saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-06-03 23:49:44 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Change CPU EP behavior with auto_pad when ConvTranspose output shape is specified. (#13311)

Browse source 
### Description
Based on the ORT spec for ConvTranspose:

```
output_shape can also be explicitly specified in which case pads values are auto generated using these equations:

total_padding[i] = stride[i] * (input_size[i] - 1) + output_padding[i] + ((kernel_shape[i] - 1) * dilations[i] + 1) - output_shape[i]
If (auto_pads == SAME_UPPER): pads[start_i] = total_padding[i]/2; pads[end_i] = total_padding[i] - (total_padding[i]/2)
Else: pads[start_i] = total_padding[i] - (total_padding[i]/2); pads[end_i] = (total_padding[i]/2).
```
However the CPU EP logic differs. Basically, unless SAME_UPPER is
specified, the default behavior (for VALID,NOTSET,SAME_LOWER) should be
SAME_LOWER.

I think this is the pragmatic fix, however it's perhaps still not
totally up to standard.
In the case tested, the operator is actually only valid if padding is
inserted. Perhaps it "should" throw some error then, if auto_pad is not
SAME_UPPER or SAME_LOWER, as the spec also mentions:

"VALID mean no padding." (For convtranspose-1 but this was removed in
convtranspose-11, making it less clear.)
"NOTSET, which means explicit padding is used" (should technically
require explicit padding then, and not generate it)

HOWEVER, changing it to throw errors could do more harm than good. For
now, probably just best to make it consistent.

### Motivation and Context
We noticed that there was a discrepancy in one of the DML tests between
CPU and DML.
auto_pad is not specified, and DML is doing SAME_LOWER behavior by
default, where CPU EP is doing SAME_UPPER behavior.

```json
    {
      "graph_name": "ConvTranspose output_shape with even strides odd kernel autopad NOTSET",
      "op_type": "ConvTranspose",
      "dilations": [1,1],
      "group": 1,
      "strides": [2,2],
      "kernel_shape": [3,3],
      "output_shape": [1,1,4,4],
      "X": {"dims": [1,1,2,2], "function": "iota"},
      "W": {"dims": [1,1,3,3], "value": [1,2,3,4,5,6,7,8,9]},
      "B": [1],
      "Y": {"dims": [1,1,4,4], "value": [1,5,6,7,5,17,15,19,11,25,16,19,17,40,25,28]},
      "T": "float32"
    }
```
This commit is contained in:
Jake Mathern 2022-10-18 12:57:47 -07:00 committed by GitHub
parent 15673b4537
commit f96f222526
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 7 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

14
onnxruntime/core/providers/cpu/nn/conv_transpose_attributes.h
View file

@ -176,15 +176,15 @@ struct ConvTransposeAttributes : public ConvAttributes {
void DistributePadding(AutoPadType pad_type, const int64_t& total_pad,
int64_t& pad_head, int64_t& pad_tail) const {
if (pad_type == AutoPadType::SAME_LOWER) { // pad more on head when total_pad is odd.
pad_head = total_pad - total_pad / 2;
pad_tail = total_pad / 2;
} else {
// for pad_type is NOTSET, SAME_LOWER or VALID
// set pad_head as total_pad/2, pad_tail as total_pad-total_pad/2.
// That said, we pad more on tail when total_pad is odd.
if (pad_type == AutoPadType::SAME_UPPER) {
// pad more on tail when total_pad is odd.
pad_head = total_pad / 2;
pad_tail = total_pad - total_pad / 2;
} else {
// When pad_type is NOTSET, SAME_LOWER or VALID,
// pad more on head when total_pad is odd.
pad_head = total_pad - total_pad / 2;
pad_tail = total_pad / 2;
}
}