[PyTorch] : implement support for replicated{1,2,3} pad #28271
Conversation
Signed-off-by: 11happy <[email protected]>
|
humble ping! |
| auto data = context.get_input(0); | ||
| auto paddings = context.const_input<std::vector<int64_t>>(1); | ||
| Output<Node> pad_value = context.mark_node(v0::Constant::create(element::f32, Shape{}, {0})); | ||
| return translate_pad_common(context, data, paddings, pad_value, "replicate"); |
There was a problem hiding this comment.
I am not sure that logic from translate_pad_common is suitable for replication operation.
That is because a format of paddings is as follows <pad_begin_axis1>, <pad_end_axis1>, <pad_begin_axis2>, <pad_end_axis2>, ... in case of tuple.
May be we need to do as follows:
- do broadcast of padding to convert to
tuplecase - create a pad vector with zeros with length equal to rank of
input - scatter elements from broadcasted padding to
padvector with known indices defined for each 1d, 2d and 3d cases
There was a problem hiding this comment.
I am not able to currently get why this logic maybe incorrect can you please give an example to explain. it would be very helpful. Thank you
There was a problem hiding this comment.
Hi @11happy,
I meant logic for preparation of pad vector can be simplified but I understand it may be out-of-scope for this PR because it is used by other translators.
Let us see tests if they pass. please resolve conflicts and I will trigger CI for your PR.
Thanks.
There was a problem hiding this comment.
resolved the conflicts.
Thank you
|
build_jenkins |
Signed-off-by: 11happy <[email protected]>
Signed-off-by: 11happy <[email protected]>
|
build_jenkins |
Overview:
This pull request fixes #23322.
Testing:
CC: