Pull Resquest regarding [Good First Issue]: Support quantized::conv1d, quantized::linear_relu #29343 #29948
+110
−8
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mvafin
requested changes
Apr 7, 2025
Comment on lines
91
to
137
| OutputVector translate_quantized_conv1d(const NodeContext& context) { | ||
| // "quantized::conv1d.new(Tensor qx, __torch__.torch.classes.quantized.Conv1dPackedParamsBase packed_weight, | ||
| // float output_scale, int output_zero_point) -> Tensor" | ||
| num_inputs_check(context, 4, 4); | ||
|
|
||
| auto input = context.get_input(0); | ||
|
|
||
| // (N, C, L) -> (N, C, L, 1) | ||
| auto unsqueeze_axis = ov::op::v0::Constant::create(element::i64, Shape{1}, {3}); | ||
| auto unsqueezed_input = context.mark_node(std::make_shared<ov::op::v0::Unsqueeze>(input, unsqueeze_axis)); | ||
|
|
||
| // Lower as if it's 2D conv with singleton spatial dim | ||
| auto conv = translate_quantized_convnd_base(context, unsqueezed_input); | ||
|
|
||
| // (N, C, L, 1) -> (N, C, L) | ||
| auto squeeze_axis = ov::op::v0::Constant::create(element::i64, Shape{1}, {3}); | ||
| auto squeezed_output = context.mark_node(std::make_shared<ov::op::v0::Squeeze>(conv, squeeze_axis)); | ||
|
|
||
| auto scale = context.get_input(2); | ||
| auto zero_point = context.get_input(3); | ||
|
|
||
| return {quantize(context, squeezed_output, scale, zero_point, input)}; | ||
| } | ||
|
|
||
| OutputVector translate_quantized_conv1d_relu(const NodeContext& context) { | ||
| num_inputs_check(context, 4, 4); | ||
|
|
||
| auto input = context.get_input(0); | ||
| auto unsqueeze_axis = ov::op::v0::Constant::create(element::i64, Shape{1}, {3}); | ||
| auto unsqueezed_input = context.mark_node(std::make_shared<ov::op::v0::Unsqueeze>(input, unsqueeze_axis)); | ||
|
|
||
| NodeContext new_context = context; | ||
| new_context.m_inputs[0] = unsqueezed_input; | ||
|
|
||
| auto conv = translate_quantized_convnd_base(new_context); | ||
| auto relu = context.mark_node(std::make_shared<v0::Relu>(conv)); | ||
|
|
||
| auto squeeze_axis = ov::op::v0::Constant::create(element::i64, Shape{1}, {3}); | ||
| auto squeezed_output = context.mark_node(std::make_shared<ov::op::v0::Squeeze>(relu, squeeze_axis)); | ||
|
|
||
| auto scale = context.get_input(2); | ||
| auto zero_point = context.get_input(3); | ||
|
|
||
| return {quantize(context, squeezed_output, scale, zero_point, input)}; | ||
| } | ||
|
|
||
|
|
There was a problem hiding this comment.
We already have convnd versions. Please use them, if they cannot support your case, modify them
|
Well... You at least need to add an operation in op_table.cpp in the list of supported operations. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Making a Conv1d and Conv1d_ReLu to support Pytorch Frontend