22050 lines
702 KiB
XML
22050 lines
702 KiB
XML
<?xml version="1.0"?>
|
|
<net name="Model0" version="11">
|
|
<layers>
|
|
<layer id="1" name="input_ids" type="Parameter" version="opset1">
|
|
<data shape="?,?" element_type="i64" />
|
|
<output>
|
|
<port id="0" precision="I64" names="input_ids">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="0" name="attention_mask" type="Parameter" version="opset1">
|
|
<data shape="?,?" element_type="i64" />
|
|
<output>
|
|
<port id="0" precision="I64" names="attention_mask">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="2" name="Constant_36913946" type="Const" version="opset1">
|
|
<data element_type="i8" shape="50265, 768" offset="0" size="38603520" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>50265</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="3" name="Convert_36913947" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>50265</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>50265</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="4" name="Constant_36913948" type="Const" version="opset1">
|
|
<data element_type="f32" shape="50265, 1" offset="38603520" size="201060" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>50265</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="5" name="__module.embeddings.word_embeddings/aten::embedding/Gather/fq_weights_0" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>50265</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>50265</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>50265</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="6" name="__module.embeddings.word_embeddings/aten::embedding/Convert" type="Convert" version="opset1">
|
|
<data destination_type="i32" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="7" name="__module.embeddings.word_embeddings/aten::embedding/Constant" type="Const" version="opset1">
|
|
<data element_type="i32" shape="" offset="38804580" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="8" name="__module.embeddings.word_embeddings/aten::embedding/Gather" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>50265</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="2" precision="I32" />
|
|
</input>
|
|
<output>
|
|
<port id="3" precision="FP32" names="83,inputs_embeds">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="9" name="Constant_36913950" type="Const" version="opset1">
|
|
<data element_type="i8" shape="1, 768" offset="38804584" size="768" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="10" name="Convert_36913951" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="11" name="Constant_36913952" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1" offset="38805352" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="12" name="__module.embeddings.token_type_embeddings/aten::embedding/Gather/fq_weights_0" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="13" name="aten::slice/Slice" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1, 514" offset="38805356" size="4112" />
|
|
<output>
|
|
<port id="0" precision="I64" names="21">
|
|
<dim>1</dim>
|
|
<dim>514</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="14" name="aten::slice/Reshape" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="38809468" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="15" name="ShapeOf_9157" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64" names="26">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="16" name="Constant_9270" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="38809476" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="17" name="Constant_9159" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="38809468" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="18" name="Gather_9160" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="2" precision="I64" />
|
|
</input>
|
|
<output>
|
|
<port id="3" precision="I64" names="13,15,16">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="19" name="aten::slice/Reshape_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="38809476" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="20" name="aten::slice/Reshape_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="38809476" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="21" name="aten::slice/Slice_1" type="Slice" version="opset8">
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
<dim>514</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="2" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="3" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="4" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="I64" names="25,buffered_token_type_ids">
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="22" name="aten::expand/Broadcast" type="Broadcast" version="opset3">
|
|
<data mode="bidirectional" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="28">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="23" name="__module.embeddings.token_type_embeddings/aten::embedding/Convert" type="Convert" version="opset1">
|
|
<data destination_type="i32" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="24" name="__module.embeddings.token_type_embeddings/aten::embedding/Constant" type="Const" version="opset1">
|
|
<data element_type="i32" shape="" offset="38804580" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="25" name="__module.embeddings.token_type_embeddings/aten::embedding/Gather" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="2" precision="I32" />
|
|
</input>
|
|
<output>
|
|
<port id="3" precision="FP32" names="85,token_type_embeddings.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="26" name="__module.embeddings/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="86_1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="27" name="Constant_36913954" type="Const" version="opset1">
|
|
<data element_type="i8" shape="514, 768" offset="38809484" size="394752" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>514</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="28" name="Convert_36913955" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>514</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>514</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="29" name="Constant_36913956" type="Const" version="opset1">
|
|
<data element_type="f32" shape="514, 1" offset="39204236" size="2056" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>514</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="30" name="__module.embeddings.position_embeddings/aten::embedding/Gather/fq_weights_0" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>514</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>514</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>514</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="31" name="Constant_8956" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1, 1" offset="38809476" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="32" name="__module.embeddings/aten::ne/NotEqual" type="NotEqual" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="BOOL" names="74">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="33" name="__module.embeddings/aten::to/Convert" type="Convert" version="opset1">
|
|
<data destination_type="i32" />
|
|
<input>
|
|
<port id="0" precision="BOOL">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I32" names="75,mask">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="34" name="68" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="38809476" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" names="68" />
|
|
</output>
|
|
</layer>
|
|
<layer id="35" name="__module.embeddings/aten::cumsum/CumSum" type="CumSum" version="opset3">
|
|
<data exclusive="false" reverse="false" />
|
|
<input>
|
|
<port id="0" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I64" />
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I32" names="76,77,78">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="36" name="__module.embeddings/aten::mul/Multiply" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I32" names="79,incremental_indices">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="37" name="__module.embeddings/aten::to/Convert_1" type="Convert" version="opset1">
|
|
<data destination_type="i64" />
|
|
<input>
|
|
<port id="0" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64" names="80">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="38" name="Constant_8957" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1, 1" offset="38809476" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="39" name="__module.embeddings/aten::add/Add_2" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="81">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="40" name="__module.embeddings.position_embeddings/aten::embedding/Convert" type="Convert" version="opset1">
|
|
<data destination_type="i32" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="41" name="__module.embeddings.position_embeddings/aten::embedding/Constant" type="Const" version="opset1">
|
|
<data element_type="i32" shape="" offset="38804580" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="42" name="__module.embeddings.position_embeddings/aten::embedding/Gather" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>514</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="2" precision="I32" />
|
|
</input>
|
|
<output>
|
|
<port id="3" precision="FP32" names="88,position_embeddings.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="43" name="__module.embeddings/aten::add_/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="86,embeddings.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="44" name="__module.embeddings.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="45" name="__module.embeddings.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="46" name="Constant_8958" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="39206296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="47" name="__module.embeddings.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="48" name="Constant_8959" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="39209368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="49" name="__module.embeddings.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="93,input.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="50" name="__module.embeddings.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="39212440" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="51" name="__module.embeddings.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="52" name="__module.embeddings.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39215512" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="53" name="__module.embeddings.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39215516" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="54" name="__module.embeddings.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39215512" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="55" name="__module.embeddings.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39215516" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="56" name="__module.embeddings.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="57" name="Constant_36913958" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="39215520" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="58" name="Convert_36913959" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="59" name="Constant_36913960" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="39805344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="60" name="__module.encoder.layer.0.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="61" name="__module.encoder.layer.0.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="62" name="Constant_8960" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="39808416" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="63" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="145,x.9">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="64" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39811488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="65" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39811492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="66" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39811488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="67" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39811492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="68" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="69" name="__module.encoder.layer.0.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="70" name="__module.encoder.layer.0.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="167,x.11">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="71" name="Constant_315" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="168">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="72" name="__module.encoder.layer.0.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="169">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="73" name="Constant_36913962" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="39811560" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="74" name="Convert_36913963" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="75" name="Constant_36913964" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="40401384" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="76" name="__module.encoder.layer.0.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="77" name="__module.encoder.layer.0.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="78" name="Constant_8961" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40404456" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="79" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="148,x.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="80" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40407528" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="81" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40407532" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="82" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40407528" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="83" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40407532" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="84" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="85" name="__module.encoder.layer.0.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="86" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="152,x.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="87" name="Constant_275" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="153">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="88" name="__module.encoder.layer.0.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="154,key_layer.1">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="89" name="__module.encoder.layer.0.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="171,attention_scores.1">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="90" name="Constant_8962" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="91" name="__module.encoder.layer.0.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="172,attention_scores.3">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="92" name="Constant_8964" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407540" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="93" name="34" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="38809476" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" names="34" />
|
|
</output>
|
|
</layer>
|
|
<layer id="94" name="aten::unsqueeze/Unsqueeze" type="Unsqueeze" version="opset1">
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I64" />
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="35">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="95" name="36" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="40407544" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" names="36" />
|
|
</output>
|
|
</layer>
|
|
<layer id="96" name="aten::unsqueeze/Unsqueeze_1" type="Unsqueeze" version="opset1">
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I64" />
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="37,42,extended_attention_mask">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="97" name="aten::to/Convert" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="47">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="98" name="Constant_8963" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407540" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="99" name="aten::rsub/Multiply" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="100" name="aten::rsub/Subtract" type="Subtract" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="50">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="101" name="Constant_8965" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407552" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="102" name="aten::mul/Multiply" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="52,attention_mask">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="103" name="__module.encoder.layer.0.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="173,input.3">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="104" name="__module.encoder.layer.0.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="174,input.5">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="105" name="Constant_36913966" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="40407556" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="106" name="Convert_36913967" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="107" name="Constant_36913968" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="40997380" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="108" name="__module.encoder.layer.0.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="109" name="__module.encoder.layer.0.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="110" name="Constant_8966" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="41000452" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="111" name="__module.encoder.layer.0.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="157,x.5">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="112" name="__module.encoder.layer.0.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="113" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="161,x.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="114" name="Constant_298" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="162">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="115" name="__module.encoder.layer.0.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="163">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="116" name="__module.encoder.layer.0.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="176,context_layer.1">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="117" name="Constant_396" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="177">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="118" name="__module.encoder.layer.0.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="178">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="119" name="__module.encoder.layer.0.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="120" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="183">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="121" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="41003548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="122" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="123" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="41006620" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="124" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="41006624" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="125" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="41006620" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="126" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="41006624" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="127" name="__module.encoder.layer.0.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="128" name="Constant_36913970" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="41006628" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="129" name="Convert_36913971" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="130" name="Constant_36913972" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="41596452" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="131" name="__module.encoder.layer.0.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="132" name="__module.encoder.layer.0.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="133" name="Constant_8967" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="41599524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="134" name="__module.encoder.layer.0.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="189,input.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="135" name="__module.encoder.layer.0.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="191">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="136" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="137" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="138" name="Constant_8968" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="41602596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="139" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="140" name="Constant_8969" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="41605668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="141" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="195,input_tensor.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="142" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="41608740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="143" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="144" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="41611812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="145" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="41611816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="146" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="41611812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="147" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="41611816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="148" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="149" name="Constant_36913974" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="41611820" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="150" name="Convert_36913975" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="151" name="Constant_36913976" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="43971116" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="152" name="__module.encoder.layer.0.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="153" name="__module.encoder.layer.0.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="154" name="Constant_8970" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="43983404" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="155" name="__module.encoder.layer.0.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="200">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="156" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="201">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="157" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="43995692" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="158" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="159" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="44007980" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="160" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="44007984" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="161" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="44007980" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="162" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="44007984" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="163" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="164" name="Constant_36913978" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="44007988" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="165" name="Convert_36913979" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="166" name="Constant_36913980" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="46367284" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="167" name="__module.encoder.layer.0.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="168" name="__module.encoder.layer.0.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="169" name="Constant_8971" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="46370356" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="170" name="__module.encoder.layer.0.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="207,input.9">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="171" name="__module.encoder.layer.0.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="209">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="172" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="173" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="174" name="Constant_8972" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="46373428" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="175" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="176" name="Constant_8973" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="46376500" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="177" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="213,input_tensor.5">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="178" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="46379572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="179" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="180" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46382644" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="181" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46382648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="182" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46382644" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="183" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46382648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="184" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="185" name="Constant_36913982" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="46382652" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="186" name="Convert_36913983" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="187" name="Constant_36913984" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="46972476" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="188" name="__module.encoder.layer.1.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="189" name="__module.encoder.layer.1.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="190" name="Constant_8974" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="46975548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="191" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="224,x.21">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="192" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46978620" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="193" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46978624" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="194" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46978620" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="195" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46978624" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="196" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="197" name="__module.encoder.layer.1.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="198" name="__module.encoder.layer.1.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="246,x.23">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="199" name="Constant_563" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="247">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="200" name="__module.encoder.layer.1.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="248">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="201" name="Constant_36913986" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="46978628" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="202" name="Convert_36913987" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="203" name="Constant_36913988" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="47568452" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="204" name="__module.encoder.layer.1.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="205" name="__module.encoder.layer.1.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="206" name="Constant_8975" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47571524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="207" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="227,x.13">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="208" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47574596" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="209" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47574600" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="210" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47574596" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="211" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47574600" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="212" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="213" name="__module.encoder.layer.1.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="214" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="231,x.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="215" name="Constant_523" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="232">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="216" name="__module.encoder.layer.1.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="233,key_layer.3">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="217" name="__module.encoder.layer.1.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="250,attention_scores.5">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="218" name="Constant_8976" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="219" name="__module.encoder.layer.1.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="251,attention_scores.7">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="220" name="__module.encoder.layer.1.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="252,input.11">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="221" name="__module.encoder.layer.1.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="253,input.13">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="222" name="Constant_36913990" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="47574604" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="223" name="Convert_36913991" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="224" name="Constant_36913992" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="48164428" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="225" name="__module.encoder.layer.1.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="226" name="__module.encoder.layer.1.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="227" name="Constant_8977" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="48167500" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="228" name="__module.encoder.layer.1.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="236,x.17">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="229" name="__module.encoder.layer.1.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="230" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="240,x.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="231" name="Constant_546" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="241">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="232" name="__module.encoder.layer.1.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="242">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="233" name="__module.encoder.layer.1.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="255,context_layer.5">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="234" name="Constant_644" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="256">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="235" name="__module.encoder.layer.1.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="257">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="236" name="__module.encoder.layer.1.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="237" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="262">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="238" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="48170572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="239" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="240" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="48173644" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="241" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="48173648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="242" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="48173644" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="243" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="48173648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="244" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="245" name="Constant_36913994" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="48173652" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="246" name="Convert_36913995" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="247" name="Constant_36913996" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="48763476" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="248" name="__module.encoder.layer.1.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="249" name="__module.encoder.layer.1.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="250" name="Constant_8978" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="48766548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="251" name="__module.encoder.layer.1.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="268,input.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="252" name="__module.encoder.layer.1.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="270">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="253" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="254" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="255" name="Constant_8979" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="48769620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="256" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="257" name="Constant_8980" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="48772692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="258" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="274,input_tensor.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="259" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="48775764" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="260" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="261" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="48778836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="262" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="48778840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="263" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="48778836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="264" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="48778840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="265" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="266" name="Constant_36913998" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="48778844" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="267" name="Convert_36913999" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="268" name="Constant_36914000" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="51138140" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="269" name="__module.encoder.layer.1.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="270" name="__module.encoder.layer.1.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="271" name="Constant_8981" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="51150428" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="272" name="__module.encoder.layer.1.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="279">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="273" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="280">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="274" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="51162716" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="275" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="276" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="51175004" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="277" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="51175008" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="278" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="51175004" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="279" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="51175008" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="280" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="281" name="Constant_36914002" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="51175012" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="282" name="Convert_36914003" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="283" name="Constant_36914004" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="53534308" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="284" name="__module.encoder.layer.1.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="285" name="__module.encoder.layer.1.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="286" name="Constant_8982" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="53537380" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="287" name="__module.encoder.layer.1.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="286,input.17">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="288" name="__module.encoder.layer.1.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="288">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="289" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="290" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="291" name="Constant_8983" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="53540452" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="292" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="293" name="Constant_8984" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="53543524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="294" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="292,input_tensor.9">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="295" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="53546596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="296" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="297" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53549668" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="298" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53549672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="299" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53549668" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="300" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53549672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="301" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="302" name="Constant_36914006" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="53549676" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="303" name="Convert_36914007" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="304" name="Constant_36914008" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="54139500" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="305" name="__module.encoder.layer.2.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="306" name="__module.encoder.layer.2.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="307" name="Constant_8985" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="54142572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="308" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="303,x.33">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="309" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54145644" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="310" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54145648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="311" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54145644" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="312" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54145648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="313" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="314" name="__module.encoder.layer.2.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="315" name="__module.encoder.layer.2.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="325,x.35">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="316" name="Constant_811" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="326">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="317" name="__module.encoder.layer.2.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="327">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="318" name="Constant_36914010" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="54145652" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="319" name="Convert_36914011" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="320" name="Constant_36914012" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="54735476" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="321" name="__module.encoder.layer.2.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="322" name="__module.encoder.layer.2.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="323" name="Constant_8986" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="54738548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="324" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="306,x.25">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="325" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54741620" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="326" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54741624" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="327" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54741620" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="328" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54741624" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="329" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="330" name="__module.encoder.layer.2.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="331" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="310,x.27">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="332" name="Constant_771" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="311">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="333" name="__module.encoder.layer.2.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="312,key_layer.5">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="334" name="__module.encoder.layer.2.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="329,attention_scores.9">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="335" name="Constant_8987" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="336" name="__module.encoder.layer.2.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="330,attention_scores.11">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="337" name="__module.encoder.layer.2.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="331,input.19">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="338" name="__module.encoder.layer.2.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="332,input.21">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="339" name="Constant_36914014" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="54741628" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="340" name="Convert_36914015" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="341" name="Constant_36914016" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="55331452" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="342" name="__module.encoder.layer.2.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="343" name="__module.encoder.layer.2.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="344" name="Constant_8988" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55334524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="345" name="__module.encoder.layer.2.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="315,x.29">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="346" name="__module.encoder.layer.2.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="347" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="319,x.31">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="348" name="Constant_794" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="320">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="349" name="__module.encoder.layer.2.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="321">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="350" name="__module.encoder.layer.2.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="334,context_layer.9">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="351" name="Constant_892" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="335">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="352" name="__module.encoder.layer.2.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="336">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="353" name="__module.encoder.layer.2.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="354" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="341">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="355" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55337596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="356" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="357" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55340668" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="358" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55340672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="359" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55340668" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="360" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55340672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="361" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="362" name="Constant_36914018" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="55340676" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="363" name="Convert_36914019" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="364" name="Constant_36914020" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="55930500" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="365" name="__module.encoder.layer.2.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="366" name="__module.encoder.layer.2.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="367" name="Constant_8989" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55933572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="368" name="__module.encoder.layer.2.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="347,input.23">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="369" name="__module.encoder.layer.2.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="349">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="370" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="371" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="372" name="Constant_8990" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55936644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="373" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="374" name="Constant_8991" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55939716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="375" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="353,input_tensor.11">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="376" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55942788" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="377" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="378" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55945860" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="379" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55945864" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="380" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55945860" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="381" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55945864" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="382" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="383" name="Constant_36914022" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="55945868" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="384" name="Convert_36914023" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="385" name="Constant_36914024" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="58305164" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="386" name="__module.encoder.layer.2.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="387" name="__module.encoder.layer.2.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="388" name="Constant_8992" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="58317452" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="389" name="__module.encoder.layer.2.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="358">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="390" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="359">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="391" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="58329740" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="392" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="393" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="58342028" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="394" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="58342032" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="395" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="58342028" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="396" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="58342032" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="397" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="398" name="Constant_36914026" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="58342036" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="399" name="Convert_36914027" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="400" name="Constant_36914028" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="60701332" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="401" name="__module.encoder.layer.2.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="402" name="__module.encoder.layer.2.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="403" name="Constant_8993" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="60704404" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="404" name="__module.encoder.layer.2.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="365,input.25">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="405" name="__module.encoder.layer.2.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="367">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="406" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="407" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="408" name="Constant_8994" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="60707476" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="409" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="410" name="Constant_8995" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="60710548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="411" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="371,input_tensor.13">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="412" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="60713620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="413" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="414" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60716692" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="415" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60716696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="416" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60716692" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="417" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60716696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="418" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="419" name="Constant_36914030" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="60716700" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="420" name="Convert_36914031" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="421" name="Constant_36914032" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="61306524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="422" name="__module.encoder.layer.3.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="423" name="__module.encoder.layer.3.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="424" name="Constant_8996" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="61309596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="425" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="382,x.45">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="426" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61312668" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="427" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61312672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="428" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61312668" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="429" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61312672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="430" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="431" name="__module.encoder.layer.3.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="432" name="__module.encoder.layer.3.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="404,x.47">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="433" name="Constant_1059" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="405">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="434" name="__module.encoder.layer.3.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="406">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="435" name="Constant_36914034" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="61312676" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="436" name="Convert_36914035" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="437" name="Constant_36914036" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="61902500" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="438" name="__module.encoder.layer.3.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="439" name="__module.encoder.layer.3.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="440" name="Constant_8997" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="61905572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="441" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="385,x.37">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="442" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61908644" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="443" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61908648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="444" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61908644" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="445" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61908648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="446" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="447" name="__module.encoder.layer.3.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="448" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="389,x.39">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="449" name="Constant_1019" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="390">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="450" name="__module.encoder.layer.3.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="391,key_layer.7">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="451" name="__module.encoder.layer.3.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="408,attention_scores.13">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="452" name="Constant_8998" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="453" name="__module.encoder.layer.3.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="409,attention_scores.15">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="454" name="__module.encoder.layer.3.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="410,input.27">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="455" name="__module.encoder.layer.3.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="411,input.29">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="456" name="Constant_36914038" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="61908652" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="457" name="Convert_36914039" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="458" name="Constant_36914040" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="62498476" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="459" name="__module.encoder.layer.3.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="460" name="__module.encoder.layer.3.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="461" name="Constant_8999" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="62501548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="462" name="__module.encoder.layer.3.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="394,x.41">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="463" name="__module.encoder.layer.3.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="464" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="398,x.43">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="465" name="Constant_1042" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="399">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="466" name="__module.encoder.layer.3.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="400">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="467" name="__module.encoder.layer.3.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="413,context_layer.13">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="468" name="Constant_1140" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="414">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="469" name="__module.encoder.layer.3.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="415">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="470" name="__module.encoder.layer.3.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="471" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="420">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="472" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="62504620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="473" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="474" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="62507692" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="475" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="62507696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="476" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="62507692" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="477" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="62507696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="478" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="479" name="Constant_36914042" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="62507700" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="480" name="Convert_36914043" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="481" name="Constant_36914044" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="63097524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="482" name="__module.encoder.layer.3.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="483" name="__module.encoder.layer.3.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="484" name="Constant_9000" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="63100596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="485" name="__module.encoder.layer.3.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="426,input.31">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="486" name="__module.encoder.layer.3.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="428">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="487" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="488" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="489" name="Constant_9001" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="63103668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="490" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="491" name="Constant_9002" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="63106740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="492" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="432,input_tensor.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="493" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="63109812" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="494" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="495" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="63112884" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="496" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="63112888" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="497" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="63112884" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="498" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="63112888" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="499" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="500" name="Constant_36914046" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="63112892" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="501" name="Convert_36914047" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="502" name="Constant_36914048" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="65472188" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="503" name="__module.encoder.layer.3.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="504" name="__module.encoder.layer.3.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="505" name="Constant_9003" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="65484476" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="506" name="__module.encoder.layer.3.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="437">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="507" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="438">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="508" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="65496764" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="509" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="510" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="65509052" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="511" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="65509056" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="512" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="65509052" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="513" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="65509056" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="514" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="515" name="Constant_36914050" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="65509060" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="516" name="Convert_36914051" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="517" name="Constant_36914052" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="67868356" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="518" name="__module.encoder.layer.3.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="519" name="__module.encoder.layer.3.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="520" name="Constant_9004" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="67871428" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="521" name="__module.encoder.layer.3.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="444,input.33">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="522" name="__module.encoder.layer.3.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="446">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="523" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="524" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="525" name="Constant_9005" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="67874500" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="526" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="527" name="Constant_9006" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="67877572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="528" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="450,input_tensor.17">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="529" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="67880644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="530" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="531" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67883716" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="532" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67883720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="533" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67883716" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="534" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67883720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="535" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="536" name="Constant_36914054" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="67883724" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="537" name="Convert_36914055" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="538" name="Constant_36914056" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="68473548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="539" name="__module.encoder.layer.4.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="540" name="__module.encoder.layer.4.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="541" name="Constant_9007" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="68476620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="542" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="461,x.57">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="543" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="68479692" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="544" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="68479696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="545" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="68479692" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="546" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="68479696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="547" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="548" name="__module.encoder.layer.4.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="549" name="__module.encoder.layer.4.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="483,x.59">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="550" name="Constant_1307" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="484">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="551" name="__module.encoder.layer.4.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="485">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="552" name="Constant_36914058" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="68479700" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="553" name="Convert_36914059" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="554" name="Constant_36914060" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="69069524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="555" name="__module.encoder.layer.4.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="556" name="__module.encoder.layer.4.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="557" name="Constant_9008" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="69072596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="558" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="464,x.49">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="559" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69075668" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="560" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69075672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="561" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69075668" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="562" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69075672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="563" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="564" name="__module.encoder.layer.4.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="565" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="468,x.51">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="566" name="Constant_1267" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="469">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="567" name="__module.encoder.layer.4.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="470,key_layer.9">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="568" name="__module.encoder.layer.4.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="487,attention_scores.17">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="569" name="Constant_9009" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="570" name="__module.encoder.layer.4.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="488,attention_scores.19">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="571" name="__module.encoder.layer.4.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="489,input.35">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="572" name="__module.encoder.layer.4.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="490,input.37">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="573" name="Constant_36914062" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="69075676" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="574" name="Convert_36914063" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="575" name="Constant_36914064" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="69665500" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="576" name="__module.encoder.layer.4.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="577" name="__module.encoder.layer.4.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="578" name="Constant_9010" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="69668572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="579" name="__module.encoder.layer.4.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="473,x.53">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="580" name="__module.encoder.layer.4.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="581" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="477,x.55">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="582" name="Constant_1290" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="478">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="583" name="__module.encoder.layer.4.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="479">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="584" name="__module.encoder.layer.4.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="492,context_layer.17">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="585" name="Constant_1388" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="493">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="586" name="__module.encoder.layer.4.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="494">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="587" name="__module.encoder.layer.4.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="588" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="499">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="589" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="69671644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="590" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="591" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69674716" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="592" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69674720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="593" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69674716" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="594" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69674720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="595" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="596" name="Constant_36914066" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="69674724" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="597" name="Convert_36914067" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="598" name="Constant_36914068" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="70264548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="599" name="__module.encoder.layer.4.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="600" name="__module.encoder.layer.4.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="601" name="Constant_9011" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="70267620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="602" name="__module.encoder.layer.4.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="505,input.39">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="603" name="__module.encoder.layer.4.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="507">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="604" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="605" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="606" name="Constant_9012" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="70270692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="607" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="608" name="Constant_9013" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="70273764" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="609" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="511,input_tensor.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="610" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="70276836" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="611" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="612" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="70279908" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="613" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="70279912" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="614" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="70279908" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="615" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="70279912" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="616" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="617" name="Constant_36914070" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="70279916" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="618" name="Convert_36914071" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="619" name="Constant_36914072" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="72639212" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="620" name="__module.encoder.layer.4.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="621" name="__module.encoder.layer.4.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="622" name="Constant_9014" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="72651500" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="623" name="__module.encoder.layer.4.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="516">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="624" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="517">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="625" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="72663788" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="626" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="627" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="72676076" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="628" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="72676080" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="629" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="72676076" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="630" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="72676080" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="631" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="632" name="Constant_36914074" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="72676084" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="633" name="Convert_36914075" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="634" name="Constant_36914076" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="75035380" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="635" name="__module.encoder.layer.4.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="636" name="__module.encoder.layer.4.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="637" name="Constant_9015" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="75038452" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="638" name="__module.encoder.layer.4.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="523,input.41">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="639" name="__module.encoder.layer.4.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="525">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="640" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="641" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="642" name="Constant_9016" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="75041524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="643" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="644" name="Constant_9017" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="75044596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="645" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="529,input_tensor.21">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="646" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="75047668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="647" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="648" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75050740" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="649" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75050744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="650" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75050740" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="651" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75050744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="652" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="653" name="Constant_36914078" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="75050748" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="654" name="Convert_36914079" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="655" name="Constant_36914080" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="75640572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="656" name="__module.encoder.layer.5.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="657" name="__module.encoder.layer.5.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="658" name="Constant_9018" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="75643644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="659" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="540,x.69">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="660" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75646716" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="661" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75646720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="662" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75646716" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="663" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75646720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="664" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="665" name="__module.encoder.layer.5.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="666" name="__module.encoder.layer.5.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="562,x.71">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="667" name="Constant_1555" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="563">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="668" name="__module.encoder.layer.5.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="564">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="669" name="Constant_36914082" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="75646724" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="670" name="Convert_36914083" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="671" name="Constant_36914084" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="76236548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="672" name="__module.encoder.layer.5.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="673" name="__module.encoder.layer.5.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="674" name="Constant_9019" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="76239620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="675" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="543,x.61">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="676" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76242692" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="677" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76242696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="678" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76242692" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="679" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76242696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="680" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="681" name="__module.encoder.layer.5.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="682" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="547,x.63">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="683" name="Constant_1515" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="548">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="684" name="__module.encoder.layer.5.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="549,key_layer.11">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="685" name="__module.encoder.layer.5.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="566,attention_scores.21">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="686" name="Constant_9020" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="687" name="__module.encoder.layer.5.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="567,attention_scores.23">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="688" name="__module.encoder.layer.5.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="568,input.43">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="689" name="__module.encoder.layer.5.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="569,input.45">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="690" name="Constant_36914086" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="76242700" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="691" name="Convert_36914087" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="692" name="Constant_36914088" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="76832524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="693" name="__module.encoder.layer.5.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="694" name="__module.encoder.layer.5.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="695" name="Constant_9021" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="76835596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="696" name="__module.encoder.layer.5.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="552,x.65">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="697" name="__module.encoder.layer.5.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="698" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="556,x.67">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="699" name="Constant_1538" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="557">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="700" name="__module.encoder.layer.5.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="558">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="701" name="__module.encoder.layer.5.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="571,context_layer.21">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="702" name="Constant_1636" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="572">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="703" name="__module.encoder.layer.5.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="573">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="704" name="__module.encoder.layer.5.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="705" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="578">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="706" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="76838668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="707" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="708" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76841740" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="709" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76841744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="710" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76841740" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="711" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76841744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="712" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="713" name="Constant_36914090" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="76841748" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="714" name="Convert_36914091" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="715" name="Constant_36914092" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="77431572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="716" name="__module.encoder.layer.5.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="717" name="__module.encoder.layer.5.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="718" name="Constant_9022" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="77434644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="719" name="__module.encoder.layer.5.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="584,input.47">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="720" name="__module.encoder.layer.5.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="586">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="721" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="722" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="723" name="Constant_9023" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="77437716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="724" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="725" name="Constant_9024" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="77440788" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="726" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="590,input_tensor.23">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="727" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="77443860" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="728" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="729" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="77446932" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="730" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="77446936" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="731" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="77446932" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="732" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="77446936" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="733" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="734" name="Constant_36914094" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="77446940" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="735" name="Convert_36914095" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="736" name="Constant_36914096" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="79806236" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="737" name="__module.encoder.layer.5.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="738" name="__module.encoder.layer.5.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="739" name="Constant_9025" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="79818524" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="740" name="__module.encoder.layer.5.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="595">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="741" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="596">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="742" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="79830812" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="743" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="744" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="79843100" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="745" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="79843104" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="746" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="79843100" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="747" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="79843104" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="748" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="749" name="Constant_36914098" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="79843108" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="750" name="Convert_36914099" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="751" name="Constant_36914100" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="82202404" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="752" name="__module.encoder.layer.5.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="753" name="__module.encoder.layer.5.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="754" name="Constant_9026" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="82205476" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="755" name="__module.encoder.layer.5.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="602,input.49">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="756" name="__module.encoder.layer.5.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="604">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="757" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="758" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="759" name="Constant_9027" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="82208548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="760" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="761" name="Constant_9028" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="82211620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="762" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="608,input_tensor.25">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="763" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="82214692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="764" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="765" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82217764" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="766" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82217768" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="767" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82217764" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="768" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82217768" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="769" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="770" name="Constant_36914102" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="82217772" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="771" name="Convert_36914103" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="772" name="Constant_36914104" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="82807596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="773" name="__module.encoder.layer.6.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="774" name="__module.encoder.layer.6.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="775" name="Constant_9029" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="82810668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="776" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="619,x.81">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="777" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82813740" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="778" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82813744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="779" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82813740" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="780" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82813744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="781" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="782" name="__module.encoder.layer.6.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="783" name="__module.encoder.layer.6.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="641,x.83">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="784" name="Constant_1803" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="642">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="785" name="__module.encoder.layer.6.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="643">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="786" name="Constant_36914106" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="82813748" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="787" name="Convert_36914107" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="788" name="Constant_36914108" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="83403572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="789" name="__module.encoder.layer.6.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="790" name="__module.encoder.layer.6.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="791" name="Constant_9030" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="83406644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="792" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="622,x.73">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="793" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="83409716" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="794" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="83409720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="795" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="83409716" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="796" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="83409720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="797" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="798" name="__module.encoder.layer.6.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="799" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="626,x.75">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="800" name="Constant_1763" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="627">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="801" name="__module.encoder.layer.6.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="628,key_layer.13">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="802" name="__module.encoder.layer.6.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="645,attention_scores.25">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="803" name="Constant_9031" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="804" name="__module.encoder.layer.6.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="646,attention_scores.27">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="805" name="__module.encoder.layer.6.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="647,input.51">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="806" name="__module.encoder.layer.6.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="648,input.53">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="807" name="Constant_36914110" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="83409724" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="808" name="Convert_36914111" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="809" name="Constant_36914112" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="83999548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="810" name="__module.encoder.layer.6.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="811" name="__module.encoder.layer.6.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="812" name="Constant_9032" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="84002620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="813" name="__module.encoder.layer.6.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="631,x.77">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="814" name="__module.encoder.layer.6.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="815" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="635,x.79">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="816" name="Constant_1786" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="636">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="817" name="__module.encoder.layer.6.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="637">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="818" name="__module.encoder.layer.6.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="650,context_layer.25">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="819" name="Constant_1884" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="651">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="820" name="__module.encoder.layer.6.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="652">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="821" name="__module.encoder.layer.6.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="822" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="657">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="823" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="84005692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="824" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="825" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="84008764" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="826" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="84008768" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="827" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="84008764" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="828" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="84008768" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="829" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="830" name="Constant_36914114" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="84008772" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="831" name="Convert_36914115" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="832" name="Constant_36914116" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="84598596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="833" name="__module.encoder.layer.6.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="834" name="__module.encoder.layer.6.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="835" name="Constant_9033" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="84601668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="836" name="__module.encoder.layer.6.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="663,input.55">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="837" name="__module.encoder.layer.6.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="665">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="838" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="839" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="840" name="Constant_9034" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="84604740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="841" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="842" name="Constant_9035" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="84607812" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="843" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="669,input_tensor.27">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="844" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="84610884" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="845" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="846" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="84613956" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="847" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="84613960" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="848" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="84613956" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="849" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="84613960" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="850" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="851" name="Constant_36914118" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="84613964" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="852" name="Convert_36914119" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="853" name="Constant_36914120" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="86973260" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="854" name="__module.encoder.layer.6.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="855" name="__module.encoder.layer.6.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="856" name="Constant_9036" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="86985548" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="857" name="__module.encoder.layer.6.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="674">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="858" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="675">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="859" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="86997836" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="860" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="861" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="87010124" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="862" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="87010128" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="863" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="87010124" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="864" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="87010128" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="865" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="866" name="Constant_36914122" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="87010132" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="867" name="Convert_36914123" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="868" name="Constant_36914124" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="89369428" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="869" name="__module.encoder.layer.6.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="870" name="__module.encoder.layer.6.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="871" name="Constant_9037" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="89372500" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="872" name="__module.encoder.layer.6.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="681,input.57">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="873" name="__module.encoder.layer.6.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="683">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="874" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="875" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="876" name="Constant_9038" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="89375572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="877" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="878" name="Constant_9039" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="89378644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="879" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="687,input_tensor.29">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="880" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="89381716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="881" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="882" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89384788" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="883" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89384792" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="884" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89384788" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="885" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89384792" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="886" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="887" name="Constant_36914126" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="89384796" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="888" name="Convert_36914127" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="889" name="Constant_36914128" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="89974620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="890" name="__module.encoder.layer.7.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="891" name="__module.encoder.layer.7.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="892" name="Constant_9040" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="89977692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="893" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="698,x.93">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="894" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89980764" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="895" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89980768" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="896" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89980764" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="897" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89980768" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="898" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="899" name="__module.encoder.layer.7.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="900" name="__module.encoder.layer.7.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="720,x.95">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="901" name="Constant_2051" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="721">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="902" name="__module.encoder.layer.7.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="722">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="903" name="Constant_36914130" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="89980772" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="904" name="Convert_36914131" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="905" name="Constant_36914132" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="90570596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="906" name="__module.encoder.layer.7.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="907" name="__module.encoder.layer.7.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="908" name="Constant_9041" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="90573668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="909" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="701,x.85">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="910" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="90576740" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="911" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="90576744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="912" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="90576740" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="913" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="90576744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="914" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="915" name="__module.encoder.layer.7.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="916" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="705,x.87">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="917" name="Constant_2011" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="706">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="918" name="__module.encoder.layer.7.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="707,key_layer.15">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="919" name="__module.encoder.layer.7.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="724,attention_scores.29">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="920" name="Constant_9042" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="921" name="__module.encoder.layer.7.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="725,attention_scores.31">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="922" name="__module.encoder.layer.7.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="726,input.59">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="923" name="__module.encoder.layer.7.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="727,input.61">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="924" name="Constant_36914134" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="90576748" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="925" name="Convert_36914135" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="926" name="Constant_36914136" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="91166572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="927" name="__module.encoder.layer.7.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="928" name="__module.encoder.layer.7.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="929" name="Constant_9043" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="91169644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="930" name="__module.encoder.layer.7.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="710,x.89">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="931" name="__module.encoder.layer.7.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="932" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="714,x.91">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="933" name="Constant_2034" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="715">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="934" name="__module.encoder.layer.7.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="716">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="935" name="__module.encoder.layer.7.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="729,context_layer.29">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="936" name="Constant_2132" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="730">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="937" name="__module.encoder.layer.7.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="731">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="938" name="__module.encoder.layer.7.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="939" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="736">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="940" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="91172716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="941" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="942" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="91175788" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="943" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="91175792" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="944" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="91175788" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="945" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="91175792" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="946" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="947" name="Constant_36914138" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="91175796" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="948" name="Convert_36914139" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="949" name="Constant_36914140" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="91765620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="950" name="__module.encoder.layer.7.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="951" name="__module.encoder.layer.7.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="952" name="Constant_9044" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="91768692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="953" name="__module.encoder.layer.7.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="742,input.63">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="954" name="__module.encoder.layer.7.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="744">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="955" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="956" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="957" name="Constant_9045" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="91771764" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="958" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="959" name="Constant_9046" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="91774836" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="960" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="748,input_tensor.31">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="961" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="91777908" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="962" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="963" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="91780980" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="964" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="91780984" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="965" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="91780980" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="966" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="91780984" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="967" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="968" name="Constant_36914142" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="91780988" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="969" name="Convert_36914143" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="970" name="Constant_36914144" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="94140284" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="971" name="__module.encoder.layer.7.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="972" name="__module.encoder.layer.7.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="973" name="Constant_9047" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="94152572" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="974" name="__module.encoder.layer.7.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="753">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="975" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="754">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="976" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="94164860" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="977" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="978" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="94177148" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="979" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="94177152" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="980" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="94177148" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="981" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="94177152" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="982" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="983" name="Constant_36914146" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="94177156" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="984" name="Convert_36914147" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="985" name="Constant_36914148" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="96536452" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="986" name="__module.encoder.layer.7.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="987" name="__module.encoder.layer.7.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="988" name="Constant_9048" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="96539524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="989" name="__module.encoder.layer.7.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="760,input.65">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="990" name="__module.encoder.layer.7.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="762">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="991" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="992" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="993" name="Constant_9049" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="96542596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="994" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="995" name="Constant_9050" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="96545668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="996" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="766,input_tensor.33">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="997" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="96548740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="998" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="999" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96551812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1000" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96551816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1001" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96551812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1002" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96551816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1003" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1004" name="Constant_36914150" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="96551820" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1005" name="Convert_36914151" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1006" name="Constant_36914152" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="97141644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1007" name="__module.encoder.layer.8.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1008" name="__module.encoder.layer.8.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1009" name="Constant_9051" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="97144716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1010" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="777,x.105">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1011" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="97147788" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1012" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="97147792" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1013" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="97147788" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1014" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="97147792" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1015" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1016" name="__module.encoder.layer.8.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1017" name="__module.encoder.layer.8.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="799,x.107">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1018" name="Constant_2299" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="800">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1019" name="__module.encoder.layer.8.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="801">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1020" name="Constant_36914154" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="97147796" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1021" name="Convert_36914155" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1022" name="Constant_36914156" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="97737620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1023" name="__module.encoder.layer.8.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1024" name="__module.encoder.layer.8.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1025" name="Constant_9052" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="97740692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1026" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="780,x.97">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1027" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="97743764" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1028" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="97743768" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1029" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="97743764" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1030" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="97743768" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1031" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1032" name="__module.encoder.layer.8.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1033" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="784,x.99">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1034" name="Constant_2259" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="785">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1035" name="__module.encoder.layer.8.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="786,key_layer.17">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1036" name="__module.encoder.layer.8.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="803,attention_scores.33">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1037" name="Constant_9053" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1038" name="__module.encoder.layer.8.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="804,attention_scores.35">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1039" name="__module.encoder.layer.8.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="805,input.67">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1040" name="__module.encoder.layer.8.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="806,input.69">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1041" name="Constant_36914158" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="97743772" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1042" name="Convert_36914159" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1043" name="Constant_36914160" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="98333596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1044" name="__module.encoder.layer.8.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1045" name="__module.encoder.layer.8.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1046" name="Constant_9054" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98336668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1047" name="__module.encoder.layer.8.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="789,x.101">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1048" name="__module.encoder.layer.8.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1049" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="793,x.103">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1050" name="Constant_2282" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="794">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1051" name="__module.encoder.layer.8.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="795">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1052" name="__module.encoder.layer.8.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="808,context_layer.33">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1053" name="Constant_2380" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="809">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1054" name="__module.encoder.layer.8.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="810">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1055" name="__module.encoder.layer.8.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1056" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="815">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1057" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98339740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1058" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1059" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98342812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1060" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98342816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1061" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98342812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1062" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98342816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1063" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1064" name="Constant_36914162" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="98342820" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1065" name="Convert_36914163" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1066" name="Constant_36914164" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="98932644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1067" name="__module.encoder.layer.8.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1068" name="__module.encoder.layer.8.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1069" name="Constant_9055" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98935716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1070" name="__module.encoder.layer.8.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="821,input.71">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1071" name="__module.encoder.layer.8.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="823">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1072" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1073" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1074" name="Constant_9056" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98938788" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1075" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1076" name="Constant_9057" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98941860" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1077" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="827,input_tensor.35">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1078" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98944932" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1079" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1080" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98948004" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1081" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98948008" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1082" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98948004" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1083" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98948008" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1084" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1085" name="Constant_36914166" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="98948012" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1086" name="Convert_36914167" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1087" name="Constant_36914168" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="101307308" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1088" name="__module.encoder.layer.8.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1089" name="__module.encoder.layer.8.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1090" name="Constant_9058" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="101319596" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1091" name="__module.encoder.layer.8.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="832">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1092" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="833">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1093" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="101331884" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1094" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1095" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="101344172" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1096" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="101344176" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1097" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="101344172" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1098" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="101344176" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1099" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1100" name="Constant_36914170" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="101344180" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1101" name="Convert_36914171" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1102" name="Constant_36914172" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="103703476" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1103" name="__module.encoder.layer.8.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1104" name="__module.encoder.layer.8.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1105" name="Constant_9059" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="103706548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1106" name="__module.encoder.layer.8.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="839,input.73">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1107" name="__module.encoder.layer.8.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="841">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1108" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1109" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1110" name="Constant_9060" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="103709620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1111" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1112" name="Constant_9061" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="103712692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1113" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="845,input_tensor.37">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1114" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="103715764" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1115" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1116" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="103718836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1117" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="103718840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1118" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="103718836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1119" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="103718840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1120" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1121" name="Constant_36914174" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="103718844" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1122" name="Convert_36914175" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1123" name="Constant_36914176" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="104308668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1124" name="__module.encoder.layer.9.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1125" name="__module.encoder.layer.9.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1126" name="Constant_9062" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="104311740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1127" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="856,x.117">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1128" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104314812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1129" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104314816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1130" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104314812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1131" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104314816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1132" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1133" name="__module.encoder.layer.9.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1134" name="__module.encoder.layer.9.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="878,x.119">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1135" name="Constant_2547" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="879">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1136" name="__module.encoder.layer.9.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="880">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1137" name="Constant_36914178" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="104314820" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1138" name="Convert_36914179" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1139" name="Constant_36914180" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="104904644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1140" name="__module.encoder.layer.9.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1141" name="__module.encoder.layer.9.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1142" name="Constant_9063" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="104907716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1143" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="859,x.109">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1144" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104910788" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1145" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104910792" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1146" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104910788" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1147" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104910792" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1148" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1149" name="__module.encoder.layer.9.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1150" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="863,x.111">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1151" name="Constant_2507" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="864">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1152" name="__module.encoder.layer.9.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="865,key_layer.19">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1153" name="__module.encoder.layer.9.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="882,attention_scores.37">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1154" name="Constant_9064" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1155" name="__module.encoder.layer.9.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="883,attention_scores.39">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1156" name="__module.encoder.layer.9.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="884,input.75">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1157" name="__module.encoder.layer.9.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="885,input.77">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1158" name="Constant_36914182" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="104910796" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1159" name="Convert_36914183" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1160" name="Constant_36914184" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="105500620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1161" name="__module.encoder.layer.9.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1162" name="__module.encoder.layer.9.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1163" name="Constant_9065" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="105503692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1164" name="__module.encoder.layer.9.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="868,x.113">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1165" name="__module.encoder.layer.9.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1166" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="872,x.115">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1167" name="Constant_2530" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="873">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1168" name="__module.encoder.layer.9.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="874">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1169" name="__module.encoder.layer.9.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="887,context_layer.37">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1170" name="Constant_2628" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="888">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1171" name="__module.encoder.layer.9.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="889">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1172" name="__module.encoder.layer.9.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1173" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="894">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1174" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="105506764" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1175" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1176" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="105509836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1177" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="105509840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1178" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="105509836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1179" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="105509840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1180" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1181" name="Constant_36914186" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="105509844" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1182" name="Convert_36914187" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1183" name="Constant_36914188" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="106099668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1184" name="__module.encoder.layer.9.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1185" name="__module.encoder.layer.9.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1186" name="Constant_9066" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="106102740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1187" name="__module.encoder.layer.9.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="900,input.79">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1188" name="__module.encoder.layer.9.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="902">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1189" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1190" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1191" name="Constant_9067" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="106105812" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1192" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1193" name="Constant_9068" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="106108884" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1194" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="906,input_tensor.39">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1195" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="106111956" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1196" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1197" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="106115028" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1198" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="106115032" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1199" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="106115028" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1200" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="106115032" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1201" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1202" name="Constant_36914190" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="106115036" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1203" name="Convert_36914191" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1204" name="Constant_36914192" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="108474332" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1205" name="__module.encoder.layer.9.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1206" name="__module.encoder.layer.9.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1207" name="Constant_9069" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="108486620" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1208" name="__module.encoder.layer.9.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="911">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1209" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="912">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1210" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="108498908" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1211" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1212" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="108511196" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1213" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="108511200" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1214" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="108511196" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1215" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="108511200" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1216" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1217" name="Constant_36914194" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="108511204" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1218" name="Convert_36914195" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1219" name="Constant_36914196" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="110870500" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1220" name="__module.encoder.layer.9.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1221" name="__module.encoder.layer.9.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1222" name="Constant_9070" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="110873572" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1223" name="__module.encoder.layer.9.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="918,input.81">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1224" name="__module.encoder.layer.9.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="920">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1225" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1226" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1227" name="Constant_9071" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="110876644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1228" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1229" name="Constant_9072" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="110879716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1230" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="924,input_tensor.41">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1231" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="110882788" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1232" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1233" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="110885860" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1234" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="110885864" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1235" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="110885860" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1236" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="110885864" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1237" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1238" name="Constant_36914198" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="110885868" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1239" name="Convert_36914199" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1240" name="Constant_36914200" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="111475692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1241" name="__module.encoder.layer.10.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1242" name="__module.encoder.layer.10.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1243" name="Constant_9073" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="111478764" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1244" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="935,x.129">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1245" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="111481836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1246" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="111481840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1247" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="111481836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1248" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="111481840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1249" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1250" name="__module.encoder.layer.10.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1251" name="__module.encoder.layer.10.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="957,x.131">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1252" name="Constant_2795" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="958">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1253" name="__module.encoder.layer.10.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="959">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1254" name="Constant_36914202" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="111481844" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1255" name="Convert_36914203" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1256" name="Constant_36914204" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="112071668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1257" name="__module.encoder.layer.10.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1258" name="__module.encoder.layer.10.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1259" name="Constant_9074" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="112074740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1260" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="938,x.121">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1261" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="112077812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1262" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="112077816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1263" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="112077812" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1264" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="112077816" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1265" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1266" name="__module.encoder.layer.10.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1267" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="942,x.123">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1268" name="Constant_2755" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="943">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1269" name="__module.encoder.layer.10.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="944,key_layer.21">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1270" name="__module.encoder.layer.10.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="961,attention_scores.41">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1271" name="Constant_9075" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1272" name="__module.encoder.layer.10.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="962,attention_scores.43">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1273" name="__module.encoder.layer.10.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="963,input.83">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1274" name="__module.encoder.layer.10.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="964,input.85">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1275" name="Constant_36914206" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="112077820" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1276" name="Convert_36914207" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1277" name="Constant_36914208" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="112667644" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1278" name="__module.encoder.layer.10.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1279" name="__module.encoder.layer.10.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1280" name="Constant_9076" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="112670716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1281" name="__module.encoder.layer.10.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="947,x.125">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1282" name="__module.encoder.layer.10.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1283" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="951,x.127">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1284" name="Constant_2778" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="952">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1285" name="__module.encoder.layer.10.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="953">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1286" name="__module.encoder.layer.10.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="966,context_layer.41">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1287" name="Constant_2876" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="967">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1288" name="__module.encoder.layer.10.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="968">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1289" name="__module.encoder.layer.10.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1290" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="973">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1291" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="112673788" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1292" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1293" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="112676860" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1294" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="112676864" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1295" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="112676860" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1296" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="112676864" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1297" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1298" name="Constant_36914210" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="112676868" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1299" name="Convert_36914211" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1300" name="Constant_36914212" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="113266692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1301" name="__module.encoder.layer.10.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1302" name="__module.encoder.layer.10.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1303" name="Constant_9077" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="113269764" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1304" name="__module.encoder.layer.10.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="979,input.87">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1305" name="__module.encoder.layer.10.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="981">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1306" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1307" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1308" name="Constant_9078" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="113272836" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1309" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1310" name="Constant_9079" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="113275908" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1311" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="985,input_tensor.43">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1312" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="113278980" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1313" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1314" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="113282052" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1315" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="113282056" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1316" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="113282052" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1317" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="113282056" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1318" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1319" name="Constant_36914214" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="113282060" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1320" name="Convert_36914215" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1321" name="Constant_36914216" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="115641356" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1322" name="__module.encoder.layer.10.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1323" name="__module.encoder.layer.10.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1324" name="Constant_9080" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="115653644" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1325" name="__module.encoder.layer.10.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="990">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1326" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="991">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1327" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="115665932" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1328" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1329" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="115678220" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1330" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="115678224" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1331" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="115678220" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1332" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="115678224" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1333" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1334" name="Constant_36914218" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="115678228" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1335" name="Convert_36914219" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1336" name="Constant_36914220" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="118037524" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1337" name="__module.encoder.layer.10.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1338" name="__module.encoder.layer.10.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1339" name="Constant_9081" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="118040596" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1340" name="__module.encoder.layer.10.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="997,input.89">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1341" name="__module.encoder.layer.10.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="999">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1342" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1343" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1344" name="Constant_9082" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="118043668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1345" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1346" name="Constant_9083" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="118046740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1347" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1003,input_tensor.45">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1348" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="118049812" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1349" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1350" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="118052884" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1351" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="118052888" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1352" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="118052884" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1353" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="118052888" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1354" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1355" name="Constant_36914222" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="118052892" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1356" name="Convert_36914223" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1357" name="Constant_36914224" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="118642716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1358" name="__module.encoder.layer.11.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1359" name="__module.encoder.layer.11.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1360" name="Constant_9084" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="118645788" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1361" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1014,x.141">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1362" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="118648860" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1363" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="118648864" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1364" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="118648860" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1365" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="118648864" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1366" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1367" name="__module.encoder.layer.11.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1368" name="__module.encoder.layer.11.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1036,x">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1369" name="Constant_3043" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="1037">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1370" name="__module.encoder.layer.11.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1038">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1371" name="Constant_36914226" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="118648868" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1372" name="Convert_36914227" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1373" name="Constant_36914228" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="119238692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1374" name="__module.encoder.layer.11.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1375" name="__module.encoder.layer.11.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1376" name="Constant_9085" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="119241764" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1377" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1017,x.133">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1378" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="119244836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1379" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="119244840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1380" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="119244836" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1381" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="119244840" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1382" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1383" name="__module.encoder.layer.11.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1384" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1021,x.135">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1385" name="Constant_3003" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="1022">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1386" name="__module.encoder.layer.11.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1023,key_layer">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1387" name="__module.encoder.layer.11.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1040,attention_scores.45">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1388" name="Constant_9086" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="40407536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1389" name="__module.encoder.layer.11.attention.self/aten::div/Divide" type="Divide" version="opset1">
|
|
<data auto_broadcast="numpy" m_pythondiv="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1041,attention_scores">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1390" name="__module.encoder.layer.11.attention.self/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1042,input.91">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1391" name="__module.encoder.layer.11.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
|
|
<data axis="-1" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="1043,input.93">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1392" name="Constant_36914230" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="119244844" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1393" name="Convert_36914231" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1394" name="Constant_36914232" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="119834668" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1395" name="__module.encoder.layer.11.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1396" name="__module.encoder.layer.11.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1397" name="Constant_9087" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="119837740" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1398" name="__module.encoder.layer.11.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1026,x.137">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1399" name="__module.encoder.layer.11.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811496" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1400" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1030,x.139">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1401" name="Constant_3026" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="1031">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1402" name="__module.encoder.layer.11.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1032">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1403" name="__module.encoder.layer.11.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="false" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1045,context_layer.45">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1404" name="Constant_3124" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="39811528" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="1046">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1405" name="__module.encoder.layer.11.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1047">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1406" name="__module.encoder.layer.11.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="41003524" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1407" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1052">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1408" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="119840812" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1409" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1410" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="119843884" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1411" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="119843888" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1412" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="119843884" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1413" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="119843888" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1414" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_3_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1415" name="Constant_36914234" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="119843892" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1416" name="Convert_36914235" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1417" name="Constant_36914236" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="120433716" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1418" name="__module.encoder.layer.11.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1419" name="__module.encoder.layer.11.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1420" name="Constant_9088" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="120436788" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1421" name="__module.encoder.layer.11.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1058,input.95">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1422" name="__module.encoder.layer.11.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1060">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1423" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1424" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1425" name="Constant_9089" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="120439860" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1426" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1427" name="Constant_9090" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="120442932" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1428" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1064,input_tensor">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1429" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="120446004" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1430" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1431" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="120449076" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1432" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="120449080" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1433" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="120449076" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1434" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="120449080" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1435" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1436" name="Constant_36914238" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="120449084" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1437" name="Convert_36914239" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1438" name="Constant_36914240" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="122808380" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1439" name="__module.encoder.layer.11.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1440" name="__module.encoder.layer.11.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1441" name="Constant_9091" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="122820668" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1442" name="__module.encoder.layer.11.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1069">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1443" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="1070">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1444" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="122832956" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1445" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1446" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="122845244" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1447" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="122845248" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1448" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="122845244" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1449" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="122845248" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1450" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1451" name="Constant_36914242" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="122845252" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1452" name="Convert_36914243" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1453" name="Constant_36914244" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="125204548" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1454" name="__module.encoder.layer.11.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1455" name="__module.encoder.layer.11.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1456" name="Constant_9092" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="125207620" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1457" name="__module.encoder.layer.11.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1076,input">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1458" name="__module.encoder.layer.11.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="1078">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1459" name="__module.encoder.layer.11.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="39206292" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1460" name="__module.encoder.layer.11.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1461" name="Constant_9093" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="125210692" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1462" name="__module.encoder.layer.11.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1463" name="Constant_9094" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="125213764" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1464" name="__module.encoder.layer.11.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="last_hidden_state">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1465" name="Result_4782" type="Result" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
</layer>
|
|
</layers>
|
|
<edges>
|
|
<edge from-layer="0" from-port="0" to-layer="94" to-port="0" />
|
|
<edge from-layer="1" from-port="0" to-layer="15" to-port="0" />
|
|
<edge from-layer="1" from-port="0" to-layer="6" to-port="0" />
|
|
<edge from-layer="1" from-port="0" to-layer="32" to-port="0" />
|
|
<edge from-layer="2" from-port="0" to-layer="3" to-port="0" />
|
|
<edge from-layer="3" from-port="1" to-layer="5" to-port="0" />
|
|
<edge from-layer="4" from-port="0" to-layer="5" to-port="1" />
|
|
<edge from-layer="5" from-port="2" to-layer="8" to-port="0" />
|
|
<edge from-layer="6" from-port="1" to-layer="8" to-port="1" />
|
|
<edge from-layer="7" from-port="0" to-layer="8" to-port="2" />
|
|
<edge from-layer="8" from-port="3" to-layer="26" to-port="0" />
|
|
<edge from-layer="9" from-port="0" to-layer="10" to-port="0" />
|
|
<edge from-layer="10" from-port="1" to-layer="12" to-port="0" />
|
|
<edge from-layer="11" from-port="0" to-layer="12" to-port="1" />
|
|
<edge from-layer="12" from-port="2" to-layer="25" to-port="0" />
|
|
<edge from-layer="13" from-port="0" to-layer="21" to-port="0" />
|
|
<edge from-layer="14" from-port="0" to-layer="21" to-port="1" />
|
|
<edge from-layer="15" from-port="1" to-layer="18" to-port="0" />
|
|
<edge from-layer="15" from-port="1" to-layer="22" to-port="1" />
|
|
<edge from-layer="16" from-port="0" to-layer="18" to-port="1" />
|
|
<edge from-layer="17" from-port="0" to-layer="18" to-port="2" />
|
|
<edge from-layer="18" from-port="3" to-layer="21" to-port="2" />
|
|
<edge from-layer="19" from-port="0" to-layer="21" to-port="3" />
|
|
<edge from-layer="20" from-port="0" to-layer="21" to-port="4" />
|
|
<edge from-layer="21" from-port="5" to-layer="22" to-port="0" />
|
|
<edge from-layer="22" from-port="2" to-layer="23" to-port="0" />
|
|
<edge from-layer="23" from-port="1" to-layer="25" to-port="1" />
|
|
<edge from-layer="24" from-port="0" to-layer="25" to-port="2" />
|
|
<edge from-layer="25" from-port="3" to-layer="26" to-port="1" />
|
|
<edge from-layer="26" from-port="2" to-layer="43" to-port="0" />
|
|
<edge from-layer="27" from-port="0" to-layer="28" to-port="0" />
|
|
<edge from-layer="28" from-port="1" to-layer="30" to-port="0" />
|
|
<edge from-layer="29" from-port="0" to-layer="30" to-port="1" />
|
|
<edge from-layer="30" from-port="2" to-layer="42" to-port="0" />
|
|
<edge from-layer="31" from-port="0" to-layer="32" to-port="1" />
|
|
<edge from-layer="32" from-port="2" to-layer="33" to-port="0" />
|
|
<edge from-layer="33" from-port="1" to-layer="35" to-port="0" />
|
|
<edge from-layer="33" from-port="1" to-layer="36" to-port="1" />
|
|
<edge from-layer="34" from-port="0" to-layer="35" to-port="1" />
|
|
<edge from-layer="35" from-port="2" to-layer="36" to-port="0" />
|
|
<edge from-layer="36" from-port="2" to-layer="37" to-port="0" />
|
|
<edge from-layer="37" from-port="1" to-layer="39" to-port="0" />
|
|
<edge from-layer="38" from-port="0" to-layer="39" to-port="1" />
|
|
<edge from-layer="39" from-port="2" to-layer="40" to-port="0" />
|
|
<edge from-layer="40" from-port="1" to-layer="42" to-port="1" />
|
|
<edge from-layer="41" from-port="0" to-layer="42" to-port="2" />
|
|
<edge from-layer="42" from-port="3" to-layer="43" to-port="1" />
|
|
<edge from-layer="43" from-port="2" to-layer="45" to-port="0" />
|
|
<edge from-layer="44" from-port="0" to-layer="45" to-port="1" />
|
|
<edge from-layer="45" from-port="2" to-layer="47" to-port="0" />
|
|
<edge from-layer="46" from-port="0" to-layer="47" to-port="1" />
|
|
<edge from-layer="47" from-port="2" to-layer="49" to-port="0" />
|
|
<edge from-layer="48" from-port="0" to-layer="49" to-port="1" />
|
|
<edge from-layer="49" from-port="2" to-layer="51" to-port="0" />
|
|
<edge from-layer="49" from-port="2" to-layer="135" to-port="1" />
|
|
<edge from-layer="50" from-port="0" to-layer="51" to-port="1" />
|
|
<edge from-layer="51" from-port="2" to-layer="56" to-port="0" />
|
|
<edge from-layer="52" from-port="0" to-layer="56" to-port="1" />
|
|
<edge from-layer="53" from-port="0" to-layer="56" to-port="2" />
|
|
<edge from-layer="54" from-port="0" to-layer="56" to-port="3" />
|
|
<edge from-layer="55" from-port="0" to-layer="56" to-port="4" />
|
|
<edge from-layer="56" from-port="5" to-layer="61" to-port="0" />
|
|
<edge from-layer="56" from-port="5" to-layer="77" to-port="0" />
|
|
<edge from-layer="56" from-port="5" to-layer="109" to-port="0" />
|
|
<edge from-layer="57" from-port="0" to-layer="58" to-port="0" />
|
|
<edge from-layer="58" from-port="1" to-layer="60" to-port="0" />
|
|
<edge from-layer="59" from-port="0" to-layer="60" to-port="1" />
|
|
<edge from-layer="60" from-port="2" to-layer="61" to-port="1" />
|
|
<edge from-layer="61" from-port="2" to-layer="63" to-port="0" />
|
|
<edge from-layer="62" from-port="0" to-layer="63" to-port="1" />
|
|
<edge from-layer="63" from-port="2" to-layer="68" to-port="0" />
|
|
<edge from-layer="64" from-port="0" to-layer="68" to-port="1" />
|
|
<edge from-layer="65" from-port="0" to-layer="68" to-port="2" />
|
|
<edge from-layer="66" from-port="0" to-layer="68" to-port="3" />
|
|
<edge from-layer="67" from-port="0" to-layer="68" to-port="4" />
|
|
<edge from-layer="68" from-port="5" to-layer="70" to-port="0" />
|
|
<edge from-layer="69" from-port="0" to-layer="70" to-port="1" />
|
|
<edge from-layer="70" from-port="2" to-layer="72" to-port="0" />
|
|
<edge from-layer="71" from-port="0" to-layer="72" to-port="1" />
|
|
<edge from-layer="72" from-port="2" to-layer="89" to-port="0" />
|
|
<edge from-layer="73" from-port="0" to-layer="74" to-port="0" />
|
|
<edge from-layer="74" from-port="1" to-layer="76" to-port="0" />
|
|
<edge from-layer="75" from-port="0" to-layer="76" to-port="1" />
|
|
<edge from-layer="76" from-port="2" to-layer="77" to-port="1" />
|
|
<edge from-layer="77" from-port="2" to-layer="79" to-port="0" />
|
|
<edge from-layer="78" from-port="0" to-layer="79" to-port="1" />
|
|
<edge from-layer="79" from-port="2" to-layer="84" to-port="0" />
|
|
<edge from-layer="80" from-port="0" to-layer="84" to-port="1" />
|
|
<edge from-layer="81" from-port="0" to-layer="84" to-port="2" />
|
|
<edge from-layer="82" from-port="0" to-layer="84" to-port="3" />
|
|
<edge from-layer="83" from-port="0" to-layer="84" to-port="4" />
|
|
<edge from-layer="84" from-port="5" to-layer="86" to-port="0" />
|
|
<edge from-layer="85" from-port="0" to-layer="86" to-port="1" />
|
|
<edge from-layer="86" from-port="2" to-layer="88" to-port="0" />
|
|
<edge from-layer="87" from-port="0" to-layer="88" to-port="1" />
|
|
<edge from-layer="88" from-port="2" to-layer="89" to-port="1" />
|
|
<edge from-layer="89" from-port="2" to-layer="91" to-port="0" />
|
|
<edge from-layer="90" from-port="0" to-layer="91" to-port="1" />
|
|
<edge from-layer="91" from-port="2" to-layer="103" to-port="0" />
|
|
<edge from-layer="92" from-port="0" to-layer="100" to-port="0" />
|
|
<edge from-layer="93" from-port="0" to-layer="94" to-port="1" />
|
|
<edge from-layer="94" from-port="2" to-layer="96" to-port="0" />
|
|
<edge from-layer="95" from-port="0" to-layer="96" to-port="1" />
|
|
<edge from-layer="96" from-port="2" to-layer="97" to-port="0" />
|
|
<edge from-layer="97" from-port="1" to-layer="99" to-port="0" />
|
|
<edge from-layer="98" from-port="0" to-layer="99" to-port="1" />
|
|
<edge from-layer="99" from-port="2" to-layer="100" to-port="1" />
|
|
<edge from-layer="100" from-port="2" to-layer="102" to-port="0" />
|
|
<edge from-layer="101" from-port="0" to-layer="102" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="103" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="922" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="805" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="688" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="1273" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="571" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="1390" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="220" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="454" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="1156" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="1039" to-port="1" />
|
|
<edge from-layer="102" from-port="2" to-layer="337" to-port="1" />
|
|
<edge from-layer="103" from-port="2" to-layer="104" to-port="0" />
|
|
<edge from-layer="104" from-port="1" to-layer="116" to-port="0" />
|
|
<edge from-layer="105" from-port="0" to-layer="106" to-port="0" />
|
|
<edge from-layer="106" from-port="1" to-layer="108" to-port="0" />
|
|
<edge from-layer="107" from-port="0" to-layer="108" to-port="1" />
|
|
<edge from-layer="108" from-port="2" to-layer="109" to-port="1" />
|
|
<edge from-layer="109" from-port="2" to-layer="111" to-port="0" />
|
|
<edge from-layer="110" from-port="0" to-layer="111" to-port="1" />
|
|
<edge from-layer="111" from-port="2" to-layer="113" to-port="0" />
|
|
<edge from-layer="112" from-port="0" to-layer="113" to-port="1" />
|
|
<edge from-layer="113" from-port="2" to-layer="115" to-port="0" />
|
|
<edge from-layer="114" from-port="0" to-layer="115" to-port="1" />
|
|
<edge from-layer="115" from-port="2" to-layer="116" to-port="1" />
|
|
<edge from-layer="116" from-port="2" to-layer="118" to-port="0" />
|
|
<edge from-layer="117" from-port="0" to-layer="118" to-port="1" />
|
|
<edge from-layer="118" from-port="2" to-layer="120" to-port="0" />
|
|
<edge from-layer="119" from-port="0" to-layer="120" to-port="1" />
|
|
<edge from-layer="120" from-port="2" to-layer="122" to-port="0" />
|
|
<edge from-layer="121" from-port="0" to-layer="122" to-port="1" />
|
|
<edge from-layer="122" from-port="2" to-layer="127" to-port="0" />
|
|
<edge from-layer="123" from-port="0" to-layer="127" to-port="1" />
|
|
<edge from-layer="124" from-port="0" to-layer="127" to-port="2" />
|
|
<edge from-layer="125" from-port="0" to-layer="127" to-port="3" />
|
|
<edge from-layer="126" from-port="0" to-layer="127" to-port="4" />
|
|
<edge from-layer="127" from-port="5" to-layer="132" to-port="0" />
|
|
<edge from-layer="128" from-port="0" to-layer="129" to-port="0" />
|
|
<edge from-layer="129" from-port="1" to-layer="131" to-port="0" />
|
|
<edge from-layer="130" from-port="0" to-layer="131" to-port="1" />
|
|
<edge from-layer="131" from-port="2" to-layer="132" to-port="1" />
|
|
<edge from-layer="132" from-port="2" to-layer="134" to-port="0" />
|
|
<edge from-layer="133" from-port="0" to-layer="134" to-port="1" />
|
|
<edge from-layer="134" from-port="2" to-layer="135" to-port="0" />
|
|
<edge from-layer="135" from-port="2" to-layer="137" to-port="0" />
|
|
<edge from-layer="136" from-port="0" to-layer="137" to-port="1" />
|
|
<edge from-layer="137" from-port="2" to-layer="139" to-port="0" />
|
|
<edge from-layer="138" from-port="0" to-layer="139" to-port="1" />
|
|
<edge from-layer="139" from-port="2" to-layer="141" to-port="0" />
|
|
<edge from-layer="140" from-port="0" to-layer="141" to-port="1" />
|
|
<edge from-layer="141" from-port="2" to-layer="143" to-port="0" />
|
|
<edge from-layer="141" from-port="2" to-layer="171" to-port="1" />
|
|
<edge from-layer="142" from-port="0" to-layer="143" to-port="1" />
|
|
<edge from-layer="143" from-port="2" to-layer="148" to-port="0" />
|
|
<edge from-layer="144" from-port="0" to-layer="148" to-port="1" />
|
|
<edge from-layer="145" from-port="0" to-layer="148" to-port="2" />
|
|
<edge from-layer="146" from-port="0" to-layer="148" to-port="3" />
|
|
<edge from-layer="147" from-port="0" to-layer="148" to-port="4" />
|
|
<edge from-layer="148" from-port="5" to-layer="153" to-port="0" />
|
|
<edge from-layer="149" from-port="0" to-layer="150" to-port="0" />
|
|
<edge from-layer="150" from-port="1" to-layer="152" to-port="0" />
|
|
<edge from-layer="151" from-port="0" to-layer="152" to-port="1" />
|
|
<edge from-layer="152" from-port="2" to-layer="153" to-port="1" />
|
|
<edge from-layer="153" from-port="2" to-layer="155" to-port="0" />
|
|
<edge from-layer="154" from-port="0" to-layer="155" to-port="1" />
|
|
<edge from-layer="155" from-port="2" to-layer="156" to-port="0" />
|
|
<edge from-layer="156" from-port="1" to-layer="158" to-port="0" />
|
|
<edge from-layer="157" from-port="0" to-layer="158" to-port="1" />
|
|
<edge from-layer="158" from-port="2" to-layer="163" to-port="0" />
|
|
<edge from-layer="159" from-port="0" to-layer="163" to-port="1" />
|
|
<edge from-layer="160" from-port="0" to-layer="163" to-port="2" />
|
|
<edge from-layer="161" from-port="0" to-layer="163" to-port="3" />
|
|
<edge from-layer="162" from-port="0" to-layer="163" to-port="4" />
|
|
<edge from-layer="163" from-port="5" to-layer="168" to-port="0" />
|
|
<edge from-layer="164" from-port="0" to-layer="165" to-port="0" />
|
|
<edge from-layer="165" from-port="1" to-layer="167" to-port="0" />
|
|
<edge from-layer="166" from-port="0" to-layer="167" to-port="1" />
|
|
<edge from-layer="167" from-port="2" to-layer="168" to-port="1" />
|
|
<edge from-layer="168" from-port="2" to-layer="170" to-port="0" />
|
|
<edge from-layer="169" from-port="0" to-layer="170" to-port="1" />
|
|
<edge from-layer="170" from-port="2" to-layer="171" to-port="0" />
|
|
<edge from-layer="171" from-port="2" to-layer="173" to-port="0" />
|
|
<edge from-layer="172" from-port="0" to-layer="173" to-port="1" />
|
|
<edge from-layer="173" from-port="2" to-layer="175" to-port="0" />
|
|
<edge from-layer="174" from-port="0" to-layer="175" to-port="1" />
|
|
<edge from-layer="175" from-port="2" to-layer="177" to-port="0" />
|
|
<edge from-layer="176" from-port="0" to-layer="177" to-port="1" />
|
|
<edge from-layer="177" from-port="2" to-layer="179" to-port="0" />
|
|
<edge from-layer="177" from-port="2" to-layer="252" to-port="1" />
|
|
<edge from-layer="178" from-port="0" to-layer="179" to-port="1" />
|
|
<edge from-layer="179" from-port="2" to-layer="184" to-port="0" />
|
|
<edge from-layer="180" from-port="0" to-layer="184" to-port="1" />
|
|
<edge from-layer="181" from-port="0" to-layer="184" to-port="2" />
|
|
<edge from-layer="182" from-port="0" to-layer="184" to-port="3" />
|
|
<edge from-layer="183" from-port="0" to-layer="184" to-port="4" />
|
|
<edge from-layer="184" from-port="5" to-layer="226" to-port="0" />
|
|
<edge from-layer="184" from-port="5" to-layer="205" to-port="0" />
|
|
<edge from-layer="184" from-port="5" to-layer="189" to-port="0" />
|
|
<edge from-layer="185" from-port="0" to-layer="186" to-port="0" />
|
|
<edge from-layer="186" from-port="1" to-layer="188" to-port="0" />
|
|
<edge from-layer="187" from-port="0" to-layer="188" to-port="1" />
|
|
<edge from-layer="188" from-port="2" to-layer="189" to-port="1" />
|
|
<edge from-layer="189" from-port="2" to-layer="191" to-port="0" />
|
|
<edge from-layer="190" from-port="0" to-layer="191" to-port="1" />
|
|
<edge from-layer="191" from-port="2" to-layer="196" to-port="0" />
|
|
<edge from-layer="192" from-port="0" to-layer="196" to-port="1" />
|
|
<edge from-layer="193" from-port="0" to-layer="196" to-port="2" />
|
|
<edge from-layer="194" from-port="0" to-layer="196" to-port="3" />
|
|
<edge from-layer="195" from-port="0" to-layer="196" to-port="4" />
|
|
<edge from-layer="196" from-port="5" to-layer="198" to-port="0" />
|
|
<edge from-layer="197" from-port="0" to-layer="198" to-port="1" />
|
|
<edge from-layer="198" from-port="2" to-layer="200" to-port="0" />
|
|
<edge from-layer="199" from-port="0" to-layer="200" to-port="1" />
|
|
<edge from-layer="200" from-port="2" to-layer="217" to-port="0" />
|
|
<edge from-layer="201" from-port="0" to-layer="202" to-port="0" />
|
|
<edge from-layer="202" from-port="1" to-layer="204" to-port="0" />
|
|
<edge from-layer="203" from-port="0" to-layer="204" to-port="1" />
|
|
<edge from-layer="204" from-port="2" to-layer="205" to-port="1" />
|
|
<edge from-layer="205" from-port="2" to-layer="207" to-port="0" />
|
|
<edge from-layer="206" from-port="0" to-layer="207" to-port="1" />
|
|
<edge from-layer="207" from-port="2" to-layer="212" to-port="0" />
|
|
<edge from-layer="208" from-port="0" to-layer="212" to-port="1" />
|
|
<edge from-layer="209" from-port="0" to-layer="212" to-port="2" />
|
|
<edge from-layer="210" from-port="0" to-layer="212" to-port="3" />
|
|
<edge from-layer="211" from-port="0" to-layer="212" to-port="4" />
|
|
<edge from-layer="212" from-port="5" to-layer="214" to-port="0" />
|
|
<edge from-layer="213" from-port="0" to-layer="214" to-port="1" />
|
|
<edge from-layer="214" from-port="2" to-layer="216" to-port="0" />
|
|
<edge from-layer="215" from-port="0" to-layer="216" to-port="1" />
|
|
<edge from-layer="216" from-port="2" to-layer="217" to-port="1" />
|
|
<edge from-layer="217" from-port="2" to-layer="219" to-port="0" />
|
|
<edge from-layer="218" from-port="0" to-layer="219" to-port="1" />
|
|
<edge from-layer="219" from-port="2" to-layer="220" to-port="0" />
|
|
<edge from-layer="220" from-port="2" to-layer="221" to-port="0" />
|
|
<edge from-layer="221" from-port="1" to-layer="233" to-port="0" />
|
|
<edge from-layer="222" from-port="0" to-layer="223" to-port="0" />
|
|
<edge from-layer="223" from-port="1" to-layer="225" to-port="0" />
|
|
<edge from-layer="224" from-port="0" to-layer="225" to-port="1" />
|
|
<edge from-layer="225" from-port="2" to-layer="226" to-port="1" />
|
|
<edge from-layer="226" from-port="2" to-layer="228" to-port="0" />
|
|
<edge from-layer="227" from-port="0" to-layer="228" to-port="1" />
|
|
<edge from-layer="228" from-port="2" to-layer="230" to-port="0" />
|
|
<edge from-layer="229" from-port="0" to-layer="230" to-port="1" />
|
|
<edge from-layer="230" from-port="2" to-layer="232" to-port="0" />
|
|
<edge from-layer="231" from-port="0" to-layer="232" to-port="1" />
|
|
<edge from-layer="232" from-port="2" to-layer="233" to-port="1" />
|
|
<edge from-layer="233" from-port="2" to-layer="235" to-port="0" />
|
|
<edge from-layer="234" from-port="0" to-layer="235" to-port="1" />
|
|
<edge from-layer="235" from-port="2" to-layer="237" to-port="0" />
|
|
<edge from-layer="236" from-port="0" to-layer="237" to-port="1" />
|
|
<edge from-layer="237" from-port="2" to-layer="239" to-port="0" />
|
|
<edge from-layer="238" from-port="0" to-layer="239" to-port="1" />
|
|
<edge from-layer="239" from-port="2" to-layer="244" to-port="0" />
|
|
<edge from-layer="240" from-port="0" to-layer="244" to-port="1" />
|
|
<edge from-layer="241" from-port="0" to-layer="244" to-port="2" />
|
|
<edge from-layer="242" from-port="0" to-layer="244" to-port="3" />
|
|
<edge from-layer="243" from-port="0" to-layer="244" to-port="4" />
|
|
<edge from-layer="244" from-port="5" to-layer="249" to-port="0" />
|
|
<edge from-layer="245" from-port="0" to-layer="246" to-port="0" />
|
|
<edge from-layer="246" from-port="1" to-layer="248" to-port="0" />
|
|
<edge from-layer="247" from-port="0" to-layer="248" to-port="1" />
|
|
<edge from-layer="248" from-port="2" to-layer="249" to-port="1" />
|
|
<edge from-layer="249" from-port="2" to-layer="251" to-port="0" />
|
|
<edge from-layer="250" from-port="0" to-layer="251" to-port="1" />
|
|
<edge from-layer="251" from-port="2" to-layer="252" to-port="0" />
|
|
<edge from-layer="252" from-port="2" to-layer="254" to-port="0" />
|
|
<edge from-layer="253" from-port="0" to-layer="254" to-port="1" />
|
|
<edge from-layer="254" from-port="2" to-layer="256" to-port="0" />
|
|
<edge from-layer="255" from-port="0" to-layer="256" to-port="1" />
|
|
<edge from-layer="256" from-port="2" to-layer="258" to-port="0" />
|
|
<edge from-layer="257" from-port="0" to-layer="258" to-port="1" />
|
|
<edge from-layer="258" from-port="2" to-layer="260" to-port="0" />
|
|
<edge from-layer="258" from-port="2" to-layer="288" to-port="1" />
|
|
<edge from-layer="259" from-port="0" to-layer="260" to-port="1" />
|
|
<edge from-layer="260" from-port="2" to-layer="265" to-port="0" />
|
|
<edge from-layer="261" from-port="0" to-layer="265" to-port="1" />
|
|
<edge from-layer="262" from-port="0" to-layer="265" to-port="2" />
|
|
<edge from-layer="263" from-port="0" to-layer="265" to-port="3" />
|
|
<edge from-layer="264" from-port="0" to-layer="265" to-port="4" />
|
|
<edge from-layer="265" from-port="5" to-layer="270" to-port="0" />
|
|
<edge from-layer="266" from-port="0" to-layer="267" to-port="0" />
|
|
<edge from-layer="267" from-port="1" to-layer="269" to-port="0" />
|
|
<edge from-layer="268" from-port="0" to-layer="269" to-port="1" />
|
|
<edge from-layer="269" from-port="2" to-layer="270" to-port="1" />
|
|
<edge from-layer="270" from-port="2" to-layer="272" to-port="0" />
|
|
<edge from-layer="271" from-port="0" to-layer="272" to-port="1" />
|
|
<edge from-layer="272" from-port="2" to-layer="273" to-port="0" />
|
|
<edge from-layer="273" from-port="1" to-layer="275" to-port="0" />
|
|
<edge from-layer="274" from-port="0" to-layer="275" to-port="1" />
|
|
<edge from-layer="275" from-port="2" to-layer="280" to-port="0" />
|
|
<edge from-layer="276" from-port="0" to-layer="280" to-port="1" />
|
|
<edge from-layer="277" from-port="0" to-layer="280" to-port="2" />
|
|
<edge from-layer="278" from-port="0" to-layer="280" to-port="3" />
|
|
<edge from-layer="279" from-port="0" to-layer="280" to-port="4" />
|
|
<edge from-layer="280" from-port="5" to-layer="285" to-port="0" />
|
|
<edge from-layer="281" from-port="0" to-layer="282" to-port="0" />
|
|
<edge from-layer="282" from-port="1" to-layer="284" to-port="0" />
|
|
<edge from-layer="283" from-port="0" to-layer="284" to-port="1" />
|
|
<edge from-layer="284" from-port="2" to-layer="285" to-port="1" />
|
|
<edge from-layer="285" from-port="2" to-layer="287" to-port="0" />
|
|
<edge from-layer="286" from-port="0" to-layer="287" to-port="1" />
|
|
<edge from-layer="287" from-port="2" to-layer="288" to-port="0" />
|
|
<edge from-layer="288" from-port="2" to-layer="290" to-port="0" />
|
|
<edge from-layer="289" from-port="0" to-layer="290" to-port="1" />
|
|
<edge from-layer="290" from-port="2" to-layer="292" to-port="0" />
|
|
<edge from-layer="291" from-port="0" to-layer="292" to-port="1" />
|
|
<edge from-layer="292" from-port="2" to-layer="294" to-port="0" />
|
|
<edge from-layer="293" from-port="0" to-layer="294" to-port="1" />
|
|
<edge from-layer="294" from-port="2" to-layer="296" to-port="0" />
|
|
<edge from-layer="294" from-port="2" to-layer="369" to-port="1" />
|
|
<edge from-layer="295" from-port="0" to-layer="296" to-port="1" />
|
|
<edge from-layer="296" from-port="2" to-layer="301" to-port="0" />
|
|
<edge from-layer="297" from-port="0" to-layer="301" to-port="1" />
|
|
<edge from-layer="298" from-port="0" to-layer="301" to-port="2" />
|
|
<edge from-layer="299" from-port="0" to-layer="301" to-port="3" />
|
|
<edge from-layer="300" from-port="0" to-layer="301" to-port="4" />
|
|
<edge from-layer="301" from-port="5" to-layer="306" to-port="0" />
|
|
<edge from-layer="301" from-port="5" to-layer="322" to-port="0" />
|
|
<edge from-layer="301" from-port="5" to-layer="343" to-port="0" />
|
|
<edge from-layer="302" from-port="0" to-layer="303" to-port="0" />
|
|
<edge from-layer="303" from-port="1" to-layer="305" to-port="0" />
|
|
<edge from-layer="304" from-port="0" to-layer="305" to-port="1" />
|
|
<edge from-layer="305" from-port="2" to-layer="306" to-port="1" />
|
|
<edge from-layer="306" from-port="2" to-layer="308" to-port="0" />
|
|
<edge from-layer="307" from-port="0" to-layer="308" to-port="1" />
|
|
<edge from-layer="308" from-port="2" to-layer="313" to-port="0" />
|
|
<edge from-layer="309" from-port="0" to-layer="313" to-port="1" />
|
|
<edge from-layer="310" from-port="0" to-layer="313" to-port="2" />
|
|
<edge from-layer="311" from-port="0" to-layer="313" to-port="3" />
|
|
<edge from-layer="312" from-port="0" to-layer="313" to-port="4" />
|
|
<edge from-layer="313" from-port="5" to-layer="315" to-port="0" />
|
|
<edge from-layer="314" from-port="0" to-layer="315" to-port="1" />
|
|
<edge from-layer="315" from-port="2" to-layer="317" to-port="0" />
|
|
<edge from-layer="316" from-port="0" to-layer="317" to-port="1" />
|
|
<edge from-layer="317" from-port="2" to-layer="334" to-port="0" />
|
|
<edge from-layer="318" from-port="0" to-layer="319" to-port="0" />
|
|
<edge from-layer="319" from-port="1" to-layer="321" to-port="0" />
|
|
<edge from-layer="320" from-port="0" to-layer="321" to-port="1" />
|
|
<edge from-layer="321" from-port="2" to-layer="322" to-port="1" />
|
|
<edge from-layer="322" from-port="2" to-layer="324" to-port="0" />
|
|
<edge from-layer="323" from-port="0" to-layer="324" to-port="1" />
|
|
<edge from-layer="324" from-port="2" to-layer="329" to-port="0" />
|
|
<edge from-layer="325" from-port="0" to-layer="329" to-port="1" />
|
|
<edge from-layer="326" from-port="0" to-layer="329" to-port="2" />
|
|
<edge from-layer="327" from-port="0" to-layer="329" to-port="3" />
|
|
<edge from-layer="328" from-port="0" to-layer="329" to-port="4" />
|
|
<edge from-layer="329" from-port="5" to-layer="331" to-port="0" />
|
|
<edge from-layer="330" from-port="0" to-layer="331" to-port="1" />
|
|
<edge from-layer="331" from-port="2" to-layer="333" to-port="0" />
|
|
<edge from-layer="332" from-port="0" to-layer="333" to-port="1" />
|
|
<edge from-layer="333" from-port="2" to-layer="334" to-port="1" />
|
|
<edge from-layer="334" from-port="2" to-layer="336" to-port="0" />
|
|
<edge from-layer="335" from-port="0" to-layer="336" to-port="1" />
|
|
<edge from-layer="336" from-port="2" to-layer="337" to-port="0" />
|
|
<edge from-layer="337" from-port="2" to-layer="338" to-port="0" />
|
|
<edge from-layer="338" from-port="1" to-layer="350" to-port="0" />
|
|
<edge from-layer="339" from-port="0" to-layer="340" to-port="0" />
|
|
<edge from-layer="340" from-port="1" to-layer="342" to-port="0" />
|
|
<edge from-layer="341" from-port="0" to-layer="342" to-port="1" />
|
|
<edge from-layer="342" from-port="2" to-layer="343" to-port="1" />
|
|
<edge from-layer="343" from-port="2" to-layer="345" to-port="0" />
|
|
<edge from-layer="344" from-port="0" to-layer="345" to-port="1" />
|
|
<edge from-layer="345" from-port="2" to-layer="347" to-port="0" />
|
|
<edge from-layer="346" from-port="0" to-layer="347" to-port="1" />
|
|
<edge from-layer="347" from-port="2" to-layer="349" to-port="0" />
|
|
<edge from-layer="348" from-port="0" to-layer="349" to-port="1" />
|
|
<edge from-layer="349" from-port="2" to-layer="350" to-port="1" />
|
|
<edge from-layer="350" from-port="2" to-layer="352" to-port="0" />
|
|
<edge from-layer="351" from-port="0" to-layer="352" to-port="1" />
|
|
<edge from-layer="352" from-port="2" to-layer="354" to-port="0" />
|
|
<edge from-layer="353" from-port="0" to-layer="354" to-port="1" />
|
|
<edge from-layer="354" from-port="2" to-layer="356" to-port="0" />
|
|
<edge from-layer="355" from-port="0" to-layer="356" to-port="1" />
|
|
<edge from-layer="356" from-port="2" to-layer="361" to-port="0" />
|
|
<edge from-layer="357" from-port="0" to-layer="361" to-port="1" />
|
|
<edge from-layer="358" from-port="0" to-layer="361" to-port="2" />
|
|
<edge from-layer="359" from-port="0" to-layer="361" to-port="3" />
|
|
<edge from-layer="360" from-port="0" to-layer="361" to-port="4" />
|
|
<edge from-layer="361" from-port="5" to-layer="366" to-port="0" />
|
|
<edge from-layer="362" from-port="0" to-layer="363" to-port="0" />
|
|
<edge from-layer="363" from-port="1" to-layer="365" to-port="0" />
|
|
<edge from-layer="364" from-port="0" to-layer="365" to-port="1" />
|
|
<edge from-layer="365" from-port="2" to-layer="366" to-port="1" />
|
|
<edge from-layer="366" from-port="2" to-layer="368" to-port="0" />
|
|
<edge from-layer="367" from-port="0" to-layer="368" to-port="1" />
|
|
<edge from-layer="368" from-port="2" to-layer="369" to-port="0" />
|
|
<edge from-layer="369" from-port="2" to-layer="371" to-port="0" />
|
|
<edge from-layer="370" from-port="0" to-layer="371" to-port="1" />
|
|
<edge from-layer="371" from-port="2" to-layer="373" to-port="0" />
|
|
<edge from-layer="372" from-port="0" to-layer="373" to-port="1" />
|
|
<edge from-layer="373" from-port="2" to-layer="375" to-port="0" />
|
|
<edge from-layer="374" from-port="0" to-layer="375" to-port="1" />
|
|
<edge from-layer="375" from-port="2" to-layer="377" to-port="0" />
|
|
<edge from-layer="375" from-port="2" to-layer="405" to-port="1" />
|
|
<edge from-layer="376" from-port="0" to-layer="377" to-port="1" />
|
|
<edge from-layer="377" from-port="2" to-layer="382" to-port="0" />
|
|
<edge from-layer="378" from-port="0" to-layer="382" to-port="1" />
|
|
<edge from-layer="379" from-port="0" to-layer="382" to-port="2" />
|
|
<edge from-layer="380" from-port="0" to-layer="382" to-port="3" />
|
|
<edge from-layer="381" from-port="0" to-layer="382" to-port="4" />
|
|
<edge from-layer="382" from-port="5" to-layer="387" to-port="0" />
|
|
<edge from-layer="383" from-port="0" to-layer="384" to-port="0" />
|
|
<edge from-layer="384" from-port="1" to-layer="386" to-port="0" />
|
|
<edge from-layer="385" from-port="0" to-layer="386" to-port="1" />
|
|
<edge from-layer="386" from-port="2" to-layer="387" to-port="1" />
|
|
<edge from-layer="387" from-port="2" to-layer="389" to-port="0" />
|
|
<edge from-layer="388" from-port="0" to-layer="389" to-port="1" />
|
|
<edge from-layer="389" from-port="2" to-layer="390" to-port="0" />
|
|
<edge from-layer="390" from-port="1" to-layer="392" to-port="0" />
|
|
<edge from-layer="391" from-port="0" to-layer="392" to-port="1" />
|
|
<edge from-layer="392" from-port="2" to-layer="397" to-port="0" />
|
|
<edge from-layer="393" from-port="0" to-layer="397" to-port="1" />
|
|
<edge from-layer="394" from-port="0" to-layer="397" to-port="2" />
|
|
<edge from-layer="395" from-port="0" to-layer="397" to-port="3" />
|
|
<edge from-layer="396" from-port="0" to-layer="397" to-port="4" />
|
|
<edge from-layer="397" from-port="5" to-layer="402" to-port="0" />
|
|
<edge from-layer="398" from-port="0" to-layer="399" to-port="0" />
|
|
<edge from-layer="399" from-port="1" to-layer="401" to-port="0" />
|
|
<edge from-layer="400" from-port="0" to-layer="401" to-port="1" />
|
|
<edge from-layer="401" from-port="2" to-layer="402" to-port="1" />
|
|
<edge from-layer="402" from-port="2" to-layer="404" to-port="0" />
|
|
<edge from-layer="403" from-port="0" to-layer="404" to-port="1" />
|
|
<edge from-layer="404" from-port="2" to-layer="405" to-port="0" />
|
|
<edge from-layer="405" from-port="2" to-layer="407" to-port="0" />
|
|
<edge from-layer="406" from-port="0" to-layer="407" to-port="1" />
|
|
<edge from-layer="407" from-port="2" to-layer="409" to-port="0" />
|
|
<edge from-layer="408" from-port="0" to-layer="409" to-port="1" />
|
|
<edge from-layer="409" from-port="2" to-layer="411" to-port="0" />
|
|
<edge from-layer="410" from-port="0" to-layer="411" to-port="1" />
|
|
<edge from-layer="411" from-port="2" to-layer="413" to-port="0" />
|
|
<edge from-layer="411" from-port="2" to-layer="486" to-port="1" />
|
|
<edge from-layer="412" from-port="0" to-layer="413" to-port="1" />
|
|
<edge from-layer="413" from-port="2" to-layer="418" to-port="0" />
|
|
<edge from-layer="414" from-port="0" to-layer="418" to-port="1" />
|
|
<edge from-layer="415" from-port="0" to-layer="418" to-port="2" />
|
|
<edge from-layer="416" from-port="0" to-layer="418" to-port="3" />
|
|
<edge from-layer="417" from-port="0" to-layer="418" to-port="4" />
|
|
<edge from-layer="418" from-port="5" to-layer="423" to-port="0" />
|
|
<edge from-layer="418" from-port="5" to-layer="439" to-port="0" />
|
|
<edge from-layer="418" from-port="5" to-layer="460" to-port="0" />
|
|
<edge from-layer="419" from-port="0" to-layer="420" to-port="0" />
|
|
<edge from-layer="420" from-port="1" to-layer="422" to-port="0" />
|
|
<edge from-layer="421" from-port="0" to-layer="422" to-port="1" />
|
|
<edge from-layer="422" from-port="2" to-layer="423" to-port="1" />
|
|
<edge from-layer="423" from-port="2" to-layer="425" to-port="0" />
|
|
<edge from-layer="424" from-port="0" to-layer="425" to-port="1" />
|
|
<edge from-layer="425" from-port="2" to-layer="430" to-port="0" />
|
|
<edge from-layer="426" from-port="0" to-layer="430" to-port="1" />
|
|
<edge from-layer="427" from-port="0" to-layer="430" to-port="2" />
|
|
<edge from-layer="428" from-port="0" to-layer="430" to-port="3" />
|
|
<edge from-layer="429" from-port="0" to-layer="430" to-port="4" />
|
|
<edge from-layer="430" from-port="5" to-layer="432" to-port="0" />
|
|
<edge from-layer="431" from-port="0" to-layer="432" to-port="1" />
|
|
<edge from-layer="432" from-port="2" to-layer="434" to-port="0" />
|
|
<edge from-layer="433" from-port="0" to-layer="434" to-port="1" />
|
|
<edge from-layer="434" from-port="2" to-layer="451" to-port="0" />
|
|
<edge from-layer="435" from-port="0" to-layer="436" to-port="0" />
|
|
<edge from-layer="436" from-port="1" to-layer="438" to-port="0" />
|
|
<edge from-layer="437" from-port="0" to-layer="438" to-port="1" />
|
|
<edge from-layer="438" from-port="2" to-layer="439" to-port="1" />
|
|
<edge from-layer="439" from-port="2" to-layer="441" to-port="0" />
|
|
<edge from-layer="440" from-port="0" to-layer="441" to-port="1" />
|
|
<edge from-layer="441" from-port="2" to-layer="446" to-port="0" />
|
|
<edge from-layer="442" from-port="0" to-layer="446" to-port="1" />
|
|
<edge from-layer="443" from-port="0" to-layer="446" to-port="2" />
|
|
<edge from-layer="444" from-port="0" to-layer="446" to-port="3" />
|
|
<edge from-layer="445" from-port="0" to-layer="446" to-port="4" />
|
|
<edge from-layer="446" from-port="5" to-layer="448" to-port="0" />
|
|
<edge from-layer="447" from-port="0" to-layer="448" to-port="1" />
|
|
<edge from-layer="448" from-port="2" to-layer="450" to-port="0" />
|
|
<edge from-layer="449" from-port="0" to-layer="450" to-port="1" />
|
|
<edge from-layer="450" from-port="2" to-layer="451" to-port="1" />
|
|
<edge from-layer="451" from-port="2" to-layer="453" to-port="0" />
|
|
<edge from-layer="452" from-port="0" to-layer="453" to-port="1" />
|
|
<edge from-layer="453" from-port="2" to-layer="454" to-port="0" />
|
|
<edge from-layer="454" from-port="2" to-layer="455" to-port="0" />
|
|
<edge from-layer="455" from-port="1" to-layer="467" to-port="0" />
|
|
<edge from-layer="456" from-port="0" to-layer="457" to-port="0" />
|
|
<edge from-layer="457" from-port="1" to-layer="459" to-port="0" />
|
|
<edge from-layer="458" from-port="0" to-layer="459" to-port="1" />
|
|
<edge from-layer="459" from-port="2" to-layer="460" to-port="1" />
|
|
<edge from-layer="460" from-port="2" to-layer="462" to-port="0" />
|
|
<edge from-layer="461" from-port="0" to-layer="462" to-port="1" />
|
|
<edge from-layer="462" from-port="2" to-layer="464" to-port="0" />
|
|
<edge from-layer="463" from-port="0" to-layer="464" to-port="1" />
|
|
<edge from-layer="464" from-port="2" to-layer="466" to-port="0" />
|
|
<edge from-layer="465" from-port="0" to-layer="466" to-port="1" />
|
|
<edge from-layer="466" from-port="2" to-layer="467" to-port="1" />
|
|
<edge from-layer="467" from-port="2" to-layer="469" to-port="0" />
|
|
<edge from-layer="468" from-port="0" to-layer="469" to-port="1" />
|
|
<edge from-layer="469" from-port="2" to-layer="471" to-port="0" />
|
|
<edge from-layer="470" from-port="0" to-layer="471" to-port="1" />
|
|
<edge from-layer="471" from-port="2" to-layer="473" to-port="0" />
|
|
<edge from-layer="472" from-port="0" to-layer="473" to-port="1" />
|
|
<edge from-layer="473" from-port="2" to-layer="478" to-port="0" />
|
|
<edge from-layer="474" from-port="0" to-layer="478" to-port="1" />
|
|
<edge from-layer="475" from-port="0" to-layer="478" to-port="2" />
|
|
<edge from-layer="476" from-port="0" to-layer="478" to-port="3" />
|
|
<edge from-layer="477" from-port="0" to-layer="478" to-port="4" />
|
|
<edge from-layer="478" from-port="5" to-layer="483" to-port="0" />
|
|
<edge from-layer="479" from-port="0" to-layer="480" to-port="0" />
|
|
<edge from-layer="480" from-port="1" to-layer="482" to-port="0" />
|
|
<edge from-layer="481" from-port="0" to-layer="482" to-port="1" />
|
|
<edge from-layer="482" from-port="2" to-layer="483" to-port="1" />
|
|
<edge from-layer="483" from-port="2" to-layer="485" to-port="0" />
|
|
<edge from-layer="484" from-port="0" to-layer="485" to-port="1" />
|
|
<edge from-layer="485" from-port="2" to-layer="486" to-port="0" />
|
|
<edge from-layer="486" from-port="2" to-layer="488" to-port="0" />
|
|
<edge from-layer="487" from-port="0" to-layer="488" to-port="1" />
|
|
<edge from-layer="488" from-port="2" to-layer="490" to-port="0" />
|
|
<edge from-layer="489" from-port="0" to-layer="490" to-port="1" />
|
|
<edge from-layer="490" from-port="2" to-layer="492" to-port="0" />
|
|
<edge from-layer="491" from-port="0" to-layer="492" to-port="1" />
|
|
<edge from-layer="492" from-port="2" to-layer="494" to-port="0" />
|
|
<edge from-layer="492" from-port="2" to-layer="522" to-port="1" />
|
|
<edge from-layer="493" from-port="0" to-layer="494" to-port="1" />
|
|
<edge from-layer="494" from-port="2" to-layer="499" to-port="0" />
|
|
<edge from-layer="495" from-port="0" to-layer="499" to-port="1" />
|
|
<edge from-layer="496" from-port="0" to-layer="499" to-port="2" />
|
|
<edge from-layer="497" from-port="0" to-layer="499" to-port="3" />
|
|
<edge from-layer="498" from-port="0" to-layer="499" to-port="4" />
|
|
<edge from-layer="499" from-port="5" to-layer="504" to-port="0" />
|
|
<edge from-layer="500" from-port="0" to-layer="501" to-port="0" />
|
|
<edge from-layer="501" from-port="1" to-layer="503" to-port="0" />
|
|
<edge from-layer="502" from-port="0" to-layer="503" to-port="1" />
|
|
<edge from-layer="503" from-port="2" to-layer="504" to-port="1" />
|
|
<edge from-layer="504" from-port="2" to-layer="506" to-port="0" />
|
|
<edge from-layer="505" from-port="0" to-layer="506" to-port="1" />
|
|
<edge from-layer="506" from-port="2" to-layer="507" to-port="0" />
|
|
<edge from-layer="507" from-port="1" to-layer="509" to-port="0" />
|
|
<edge from-layer="508" from-port="0" to-layer="509" to-port="1" />
|
|
<edge from-layer="509" from-port="2" to-layer="514" to-port="0" />
|
|
<edge from-layer="510" from-port="0" to-layer="514" to-port="1" />
|
|
<edge from-layer="511" from-port="0" to-layer="514" to-port="2" />
|
|
<edge from-layer="512" from-port="0" to-layer="514" to-port="3" />
|
|
<edge from-layer="513" from-port="0" to-layer="514" to-port="4" />
|
|
<edge from-layer="514" from-port="5" to-layer="519" to-port="0" />
|
|
<edge from-layer="515" from-port="0" to-layer="516" to-port="0" />
|
|
<edge from-layer="516" from-port="1" to-layer="518" to-port="0" />
|
|
<edge from-layer="517" from-port="0" to-layer="518" to-port="1" />
|
|
<edge from-layer="518" from-port="2" to-layer="519" to-port="1" />
|
|
<edge from-layer="519" from-port="2" to-layer="521" to-port="0" />
|
|
<edge from-layer="520" from-port="0" to-layer="521" to-port="1" />
|
|
<edge from-layer="521" from-port="2" to-layer="522" to-port="0" />
|
|
<edge from-layer="522" from-port="2" to-layer="524" to-port="0" />
|
|
<edge from-layer="523" from-port="0" to-layer="524" to-port="1" />
|
|
<edge from-layer="524" from-port="2" to-layer="526" to-port="0" />
|
|
<edge from-layer="525" from-port="0" to-layer="526" to-port="1" />
|
|
<edge from-layer="526" from-port="2" to-layer="528" to-port="0" />
|
|
<edge from-layer="527" from-port="0" to-layer="528" to-port="1" />
|
|
<edge from-layer="528" from-port="2" to-layer="530" to-port="0" />
|
|
<edge from-layer="528" from-port="2" to-layer="603" to-port="1" />
|
|
<edge from-layer="529" from-port="0" to-layer="530" to-port="1" />
|
|
<edge from-layer="530" from-port="2" to-layer="535" to-port="0" />
|
|
<edge from-layer="531" from-port="0" to-layer="535" to-port="1" />
|
|
<edge from-layer="532" from-port="0" to-layer="535" to-port="2" />
|
|
<edge from-layer="533" from-port="0" to-layer="535" to-port="3" />
|
|
<edge from-layer="534" from-port="0" to-layer="535" to-port="4" />
|
|
<edge from-layer="535" from-port="5" to-layer="540" to-port="0" />
|
|
<edge from-layer="535" from-port="5" to-layer="556" to-port="0" />
|
|
<edge from-layer="535" from-port="5" to-layer="577" to-port="0" />
|
|
<edge from-layer="536" from-port="0" to-layer="537" to-port="0" />
|
|
<edge from-layer="537" from-port="1" to-layer="539" to-port="0" />
|
|
<edge from-layer="538" from-port="0" to-layer="539" to-port="1" />
|
|
<edge from-layer="539" from-port="2" to-layer="540" to-port="1" />
|
|
<edge from-layer="540" from-port="2" to-layer="542" to-port="0" />
|
|
<edge from-layer="541" from-port="0" to-layer="542" to-port="1" />
|
|
<edge from-layer="542" from-port="2" to-layer="547" to-port="0" />
|
|
<edge from-layer="543" from-port="0" to-layer="547" to-port="1" />
|
|
<edge from-layer="544" from-port="0" to-layer="547" to-port="2" />
|
|
<edge from-layer="545" from-port="0" to-layer="547" to-port="3" />
|
|
<edge from-layer="546" from-port="0" to-layer="547" to-port="4" />
|
|
<edge from-layer="547" from-port="5" to-layer="549" to-port="0" />
|
|
<edge from-layer="548" from-port="0" to-layer="549" to-port="1" />
|
|
<edge from-layer="549" from-port="2" to-layer="551" to-port="0" />
|
|
<edge from-layer="550" from-port="0" to-layer="551" to-port="1" />
|
|
<edge from-layer="551" from-port="2" to-layer="568" to-port="0" />
|
|
<edge from-layer="552" from-port="0" to-layer="553" to-port="0" />
|
|
<edge from-layer="553" from-port="1" to-layer="555" to-port="0" />
|
|
<edge from-layer="554" from-port="0" to-layer="555" to-port="1" />
|
|
<edge from-layer="555" from-port="2" to-layer="556" to-port="1" />
|
|
<edge from-layer="556" from-port="2" to-layer="558" to-port="0" />
|
|
<edge from-layer="557" from-port="0" to-layer="558" to-port="1" />
|
|
<edge from-layer="558" from-port="2" to-layer="563" to-port="0" />
|
|
<edge from-layer="559" from-port="0" to-layer="563" to-port="1" />
|
|
<edge from-layer="560" from-port="0" to-layer="563" to-port="2" />
|
|
<edge from-layer="561" from-port="0" to-layer="563" to-port="3" />
|
|
<edge from-layer="562" from-port="0" to-layer="563" to-port="4" />
|
|
<edge from-layer="563" from-port="5" to-layer="565" to-port="0" />
|
|
<edge from-layer="564" from-port="0" to-layer="565" to-port="1" />
|
|
<edge from-layer="565" from-port="2" to-layer="567" to-port="0" />
|
|
<edge from-layer="566" from-port="0" to-layer="567" to-port="1" />
|
|
<edge from-layer="567" from-port="2" to-layer="568" to-port="1" />
|
|
<edge from-layer="568" from-port="2" to-layer="570" to-port="0" />
|
|
<edge from-layer="569" from-port="0" to-layer="570" to-port="1" />
|
|
<edge from-layer="570" from-port="2" to-layer="571" to-port="0" />
|
|
<edge from-layer="571" from-port="2" to-layer="572" to-port="0" />
|
|
<edge from-layer="572" from-port="1" to-layer="584" to-port="0" />
|
|
<edge from-layer="573" from-port="0" to-layer="574" to-port="0" />
|
|
<edge from-layer="574" from-port="1" to-layer="576" to-port="0" />
|
|
<edge from-layer="575" from-port="0" to-layer="576" to-port="1" />
|
|
<edge from-layer="576" from-port="2" to-layer="577" to-port="1" />
|
|
<edge from-layer="577" from-port="2" to-layer="579" to-port="0" />
|
|
<edge from-layer="578" from-port="0" to-layer="579" to-port="1" />
|
|
<edge from-layer="579" from-port="2" to-layer="581" to-port="0" />
|
|
<edge from-layer="580" from-port="0" to-layer="581" to-port="1" />
|
|
<edge from-layer="581" from-port="2" to-layer="583" to-port="0" />
|
|
<edge from-layer="582" from-port="0" to-layer="583" to-port="1" />
|
|
<edge from-layer="583" from-port="2" to-layer="584" to-port="1" />
|
|
<edge from-layer="584" from-port="2" to-layer="586" to-port="0" />
|
|
<edge from-layer="585" from-port="0" to-layer="586" to-port="1" />
|
|
<edge from-layer="586" from-port="2" to-layer="588" to-port="0" />
|
|
<edge from-layer="587" from-port="0" to-layer="588" to-port="1" />
|
|
<edge from-layer="588" from-port="2" to-layer="590" to-port="0" />
|
|
<edge from-layer="589" from-port="0" to-layer="590" to-port="1" />
|
|
<edge from-layer="590" from-port="2" to-layer="595" to-port="0" />
|
|
<edge from-layer="591" from-port="0" to-layer="595" to-port="1" />
|
|
<edge from-layer="592" from-port="0" to-layer="595" to-port="2" />
|
|
<edge from-layer="593" from-port="0" to-layer="595" to-port="3" />
|
|
<edge from-layer="594" from-port="0" to-layer="595" to-port="4" />
|
|
<edge from-layer="595" from-port="5" to-layer="600" to-port="0" />
|
|
<edge from-layer="596" from-port="0" to-layer="597" to-port="0" />
|
|
<edge from-layer="597" from-port="1" to-layer="599" to-port="0" />
|
|
<edge from-layer="598" from-port="0" to-layer="599" to-port="1" />
|
|
<edge from-layer="599" from-port="2" to-layer="600" to-port="1" />
|
|
<edge from-layer="600" from-port="2" to-layer="602" to-port="0" />
|
|
<edge from-layer="601" from-port="0" to-layer="602" to-port="1" />
|
|
<edge from-layer="602" from-port="2" to-layer="603" to-port="0" />
|
|
<edge from-layer="603" from-port="2" to-layer="605" to-port="0" />
|
|
<edge from-layer="604" from-port="0" to-layer="605" to-port="1" />
|
|
<edge from-layer="605" from-port="2" to-layer="607" to-port="0" />
|
|
<edge from-layer="606" from-port="0" to-layer="607" to-port="1" />
|
|
<edge from-layer="607" from-port="2" to-layer="609" to-port="0" />
|
|
<edge from-layer="608" from-port="0" to-layer="609" to-port="1" />
|
|
<edge from-layer="609" from-port="2" to-layer="611" to-port="0" />
|
|
<edge from-layer="609" from-port="2" to-layer="639" to-port="1" />
|
|
<edge from-layer="610" from-port="0" to-layer="611" to-port="1" />
|
|
<edge from-layer="611" from-port="2" to-layer="616" to-port="0" />
|
|
<edge from-layer="612" from-port="0" to-layer="616" to-port="1" />
|
|
<edge from-layer="613" from-port="0" to-layer="616" to-port="2" />
|
|
<edge from-layer="614" from-port="0" to-layer="616" to-port="3" />
|
|
<edge from-layer="615" from-port="0" to-layer="616" to-port="4" />
|
|
<edge from-layer="616" from-port="5" to-layer="621" to-port="0" />
|
|
<edge from-layer="617" from-port="0" to-layer="618" to-port="0" />
|
|
<edge from-layer="618" from-port="1" to-layer="620" to-port="0" />
|
|
<edge from-layer="619" from-port="0" to-layer="620" to-port="1" />
|
|
<edge from-layer="620" from-port="2" to-layer="621" to-port="1" />
|
|
<edge from-layer="621" from-port="2" to-layer="623" to-port="0" />
|
|
<edge from-layer="622" from-port="0" to-layer="623" to-port="1" />
|
|
<edge from-layer="623" from-port="2" to-layer="624" to-port="0" />
|
|
<edge from-layer="624" from-port="1" to-layer="626" to-port="0" />
|
|
<edge from-layer="625" from-port="0" to-layer="626" to-port="1" />
|
|
<edge from-layer="626" from-port="2" to-layer="631" to-port="0" />
|
|
<edge from-layer="627" from-port="0" to-layer="631" to-port="1" />
|
|
<edge from-layer="628" from-port="0" to-layer="631" to-port="2" />
|
|
<edge from-layer="629" from-port="0" to-layer="631" to-port="3" />
|
|
<edge from-layer="630" from-port="0" to-layer="631" to-port="4" />
|
|
<edge from-layer="631" from-port="5" to-layer="636" to-port="0" />
|
|
<edge from-layer="632" from-port="0" to-layer="633" to-port="0" />
|
|
<edge from-layer="633" from-port="1" to-layer="635" to-port="0" />
|
|
<edge from-layer="634" from-port="0" to-layer="635" to-port="1" />
|
|
<edge from-layer="635" from-port="2" to-layer="636" to-port="1" />
|
|
<edge from-layer="636" from-port="2" to-layer="638" to-port="0" />
|
|
<edge from-layer="637" from-port="0" to-layer="638" to-port="1" />
|
|
<edge from-layer="638" from-port="2" to-layer="639" to-port="0" />
|
|
<edge from-layer="639" from-port="2" to-layer="641" to-port="0" />
|
|
<edge from-layer="640" from-port="0" to-layer="641" to-port="1" />
|
|
<edge from-layer="641" from-port="2" to-layer="643" to-port="0" />
|
|
<edge from-layer="642" from-port="0" to-layer="643" to-port="1" />
|
|
<edge from-layer="643" from-port="2" to-layer="645" to-port="0" />
|
|
<edge from-layer="644" from-port="0" to-layer="645" to-port="1" />
|
|
<edge from-layer="645" from-port="2" to-layer="647" to-port="0" />
|
|
<edge from-layer="645" from-port="2" to-layer="720" to-port="1" />
|
|
<edge from-layer="646" from-port="0" to-layer="647" to-port="1" />
|
|
<edge from-layer="647" from-port="2" to-layer="652" to-port="0" />
|
|
<edge from-layer="648" from-port="0" to-layer="652" to-port="1" />
|
|
<edge from-layer="649" from-port="0" to-layer="652" to-port="2" />
|
|
<edge from-layer="650" from-port="0" to-layer="652" to-port="3" />
|
|
<edge from-layer="651" from-port="0" to-layer="652" to-port="4" />
|
|
<edge from-layer="652" from-port="5" to-layer="657" to-port="0" />
|
|
<edge from-layer="652" from-port="5" to-layer="673" to-port="0" />
|
|
<edge from-layer="652" from-port="5" to-layer="694" to-port="0" />
|
|
<edge from-layer="653" from-port="0" to-layer="654" to-port="0" />
|
|
<edge from-layer="654" from-port="1" to-layer="656" to-port="0" />
|
|
<edge from-layer="655" from-port="0" to-layer="656" to-port="1" />
|
|
<edge from-layer="656" from-port="2" to-layer="657" to-port="1" />
|
|
<edge from-layer="657" from-port="2" to-layer="659" to-port="0" />
|
|
<edge from-layer="658" from-port="0" to-layer="659" to-port="1" />
|
|
<edge from-layer="659" from-port="2" to-layer="664" to-port="0" />
|
|
<edge from-layer="660" from-port="0" to-layer="664" to-port="1" />
|
|
<edge from-layer="661" from-port="0" to-layer="664" to-port="2" />
|
|
<edge from-layer="662" from-port="0" to-layer="664" to-port="3" />
|
|
<edge from-layer="663" from-port="0" to-layer="664" to-port="4" />
|
|
<edge from-layer="664" from-port="5" to-layer="666" to-port="0" />
|
|
<edge from-layer="665" from-port="0" to-layer="666" to-port="1" />
|
|
<edge from-layer="666" from-port="2" to-layer="668" to-port="0" />
|
|
<edge from-layer="667" from-port="0" to-layer="668" to-port="1" />
|
|
<edge from-layer="668" from-port="2" to-layer="685" to-port="0" />
|
|
<edge from-layer="669" from-port="0" to-layer="670" to-port="0" />
|
|
<edge from-layer="670" from-port="1" to-layer="672" to-port="0" />
|
|
<edge from-layer="671" from-port="0" to-layer="672" to-port="1" />
|
|
<edge from-layer="672" from-port="2" to-layer="673" to-port="1" />
|
|
<edge from-layer="673" from-port="2" to-layer="675" to-port="0" />
|
|
<edge from-layer="674" from-port="0" to-layer="675" to-port="1" />
|
|
<edge from-layer="675" from-port="2" to-layer="680" to-port="0" />
|
|
<edge from-layer="676" from-port="0" to-layer="680" to-port="1" />
|
|
<edge from-layer="677" from-port="0" to-layer="680" to-port="2" />
|
|
<edge from-layer="678" from-port="0" to-layer="680" to-port="3" />
|
|
<edge from-layer="679" from-port="0" to-layer="680" to-port="4" />
|
|
<edge from-layer="680" from-port="5" to-layer="682" to-port="0" />
|
|
<edge from-layer="681" from-port="0" to-layer="682" to-port="1" />
|
|
<edge from-layer="682" from-port="2" to-layer="684" to-port="0" />
|
|
<edge from-layer="683" from-port="0" to-layer="684" to-port="1" />
|
|
<edge from-layer="684" from-port="2" to-layer="685" to-port="1" />
|
|
<edge from-layer="685" from-port="2" to-layer="687" to-port="0" />
|
|
<edge from-layer="686" from-port="0" to-layer="687" to-port="1" />
|
|
<edge from-layer="687" from-port="2" to-layer="688" to-port="0" />
|
|
<edge from-layer="688" from-port="2" to-layer="689" to-port="0" />
|
|
<edge from-layer="689" from-port="1" to-layer="701" to-port="0" />
|
|
<edge from-layer="690" from-port="0" to-layer="691" to-port="0" />
|
|
<edge from-layer="691" from-port="1" to-layer="693" to-port="0" />
|
|
<edge from-layer="692" from-port="0" to-layer="693" to-port="1" />
|
|
<edge from-layer="693" from-port="2" to-layer="694" to-port="1" />
|
|
<edge from-layer="694" from-port="2" to-layer="696" to-port="0" />
|
|
<edge from-layer="695" from-port="0" to-layer="696" to-port="1" />
|
|
<edge from-layer="696" from-port="2" to-layer="698" to-port="0" />
|
|
<edge from-layer="697" from-port="0" to-layer="698" to-port="1" />
|
|
<edge from-layer="698" from-port="2" to-layer="700" to-port="0" />
|
|
<edge from-layer="699" from-port="0" to-layer="700" to-port="1" />
|
|
<edge from-layer="700" from-port="2" to-layer="701" to-port="1" />
|
|
<edge from-layer="701" from-port="2" to-layer="703" to-port="0" />
|
|
<edge from-layer="702" from-port="0" to-layer="703" to-port="1" />
|
|
<edge from-layer="703" from-port="2" to-layer="705" to-port="0" />
|
|
<edge from-layer="704" from-port="0" to-layer="705" to-port="1" />
|
|
<edge from-layer="705" from-port="2" to-layer="707" to-port="0" />
|
|
<edge from-layer="706" from-port="0" to-layer="707" to-port="1" />
|
|
<edge from-layer="707" from-port="2" to-layer="712" to-port="0" />
|
|
<edge from-layer="708" from-port="0" to-layer="712" to-port="1" />
|
|
<edge from-layer="709" from-port="0" to-layer="712" to-port="2" />
|
|
<edge from-layer="710" from-port="0" to-layer="712" to-port="3" />
|
|
<edge from-layer="711" from-port="0" to-layer="712" to-port="4" />
|
|
<edge from-layer="712" from-port="5" to-layer="717" to-port="0" />
|
|
<edge from-layer="713" from-port="0" to-layer="714" to-port="0" />
|
|
<edge from-layer="714" from-port="1" to-layer="716" to-port="0" />
|
|
<edge from-layer="715" from-port="0" to-layer="716" to-port="1" />
|
|
<edge from-layer="716" from-port="2" to-layer="717" to-port="1" />
|
|
<edge from-layer="717" from-port="2" to-layer="719" to-port="0" />
|
|
<edge from-layer="718" from-port="0" to-layer="719" to-port="1" />
|
|
<edge from-layer="719" from-port="2" to-layer="720" to-port="0" />
|
|
<edge from-layer="720" from-port="2" to-layer="722" to-port="0" />
|
|
<edge from-layer="721" from-port="0" to-layer="722" to-port="1" />
|
|
<edge from-layer="722" from-port="2" to-layer="724" to-port="0" />
|
|
<edge from-layer="723" from-port="0" to-layer="724" to-port="1" />
|
|
<edge from-layer="724" from-port="2" to-layer="726" to-port="0" />
|
|
<edge from-layer="725" from-port="0" to-layer="726" to-port="1" />
|
|
<edge from-layer="726" from-port="2" to-layer="756" to-port="1" />
|
|
<edge from-layer="726" from-port="2" to-layer="728" to-port="0" />
|
|
<edge from-layer="727" from-port="0" to-layer="728" to-port="1" />
|
|
<edge from-layer="728" from-port="2" to-layer="733" to-port="0" />
|
|
<edge from-layer="729" from-port="0" to-layer="733" to-port="1" />
|
|
<edge from-layer="730" from-port="0" to-layer="733" to-port="2" />
|
|
<edge from-layer="731" from-port="0" to-layer="733" to-port="3" />
|
|
<edge from-layer="732" from-port="0" to-layer="733" to-port="4" />
|
|
<edge from-layer="733" from-port="5" to-layer="738" to-port="0" />
|
|
<edge from-layer="734" from-port="0" to-layer="735" to-port="0" />
|
|
<edge from-layer="735" from-port="1" to-layer="737" to-port="0" />
|
|
<edge from-layer="736" from-port="0" to-layer="737" to-port="1" />
|
|
<edge from-layer="737" from-port="2" to-layer="738" to-port="1" />
|
|
<edge from-layer="738" from-port="2" to-layer="740" to-port="0" />
|
|
<edge from-layer="739" from-port="0" to-layer="740" to-port="1" />
|
|
<edge from-layer="740" from-port="2" to-layer="741" to-port="0" />
|
|
<edge from-layer="741" from-port="1" to-layer="743" to-port="0" />
|
|
<edge from-layer="742" from-port="0" to-layer="743" to-port="1" />
|
|
<edge from-layer="743" from-port="2" to-layer="748" to-port="0" />
|
|
<edge from-layer="744" from-port="0" to-layer="748" to-port="1" />
|
|
<edge from-layer="745" from-port="0" to-layer="748" to-port="2" />
|
|
<edge from-layer="746" from-port="0" to-layer="748" to-port="3" />
|
|
<edge from-layer="747" from-port="0" to-layer="748" to-port="4" />
|
|
<edge from-layer="748" from-port="5" to-layer="753" to-port="0" />
|
|
<edge from-layer="749" from-port="0" to-layer="750" to-port="0" />
|
|
<edge from-layer="750" from-port="1" to-layer="752" to-port="0" />
|
|
<edge from-layer="751" from-port="0" to-layer="752" to-port="1" />
|
|
<edge from-layer="752" from-port="2" to-layer="753" to-port="1" />
|
|
<edge from-layer="753" from-port="2" to-layer="755" to-port="0" />
|
|
<edge from-layer="754" from-port="0" to-layer="755" to-port="1" />
|
|
<edge from-layer="755" from-port="2" to-layer="756" to-port="0" />
|
|
<edge from-layer="756" from-port="2" to-layer="758" to-port="0" />
|
|
<edge from-layer="757" from-port="0" to-layer="758" to-port="1" />
|
|
<edge from-layer="758" from-port="2" to-layer="760" to-port="0" />
|
|
<edge from-layer="759" from-port="0" to-layer="760" to-port="1" />
|
|
<edge from-layer="760" from-port="2" to-layer="762" to-port="0" />
|
|
<edge from-layer="761" from-port="0" to-layer="762" to-port="1" />
|
|
<edge from-layer="762" from-port="2" to-layer="764" to-port="0" />
|
|
<edge from-layer="762" from-port="2" to-layer="837" to-port="1" />
|
|
<edge from-layer="763" from-port="0" to-layer="764" to-port="1" />
|
|
<edge from-layer="764" from-port="2" to-layer="769" to-port="0" />
|
|
<edge from-layer="765" from-port="0" to-layer="769" to-port="1" />
|
|
<edge from-layer="766" from-port="0" to-layer="769" to-port="2" />
|
|
<edge from-layer="767" from-port="0" to-layer="769" to-port="3" />
|
|
<edge from-layer="768" from-port="0" to-layer="769" to-port="4" />
|
|
<edge from-layer="769" from-port="5" to-layer="774" to-port="0" />
|
|
<edge from-layer="769" from-port="5" to-layer="790" to-port="0" />
|
|
<edge from-layer="769" from-port="5" to-layer="811" to-port="0" />
|
|
<edge from-layer="770" from-port="0" to-layer="771" to-port="0" />
|
|
<edge from-layer="771" from-port="1" to-layer="773" to-port="0" />
|
|
<edge from-layer="772" from-port="0" to-layer="773" to-port="1" />
|
|
<edge from-layer="773" from-port="2" to-layer="774" to-port="1" />
|
|
<edge from-layer="774" from-port="2" to-layer="776" to-port="0" />
|
|
<edge from-layer="775" from-port="0" to-layer="776" to-port="1" />
|
|
<edge from-layer="776" from-port="2" to-layer="781" to-port="0" />
|
|
<edge from-layer="777" from-port="0" to-layer="781" to-port="1" />
|
|
<edge from-layer="778" from-port="0" to-layer="781" to-port="2" />
|
|
<edge from-layer="779" from-port="0" to-layer="781" to-port="3" />
|
|
<edge from-layer="780" from-port="0" to-layer="781" to-port="4" />
|
|
<edge from-layer="781" from-port="5" to-layer="783" to-port="0" />
|
|
<edge from-layer="782" from-port="0" to-layer="783" to-port="1" />
|
|
<edge from-layer="783" from-port="2" to-layer="785" to-port="0" />
|
|
<edge from-layer="784" from-port="0" to-layer="785" to-port="1" />
|
|
<edge from-layer="785" from-port="2" to-layer="802" to-port="0" />
|
|
<edge from-layer="786" from-port="0" to-layer="787" to-port="0" />
|
|
<edge from-layer="787" from-port="1" to-layer="789" to-port="0" />
|
|
<edge from-layer="788" from-port="0" to-layer="789" to-port="1" />
|
|
<edge from-layer="789" from-port="2" to-layer="790" to-port="1" />
|
|
<edge from-layer="790" from-port="2" to-layer="792" to-port="0" />
|
|
<edge from-layer="791" from-port="0" to-layer="792" to-port="1" />
|
|
<edge from-layer="792" from-port="2" to-layer="797" to-port="0" />
|
|
<edge from-layer="793" from-port="0" to-layer="797" to-port="1" />
|
|
<edge from-layer="794" from-port="0" to-layer="797" to-port="2" />
|
|
<edge from-layer="795" from-port="0" to-layer="797" to-port="3" />
|
|
<edge from-layer="796" from-port="0" to-layer="797" to-port="4" />
|
|
<edge from-layer="797" from-port="5" to-layer="799" to-port="0" />
|
|
<edge from-layer="798" from-port="0" to-layer="799" to-port="1" />
|
|
<edge from-layer="799" from-port="2" to-layer="801" to-port="0" />
|
|
<edge from-layer="800" from-port="0" to-layer="801" to-port="1" />
|
|
<edge from-layer="801" from-port="2" to-layer="802" to-port="1" />
|
|
<edge from-layer="802" from-port="2" to-layer="804" to-port="0" />
|
|
<edge from-layer="803" from-port="0" to-layer="804" to-port="1" />
|
|
<edge from-layer="804" from-port="2" to-layer="805" to-port="0" />
|
|
<edge from-layer="805" from-port="2" to-layer="806" to-port="0" />
|
|
<edge from-layer="806" from-port="1" to-layer="818" to-port="0" />
|
|
<edge from-layer="807" from-port="0" to-layer="808" to-port="0" />
|
|
<edge from-layer="808" from-port="1" to-layer="810" to-port="0" />
|
|
<edge from-layer="809" from-port="0" to-layer="810" to-port="1" />
|
|
<edge from-layer="810" from-port="2" to-layer="811" to-port="1" />
|
|
<edge from-layer="811" from-port="2" to-layer="813" to-port="0" />
|
|
<edge from-layer="812" from-port="0" to-layer="813" to-port="1" />
|
|
<edge from-layer="813" from-port="2" to-layer="815" to-port="0" />
|
|
<edge from-layer="814" from-port="0" to-layer="815" to-port="1" />
|
|
<edge from-layer="815" from-port="2" to-layer="817" to-port="0" />
|
|
<edge from-layer="816" from-port="0" to-layer="817" to-port="1" />
|
|
<edge from-layer="817" from-port="2" to-layer="818" to-port="1" />
|
|
<edge from-layer="818" from-port="2" to-layer="820" to-port="0" />
|
|
<edge from-layer="819" from-port="0" to-layer="820" to-port="1" />
|
|
<edge from-layer="820" from-port="2" to-layer="822" to-port="0" />
|
|
<edge from-layer="821" from-port="0" to-layer="822" to-port="1" />
|
|
<edge from-layer="822" from-port="2" to-layer="824" to-port="0" />
|
|
<edge from-layer="823" from-port="0" to-layer="824" to-port="1" />
|
|
<edge from-layer="824" from-port="2" to-layer="829" to-port="0" />
|
|
<edge from-layer="825" from-port="0" to-layer="829" to-port="1" />
|
|
<edge from-layer="826" from-port="0" to-layer="829" to-port="2" />
|
|
<edge from-layer="827" from-port="0" to-layer="829" to-port="3" />
|
|
<edge from-layer="828" from-port="0" to-layer="829" to-port="4" />
|
|
<edge from-layer="829" from-port="5" to-layer="834" to-port="0" />
|
|
<edge from-layer="830" from-port="0" to-layer="831" to-port="0" />
|
|
<edge from-layer="831" from-port="1" to-layer="833" to-port="0" />
|
|
<edge from-layer="832" from-port="0" to-layer="833" to-port="1" />
|
|
<edge from-layer="833" from-port="2" to-layer="834" to-port="1" />
|
|
<edge from-layer="834" from-port="2" to-layer="836" to-port="0" />
|
|
<edge from-layer="835" from-port="0" to-layer="836" to-port="1" />
|
|
<edge from-layer="836" from-port="2" to-layer="837" to-port="0" />
|
|
<edge from-layer="837" from-port="2" to-layer="839" to-port="0" />
|
|
<edge from-layer="838" from-port="0" to-layer="839" to-port="1" />
|
|
<edge from-layer="839" from-port="2" to-layer="841" to-port="0" />
|
|
<edge from-layer="840" from-port="0" to-layer="841" to-port="1" />
|
|
<edge from-layer="841" from-port="2" to-layer="843" to-port="0" />
|
|
<edge from-layer="842" from-port="0" to-layer="843" to-port="1" />
|
|
<edge from-layer="843" from-port="2" to-layer="845" to-port="0" />
|
|
<edge from-layer="843" from-port="2" to-layer="873" to-port="1" />
|
|
<edge from-layer="844" from-port="0" to-layer="845" to-port="1" />
|
|
<edge from-layer="845" from-port="2" to-layer="850" to-port="0" />
|
|
<edge from-layer="846" from-port="0" to-layer="850" to-port="1" />
|
|
<edge from-layer="847" from-port="0" to-layer="850" to-port="2" />
|
|
<edge from-layer="848" from-port="0" to-layer="850" to-port="3" />
|
|
<edge from-layer="849" from-port="0" to-layer="850" to-port="4" />
|
|
<edge from-layer="850" from-port="5" to-layer="855" to-port="0" />
|
|
<edge from-layer="851" from-port="0" to-layer="852" to-port="0" />
|
|
<edge from-layer="852" from-port="1" to-layer="854" to-port="0" />
|
|
<edge from-layer="853" from-port="0" to-layer="854" to-port="1" />
|
|
<edge from-layer="854" from-port="2" to-layer="855" to-port="1" />
|
|
<edge from-layer="855" from-port="2" to-layer="857" to-port="0" />
|
|
<edge from-layer="856" from-port="0" to-layer="857" to-port="1" />
|
|
<edge from-layer="857" from-port="2" to-layer="858" to-port="0" />
|
|
<edge from-layer="858" from-port="1" to-layer="860" to-port="0" />
|
|
<edge from-layer="859" from-port="0" to-layer="860" to-port="1" />
|
|
<edge from-layer="860" from-port="2" to-layer="865" to-port="0" />
|
|
<edge from-layer="861" from-port="0" to-layer="865" to-port="1" />
|
|
<edge from-layer="862" from-port="0" to-layer="865" to-port="2" />
|
|
<edge from-layer="863" from-port="0" to-layer="865" to-port="3" />
|
|
<edge from-layer="864" from-port="0" to-layer="865" to-port="4" />
|
|
<edge from-layer="865" from-port="5" to-layer="870" to-port="0" />
|
|
<edge from-layer="866" from-port="0" to-layer="867" to-port="0" />
|
|
<edge from-layer="867" from-port="1" to-layer="869" to-port="0" />
|
|
<edge from-layer="868" from-port="0" to-layer="869" to-port="1" />
|
|
<edge from-layer="869" from-port="2" to-layer="870" to-port="1" />
|
|
<edge from-layer="870" from-port="2" to-layer="872" to-port="0" />
|
|
<edge from-layer="871" from-port="0" to-layer="872" to-port="1" />
|
|
<edge from-layer="872" from-port="2" to-layer="873" to-port="0" />
|
|
<edge from-layer="873" from-port="2" to-layer="875" to-port="0" />
|
|
<edge from-layer="874" from-port="0" to-layer="875" to-port="1" />
|
|
<edge from-layer="875" from-port="2" to-layer="877" to-port="0" />
|
|
<edge from-layer="876" from-port="0" to-layer="877" to-port="1" />
|
|
<edge from-layer="877" from-port="2" to-layer="879" to-port="0" />
|
|
<edge from-layer="878" from-port="0" to-layer="879" to-port="1" />
|
|
<edge from-layer="879" from-port="2" to-layer="881" to-port="0" />
|
|
<edge from-layer="879" from-port="2" to-layer="954" to-port="1" />
|
|
<edge from-layer="880" from-port="0" to-layer="881" to-port="1" />
|
|
<edge from-layer="881" from-port="2" to-layer="886" to-port="0" />
|
|
<edge from-layer="882" from-port="0" to-layer="886" to-port="1" />
|
|
<edge from-layer="883" from-port="0" to-layer="886" to-port="2" />
|
|
<edge from-layer="884" from-port="0" to-layer="886" to-port="3" />
|
|
<edge from-layer="885" from-port="0" to-layer="886" to-port="4" />
|
|
<edge from-layer="886" from-port="5" to-layer="891" to-port="0" />
|
|
<edge from-layer="886" from-port="5" to-layer="907" to-port="0" />
|
|
<edge from-layer="886" from-port="5" to-layer="928" to-port="0" />
|
|
<edge from-layer="887" from-port="0" to-layer="888" to-port="0" />
|
|
<edge from-layer="888" from-port="1" to-layer="890" to-port="0" />
|
|
<edge from-layer="889" from-port="0" to-layer="890" to-port="1" />
|
|
<edge from-layer="890" from-port="2" to-layer="891" to-port="1" />
|
|
<edge from-layer="891" from-port="2" to-layer="893" to-port="0" />
|
|
<edge from-layer="892" from-port="0" to-layer="893" to-port="1" />
|
|
<edge from-layer="893" from-port="2" to-layer="898" to-port="0" />
|
|
<edge from-layer="894" from-port="0" to-layer="898" to-port="1" />
|
|
<edge from-layer="895" from-port="0" to-layer="898" to-port="2" />
|
|
<edge from-layer="896" from-port="0" to-layer="898" to-port="3" />
|
|
<edge from-layer="897" from-port="0" to-layer="898" to-port="4" />
|
|
<edge from-layer="898" from-port="5" to-layer="900" to-port="0" />
|
|
<edge from-layer="899" from-port="0" to-layer="900" to-port="1" />
|
|
<edge from-layer="900" from-port="2" to-layer="902" to-port="0" />
|
|
<edge from-layer="901" from-port="0" to-layer="902" to-port="1" />
|
|
<edge from-layer="902" from-port="2" to-layer="919" to-port="0" />
|
|
<edge from-layer="903" from-port="0" to-layer="904" to-port="0" />
|
|
<edge from-layer="904" from-port="1" to-layer="906" to-port="0" />
|
|
<edge from-layer="905" from-port="0" to-layer="906" to-port="1" />
|
|
<edge from-layer="906" from-port="2" to-layer="907" to-port="1" />
|
|
<edge from-layer="907" from-port="2" to-layer="909" to-port="0" />
|
|
<edge from-layer="908" from-port="0" to-layer="909" to-port="1" />
|
|
<edge from-layer="909" from-port="2" to-layer="914" to-port="0" />
|
|
<edge from-layer="910" from-port="0" to-layer="914" to-port="1" />
|
|
<edge from-layer="911" from-port="0" to-layer="914" to-port="2" />
|
|
<edge from-layer="912" from-port="0" to-layer="914" to-port="3" />
|
|
<edge from-layer="913" from-port="0" to-layer="914" to-port="4" />
|
|
<edge from-layer="914" from-port="5" to-layer="916" to-port="0" />
|
|
<edge from-layer="915" from-port="0" to-layer="916" to-port="1" />
|
|
<edge from-layer="916" from-port="2" to-layer="918" to-port="0" />
|
|
<edge from-layer="917" from-port="0" to-layer="918" to-port="1" />
|
|
<edge from-layer="918" from-port="2" to-layer="919" to-port="1" />
|
|
<edge from-layer="919" from-port="2" to-layer="921" to-port="0" />
|
|
<edge from-layer="920" from-port="0" to-layer="921" to-port="1" />
|
|
<edge from-layer="921" from-port="2" to-layer="922" to-port="0" />
|
|
<edge from-layer="922" from-port="2" to-layer="923" to-port="0" />
|
|
<edge from-layer="923" from-port="1" to-layer="935" to-port="0" />
|
|
<edge from-layer="924" from-port="0" to-layer="925" to-port="0" />
|
|
<edge from-layer="925" from-port="1" to-layer="927" to-port="0" />
|
|
<edge from-layer="926" from-port="0" to-layer="927" to-port="1" />
|
|
<edge from-layer="927" from-port="2" to-layer="928" to-port="1" />
|
|
<edge from-layer="928" from-port="2" to-layer="930" to-port="0" />
|
|
<edge from-layer="929" from-port="0" to-layer="930" to-port="1" />
|
|
<edge from-layer="930" from-port="2" to-layer="932" to-port="0" />
|
|
<edge from-layer="931" from-port="0" to-layer="932" to-port="1" />
|
|
<edge from-layer="932" from-port="2" to-layer="934" to-port="0" />
|
|
<edge from-layer="933" from-port="0" to-layer="934" to-port="1" />
|
|
<edge from-layer="934" from-port="2" to-layer="935" to-port="1" />
|
|
<edge from-layer="935" from-port="2" to-layer="937" to-port="0" />
|
|
<edge from-layer="936" from-port="0" to-layer="937" to-port="1" />
|
|
<edge from-layer="937" from-port="2" to-layer="939" to-port="0" />
|
|
<edge from-layer="938" from-port="0" to-layer="939" to-port="1" />
|
|
<edge from-layer="939" from-port="2" to-layer="941" to-port="0" />
|
|
<edge from-layer="940" from-port="0" to-layer="941" to-port="1" />
|
|
<edge from-layer="941" from-port="2" to-layer="946" to-port="0" />
|
|
<edge from-layer="942" from-port="0" to-layer="946" to-port="1" />
|
|
<edge from-layer="943" from-port="0" to-layer="946" to-port="2" />
|
|
<edge from-layer="944" from-port="0" to-layer="946" to-port="3" />
|
|
<edge from-layer="945" from-port="0" to-layer="946" to-port="4" />
|
|
<edge from-layer="946" from-port="5" to-layer="951" to-port="0" />
|
|
<edge from-layer="947" from-port="0" to-layer="948" to-port="0" />
|
|
<edge from-layer="948" from-port="1" to-layer="950" to-port="0" />
|
|
<edge from-layer="949" from-port="0" to-layer="950" to-port="1" />
|
|
<edge from-layer="950" from-port="2" to-layer="951" to-port="1" />
|
|
<edge from-layer="951" from-port="2" to-layer="953" to-port="0" />
|
|
<edge from-layer="952" from-port="0" to-layer="953" to-port="1" />
|
|
<edge from-layer="953" from-port="2" to-layer="954" to-port="0" />
|
|
<edge from-layer="954" from-port="2" to-layer="956" to-port="0" />
|
|
<edge from-layer="955" from-port="0" to-layer="956" to-port="1" />
|
|
<edge from-layer="956" from-port="2" to-layer="958" to-port="0" />
|
|
<edge from-layer="957" from-port="0" to-layer="958" to-port="1" />
|
|
<edge from-layer="958" from-port="2" to-layer="960" to-port="0" />
|
|
<edge from-layer="959" from-port="0" to-layer="960" to-port="1" />
|
|
<edge from-layer="960" from-port="2" to-layer="962" to-port="0" />
|
|
<edge from-layer="960" from-port="2" to-layer="990" to-port="1" />
|
|
<edge from-layer="961" from-port="0" to-layer="962" to-port="1" />
|
|
<edge from-layer="962" from-port="2" to-layer="967" to-port="0" />
|
|
<edge from-layer="963" from-port="0" to-layer="967" to-port="1" />
|
|
<edge from-layer="964" from-port="0" to-layer="967" to-port="2" />
|
|
<edge from-layer="965" from-port="0" to-layer="967" to-port="3" />
|
|
<edge from-layer="966" from-port="0" to-layer="967" to-port="4" />
|
|
<edge from-layer="967" from-port="5" to-layer="972" to-port="0" />
|
|
<edge from-layer="968" from-port="0" to-layer="969" to-port="0" />
|
|
<edge from-layer="969" from-port="1" to-layer="971" to-port="0" />
|
|
<edge from-layer="970" from-port="0" to-layer="971" to-port="1" />
|
|
<edge from-layer="971" from-port="2" to-layer="972" to-port="1" />
|
|
<edge from-layer="972" from-port="2" to-layer="974" to-port="0" />
|
|
<edge from-layer="973" from-port="0" to-layer="974" to-port="1" />
|
|
<edge from-layer="974" from-port="2" to-layer="975" to-port="0" />
|
|
<edge from-layer="975" from-port="1" to-layer="977" to-port="0" />
|
|
<edge from-layer="976" from-port="0" to-layer="977" to-port="1" />
|
|
<edge from-layer="977" from-port="2" to-layer="982" to-port="0" />
|
|
<edge from-layer="978" from-port="0" to-layer="982" to-port="1" />
|
|
<edge from-layer="979" from-port="0" to-layer="982" to-port="2" />
|
|
<edge from-layer="980" from-port="0" to-layer="982" to-port="3" />
|
|
<edge from-layer="981" from-port="0" to-layer="982" to-port="4" />
|
|
<edge from-layer="982" from-port="5" to-layer="987" to-port="0" />
|
|
<edge from-layer="983" from-port="0" to-layer="984" to-port="0" />
|
|
<edge from-layer="984" from-port="1" to-layer="986" to-port="0" />
|
|
<edge from-layer="985" from-port="0" to-layer="986" to-port="1" />
|
|
<edge from-layer="986" from-port="2" to-layer="987" to-port="1" />
|
|
<edge from-layer="987" from-port="2" to-layer="989" to-port="0" />
|
|
<edge from-layer="988" from-port="0" to-layer="989" to-port="1" />
|
|
<edge from-layer="989" from-port="2" to-layer="990" to-port="0" />
|
|
<edge from-layer="990" from-port="2" to-layer="992" to-port="0" />
|
|
<edge from-layer="991" from-port="0" to-layer="992" to-port="1" />
|
|
<edge from-layer="992" from-port="2" to-layer="994" to-port="0" />
|
|
<edge from-layer="993" from-port="0" to-layer="994" to-port="1" />
|
|
<edge from-layer="994" from-port="2" to-layer="996" to-port="0" />
|
|
<edge from-layer="995" from-port="0" to-layer="996" to-port="1" />
|
|
<edge from-layer="996" from-port="2" to-layer="998" to-port="0" />
|
|
<edge from-layer="996" from-port="2" to-layer="1071" to-port="1" />
|
|
<edge from-layer="997" from-port="0" to-layer="998" to-port="1" />
|
|
<edge from-layer="998" from-port="2" to-layer="1003" to-port="0" />
|
|
<edge from-layer="999" from-port="0" to-layer="1003" to-port="1" />
|
|
<edge from-layer="1000" from-port="0" to-layer="1003" to-port="2" />
|
|
<edge from-layer="1001" from-port="0" to-layer="1003" to-port="3" />
|
|
<edge from-layer="1002" from-port="0" to-layer="1003" to-port="4" />
|
|
<edge from-layer="1003" from-port="5" to-layer="1008" to-port="0" />
|
|
<edge from-layer="1003" from-port="5" to-layer="1024" to-port="0" />
|
|
<edge from-layer="1003" from-port="5" to-layer="1045" to-port="0" />
|
|
<edge from-layer="1004" from-port="0" to-layer="1005" to-port="0" />
|
|
<edge from-layer="1005" from-port="1" to-layer="1007" to-port="0" />
|
|
<edge from-layer="1006" from-port="0" to-layer="1007" to-port="1" />
|
|
<edge from-layer="1007" from-port="2" to-layer="1008" to-port="1" />
|
|
<edge from-layer="1008" from-port="2" to-layer="1010" to-port="0" />
|
|
<edge from-layer="1009" from-port="0" to-layer="1010" to-port="1" />
|
|
<edge from-layer="1010" from-port="2" to-layer="1015" to-port="0" />
|
|
<edge from-layer="1011" from-port="0" to-layer="1015" to-port="1" />
|
|
<edge from-layer="1012" from-port="0" to-layer="1015" to-port="2" />
|
|
<edge from-layer="1013" from-port="0" to-layer="1015" to-port="3" />
|
|
<edge from-layer="1014" from-port="0" to-layer="1015" to-port="4" />
|
|
<edge from-layer="1015" from-port="5" to-layer="1017" to-port="0" />
|
|
<edge from-layer="1016" from-port="0" to-layer="1017" to-port="1" />
|
|
<edge from-layer="1017" from-port="2" to-layer="1019" to-port="0" />
|
|
<edge from-layer="1018" from-port="0" to-layer="1019" to-port="1" />
|
|
<edge from-layer="1019" from-port="2" to-layer="1036" to-port="0" />
|
|
<edge from-layer="1020" from-port="0" to-layer="1021" to-port="0" />
|
|
<edge from-layer="1021" from-port="1" to-layer="1023" to-port="0" />
|
|
<edge from-layer="1022" from-port="0" to-layer="1023" to-port="1" />
|
|
<edge from-layer="1023" from-port="2" to-layer="1024" to-port="1" />
|
|
<edge from-layer="1024" from-port="2" to-layer="1026" to-port="0" />
|
|
<edge from-layer="1025" from-port="0" to-layer="1026" to-port="1" />
|
|
<edge from-layer="1026" from-port="2" to-layer="1031" to-port="0" />
|
|
<edge from-layer="1027" from-port="0" to-layer="1031" to-port="1" />
|
|
<edge from-layer="1028" from-port="0" to-layer="1031" to-port="2" />
|
|
<edge from-layer="1029" from-port="0" to-layer="1031" to-port="3" />
|
|
<edge from-layer="1030" from-port="0" to-layer="1031" to-port="4" />
|
|
<edge from-layer="1031" from-port="5" to-layer="1033" to-port="0" />
|
|
<edge from-layer="1032" from-port="0" to-layer="1033" to-port="1" />
|
|
<edge from-layer="1033" from-port="2" to-layer="1035" to-port="0" />
|
|
<edge from-layer="1034" from-port="0" to-layer="1035" to-port="1" />
|
|
<edge from-layer="1035" from-port="2" to-layer="1036" to-port="1" />
|
|
<edge from-layer="1036" from-port="2" to-layer="1038" to-port="0" />
|
|
<edge from-layer="1037" from-port="0" to-layer="1038" to-port="1" />
|
|
<edge from-layer="1038" from-port="2" to-layer="1039" to-port="0" />
|
|
<edge from-layer="1039" from-port="2" to-layer="1040" to-port="0" />
|
|
<edge from-layer="1040" from-port="1" to-layer="1052" to-port="0" />
|
|
<edge from-layer="1041" from-port="0" to-layer="1042" to-port="0" />
|
|
<edge from-layer="1042" from-port="1" to-layer="1044" to-port="0" />
|
|
<edge from-layer="1043" from-port="0" to-layer="1044" to-port="1" />
|
|
<edge from-layer="1044" from-port="2" to-layer="1045" to-port="1" />
|
|
<edge from-layer="1045" from-port="2" to-layer="1047" to-port="0" />
|
|
<edge from-layer="1046" from-port="0" to-layer="1047" to-port="1" />
|
|
<edge from-layer="1047" from-port="2" to-layer="1049" to-port="0" />
|
|
<edge from-layer="1048" from-port="0" to-layer="1049" to-port="1" />
|
|
<edge from-layer="1049" from-port="2" to-layer="1051" to-port="0" />
|
|
<edge from-layer="1050" from-port="0" to-layer="1051" to-port="1" />
|
|
<edge from-layer="1051" from-port="2" to-layer="1052" to-port="1" />
|
|
<edge from-layer="1052" from-port="2" to-layer="1054" to-port="0" />
|
|
<edge from-layer="1053" from-port="0" to-layer="1054" to-port="1" />
|
|
<edge from-layer="1054" from-port="2" to-layer="1056" to-port="0" />
|
|
<edge from-layer="1055" from-port="0" to-layer="1056" to-port="1" />
|
|
<edge from-layer="1056" from-port="2" to-layer="1058" to-port="0" />
|
|
<edge from-layer="1057" from-port="0" to-layer="1058" to-port="1" />
|
|
<edge from-layer="1058" from-port="2" to-layer="1063" to-port="0" />
|
|
<edge from-layer="1059" from-port="0" to-layer="1063" to-port="1" />
|
|
<edge from-layer="1060" from-port="0" to-layer="1063" to-port="2" />
|
|
<edge from-layer="1061" from-port="0" to-layer="1063" to-port="3" />
|
|
<edge from-layer="1062" from-port="0" to-layer="1063" to-port="4" />
|
|
<edge from-layer="1063" from-port="5" to-layer="1068" to-port="0" />
|
|
<edge from-layer="1064" from-port="0" to-layer="1065" to-port="0" />
|
|
<edge from-layer="1065" from-port="1" to-layer="1067" to-port="0" />
|
|
<edge from-layer="1066" from-port="0" to-layer="1067" to-port="1" />
|
|
<edge from-layer="1067" from-port="2" to-layer="1068" to-port="1" />
|
|
<edge from-layer="1068" from-port="2" to-layer="1070" to-port="0" />
|
|
<edge from-layer="1069" from-port="0" to-layer="1070" to-port="1" />
|
|
<edge from-layer="1070" from-port="2" to-layer="1071" to-port="0" />
|
|
<edge from-layer="1071" from-port="2" to-layer="1073" to-port="0" />
|
|
<edge from-layer="1072" from-port="0" to-layer="1073" to-port="1" />
|
|
<edge from-layer="1073" from-port="2" to-layer="1075" to-port="0" />
|
|
<edge from-layer="1074" from-port="0" to-layer="1075" to-port="1" />
|
|
<edge from-layer="1075" from-port="2" to-layer="1077" to-port="0" />
|
|
<edge from-layer="1076" from-port="0" to-layer="1077" to-port="1" />
|
|
<edge from-layer="1077" from-port="2" to-layer="1079" to-port="0" />
|
|
<edge from-layer="1077" from-port="2" to-layer="1107" to-port="1" />
|
|
<edge from-layer="1078" from-port="0" to-layer="1079" to-port="1" />
|
|
<edge from-layer="1079" from-port="2" to-layer="1084" to-port="0" />
|
|
<edge from-layer="1080" from-port="0" to-layer="1084" to-port="1" />
|
|
<edge from-layer="1081" from-port="0" to-layer="1084" to-port="2" />
|
|
<edge from-layer="1082" from-port="0" to-layer="1084" to-port="3" />
|
|
<edge from-layer="1083" from-port="0" to-layer="1084" to-port="4" />
|
|
<edge from-layer="1084" from-port="5" to-layer="1089" to-port="0" />
|
|
<edge from-layer="1085" from-port="0" to-layer="1086" to-port="0" />
|
|
<edge from-layer="1086" from-port="1" to-layer="1088" to-port="0" />
|
|
<edge from-layer="1087" from-port="0" to-layer="1088" to-port="1" />
|
|
<edge from-layer="1088" from-port="2" to-layer="1089" to-port="1" />
|
|
<edge from-layer="1089" from-port="2" to-layer="1091" to-port="0" />
|
|
<edge from-layer="1090" from-port="0" to-layer="1091" to-port="1" />
|
|
<edge from-layer="1091" from-port="2" to-layer="1092" to-port="0" />
|
|
<edge from-layer="1092" from-port="1" to-layer="1094" to-port="0" />
|
|
<edge from-layer="1093" from-port="0" to-layer="1094" to-port="1" />
|
|
<edge from-layer="1094" from-port="2" to-layer="1099" to-port="0" />
|
|
<edge from-layer="1095" from-port="0" to-layer="1099" to-port="1" />
|
|
<edge from-layer="1096" from-port="0" to-layer="1099" to-port="2" />
|
|
<edge from-layer="1097" from-port="0" to-layer="1099" to-port="3" />
|
|
<edge from-layer="1098" from-port="0" to-layer="1099" to-port="4" />
|
|
<edge from-layer="1099" from-port="5" to-layer="1104" to-port="0" />
|
|
<edge from-layer="1100" from-port="0" to-layer="1101" to-port="0" />
|
|
<edge from-layer="1101" from-port="1" to-layer="1103" to-port="0" />
|
|
<edge from-layer="1102" from-port="0" to-layer="1103" to-port="1" />
|
|
<edge from-layer="1103" from-port="2" to-layer="1104" to-port="1" />
|
|
<edge from-layer="1104" from-port="2" to-layer="1106" to-port="0" />
|
|
<edge from-layer="1105" from-port="0" to-layer="1106" to-port="1" />
|
|
<edge from-layer="1106" from-port="2" to-layer="1107" to-port="0" />
|
|
<edge from-layer="1107" from-port="2" to-layer="1109" to-port="0" />
|
|
<edge from-layer="1108" from-port="0" to-layer="1109" to-port="1" />
|
|
<edge from-layer="1109" from-port="2" to-layer="1111" to-port="0" />
|
|
<edge from-layer="1110" from-port="0" to-layer="1111" to-port="1" />
|
|
<edge from-layer="1111" from-port="2" to-layer="1113" to-port="0" />
|
|
<edge from-layer="1112" from-port="0" to-layer="1113" to-port="1" />
|
|
<edge from-layer="1113" from-port="2" to-layer="1115" to-port="0" />
|
|
<edge from-layer="1113" from-port="2" to-layer="1188" to-port="1" />
|
|
<edge from-layer="1114" from-port="0" to-layer="1115" to-port="1" />
|
|
<edge from-layer="1115" from-port="2" to-layer="1120" to-port="0" />
|
|
<edge from-layer="1116" from-port="0" to-layer="1120" to-port="1" />
|
|
<edge from-layer="1117" from-port="0" to-layer="1120" to-port="2" />
|
|
<edge from-layer="1118" from-port="0" to-layer="1120" to-port="3" />
|
|
<edge from-layer="1119" from-port="0" to-layer="1120" to-port="4" />
|
|
<edge from-layer="1120" from-port="5" to-layer="1125" to-port="0" />
|
|
<edge from-layer="1120" from-port="5" to-layer="1162" to-port="0" />
|
|
<edge from-layer="1120" from-port="5" to-layer="1141" to-port="0" />
|
|
<edge from-layer="1121" from-port="0" to-layer="1122" to-port="0" />
|
|
<edge from-layer="1122" from-port="1" to-layer="1124" to-port="0" />
|
|
<edge from-layer="1123" from-port="0" to-layer="1124" to-port="1" />
|
|
<edge from-layer="1124" from-port="2" to-layer="1125" to-port="1" />
|
|
<edge from-layer="1125" from-port="2" to-layer="1127" to-port="0" />
|
|
<edge from-layer="1126" from-port="0" to-layer="1127" to-port="1" />
|
|
<edge from-layer="1127" from-port="2" to-layer="1132" to-port="0" />
|
|
<edge from-layer="1128" from-port="0" to-layer="1132" to-port="1" />
|
|
<edge from-layer="1129" from-port="0" to-layer="1132" to-port="2" />
|
|
<edge from-layer="1130" from-port="0" to-layer="1132" to-port="3" />
|
|
<edge from-layer="1131" from-port="0" to-layer="1132" to-port="4" />
|
|
<edge from-layer="1132" from-port="5" to-layer="1134" to-port="0" />
|
|
<edge from-layer="1133" from-port="0" to-layer="1134" to-port="1" />
|
|
<edge from-layer="1134" from-port="2" to-layer="1136" to-port="0" />
|
|
<edge from-layer="1135" from-port="0" to-layer="1136" to-port="1" />
|
|
<edge from-layer="1136" from-port="2" to-layer="1153" to-port="0" />
|
|
<edge from-layer="1137" from-port="0" to-layer="1138" to-port="0" />
|
|
<edge from-layer="1138" from-port="1" to-layer="1140" to-port="0" />
|
|
<edge from-layer="1139" from-port="0" to-layer="1140" to-port="1" />
|
|
<edge from-layer="1140" from-port="2" to-layer="1141" to-port="1" />
|
|
<edge from-layer="1141" from-port="2" to-layer="1143" to-port="0" />
|
|
<edge from-layer="1142" from-port="0" to-layer="1143" to-port="1" />
|
|
<edge from-layer="1143" from-port="2" to-layer="1148" to-port="0" />
|
|
<edge from-layer="1144" from-port="0" to-layer="1148" to-port="1" />
|
|
<edge from-layer="1145" from-port="0" to-layer="1148" to-port="2" />
|
|
<edge from-layer="1146" from-port="0" to-layer="1148" to-port="3" />
|
|
<edge from-layer="1147" from-port="0" to-layer="1148" to-port="4" />
|
|
<edge from-layer="1148" from-port="5" to-layer="1150" to-port="0" />
|
|
<edge from-layer="1149" from-port="0" to-layer="1150" to-port="1" />
|
|
<edge from-layer="1150" from-port="2" to-layer="1152" to-port="0" />
|
|
<edge from-layer="1151" from-port="0" to-layer="1152" to-port="1" />
|
|
<edge from-layer="1152" from-port="2" to-layer="1153" to-port="1" />
|
|
<edge from-layer="1153" from-port="2" to-layer="1155" to-port="0" />
|
|
<edge from-layer="1154" from-port="0" to-layer="1155" to-port="1" />
|
|
<edge from-layer="1155" from-port="2" to-layer="1156" to-port="0" />
|
|
<edge from-layer="1156" from-port="2" to-layer="1157" to-port="0" />
|
|
<edge from-layer="1157" from-port="1" to-layer="1169" to-port="0" />
|
|
<edge from-layer="1158" from-port="0" to-layer="1159" to-port="0" />
|
|
<edge from-layer="1159" from-port="1" to-layer="1161" to-port="0" />
|
|
<edge from-layer="1160" from-port="0" to-layer="1161" to-port="1" />
|
|
<edge from-layer="1161" from-port="2" to-layer="1162" to-port="1" />
|
|
<edge from-layer="1162" from-port="2" to-layer="1164" to-port="0" />
|
|
<edge from-layer="1163" from-port="0" to-layer="1164" to-port="1" />
|
|
<edge from-layer="1164" from-port="2" to-layer="1166" to-port="0" />
|
|
<edge from-layer="1165" from-port="0" to-layer="1166" to-port="1" />
|
|
<edge from-layer="1166" from-port="2" to-layer="1168" to-port="0" />
|
|
<edge from-layer="1167" from-port="0" to-layer="1168" to-port="1" />
|
|
<edge from-layer="1168" from-port="2" to-layer="1169" to-port="1" />
|
|
<edge from-layer="1169" from-port="2" to-layer="1171" to-port="0" />
|
|
<edge from-layer="1170" from-port="0" to-layer="1171" to-port="1" />
|
|
<edge from-layer="1171" from-port="2" to-layer="1173" to-port="0" />
|
|
<edge from-layer="1172" from-port="0" to-layer="1173" to-port="1" />
|
|
<edge from-layer="1173" from-port="2" to-layer="1175" to-port="0" />
|
|
<edge from-layer="1174" from-port="0" to-layer="1175" to-port="1" />
|
|
<edge from-layer="1175" from-port="2" to-layer="1180" to-port="0" />
|
|
<edge from-layer="1176" from-port="0" to-layer="1180" to-port="1" />
|
|
<edge from-layer="1177" from-port="0" to-layer="1180" to-port="2" />
|
|
<edge from-layer="1178" from-port="0" to-layer="1180" to-port="3" />
|
|
<edge from-layer="1179" from-port="0" to-layer="1180" to-port="4" />
|
|
<edge from-layer="1180" from-port="5" to-layer="1185" to-port="0" />
|
|
<edge from-layer="1181" from-port="0" to-layer="1182" to-port="0" />
|
|
<edge from-layer="1182" from-port="1" to-layer="1184" to-port="0" />
|
|
<edge from-layer="1183" from-port="0" to-layer="1184" to-port="1" />
|
|
<edge from-layer="1184" from-port="2" to-layer="1185" to-port="1" />
|
|
<edge from-layer="1185" from-port="2" to-layer="1187" to-port="0" />
|
|
<edge from-layer="1186" from-port="0" to-layer="1187" to-port="1" />
|
|
<edge from-layer="1187" from-port="2" to-layer="1188" to-port="0" />
|
|
<edge from-layer="1188" from-port="2" to-layer="1190" to-port="0" />
|
|
<edge from-layer="1189" from-port="0" to-layer="1190" to-port="1" />
|
|
<edge from-layer="1190" from-port="2" to-layer="1192" to-port="0" />
|
|
<edge from-layer="1191" from-port="0" to-layer="1192" to-port="1" />
|
|
<edge from-layer="1192" from-port="2" to-layer="1194" to-port="0" />
|
|
<edge from-layer="1193" from-port="0" to-layer="1194" to-port="1" />
|
|
<edge from-layer="1194" from-port="2" to-layer="1196" to-port="0" />
|
|
<edge from-layer="1194" from-port="2" to-layer="1224" to-port="1" />
|
|
<edge from-layer="1195" from-port="0" to-layer="1196" to-port="1" />
|
|
<edge from-layer="1196" from-port="2" to-layer="1201" to-port="0" />
|
|
<edge from-layer="1197" from-port="0" to-layer="1201" to-port="1" />
|
|
<edge from-layer="1198" from-port="0" to-layer="1201" to-port="2" />
|
|
<edge from-layer="1199" from-port="0" to-layer="1201" to-port="3" />
|
|
<edge from-layer="1200" from-port="0" to-layer="1201" to-port="4" />
|
|
<edge from-layer="1201" from-port="5" to-layer="1206" to-port="0" />
|
|
<edge from-layer="1202" from-port="0" to-layer="1203" to-port="0" />
|
|
<edge from-layer="1203" from-port="1" to-layer="1205" to-port="0" />
|
|
<edge from-layer="1204" from-port="0" to-layer="1205" to-port="1" />
|
|
<edge from-layer="1205" from-port="2" to-layer="1206" to-port="1" />
|
|
<edge from-layer="1206" from-port="2" to-layer="1208" to-port="0" />
|
|
<edge from-layer="1207" from-port="0" to-layer="1208" to-port="1" />
|
|
<edge from-layer="1208" from-port="2" to-layer="1209" to-port="0" />
|
|
<edge from-layer="1209" from-port="1" to-layer="1211" to-port="0" />
|
|
<edge from-layer="1210" from-port="0" to-layer="1211" to-port="1" />
|
|
<edge from-layer="1211" from-port="2" to-layer="1216" to-port="0" />
|
|
<edge from-layer="1212" from-port="0" to-layer="1216" to-port="1" />
|
|
<edge from-layer="1213" from-port="0" to-layer="1216" to-port="2" />
|
|
<edge from-layer="1214" from-port="0" to-layer="1216" to-port="3" />
|
|
<edge from-layer="1215" from-port="0" to-layer="1216" to-port="4" />
|
|
<edge from-layer="1216" from-port="5" to-layer="1221" to-port="0" />
|
|
<edge from-layer="1217" from-port="0" to-layer="1218" to-port="0" />
|
|
<edge from-layer="1218" from-port="1" to-layer="1220" to-port="0" />
|
|
<edge from-layer="1219" from-port="0" to-layer="1220" to-port="1" />
|
|
<edge from-layer="1220" from-port="2" to-layer="1221" to-port="1" />
|
|
<edge from-layer="1221" from-port="2" to-layer="1223" to-port="0" />
|
|
<edge from-layer="1222" from-port="0" to-layer="1223" to-port="1" />
|
|
<edge from-layer="1223" from-port="2" to-layer="1224" to-port="0" />
|
|
<edge from-layer="1224" from-port="2" to-layer="1226" to-port="0" />
|
|
<edge from-layer="1225" from-port="0" to-layer="1226" to-port="1" />
|
|
<edge from-layer="1226" from-port="2" to-layer="1228" to-port="0" />
|
|
<edge from-layer="1227" from-port="0" to-layer="1228" to-port="1" />
|
|
<edge from-layer="1228" from-port="2" to-layer="1230" to-port="0" />
|
|
<edge from-layer="1229" from-port="0" to-layer="1230" to-port="1" />
|
|
<edge from-layer="1230" from-port="2" to-layer="1232" to-port="0" />
|
|
<edge from-layer="1230" from-port="2" to-layer="1305" to-port="1" />
|
|
<edge from-layer="1231" from-port="0" to-layer="1232" to-port="1" />
|
|
<edge from-layer="1232" from-port="2" to-layer="1237" to-port="0" />
|
|
<edge from-layer="1233" from-port="0" to-layer="1237" to-port="1" />
|
|
<edge from-layer="1234" from-port="0" to-layer="1237" to-port="2" />
|
|
<edge from-layer="1235" from-port="0" to-layer="1237" to-port="3" />
|
|
<edge from-layer="1236" from-port="0" to-layer="1237" to-port="4" />
|
|
<edge from-layer="1237" from-port="5" to-layer="1242" to-port="0" />
|
|
<edge from-layer="1237" from-port="5" to-layer="1279" to-port="0" />
|
|
<edge from-layer="1237" from-port="5" to-layer="1258" to-port="0" />
|
|
<edge from-layer="1238" from-port="0" to-layer="1239" to-port="0" />
|
|
<edge from-layer="1239" from-port="1" to-layer="1241" to-port="0" />
|
|
<edge from-layer="1240" from-port="0" to-layer="1241" to-port="1" />
|
|
<edge from-layer="1241" from-port="2" to-layer="1242" to-port="1" />
|
|
<edge from-layer="1242" from-port="2" to-layer="1244" to-port="0" />
|
|
<edge from-layer="1243" from-port="0" to-layer="1244" to-port="1" />
|
|
<edge from-layer="1244" from-port="2" to-layer="1249" to-port="0" />
|
|
<edge from-layer="1245" from-port="0" to-layer="1249" to-port="1" />
|
|
<edge from-layer="1246" from-port="0" to-layer="1249" to-port="2" />
|
|
<edge from-layer="1247" from-port="0" to-layer="1249" to-port="3" />
|
|
<edge from-layer="1248" from-port="0" to-layer="1249" to-port="4" />
|
|
<edge from-layer="1249" from-port="5" to-layer="1251" to-port="0" />
|
|
<edge from-layer="1250" from-port="0" to-layer="1251" to-port="1" />
|
|
<edge from-layer="1251" from-port="2" to-layer="1253" to-port="0" />
|
|
<edge from-layer="1252" from-port="0" to-layer="1253" to-port="1" />
|
|
<edge from-layer="1253" from-port="2" to-layer="1270" to-port="0" />
|
|
<edge from-layer="1254" from-port="0" to-layer="1255" to-port="0" />
|
|
<edge from-layer="1255" from-port="1" to-layer="1257" to-port="0" />
|
|
<edge from-layer="1256" from-port="0" to-layer="1257" to-port="1" />
|
|
<edge from-layer="1257" from-port="2" to-layer="1258" to-port="1" />
|
|
<edge from-layer="1258" from-port="2" to-layer="1260" to-port="0" />
|
|
<edge from-layer="1259" from-port="0" to-layer="1260" to-port="1" />
|
|
<edge from-layer="1260" from-port="2" to-layer="1265" to-port="0" />
|
|
<edge from-layer="1261" from-port="0" to-layer="1265" to-port="1" />
|
|
<edge from-layer="1262" from-port="0" to-layer="1265" to-port="2" />
|
|
<edge from-layer="1263" from-port="0" to-layer="1265" to-port="3" />
|
|
<edge from-layer="1264" from-port="0" to-layer="1265" to-port="4" />
|
|
<edge from-layer="1265" from-port="5" to-layer="1267" to-port="0" />
|
|
<edge from-layer="1266" from-port="0" to-layer="1267" to-port="1" />
|
|
<edge from-layer="1267" from-port="2" to-layer="1269" to-port="0" />
|
|
<edge from-layer="1268" from-port="0" to-layer="1269" to-port="1" />
|
|
<edge from-layer="1269" from-port="2" to-layer="1270" to-port="1" />
|
|
<edge from-layer="1270" from-port="2" to-layer="1272" to-port="0" />
|
|
<edge from-layer="1271" from-port="0" to-layer="1272" to-port="1" />
|
|
<edge from-layer="1272" from-port="2" to-layer="1273" to-port="0" />
|
|
<edge from-layer="1273" from-port="2" to-layer="1274" to-port="0" />
|
|
<edge from-layer="1274" from-port="1" to-layer="1286" to-port="0" />
|
|
<edge from-layer="1275" from-port="0" to-layer="1276" to-port="0" />
|
|
<edge from-layer="1276" from-port="1" to-layer="1278" to-port="0" />
|
|
<edge from-layer="1277" from-port="0" to-layer="1278" to-port="1" />
|
|
<edge from-layer="1278" from-port="2" to-layer="1279" to-port="1" />
|
|
<edge from-layer="1279" from-port="2" to-layer="1281" to-port="0" />
|
|
<edge from-layer="1280" from-port="0" to-layer="1281" to-port="1" />
|
|
<edge from-layer="1281" from-port="2" to-layer="1283" to-port="0" />
|
|
<edge from-layer="1282" from-port="0" to-layer="1283" to-port="1" />
|
|
<edge from-layer="1283" from-port="2" to-layer="1285" to-port="0" />
|
|
<edge from-layer="1284" from-port="0" to-layer="1285" to-port="1" />
|
|
<edge from-layer="1285" from-port="2" to-layer="1286" to-port="1" />
|
|
<edge from-layer="1286" from-port="2" to-layer="1288" to-port="0" />
|
|
<edge from-layer="1287" from-port="0" to-layer="1288" to-port="1" />
|
|
<edge from-layer="1288" from-port="2" to-layer="1290" to-port="0" />
|
|
<edge from-layer="1289" from-port="0" to-layer="1290" to-port="1" />
|
|
<edge from-layer="1290" from-port="2" to-layer="1292" to-port="0" />
|
|
<edge from-layer="1291" from-port="0" to-layer="1292" to-port="1" />
|
|
<edge from-layer="1292" from-port="2" to-layer="1297" to-port="0" />
|
|
<edge from-layer="1293" from-port="0" to-layer="1297" to-port="1" />
|
|
<edge from-layer="1294" from-port="0" to-layer="1297" to-port="2" />
|
|
<edge from-layer="1295" from-port="0" to-layer="1297" to-port="3" />
|
|
<edge from-layer="1296" from-port="0" to-layer="1297" to-port="4" />
|
|
<edge from-layer="1297" from-port="5" to-layer="1302" to-port="0" />
|
|
<edge from-layer="1298" from-port="0" to-layer="1299" to-port="0" />
|
|
<edge from-layer="1299" from-port="1" to-layer="1301" to-port="0" />
|
|
<edge from-layer="1300" from-port="0" to-layer="1301" to-port="1" />
|
|
<edge from-layer="1301" from-port="2" to-layer="1302" to-port="1" />
|
|
<edge from-layer="1302" from-port="2" to-layer="1304" to-port="0" />
|
|
<edge from-layer="1303" from-port="0" to-layer="1304" to-port="1" />
|
|
<edge from-layer="1304" from-port="2" to-layer="1305" to-port="0" />
|
|
<edge from-layer="1305" from-port="2" to-layer="1307" to-port="0" />
|
|
<edge from-layer="1306" from-port="0" to-layer="1307" to-port="1" />
|
|
<edge from-layer="1307" from-port="2" to-layer="1309" to-port="0" />
|
|
<edge from-layer="1308" from-port="0" to-layer="1309" to-port="1" />
|
|
<edge from-layer="1309" from-port="2" to-layer="1311" to-port="0" />
|
|
<edge from-layer="1310" from-port="0" to-layer="1311" to-port="1" />
|
|
<edge from-layer="1311" from-port="2" to-layer="1313" to-port="0" />
|
|
<edge from-layer="1311" from-port="2" to-layer="1341" to-port="1" />
|
|
<edge from-layer="1312" from-port="0" to-layer="1313" to-port="1" />
|
|
<edge from-layer="1313" from-port="2" to-layer="1318" to-port="0" />
|
|
<edge from-layer="1314" from-port="0" to-layer="1318" to-port="1" />
|
|
<edge from-layer="1315" from-port="0" to-layer="1318" to-port="2" />
|
|
<edge from-layer="1316" from-port="0" to-layer="1318" to-port="3" />
|
|
<edge from-layer="1317" from-port="0" to-layer="1318" to-port="4" />
|
|
<edge from-layer="1318" from-port="5" to-layer="1323" to-port="0" />
|
|
<edge from-layer="1319" from-port="0" to-layer="1320" to-port="0" />
|
|
<edge from-layer="1320" from-port="1" to-layer="1322" to-port="0" />
|
|
<edge from-layer="1321" from-port="0" to-layer="1322" to-port="1" />
|
|
<edge from-layer="1322" from-port="2" to-layer="1323" to-port="1" />
|
|
<edge from-layer="1323" from-port="2" to-layer="1325" to-port="0" />
|
|
<edge from-layer="1324" from-port="0" to-layer="1325" to-port="1" />
|
|
<edge from-layer="1325" from-port="2" to-layer="1326" to-port="0" />
|
|
<edge from-layer="1326" from-port="1" to-layer="1328" to-port="0" />
|
|
<edge from-layer="1327" from-port="0" to-layer="1328" to-port="1" />
|
|
<edge from-layer="1328" from-port="2" to-layer="1333" to-port="0" />
|
|
<edge from-layer="1329" from-port="0" to-layer="1333" to-port="1" />
|
|
<edge from-layer="1330" from-port="0" to-layer="1333" to-port="2" />
|
|
<edge from-layer="1331" from-port="0" to-layer="1333" to-port="3" />
|
|
<edge from-layer="1332" from-port="0" to-layer="1333" to-port="4" />
|
|
<edge from-layer="1333" from-port="5" to-layer="1338" to-port="0" />
|
|
<edge from-layer="1334" from-port="0" to-layer="1335" to-port="0" />
|
|
<edge from-layer="1335" from-port="1" to-layer="1337" to-port="0" />
|
|
<edge from-layer="1336" from-port="0" to-layer="1337" to-port="1" />
|
|
<edge from-layer="1337" from-port="2" to-layer="1338" to-port="1" />
|
|
<edge from-layer="1338" from-port="2" to-layer="1340" to-port="0" />
|
|
<edge from-layer="1339" from-port="0" to-layer="1340" to-port="1" />
|
|
<edge from-layer="1340" from-port="2" to-layer="1341" to-port="0" />
|
|
<edge from-layer="1341" from-port="2" to-layer="1343" to-port="0" />
|
|
<edge from-layer="1342" from-port="0" to-layer="1343" to-port="1" />
|
|
<edge from-layer="1343" from-port="2" to-layer="1345" to-port="0" />
|
|
<edge from-layer="1344" from-port="0" to-layer="1345" to-port="1" />
|
|
<edge from-layer="1345" from-port="2" to-layer="1347" to-port="0" />
|
|
<edge from-layer="1346" from-port="0" to-layer="1347" to-port="1" />
|
|
<edge from-layer="1347" from-port="2" to-layer="1349" to-port="0" />
|
|
<edge from-layer="1347" from-port="2" to-layer="1422" to-port="1" />
|
|
<edge from-layer="1348" from-port="0" to-layer="1349" to-port="1" />
|
|
<edge from-layer="1349" from-port="2" to-layer="1354" to-port="0" />
|
|
<edge from-layer="1350" from-port="0" to-layer="1354" to-port="1" />
|
|
<edge from-layer="1351" from-port="0" to-layer="1354" to-port="2" />
|
|
<edge from-layer="1352" from-port="0" to-layer="1354" to-port="3" />
|
|
<edge from-layer="1353" from-port="0" to-layer="1354" to-port="4" />
|
|
<edge from-layer="1354" from-port="5" to-layer="1359" to-port="0" />
|
|
<edge from-layer="1354" from-port="5" to-layer="1375" to-port="0" />
|
|
<edge from-layer="1354" from-port="5" to-layer="1396" to-port="0" />
|
|
<edge from-layer="1355" from-port="0" to-layer="1356" to-port="0" />
|
|
<edge from-layer="1356" from-port="1" to-layer="1358" to-port="0" />
|
|
<edge from-layer="1357" from-port="0" to-layer="1358" to-port="1" />
|
|
<edge from-layer="1358" from-port="2" to-layer="1359" to-port="1" />
|
|
<edge from-layer="1359" from-port="2" to-layer="1361" to-port="0" />
|
|
<edge from-layer="1360" from-port="0" to-layer="1361" to-port="1" />
|
|
<edge from-layer="1361" from-port="2" to-layer="1366" to-port="0" />
|
|
<edge from-layer="1362" from-port="0" to-layer="1366" to-port="1" />
|
|
<edge from-layer="1363" from-port="0" to-layer="1366" to-port="2" />
|
|
<edge from-layer="1364" from-port="0" to-layer="1366" to-port="3" />
|
|
<edge from-layer="1365" from-port="0" to-layer="1366" to-port="4" />
|
|
<edge from-layer="1366" from-port="5" to-layer="1368" to-port="0" />
|
|
<edge from-layer="1367" from-port="0" to-layer="1368" to-port="1" />
|
|
<edge from-layer="1368" from-port="2" to-layer="1370" to-port="0" />
|
|
<edge from-layer="1369" from-port="0" to-layer="1370" to-port="1" />
|
|
<edge from-layer="1370" from-port="2" to-layer="1387" to-port="0" />
|
|
<edge from-layer="1371" from-port="0" to-layer="1372" to-port="0" />
|
|
<edge from-layer="1372" from-port="1" to-layer="1374" to-port="0" />
|
|
<edge from-layer="1373" from-port="0" to-layer="1374" to-port="1" />
|
|
<edge from-layer="1374" from-port="2" to-layer="1375" to-port="1" />
|
|
<edge from-layer="1375" from-port="2" to-layer="1377" to-port="0" />
|
|
<edge from-layer="1376" from-port="0" to-layer="1377" to-port="1" />
|
|
<edge from-layer="1377" from-port="2" to-layer="1382" to-port="0" />
|
|
<edge from-layer="1378" from-port="0" to-layer="1382" to-port="1" />
|
|
<edge from-layer="1379" from-port="0" to-layer="1382" to-port="2" />
|
|
<edge from-layer="1380" from-port="0" to-layer="1382" to-port="3" />
|
|
<edge from-layer="1381" from-port="0" to-layer="1382" to-port="4" />
|
|
<edge from-layer="1382" from-port="5" to-layer="1384" to-port="0" />
|
|
<edge from-layer="1383" from-port="0" to-layer="1384" to-port="1" />
|
|
<edge from-layer="1384" from-port="2" to-layer="1386" to-port="0" />
|
|
<edge from-layer="1385" from-port="0" to-layer="1386" to-port="1" />
|
|
<edge from-layer="1386" from-port="2" to-layer="1387" to-port="1" />
|
|
<edge from-layer="1387" from-port="2" to-layer="1389" to-port="0" />
|
|
<edge from-layer="1388" from-port="0" to-layer="1389" to-port="1" />
|
|
<edge from-layer="1389" from-port="2" to-layer="1390" to-port="0" />
|
|
<edge from-layer="1390" from-port="2" to-layer="1391" to-port="0" />
|
|
<edge from-layer="1391" from-port="1" to-layer="1403" to-port="0" />
|
|
<edge from-layer="1392" from-port="0" to-layer="1393" to-port="0" />
|
|
<edge from-layer="1393" from-port="1" to-layer="1395" to-port="0" />
|
|
<edge from-layer="1394" from-port="0" to-layer="1395" to-port="1" />
|
|
<edge from-layer="1395" from-port="2" to-layer="1396" to-port="1" />
|
|
<edge from-layer="1396" from-port="2" to-layer="1398" to-port="0" />
|
|
<edge from-layer="1397" from-port="0" to-layer="1398" to-port="1" />
|
|
<edge from-layer="1398" from-port="2" to-layer="1400" to-port="0" />
|
|
<edge from-layer="1399" from-port="0" to-layer="1400" to-port="1" />
|
|
<edge from-layer="1400" from-port="2" to-layer="1402" to-port="0" />
|
|
<edge from-layer="1401" from-port="0" to-layer="1402" to-port="1" />
|
|
<edge from-layer="1402" from-port="2" to-layer="1403" to-port="1" />
|
|
<edge from-layer="1403" from-port="2" to-layer="1405" to-port="0" />
|
|
<edge from-layer="1404" from-port="0" to-layer="1405" to-port="1" />
|
|
<edge from-layer="1405" from-port="2" to-layer="1407" to-port="0" />
|
|
<edge from-layer="1406" from-port="0" to-layer="1407" to-port="1" />
|
|
<edge from-layer="1407" from-port="2" to-layer="1409" to-port="0" />
|
|
<edge from-layer="1408" from-port="0" to-layer="1409" to-port="1" />
|
|
<edge from-layer="1409" from-port="2" to-layer="1414" to-port="0" />
|
|
<edge from-layer="1410" from-port="0" to-layer="1414" to-port="1" />
|
|
<edge from-layer="1411" from-port="0" to-layer="1414" to-port="2" />
|
|
<edge from-layer="1412" from-port="0" to-layer="1414" to-port="3" />
|
|
<edge from-layer="1413" from-port="0" to-layer="1414" to-port="4" />
|
|
<edge from-layer="1414" from-port="5" to-layer="1419" to-port="0" />
|
|
<edge from-layer="1415" from-port="0" to-layer="1416" to-port="0" />
|
|
<edge from-layer="1416" from-port="1" to-layer="1418" to-port="0" />
|
|
<edge from-layer="1417" from-port="0" to-layer="1418" to-port="1" />
|
|
<edge from-layer="1418" from-port="2" to-layer="1419" to-port="1" />
|
|
<edge from-layer="1419" from-port="2" to-layer="1421" to-port="0" />
|
|
<edge from-layer="1420" from-port="0" to-layer="1421" to-port="1" />
|
|
<edge from-layer="1421" from-port="2" to-layer="1422" to-port="0" />
|
|
<edge from-layer="1422" from-port="2" to-layer="1424" to-port="0" />
|
|
<edge from-layer="1423" from-port="0" to-layer="1424" to-port="1" />
|
|
<edge from-layer="1424" from-port="2" to-layer="1426" to-port="0" />
|
|
<edge from-layer="1425" from-port="0" to-layer="1426" to-port="1" />
|
|
<edge from-layer="1426" from-port="2" to-layer="1428" to-port="0" />
|
|
<edge from-layer="1427" from-port="0" to-layer="1428" to-port="1" />
|
|
<edge from-layer="1428" from-port="2" to-layer="1430" to-port="0" />
|
|
<edge from-layer="1428" from-port="2" to-layer="1458" to-port="1" />
|
|
<edge from-layer="1429" from-port="0" to-layer="1430" to-port="1" />
|
|
<edge from-layer="1430" from-port="2" to-layer="1435" to-port="0" />
|
|
<edge from-layer="1431" from-port="0" to-layer="1435" to-port="1" />
|
|
<edge from-layer="1432" from-port="0" to-layer="1435" to-port="2" />
|
|
<edge from-layer="1433" from-port="0" to-layer="1435" to-port="3" />
|
|
<edge from-layer="1434" from-port="0" to-layer="1435" to-port="4" />
|
|
<edge from-layer="1435" from-port="5" to-layer="1440" to-port="0" />
|
|
<edge from-layer="1436" from-port="0" to-layer="1437" to-port="0" />
|
|
<edge from-layer="1437" from-port="1" to-layer="1439" to-port="0" />
|
|
<edge from-layer="1438" from-port="0" to-layer="1439" to-port="1" />
|
|
<edge from-layer="1439" from-port="2" to-layer="1440" to-port="1" />
|
|
<edge from-layer="1440" from-port="2" to-layer="1442" to-port="0" />
|
|
<edge from-layer="1441" from-port="0" to-layer="1442" to-port="1" />
|
|
<edge from-layer="1442" from-port="2" to-layer="1443" to-port="0" />
|
|
<edge from-layer="1443" from-port="1" to-layer="1445" to-port="0" />
|
|
<edge from-layer="1444" from-port="0" to-layer="1445" to-port="1" />
|
|
<edge from-layer="1445" from-port="2" to-layer="1450" to-port="0" />
|
|
<edge from-layer="1446" from-port="0" to-layer="1450" to-port="1" />
|
|
<edge from-layer="1447" from-port="0" to-layer="1450" to-port="2" />
|
|
<edge from-layer="1448" from-port="0" to-layer="1450" to-port="3" />
|
|
<edge from-layer="1449" from-port="0" to-layer="1450" to-port="4" />
|
|
<edge from-layer="1450" from-port="5" to-layer="1455" to-port="0" />
|
|
<edge from-layer="1451" from-port="0" to-layer="1452" to-port="0" />
|
|
<edge from-layer="1452" from-port="1" to-layer="1454" to-port="0" />
|
|
<edge from-layer="1453" from-port="0" to-layer="1454" to-port="1" />
|
|
<edge from-layer="1454" from-port="2" to-layer="1455" to-port="1" />
|
|
<edge from-layer="1455" from-port="2" to-layer="1457" to-port="0" />
|
|
<edge from-layer="1456" from-port="0" to-layer="1457" to-port="1" />
|
|
<edge from-layer="1457" from-port="2" to-layer="1458" to-port="0" />
|
|
<edge from-layer="1458" from-port="2" to-layer="1460" to-port="0" />
|
|
<edge from-layer="1459" from-port="0" to-layer="1460" to-port="1" />
|
|
<edge from-layer="1460" from-port="2" to-layer="1462" to-port="0" />
|
|
<edge from-layer="1461" from-port="0" to-layer="1462" to-port="1" />
|
|
<edge from-layer="1462" from-port="2" to-layer="1464" to-port="0" />
|
|
<edge from-layer="1463" from-port="0" to-layer="1464" to-port="1" />
|
|
<edge from-layer="1464" from-port="2" to-layer="1465" to-port="0" />
|
|
</edges>
|
|
<rt_info>
|
|
<Runtime_version value="2024.4.1-16618-643f23d1318-releases/2024/4" />
|
|
<conversion_parameters>
|
|
<framework value="pytorch" />
|
|
<is_python_object value="True" />
|
|
</conversion_parameters>
|
|
<nncf>
|
|
<friendly_names_were_updated value="True" />
|
|
<quantization>
|
|
<advanced_parameters value="{'overflow_fix': 'disable', 'quantize_outputs': False, 'inplace_statistics': True, 'disable_channel_alignment': True, 'disable_bias_correction': False, 'batchwise_statistics': None, 'activations_quantization_params': None, 'weights_quantization_params': None, 'activations_range_estimator_params': {'min': {'statistics_type': None, 'aggregator_type': None, 'clipping_value': None, 'quantile_outlier_prob': 0.0001}, 'max': {'statistics_type': None, 'aggregator_type': None, 'clipping_value': None, 'quantile_outlier_prob': 0.0001}}, 'weights_range_estimator_params': {'min': {'statistics_type': None, 'aggregator_type': None, 'clipping_value': None, 'quantile_outlier_prob': 0.0001}, 'max': {'statistics_type': None, 'aggregator_type': None, 'clipping_value': None, 'quantile_outlier_prob': 0.0001}}, 'bias_correction_params': {'apply_for_all_nodes': False, 'threshold': None}, 'smooth_quant_alphas': {'convolution': -1, 'matmul': 0.95}, 'smooth_quant_alpha': None, 'backend_params': {}}" />
|
|
<fast_bias_correction value="True" />
|
|
<ignored_scope>
|
|
<types value="['GroupNormalization']" />
|
|
</ignored_scope>
|
|
<model_type value="transformer" />
|
|
<preset value="mixed" />
|
|
<subset_size value="300" />
|
|
<target_device value="ANY" />
|
|
</quantization>
|
|
</nncf>
|
|
<optimum>
|
|
<optimum_intel_version value="1.20.0.dev0+b31524c" />
|
|
<optimum_version value="1.23.0" />
|
|
<pytorch_version value="2.5.0.dev20240807+cu121" />
|
|
<transformers_version value="4.43.4" />
|
|
</optimum>
|
|
</rt_info>
|
|
</net>
|