Loading commit data...
arg_reduce.in.cpp Loading commit data...
asin.in.cpp Loading commit data...
atan.in.cpp Loading commit data...
atan2.in.cpp Loading commit data...
auto_broadcast.in.cpp Loading commit data...
autodiff.in.cpp Loading commit data...
batch_mat_mul.in.cpp Loading commit data...
batch_norm.in.cpp Loading commit data...
broadcast.in.cpp Loading commit data...
builder_flatten.in.cpp Loading commit data...
ceiling.in.cpp Loading commit data...
comparison.in.cpp Loading commit data...
computation_reuse.in.cpp Loading commit data...
concat.in.cpp Loading commit data...
constant.in.cpp Loading commit data...
convert.in.cpp Loading commit data...
convolution.in.cpp Loading commit data...
convolution_reference.in.cpp Loading commit data...
cos.in.cpp Loading commit data...
cosh.in.cpp Loading commit data...
cum_sum.in.cpp Loading commit data...
distributed.in.cpp Loading commit data...
divide.in.cpp Loading commit data...
dot.in.cpp Loading commit data...
dyn_broadcast.in.cpp Loading commit data...
dyn_replace_slice_reference.in.cpp Loading commit data...
dyn_reshape.in.cpp Loading commit data...
dyn_slice_reference.in.cpp Loading commit data...
dynamic.in.cpp Loading commit data...
embedding_lookup.in.cpp Loading commit data...
erf.in.cpp Loading commit data...
exp.in.cpp Loading commit data...
floor.in.cpp Loading commit data...
function_name.in.cpp Loading commit data...
fused_op.in.cpp Loading commit data...
gather.in.cpp Loading commit data...
gelu.in.cpp Loading commit data...
generate_mask.in.cpp Loading commit data...
graph_comparison.in.cpp Loading commit data...
group_convolution.in.cpp Loading commit data...
layer_norm.in.cpp Loading commit data...
log.in.cpp Loading commit data...
logical_and.in.cpp Loading commit data...
logical_or.in.cpp Loading commit data...
logical_xor.in.cpp Loading commit data...
lrn.in.cpp Loading commit data...
matmul.in.cpp Loading commit data...
max.in.cpp Loading commit data...
maximum.in.cpp Loading commit data...
min.in.cpp Loading commit data...
minimum.in.cpp Loading commit data...
mlir.in.cpp Loading commit data...
multiple_backends.in.cpp Loading commit data...
multiple_result.in.cpp Loading commit data...
multiply.in.cpp Loading commit data...
negative.in.cpp Loading commit data...
node_name.in.cpp Loading commit data...
not.in.cpp Loading commit data...
numeric.in.cpp Loading commit data...
one_hot.in.cpp Loading commit data...
pad.in.cpp Loading commit data...
parameter_as_output.in.cpp Loading commit data...
partial_slice.in.cpp Loading commit data...
pool.in.cpp Loading commit data...
power.in.cpp Loading commit data...
product.in.cpp Loading commit data...
quantize_dequantize.in.cpp Loading commit data...
quantized_convolution.in.cpp Loading commit data...
quantized_dot.in.cpp Loading commit data...
random_uniform.in.cpp Loading commit data...
range.in.cpp Loading commit data...
relu.in.cpp Loading commit data...
replace_slice.in.cpp Loading commit data...
reshape.in.cpp Loading commit data...
reverse.in.cpp Loading commit data...
reverse_sequence.in.cpp Loading commit data...
round.in.cpp Loading commit data...
scatter.in.cpp Loading commit data...
select.in.cpp Loading commit data...
shape_of.in.cpp Loading commit data...
sigmoid.in.cpp Loading commit data...
sign.in.cpp Loading commit data...
sin.in.cpp Loading commit data...
sinh.in.cpp Loading commit data...
slice.in.cpp Loading commit data...
softmax.in.cpp Loading commit data...
sqrt.in.cpp Loading commit data...
strided_slice.in.cpp Loading commit data...
subtract.in.cpp Loading commit data...
sum.in.cpp Loading commit data...
tan.in.cpp Loading commit data...
tanh.in.cpp Loading commit data...
tile.in.cpp Loading commit data...
topk.in.cpp Loading commit data...
transpose.in.cpp Loading commit data...
unhandled_op.in.cpp Loading commit data...
validate_call.in.cpp Loading commit data...
zero_sized.in.cpp Loading commit data...