11320 lines
364 KiB
XML
11320 lines
364 KiB
XML
<?xml version="1.0"?>
|
|
<net name="Model314" version="11">
|
|
<layers>
|
|
<layer id="2" 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="1" 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="0" name="token_type_ids" type="Parameter" version="opset1">
|
|
<data shape="?,?" element_type="i64" />
|
|
<output>
|
|
<port id="0" precision="I64" names="token_type_ids">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="3" name="Constant_692088" type="Const" version="opset1">
|
|
<data element_type="i8" shape="30522, 768" offset="0" size="23440896" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>30522</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="4" name="Convert_692089" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>30522</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>30522</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="5" name="Constant_692090" type="Const" version="opset1">
|
|
<data element_type="f32" shape="30522, 1" offset="23440896" size="122088" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>30522</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="6" name="__module.bert.embeddings.word_embeddings/aten::embedding/Gather/fq_weights_0" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>30522</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>30522</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>30522</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="7" name="__module.bert.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="8" name="__module.bert.embeddings.word_embeddings/aten::embedding/Constant" type="Const" version="opset1">
|
|
<data element_type="i32" shape="" offset="23562984" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="9" name="__module.bert.embeddings.word_embeddings/aten::embedding/Gather" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>30522</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="48,inputs_embeds">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="10" name="Constant_692092" type="Const" version="opset1">
|
|
<data element_type="i8" shape="2, 768" offset="23562988" size="1536" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>2</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="11" name="Convert_692093" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>2</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>2</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="12" name="Constant_692094" type="Const" version="opset1">
|
|
<data element_type="f32" shape="2, 1" offset="23564524" size="8" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>2</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="13" name="__module.bert.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>2</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>2</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>2</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="14" name="__module.bert.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="15" name="__module.bert.embeddings.token_type_embeddings/aten::embedding/Constant" type="Const" version="opset1">
|
|
<data element_type="i32" shape="" offset="23562984" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="16" name="__module.bert.embeddings.token_type_embeddings/aten::embedding/Gather" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>2</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="50,token_type_embeddings.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="17" name="__module.bert.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="51_1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="18" name="Constant_692096" type="Const" version="opset1">
|
|
<data element_type="i8" shape="512, 768" offset="23564532" size="393216" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>512</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="19" name="Convert_692097" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>512</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>512</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="20" name="Constant_692098" type="Const" version="opset1">
|
|
<data element_type="f32" shape="512, 1" offset="23957748" size="2048" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>512</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="21" name="__module.bert.embeddings.position_embeddings/aten::embedding/Gather/fq_weights_0" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>512</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>512</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>512</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="22" name="__module.bert.embeddings/aten::slice/Slice" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1, 512" offset="23959796" size="4096" />
|
|
<output>
|
|
<port id="0" precision="I64" names="45">
|
|
<dim>1</dim>
|
|
<dim>512</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="23" name="__module.bert.embeddings/aten::slice/Reshape" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="24" name="ShapeOf_477755" 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">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="25" name="Constant_477874" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="23963900" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="26" name="Constant_477757" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="27" name="Gather_477758" 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="35,41,43,44,61">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="28" name="__module.bert.embeddings/aten::slice/Reshape_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="23963900" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="29" name="__module.bert.embeddings/aten::slice/Reshape_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="23963900" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="30" name="__module.bert.embeddings/aten::slice/Slice_1" type="Slice" version="opset8">
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
<dim>512</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="46">
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="31" name="__module.bert.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="32" name="__module.bert.embeddings.position_embeddings/aten::embedding/Constant" type="Const" version="opset1">
|
|
<data element_type="i32" shape="" offset="23562984" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="33" name="__module.bert.embeddings.position_embeddings/aten::embedding/Gather" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>512</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="53,position_embeddings.1">
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="34" name="__module.bert.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="51,embeddings.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="35" name="__module.bert.embeddings.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="36" name="__module.bert.embeddings.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="37" name="Constant_477628" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="23963912" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="38" name="__module.bert.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="39" name="Constant_477629" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="23966984" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="40" name="__module.bert.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="58,input.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="41" name="__module.bert.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="23970056" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="42" name="__module.bert.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="43" name="__module.bert.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="23973128" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="44" name="__module.bert.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="23973132" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="45" name="__module.bert.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="23973128" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="46" name="__module.bert.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="23973132" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="47" name="__module.bert.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="48" name="Constant_692100" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="23973136" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="49" name="Convert_692101" 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="50" name="Constant_692102" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="24562960" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="51" name="__module.bert.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="52" name="__module.bert.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="53" name="Constant_477630" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="24566032" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="54" name="__module.bert.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="96,x.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="55" name="__module.bert.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="24569104" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="56" name="__module.bert.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="24569108" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="57" name="__module.bert.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="24569104" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="58" name="__module.bert.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="24569108" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="59" name="__module.bert.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="60" name="__module.bert.encoder.layer.0.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="61" name="__module.bert.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="100,x.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="62" name="Constant_471777" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="101">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="63" name="__module.bert.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="102">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="64" name="Constant_692104" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="24569176" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="65" name="Convert_692105" 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="66" name="Constant_692106" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="25159000" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="67" name="__module.bert.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="68" name="__module.bert.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="69" name="Constant_477631" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="25162072" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="70" name="__module.bert.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="105,x.5">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="71" name="__module.bert.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25165144" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="72" name="__module.bert.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25165148" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="73" name="__module.bert.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25165144" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="74" name="__module.bert.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25165148" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="75" name="__module.bert.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="76" name="__module.bert.encoder.layer.0.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="77" name="__module.bert.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="109,x.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="78" name="Constant_471802" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="110">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="79" name="__module.bert.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="111">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="80" name="Constant_692108" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="25165152" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="81" name="Convert_692109" 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="82" name="Constant_692110" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="25754976" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="83" name="__module.bert.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="84" name="__module.bert.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="85" name="Constant_477632" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="25758048" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="86" name="__module.bert.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="114,x.9">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="87" name="__module.bert.encoder.layer.0.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="88" name="__module.bert.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="118,x.11">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="89" name="Constant_471827" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="119">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="90" name="__module.bert.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="120">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="91" name="Constant_477634" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="25761120" 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="92" name="31" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963900" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" names="31" />
|
|
</output>
|
|
</layer>
|
|
<layer id="93" name="__module.bert/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="63">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="94" name="21" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="25761124" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" names="21" />
|
|
</output>
|
|
</layer>
|
|
<layer id="95" name="__module.bert/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="64,65">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="96" name="Constant_477764" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="97" name="Constant_477765" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="98" name="Gather_477766" 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="60">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="99" name="Constant_476860" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="23963900" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="100" name="Constant_477876" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761132" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="101" name="Constant_477877" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="102" name="Gather_477878" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="2" precision="I64" />
|
|
</input>
|
|
<output>
|
|
<port id="3" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="103" name="__module.bert/prim::ListConstruct/Concat" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="2" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="3" precision="I64" names="66">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="104" name="__module.bert/aten::expand/Broadcast" type="Broadcast" version="opset3">
|
|
<data mode="bidirectional" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="67">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="105" name="__module.bert/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="68">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="106" name="Constant_477633" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 1, 1" offset="25761120" 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="107" name="__module.bert/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="108" name="__module.bert/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="69,inverted_mask">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="109" name="__module.bert/aten::to/Convert_1" type="Convert" version="opset1">
|
|
<data destination_type="boolean" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="BOOL" names="70">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="110" name="__module.bert/aten::masked_fill/ConvertLike" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25761148" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="111" name="__module.bert/aten::masked_fill/Select" type="Select" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="BOOL">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="3" precision="FP32" names="71">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="112" name="__module.bert.encoder.layer.0.attention.self/aten::scaled_dot_product_attention/ScaledDotProductAttention" type="ScaledDotProductAttention" version="opset13">
|
|
<data causal="false" />
|
|
<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>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="3" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="4" precision="FP32" names="121,attn_output.1">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="113" name="__module.bert.encoder.layer.0.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="114" name="__module.bert.encoder.layer.0.attention.self/aten::transpose/Transpose" 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="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="122,attn_output.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="115" name="Constant_477771" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="25761168" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="116" name="__module.bert.encoder.layer.0.attention.self/aten::reshape/Reshape" 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="124">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="117" name="__module.bert.encoder.layer.0.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="25761192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="118" name="__module.bert.encoder.layer.0.attention.self/aten::reshape/Reshape_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="119" name="__module.bert.encoder.layer.0.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25764264" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="120" name="__module.bert.encoder.layer.0.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25764268" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="121" name="__module.bert.encoder.layer.0.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25764264" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="122" name="__module.bert.encoder.layer.0.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25764268" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="123" name="__module.bert.encoder.layer.0.attention.self/aten::reshape/Reshape_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="124" name="Constant_692112" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="25764272" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="125" name="Convert_692113" 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="126" name="Constant_692114" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="26354096" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="127" name="__module.bert.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="128" name="__module.bert.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="129" name="Constant_477635" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="26357168" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="130" name="__module.bert.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="129,input.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="131" name="__module.bert.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="131">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="132" name="__module.bert.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="133" name="__module.bert.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="134" name="Constant_477636" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="26360240" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="135" name="__module.bert.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="136" name="Constant_477637" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="26363312" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="137" name="__module.bert.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="135,input_tensor.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="138" name="__module.bert.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="26366384" 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.bert.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="140" name="__module.bert.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="26369456" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="141" name="__module.bert.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="26369460" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="142" name="__module.bert.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="26369456" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="143" name="__module.bert.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="26369460" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="144" name="__module.bert.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="145" name="Constant_692116" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="26369464" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="146" name="Convert_692117" 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="147" name="Constant_692118" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="28728760" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="148" name="__module.bert.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="149" name="__module.bert.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="150" name="Constant_477638" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="28741048" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="151" name="__module.bert.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="139">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="152" name="__module.bert.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="140">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="153" name="__module.bert.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="28753336" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="154" name="__module.bert.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="155" name="__module.bert.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="28765624" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="156" name="__module.bert.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="28765628" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="157" name="__module.bert.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="28765624" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="158" name="__module.bert.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="28765628" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="159" name="__module.bert.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="160" name="Constant_692120" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="28765632" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="161" name="Convert_692121" 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="162" name="Constant_692122" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="31124928" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="163" name="__module.bert.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="164" name="__module.bert.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="165" name="Constant_477639" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="31128000" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="166" name="__module.bert.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="145,input.5">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="167" name="__module.bert.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="147">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="168" name="__module.bert.encoder.layer.0.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="169" name="__module.bert.encoder.layer.0.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="170" name="Constant_477640" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="31131072" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="171" name="__module.bert.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="172" name="Constant_477641" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="31134144" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="173" name="__module.bert.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="151,hidden_states.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="174" name="__module.bert.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="31137216" 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.bert.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="176" name="__module.bert.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="31140288" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="177" name="__module.bert.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="31140292" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="178" name="__module.bert.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="31140288" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="179" name="__module.bert.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="31140292" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="180" name="__module.bert.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="181" name="Constant_692124" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="31140296" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="182" name="Convert_692125" 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="183" name="Constant_692126" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="31730120" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="184" name="__module.bert.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="185" name="__module.bert.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="186" name="Constant_477642" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="31733192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="187" name="__module.bert.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="164,x.13">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="188" name="__module.bert.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31736264" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="189" name="__module.bert.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31736268" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="190" name="__module.bert.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31736264" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="191" name="__module.bert.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31736268" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="192" name="__module.bert.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="193" name="__module.bert.encoder.layer.1.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="194" name="__module.bert.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="168,x.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="195" name="Constant_472006" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="169">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="196" name="__module.bert.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="170">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="197" name="Constant_692128" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="31736272" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="198" name="Convert_692129" 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="199" name="Constant_692130" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="32326096" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="200" name="__module.bert.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="201" name="__module.bert.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="202" name="Constant_477643" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="32329168" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="203" name="__module.bert.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="173,x.17">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="204" name="__module.bert.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32332240" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="205" name="__module.bert.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32332244" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="206" name="__module.bert.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32332240" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="207" name="__module.bert.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32332244" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="208" name="__module.bert.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="209" name="__module.bert.encoder.layer.1.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="210" name="__module.bert.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="177,x.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="211" name="Constant_472029" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="178">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="212" name="__module.bert.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="179">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="213" name="Constant_692132" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="32332248" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="214" name="Convert_692133" 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="215" name="Constant_692134" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="32922072" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="216" name="__module.bert.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="217" name="__module.bert.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="218" name="Constant_477644" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="32925144" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="219" name="__module.bert.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="182,x.21">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="220" name="__module.bert.encoder.layer.1.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="221" name="__module.bert.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="186,x.23">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="222" name="Constant_472052" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="187">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="223" name="__module.bert.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="188">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="224" name="__module.bert.encoder.layer.1.attention.self/aten::scaled_dot_product_attention/ScaledDotProductAttention" type="ScaledDotProductAttention" version="opset13">
|
|
<data causal="false" />
|
|
<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>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="3" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="4" precision="FP32" names="189,attn_output.5">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="225" name="__module.bert.encoder.layer.1.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="226" name="__module.bert.encoder.layer.1.attention.self/aten::transpose/Transpose" 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="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="190,attn_output.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="227" name="Constant_477772" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="25761168" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="228" name="__module.bert.encoder.layer.1.attention.self/aten::reshape/Reshape" 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="192">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="229" name="__module.bert.encoder.layer.1.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="32928216" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="230" name="__module.bert.encoder.layer.1.attention.self/aten::reshape/Reshape_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="231" name="__module.bert.encoder.layer.1.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32931288" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="232" name="__module.bert.encoder.layer.1.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32931292" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="233" name="__module.bert.encoder.layer.1.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32931288" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="234" name="__module.bert.encoder.layer.1.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32931292" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="235" name="__module.bert.encoder.layer.1.attention.self/aten::reshape/Reshape_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="236" name="Constant_692136" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="32931296" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="237" name="Convert_692137" 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="238" name="Constant_692138" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="33521120" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="239" name="__module.bert.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="240" name="__module.bert.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="241" name="Constant_477645" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="33524192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="242" name="__module.bert.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="197,input.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="243" name="__module.bert.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="199">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="244" name="__module.bert.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="245" name="__module.bert.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="246" name="Constant_477646" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="33527264" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="247" name="__module.bert.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="248" name="Constant_477647" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="33530336" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="249" name="__module.bert.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="203,input_tensor.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="250" name="__module.bert.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="33533408" 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.bert.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="252" name="__module.bert.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="33536480" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="253" name="__module.bert.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="33536484" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="254" name="__module.bert.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="33536480" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="255" name="__module.bert.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="33536484" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="256" name="__module.bert.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="257" name="Constant_692140" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="33536488" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="258" name="Convert_692141" 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="259" name="Constant_692142" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="35895784" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="260" name="__module.bert.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="261" name="__module.bert.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="262" name="Constant_477648" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="35908072" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="263" name="__module.bert.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="207">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="264" name="__module.bert.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="208">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="265" name="__module.bert.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="35920360" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="266" name="__module.bert.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="267" name="__module.bert.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="35932648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="268" name="__module.bert.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="35932652" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="269" name="__module.bert.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="35932648" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="270" name="__module.bert.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="35932652" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="271" name="__module.bert.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="272" name="Constant_692144" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="35932656" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="273" name="Convert_692145" 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="274" name="Constant_692146" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="38291952" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="275" name="__module.bert.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="276" name="__module.bert.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="277" name="Constant_477649" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="38295024" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="278" name="__module.bert.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="213,input.9">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="279" name="__module.bert.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="215">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="280" name="__module.bert.encoder.layer.1.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="281" name="__module.bert.encoder.layer.1.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="282" name="Constant_477650" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="38298096" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="283" name="__module.bert.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="284" name="Constant_477651" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="38301168" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="285" name="__module.bert.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="219,hidden_states.13">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="286" name="__module.bert.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="38304240" 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.bert.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="288" name="__module.bert.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="38307312" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="289" name="__module.bert.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="38307316" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="290" name="__module.bert.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="38307312" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="291" name="__module.bert.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="38307316" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="292" name="__module.bert.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="293" name="Constant_692148" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="38307320" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="294" name="Convert_692149" 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="295" name="Constant_692150" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="38897144" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="296" name="__module.bert.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="297" name="__module.bert.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="298" name="Constant_477652" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="38900216" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="299" name="__module.bert.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="232,x.25">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="300" name="__module.bert.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38903288" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="301" name="__module.bert.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38903292" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="302" name="__module.bert.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38903288" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="303" name="__module.bert.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38903292" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="304" name="__module.bert.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="305" name="__module.bert.encoder.layer.2.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="306" name="__module.bert.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="236,x.27">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="307" name="Constant_472229" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="237">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="308" name="__module.bert.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="238">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="309" name="Constant_692152" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="38903296" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="310" name="Convert_692153" 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="311" name="Constant_692154" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="39493120" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="312" name="__module.bert.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="313" name="__module.bert.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="314" name="Constant_477653" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="39496192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="315" name="__module.bert.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="241,x.29">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="316" name="__module.bert.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39499264" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="317" name="__module.bert.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39499268" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="318" name="__module.bert.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39499264" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="319" name="__module.bert.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39499268" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="320" name="__module.bert.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="321" name="__module.bert.encoder.layer.2.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="322" name="__module.bert.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="245,x.31">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="323" name="Constant_472252" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="246">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="324" name="__module.bert.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="247">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="325" name="Constant_692156" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="39499272" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="326" name="Convert_692157" 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="327" name="Constant_692158" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="40089096" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="328" name="__module.bert.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="329" name="__module.bert.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="330" name="Constant_477654" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40092168" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="331" name="__module.bert.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="250,x.33">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="332" name="__module.bert.encoder.layer.2.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="333" name="__module.bert.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="254,x.35">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="334" name="Constant_472275" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="255">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="335" name="__module.bert.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="256">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="336" name="__module.bert.encoder.layer.2.attention.self/aten::scaled_dot_product_attention/ScaledDotProductAttention" type="ScaledDotProductAttention" version="opset13">
|
|
<data causal="false" />
|
|
<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>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="3" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="4" precision="FP32" names="257,attn_output.9">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="337" name="__module.bert.encoder.layer.2.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="338" name="__module.bert.encoder.layer.2.attention.self/aten::transpose/Transpose" 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="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="258,attn_output.11">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="339" name="Constant_477773" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="25761168" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="340" name="__module.bert.encoder.layer.2.attention.self/aten::reshape/Reshape" 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="260">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="341" name="__module.bert.encoder.layer.2.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40095240" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="342" name="__module.bert.encoder.layer.2.attention.self/aten::reshape/Reshape_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="343" name="__module.bert.encoder.layer.2.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40098312" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="344" name="__module.bert.encoder.layer.2.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40098316" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="345" name="__module.bert.encoder.layer.2.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40098312" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="346" name="__module.bert.encoder.layer.2.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40098316" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="347" name="__module.bert.encoder.layer.2.attention.self/aten::reshape/Reshape_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="348" name="Constant_692160" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="40098320" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="349" name="Convert_692161" 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="350" name="Constant_692162" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="40688144" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="351" name="__module.bert.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="352" name="__module.bert.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="353" name="Constant_477655" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40691216" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="354" name="__module.bert.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="265,input.11">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="355" name="__module.bert.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="267">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="356" name="__module.bert.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="357" name="__module.bert.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="358" name="Constant_477656" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40694288" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="359" name="__module.bert.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="360" name="Constant_477657" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40697360" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="361" name="__module.bert.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="271,input_tensor.5">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="362" name="__module.bert.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="40700432" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="363" name="__module.bert.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="364" name="__module.bert.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="40703504" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="365" name="__module.bert.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="40703508" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="366" name="__module.bert.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="40703504" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="367" name="__module.bert.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="40703508" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="368" name="__module.bert.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="369" name="Constant_692164" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="40703512" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="370" name="Convert_692165" 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="371" name="Constant_692166" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="43062808" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="372" name="__module.bert.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="373" name="__module.bert.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="374" name="Constant_477658" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="43075096" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="375" name="__module.bert.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="275">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="376" name="__module.bert.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="276">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="377" name="__module.bert.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="43087384" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="378" name="__module.bert.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="379" name="__module.bert.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="43099672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="380" name="__module.bert.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="43099676" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="381" name="__module.bert.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="43099672" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="382" name="__module.bert.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="43099676" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="383" name="__module.bert.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="384" name="Constant_692168" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="43099680" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="385" name="Convert_692169" 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="386" name="Constant_692170" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="45458976" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="387" name="__module.bert.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="388" name="__module.bert.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="389" name="Constant_477659" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="45462048" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="390" name="__module.bert.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="281,input.13">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="391" name="__module.bert.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="283">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="392" name="__module.bert.encoder.layer.2.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="393" name="__module.bert.encoder.layer.2.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="394" name="Constant_477660" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="45465120" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="395" name="__module.bert.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="396" name="Constant_477661" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="45468192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="397" name="__module.bert.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="287,hidden_states.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="398" name="__module.bert.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="45471264" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="399" name="__module.bert.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="400" name="__module.bert.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="45474336" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="401" name="__module.bert.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="45474340" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="402" name="__module.bert.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="45474336" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="403" name="__module.bert.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="45474340" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="404" name="__module.bert.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="405" name="Constant_692172" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="45474344" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="406" name="Convert_692173" 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="407" name="Constant_692174" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="46064168" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="408" name="__module.bert.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="409" name="__module.bert.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="410" name="Constant_477662" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="46067240" 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.bert.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="300,x.37">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="412" name="__module.bert.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46070312" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="413" name="__module.bert.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46070316" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="414" name="__module.bert.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46070312" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="415" name="__module.bert.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46070316" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="416" name="__module.bert.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="417" name="__module.bert.encoder.layer.3.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="418" name="__module.bert.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="304,x.39">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="419" name="Constant_472452" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="305">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="420" name="__module.bert.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="306">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="421" name="Constant_692176" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="46070320" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="422" name="Convert_692177" 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="423" name="Constant_692178" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="46660144" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="424" name="__module.bert.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="425" name="__module.bert.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="426" name="Constant_477663" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="46663216" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="427" name="__module.bert.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="309,x.41">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="428" name="__module.bert.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46666288" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="429" name="__module.bert.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46666292" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="430" name="__module.bert.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46666288" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="431" name="__module.bert.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46666292" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="432" name="__module.bert.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="433" name="__module.bert.encoder.layer.3.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="434" name="__module.bert.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="313,x.43">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="435" name="Constant_472475" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="314">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="436" name="__module.bert.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="315">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="437" name="Constant_692180" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="46666296" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="438" name="Convert_692181" 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="439" name="Constant_692182" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="47256120" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="440" name="__module.bert.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="441" name="__module.bert.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="442" name="Constant_477664" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47259192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="443" name="__module.bert.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="318,x.45">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="444" name="__module.bert.encoder.layer.3.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="445" name="__module.bert.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="322,x.47">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="446" name="Constant_472498" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="323">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="447" name="__module.bert.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="324">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="448" name="__module.bert.encoder.layer.3.attention.self/aten::scaled_dot_product_attention/ScaledDotProductAttention" type="ScaledDotProductAttention" version="opset13">
|
|
<data causal="false" />
|
|
<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>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="3" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="4" precision="FP32" names="325,attn_output.13">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="449" name="__module.bert.encoder.layer.3.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="450" name="__module.bert.encoder.layer.3.attention.self/aten::transpose/Transpose" 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="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="326,attn_output.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="451" name="Constant_477774" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="25761168" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="452" name="__module.bert.encoder.layer.3.attention.self/aten::reshape/Reshape" 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="328">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="453" name="__module.bert.encoder.layer.3.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47262264" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="454" name="__module.bert.encoder.layer.3.attention.self/aten::reshape/Reshape_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="455" name="__module.bert.encoder.layer.3.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47265336" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="456" name="__module.bert.encoder.layer.3.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47265340" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="457" name="__module.bert.encoder.layer.3.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47265336" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="458" name="__module.bert.encoder.layer.3.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47265340" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="459" name="__module.bert.encoder.layer.3.attention.self/aten::reshape/Reshape_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="460" name="Constant_692184" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="47265344" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="461" name="Convert_692185" 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="462" name="Constant_692186" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="47855168" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="463" name="__module.bert.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="464" name="__module.bert.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="465" name="Constant_477665" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47858240" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="466" name="__module.bert.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="333,input.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="467" name="__module.bert.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="335">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="468" name="__module.bert.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="469" name="__module.bert.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="470" name="Constant_477666" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47861312" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="471" name="__module.bert.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="472" name="Constant_477667" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47864384" 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.bert.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="339,input_tensor.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="474" name="__module.bert.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="47867456" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="475" name="__module.bert.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="476" name="__module.bert.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="47870528" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="477" name="__module.bert.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="47870532" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="478" name="__module.bert.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="47870528" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="479" name="__module.bert.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="47870532" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="480" name="__module.bert.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="481" name="Constant_692188" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="47870536" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="482" name="Convert_692189" 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="483" name="Constant_692190" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="50229832" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="484" name="__module.bert.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="485" name="__module.bert.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="486" name="Constant_477668" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="50242120" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="487" name="__module.bert.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="343">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="488" name="__module.bert.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="344">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="489" name="__module.bert.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="50254408" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="490" name="__module.bert.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="491" name="__module.bert.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="50266696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="492" name="__module.bert.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="50266700" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="493" name="__module.bert.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="50266696" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="494" name="__module.bert.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="50266700" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="495" name="__module.bert.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="496" name="Constant_692192" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="50266704" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="497" name="Convert_692193" 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="498" name="Constant_692194" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="52626000" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="499" name="__module.bert.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="500" name="__module.bert.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="501" name="Constant_477669" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="52629072" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="502" name="__module.bert.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="349,input.17">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="503" name="__module.bert.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="351">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="504" name="__module.bert.encoder.layer.3.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="505" name="__module.bert.encoder.layer.3.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="506" name="Constant_477670" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="52632144" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="507" name="__module.bert.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="508" name="Constant_477671" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="52635216" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="509" name="__module.bert.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="355,hidden_states.25">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="510" name="__module.bert.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="52638288" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="511" name="__module.bert.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="512" name="__module.bert.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="52641360" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="513" name="__module.bert.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="52641364" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="514" name="__module.bert.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="52641360" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="515" name="__module.bert.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="52641364" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="516" name="__module.bert.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="517" name="Constant_692196" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="52641368" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="518" name="Convert_692197" 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="519" name="Constant_692198" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="53231192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="520" name="__module.bert.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="521" name="__module.bert.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="522" name="Constant_477672" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="53234264" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="523" name="__module.bert.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="368,x.49">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="524" name="__module.bert.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53237336" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="525" name="__module.bert.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53237340" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="526" name="__module.bert.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53237336" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="527" name="__module.bert.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53237340" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="528" name="__module.bert.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="529" name="__module.bert.encoder.layer.4.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="530" name="__module.bert.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="372,x.51">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="531" name="Constant_472675" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="373">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="532" name="__module.bert.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="374">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="533" name="Constant_692200" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="53237344" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="534" name="Convert_692201" 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="535" name="Constant_692202" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="53827168" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="536" name="__module.bert.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="537" name="__module.bert.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="538" name="Constant_477673" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="53830240" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="539" name="__module.bert.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="377,x.53">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="540" name="__module.bert.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53833312" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="541" name="__module.bert.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53833316" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="542" name="__module.bert.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53833312" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="543" name="__module.bert.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53833316" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="544" name="__module.bert.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="545" name="__module.bert.encoder.layer.4.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="546" name="__module.bert.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="381,x.55">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="547" name="Constant_472698" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="382">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="548" name="__module.bert.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="383">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="549" name="Constant_692204" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="53833320" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="550" name="Convert_692205" 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="551" name="Constant_692206" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="54423144" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="552" name="__module.bert.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="553" name="__module.bert.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="554" name="Constant_477674" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="54426216" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="555" name="__module.bert.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="386,x.57">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="556" name="__module.bert.encoder.layer.4.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="557" name="__module.bert.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="390,x.59">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="558" name="Constant_472721" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="391">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="559" name="__module.bert.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="392">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="560" name="__module.bert.encoder.layer.4.attention.self/aten::scaled_dot_product_attention/ScaledDotProductAttention" type="ScaledDotProductAttention" version="opset13">
|
|
<data causal="false" />
|
|
<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>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="3" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="4" precision="FP32" names="393,attn_output.17">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="561" name="__module.bert.encoder.layer.4.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="562" name="__module.bert.encoder.layer.4.attention.self/aten::transpose/Transpose" 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="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="394,attn_output.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="563" name="Constant_477775" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="25761168" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="564" name="__module.bert.encoder.layer.4.attention.self/aten::reshape/Reshape" 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="396">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="565" name="__module.bert.encoder.layer.4.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="54429288" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="566" name="__module.bert.encoder.layer.4.attention.self/aten::reshape/Reshape_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="567" name="__module.bert.encoder.layer.4.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54432360" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="568" name="__module.bert.encoder.layer.4.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54432364" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="569" name="__module.bert.encoder.layer.4.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54432360" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="570" name="__module.bert.encoder.layer.4.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="54432364" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="571" name="__module.bert.encoder.layer.4.attention.self/aten::reshape/Reshape_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="572" name="Constant_692208" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="54432368" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="573" name="Convert_692209" 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="574" name="Constant_692210" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="55022192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="575" name="__module.bert.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="576" name="__module.bert.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="577" name="Constant_477675" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55025264" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="578" name="__module.bert.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="401,input.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="579" name="__module.bert.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="403">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="580" name="__module.bert.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="581" name="__module.bert.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="582" name="Constant_477676" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55028336" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="583" name="__module.bert.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="584" name="Constant_477677" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55031408" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="585" name="__module.bert.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="407,input_tensor.9">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="586" name="__module.bert.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="55034480" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="587" name="__module.bert.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="588" name="__module.bert.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="55037552" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="589" name="__module.bert.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="55037556" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="590" name="__module.bert.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="55037552" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="591" name="__module.bert.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="55037556" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="592" name="__module.bert.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="593" name="Constant_692212" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="55037560" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="594" name="Convert_692213" 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="595" name="Constant_692214" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="57396856" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="596" name="__module.bert.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="597" name="__module.bert.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="598" name="Constant_477678" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="57409144" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="599" name="__module.bert.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="411">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="600" name="__module.bert.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="412">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="601" name="__module.bert.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="57421432" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="602" name="__module.bert.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="603" name="__module.bert.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="57433720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="604" name="__module.bert.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="57433724" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="605" name="__module.bert.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="57433720" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="606" name="__module.bert.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="57433724" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="607" name="__module.bert.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="608" name="Constant_692216" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="57433728" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="609" name="Convert_692217" 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="610" name="Constant_692218" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="59793024" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="611" name="__module.bert.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="612" name="__module.bert.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="613" name="Constant_477679" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="59796096" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="614" name="__module.bert.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="417,input.21">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="615" name="__module.bert.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="419">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="616" name="__module.bert.encoder.layer.4.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="617" name="__module.bert.encoder.layer.4.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="618" name="Constant_477680" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="59799168" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="619" name="__module.bert.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="620" name="Constant_477681" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="59802240" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="621" name="__module.bert.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="423,hidden_states.31">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="622" name="__module.bert.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="59805312" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="623" name="__module.bert.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="624" name="__module.bert.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="59808384" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="625" name="__module.bert.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="59808388" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="626" name="__module.bert.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="59808384" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="627" name="__module.bert.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="59808388" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="628" name="__module.bert.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="629" name="Constant_692220" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="59808392" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="630" name="Convert_692221" 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="631" name="Constant_692222" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="60398216" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="632" name="__module.bert.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="633" name="__module.bert.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="634" name="Constant_477682" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="60401288" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="635" name="__module.bert.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="436,x.61">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="636" name="__module.bert.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60404360" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="637" name="__module.bert.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60404364" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="638" name="__module.bert.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60404360" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="639" name="__module.bert.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60404364" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="640" name="__module.bert.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="641" name="__module.bert.encoder.layer.5.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="642" name="__module.bert.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="440,x.63">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="643" name="Constant_472898" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="441">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="644" name="__module.bert.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="442">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="645" name="Constant_692224" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="60404368" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="646" name="Convert_692225" 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="647" name="Constant_692226" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="60994192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="648" name="__module.bert.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="649" name="__module.bert.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="650" name="Constant_477683" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="60997264" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="651" name="__module.bert.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="445,x.65">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="652" name="__module.bert.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61000336" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="653" name="__module.bert.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61000340" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="654" name="__module.bert.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61000336" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="655" name="__module.bert.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61000340" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="656" name="__module.bert.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="657" name="__module.bert.encoder.layer.5.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="658" name="__module.bert.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="449,x.67">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="659" name="Constant_472921" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="450">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="660" name="__module.bert.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="451">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="661" name="Constant_692228" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="61000344" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="662" name="Convert_692229" 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="663" name="Constant_692230" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="61590168" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="664" name="__module.bert.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="665" name="__module.bert.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="666" name="Constant_477684" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="61593240" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="667" name="__module.bert.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="454,x.69">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="668" name="__module.bert.encoder.layer.5.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="669" name="__module.bert.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="458,x">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="670" name="Constant_472944" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="459">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="671" name="__module.bert.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="460">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="672" name="__module.bert.encoder.layer.5.attention.self/aten::scaled_dot_product_attention/ScaledDotProductAttention" type="ScaledDotProductAttention" version="opset13">
|
|
<data causal="false" />
|
|
<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>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="3" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="4" precision="FP32" names="461,attn_output.21">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="673" name="__module.bert.encoder.layer.5.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="674" name="__module.bert.encoder.layer.5.attention.self/aten::transpose/Transpose" 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="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="462,attn_output">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="675" name="Constant_477776" type="Const" version="opset1">
|
|
<data element_type="i64" shape="3" offset="25761168" size="24" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="676" name="__module.bert.encoder.layer.5.attention.self/aten::reshape/Reshape" 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="464">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="677" name="__module.bert.encoder.layer.5.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="61596312" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="678" name="__module.bert.encoder.layer.5.attention.self/aten::reshape/Reshape_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="679" name="__module.bert.encoder.layer.5.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61599384" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="680" name="__module.bert.encoder.layer.5.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61599388" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="681" name="__module.bert.encoder.layer.5.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61599384" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="682" name="__module.bert.encoder.layer.5.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61599388" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="683" name="__module.bert.encoder.layer.5.attention.self/aten::reshape/Reshape_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="684" name="Constant_692232" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="61599392" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="685" name="Convert_692233" 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="686" name="Constant_692234" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="62189216" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="687" name="__module.bert.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="688" name="__module.bert.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="689" name="Constant_477685" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="62192288" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="690" name="__module.bert.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="469,input.23">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="691" name="__module.bert.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="471">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="692" name="__module.bert.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="693" name="__module.bert.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="694" name="Constant_477686" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="62195360" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="695" name="__module.bert.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="696" name="Constant_477687" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="62198432" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="697" name="__module.bert.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="475,input_tensor">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="698" name="__module.bert.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="62201504" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="699" name="__module.bert.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="700" name="__module.bert.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="62204576" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="701" name="__module.bert.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="62204580" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="702" name="__module.bert.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="62204576" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="703" name="__module.bert.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="62204580" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="704" name="__module.bert.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="705" name="Constant_692236" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="62204584" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="706" name="Convert_692237" 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="707" name="Constant_692238" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="64563880" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="708" name="__module.bert.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="709" name="__module.bert.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="710" name="Constant_477688" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="64576168" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="711" name="__module.bert.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="479">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="712" name="__module.bert.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="480">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="713" name="__module.bert.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="64588456" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="714" name="__module.bert.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="715" name="__module.bert.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="64600744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="716" name="__module.bert.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="64600748" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="717" name="__module.bert.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="64600744" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="718" name="__module.bert.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="64600748" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="719" name="__module.bert.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="720" name="Constant_692240" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="64600752" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="721" name="Convert_692241" 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="722" name="Constant_692242" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="66960048" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="723" name="__module.bert.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="724" name="__module.bert.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="725" name="Constant_477689" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="66963120" 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.bert.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="485,input.25">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="727" name="__module.bert.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="487">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="728" name="__module.bert.encoder.layer.5.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="729" name="__module.bert.encoder.layer.5.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-13" 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="730" name="Constant_477690" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="66966192" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="731" name="__module.bert.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="732" name="Constant_477691" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="66969264" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="733" name="__module.bert.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="491,493,hidden_states">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="734" name="23" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" names="23" />
|
|
</output>
|
|
</layer>
|
|
<layer id="735" name="__module.bert.pooler/aten::select/Gather" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64" />
|
|
<port id="2" precision="I64" />
|
|
</input>
|
|
<output>
|
|
<port id="3" precision="FP32" names="494">
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="736" name="__module.bert.pooler/aten::select/Gather_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 768" offset="66972336" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="737" name="__module.bert.pooler/aten::select/Gather_0_0/nncf_smooth_quant" 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>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="738" name="__module.bert.pooler/aten::select/Gather_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="66975408" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="739" name="__module.bert.pooler/aten::select/Gather_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="66975412" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="740" name="__module.bert.pooler/aten::select/Gather_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="66975408" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="741" name="__module.bert.pooler/aten::select/Gather_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="66975412" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="742" name="__module.bert.pooler/aten::select/Gather_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>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>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="743" name="Constant_692244" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="66975416" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="744" name="Convert_692245" 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="745" name="Constant_692246" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="67565240" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="746" name="__module.bert.pooler.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="747" name="__module.bert.pooler.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>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>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="748" name="Constant_477692" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 768" offset="67568312" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="749" name="__module.bert.pooler.dense/aten::linear/Add" type="Add" 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>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="497">
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="750" name="__module.bert.pooler.activation/aten::tanh/Tanh" type="Tanh" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="498,input">
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="751" name="__module.bert.pooler.activation/aten::tanh/Tanh_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 768" offset="67571384" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="752" name="__module.bert.pooler.activation/aten::tanh/Tanh_0_0/nncf_smooth_quant" 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>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="753" name="__module.bert.pooler.activation/aten::tanh/Tanh_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67574456" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="754" name="__module.bert.pooler.activation/aten::tanh/Tanh_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67574460" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="755" name="__module.bert.pooler.activation/aten::tanh/Tanh_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67574456" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="756" name="__module.bert.pooler.activation/aten::tanh/Tanh_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67574460" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="757" name="__module.bert.pooler.activation/aten::tanh/Tanh_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>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>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="758" name="Constant_692248" type="Const" version="opset1">
|
|
<data element_type="i8" shape="1, 768" offset="67574464" size="768" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="759" name="Convert_692249" 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="760" name="Constant_692250" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1" offset="67575232" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="761" name="__module.classifier/aten::linear/MatMul/fq_weights_1" 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="762" name="__module.classifier/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="763" name="Constant_477693" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1" offset="67575236" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="764" name="__module.classifier/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="logits">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="765" name="Result_474182" type="Result" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
</layer>
|
|
</layers>
|
|
<edges>
|
|
<edge from-layer="0" from-port="0" to-layer="14" to-port="0" />
|
|
<edge from-layer="1" from-port="0" to-layer="93" to-port="0" />
|
|
<edge from-layer="2" from-port="0" to-layer="7" to-port="0" />
|
|
<edge from-layer="2" from-port="0" to-layer="24" to-port="0" />
|
|
<edge from-layer="3" from-port="0" to-layer="4" to-port="0" />
|
|
<edge from-layer="4" from-port="1" to-layer="6" to-port="0" />
|
|
<edge from-layer="5" from-port="0" to-layer="6" to-port="1" />
|
|
<edge from-layer="6" from-port="2" to-layer="9" to-port="0" />
|
|
<edge from-layer="7" from-port="1" to-layer="9" to-port="1" />
|
|
<edge from-layer="8" from-port="0" to-layer="9" to-port="2" />
|
|
<edge from-layer="9" from-port="3" to-layer="17" to-port="0" />
|
|
<edge from-layer="10" from-port="0" to-layer="11" to-port="0" />
|
|
<edge from-layer="11" from-port="1" to-layer="13" to-port="0" />
|
|
<edge from-layer="12" from-port="0" to-layer="13" to-port="1" />
|
|
<edge from-layer="13" from-port="2" to-layer="16" to-port="0" />
|
|
<edge from-layer="14" from-port="1" to-layer="16" to-port="1" />
|
|
<edge from-layer="15" from-port="0" to-layer="16" to-port="2" />
|
|
<edge from-layer="16" from-port="3" to-layer="17" to-port="1" />
|
|
<edge from-layer="17" from-port="2" to-layer="34" to-port="0" />
|
|
<edge from-layer="18" from-port="0" to-layer="19" to-port="0" />
|
|
<edge from-layer="19" from-port="1" to-layer="21" to-port="0" />
|
|
<edge from-layer="20" from-port="0" to-layer="21" to-port="1" />
|
|
<edge from-layer="21" from-port="2" to-layer="33" to-port="0" />
|
|
<edge from-layer="22" from-port="0" to-layer="30" to-port="0" />
|
|
<edge from-layer="23" from-port="0" to-layer="30" to-port="1" />
|
|
<edge from-layer="24" from-port="1" to-layer="102" to-port="0" />
|
|
<edge from-layer="24" from-port="1" to-layer="27" to-port="0" />
|
|
<edge from-layer="24" from-port="1" to-layer="98" to-port="0" />
|
|
<edge from-layer="25" from-port="0" to-layer="27" to-port="1" />
|
|
<edge from-layer="26" from-port="0" to-layer="27" to-port="2" />
|
|
<edge from-layer="27" from-port="3" to-layer="30" to-port="2" />
|
|
<edge from-layer="28" from-port="0" to-layer="30" to-port="3" />
|
|
<edge from-layer="29" from-port="0" to-layer="30" to-port="4" />
|
|
<edge from-layer="30" from-port="5" to-layer="31" to-port="0" />
|
|
<edge from-layer="31" from-port="1" to-layer="33" to-port="1" />
|
|
<edge from-layer="32" from-port="0" to-layer="33" to-port="2" />
|
|
<edge from-layer="33" from-port="3" to-layer="34" to-port="1" />
|
|
<edge from-layer="34" from-port="2" to-layer="36" to-port="0" />
|
|
<edge from-layer="35" from-port="0" to-layer="36" to-port="1" />
|
|
<edge from-layer="36" from-port="2" to-layer="38" to-port="0" />
|
|
<edge from-layer="37" from-port="0" to-layer="38" to-port="1" />
|
|
<edge from-layer="38" from-port="2" to-layer="40" to-port="0" />
|
|
<edge from-layer="39" from-port="0" to-layer="40" to-port="1" />
|
|
<edge from-layer="40" from-port="2" to-layer="42" to-port="0" />
|
|
<edge from-layer="40" from-port="2" to-layer="131" to-port="1" />
|
|
<edge from-layer="41" from-port="0" to-layer="42" to-port="1" />
|
|
<edge from-layer="42" from-port="2" to-layer="47" to-port="0" />
|
|
<edge from-layer="43" from-port="0" to-layer="47" to-port="1" />
|
|
<edge from-layer="44" from-port="0" to-layer="47" to-port="2" />
|
|
<edge from-layer="45" from-port="0" to-layer="47" to-port="3" />
|
|
<edge from-layer="46" from-port="0" to-layer="47" to-port="4" />
|
|
<edge from-layer="47" from-port="5" to-layer="52" to-port="0" />
|
|
<edge from-layer="47" from-port="5" to-layer="68" to-port="0" />
|
|
<edge from-layer="47" from-port="5" to-layer="84" to-port="0" />
|
|
<edge from-layer="48" from-port="0" to-layer="49" to-port="0" />
|
|
<edge from-layer="49" from-port="1" to-layer="51" to-port="0" />
|
|
<edge from-layer="50" from-port="0" to-layer="51" to-port="1" />
|
|
<edge from-layer="51" from-port="2" to-layer="52" to-port="1" />
|
|
<edge from-layer="52" from-port="2" to-layer="54" to-port="0" />
|
|
<edge from-layer="53" from-port="0" to-layer="54" to-port="1" />
|
|
<edge from-layer="54" from-port="2" to-layer="59" to-port="0" />
|
|
<edge from-layer="55" from-port="0" to-layer="59" to-port="1" />
|
|
<edge from-layer="56" from-port="0" to-layer="59" to-port="2" />
|
|
<edge from-layer="57" from-port="0" to-layer="59" to-port="3" />
|
|
<edge from-layer="58" from-port="0" to-layer="59" to-port="4" />
|
|
<edge from-layer="59" from-port="5" to-layer="61" to-port="0" />
|
|
<edge from-layer="60" from-port="0" 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="112" to-port="0" />
|
|
<edge from-layer="64" from-port="0" to-layer="65" to-port="0" />
|
|
<edge from-layer="65" from-port="1" to-layer="67" to-port="0" />
|
|
<edge from-layer="66" from-port="0" to-layer="67" to-port="1" />
|
|
<edge from-layer="67" from-port="2" to-layer="68" to-port="1" />
|
|
<edge from-layer="68" from-port="2" 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="75" to-port="0" />
|
|
<edge from-layer="71" from-port="0" to-layer="75" to-port="1" />
|
|
<edge from-layer="72" from-port="0" to-layer="75" to-port="2" />
|
|
<edge from-layer="73" from-port="0" to-layer="75" to-port="3" />
|
|
<edge from-layer="74" from-port="0" to-layer="75" to-port="4" />
|
|
<edge from-layer="75" from-port="5" to-layer="77" to-port="0" />
|
|
<edge from-layer="76" from-port="0" 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="112" to-port="1" />
|
|
<edge from-layer="80" from-port="0" to-layer="81" to-port="0" />
|
|
<edge from-layer="81" from-port="1" to-layer="83" to-port="0" />
|
|
<edge from-layer="82" from-port="0" to-layer="83" to-port="1" />
|
|
<edge from-layer="83" from-port="2" to-layer="84" to-port="1" />
|
|
<edge from-layer="84" from-port="2" 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="90" to-port="0" />
|
|
<edge from-layer="89" from-port="0" to-layer="90" to-port="1" />
|
|
<edge from-layer="90" from-port="2" to-layer="112" to-port="2" />
|
|
<edge from-layer="91" from-port="0" to-layer="108" to-port="0" />
|
|
<edge from-layer="92" from-port="0" to-layer="735" to-port="2" />
|
|
<edge from-layer="92" from-port="0" to-layer="93" to-port="1" />
|
|
<edge from-layer="93" from-port="2" to-layer="95" to-port="0" />
|
|
<edge from-layer="94" from-port="0" to-layer="95" to-port="1" />
|
|
<edge from-layer="95" from-port="2" to-layer="104" to-port="0" />
|
|
<edge from-layer="96" from-port="0" to-layer="98" to-port="1" />
|
|
<edge from-layer="97" from-port="0" to-layer="98" to-port="2" />
|
|
<edge from-layer="98" from-port="3" to-layer="103" to-port="0" />
|
|
<edge from-layer="99" from-port="0" to-layer="103" to-port="1" />
|
|
<edge from-layer="100" from-port="0" to-layer="102" to-port="1" />
|
|
<edge from-layer="101" from-port="0" to-layer="102" to-port="2" />
|
|
<edge from-layer="102" from-port="3" to-layer="103" to-port="2" />
|
|
<edge from-layer="103" from-port="3" to-layer="104" to-port="1" />
|
|
<edge from-layer="104" from-port="2" to-layer="105" to-port="0" />
|
|
<edge from-layer="105" from-port="1" to-layer="107" to-port="0" />
|
|
<edge from-layer="106" from-port="0" to-layer="107" to-port="1" />
|
|
<edge from-layer="107" from-port="2" to-layer="108" to-port="1" />
|
|
<edge from-layer="108" from-port="2" to-layer="111" to-port="2" />
|
|
<edge from-layer="108" from-port="2" to-layer="109" to-port="0" />
|
|
<edge from-layer="109" from-port="1" 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="3" to-layer="112" to-port="3" />
|
|
<edge from-layer="111" from-port="3" to-layer="560" to-port="3" />
|
|
<edge from-layer="111" from-port="3" to-layer="672" to-port="3" />
|
|
<edge from-layer="111" from-port="3" to-layer="224" to-port="3" />
|
|
<edge from-layer="111" from-port="3" to-layer="336" to-port="3" />
|
|
<edge from-layer="111" from-port="3" to-layer="448" to-port="3" />
|
|
<edge from-layer="112" from-port="4" to-layer="114" to-port="0" />
|
|
<edge from-layer="113" from-port="0" to-layer="114" to-port="1" />
|
|
<edge from-layer="114" from-port="2" to-layer="116" to-port="0" />
|
|
<edge from-layer="115" from-port="0" 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="123" to-port="0" />
|
|
<edge from-layer="119" from-port="0" to-layer="123" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="123" to-port="2" />
|
|
<edge from-layer="121" from-port="0" to-layer="123" to-port="3" />
|
|
<edge from-layer="122" from-port="0" to-layer="123" to-port="4" />
|
|
<edge from-layer="123" from-port="5" to-layer="128" to-port="0" />
|
|
<edge from-layer="124" from-port="0" to-layer="125" to-port="0" />
|
|
<edge from-layer="125" from-port="1" to-layer="127" to-port="0" />
|
|
<edge from-layer="126" from-port="0" to-layer="127" to-port="1" />
|
|
<edge from-layer="127" from-port="2" to-layer="128" to-port="1" />
|
|
<edge from-layer="128" from-port="2" to-layer="130" to-port="0" />
|
|
<edge from-layer="129" from-port="0" to-layer="130" to-port="1" />
|
|
<edge from-layer="130" from-port="2" to-layer="131" to-port="0" />
|
|
<edge from-layer="131" from-port="2" to-layer="133" to-port="0" />
|
|
<edge from-layer="132" from-port="0" to-layer="133" to-port="1" />
|
|
<edge from-layer="133" from-port="2" to-layer="135" to-port="0" />
|
|
<edge from-layer="134" from-port="0" to-layer="135" to-port="1" />
|
|
<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="167" 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="144" to-port="0" />
|
|
<edge from-layer="140" from-port="0" to-layer="144" to-port="1" />
|
|
<edge from-layer="141" from-port="0" to-layer="144" to-port="2" />
|
|
<edge from-layer="142" from-port="0" to-layer="144" to-port="3" />
|
|
<edge from-layer="143" from-port="0" to-layer="144" to-port="4" />
|
|
<edge from-layer="144" from-port="5" to-layer="149" to-port="0" />
|
|
<edge from-layer="145" from-port="0" to-layer="146" to-port="0" />
|
|
<edge from-layer="146" from-port="1" to-layer="148" to-port="0" />
|
|
<edge from-layer="147" from-port="0" to-layer="148" to-port="1" />
|
|
<edge from-layer="148" from-port="2" to-layer="149" to-port="1" />
|
|
<edge from-layer="149" from-port="2" to-layer="151" to-port="0" />
|
|
<edge from-layer="150" from-port="0" to-layer="151" to-port="1" />
|
|
<edge from-layer="151" from-port="2" to-layer="152" to-port="0" />
|
|
<edge from-layer="152" from-port="1" to-layer="154" to-port="0" />
|
|
<edge from-layer="153" from-port="0" to-layer="154" to-port="1" />
|
|
<edge from-layer="154" from-port="2" to-layer="159" to-port="0" />
|
|
<edge from-layer="155" from-port="0" to-layer="159" to-port="1" />
|
|
<edge from-layer="156" from-port="0" to-layer="159" to-port="2" />
|
|
<edge from-layer="157" from-port="0" to-layer="159" to-port="3" />
|
|
<edge from-layer="158" from-port="0" to-layer="159" to-port="4" />
|
|
<edge from-layer="159" from-port="5" to-layer="164" to-port="0" />
|
|
<edge from-layer="160" from-port="0" to-layer="161" to-port="0" />
|
|
<edge from-layer="161" from-port="1" to-layer="163" to-port="0" />
|
|
<edge from-layer="162" from-port="0" to-layer="163" to-port="1" />
|
|
<edge from-layer="163" from-port="2" to-layer="164" to-port="1" />
|
|
<edge from-layer="164" from-port="2" to-layer="166" to-port="0" />
|
|
<edge from-layer="165" from-port="0" to-layer="166" to-port="1" />
|
|
<edge from-layer="166" from-port="2" to-layer="167" to-port="0" />
|
|
<edge from-layer="167" from-port="2" to-layer="169" to-port="0" />
|
|
<edge from-layer="168" from-port="0" to-layer="169" to-port="1" />
|
|
<edge from-layer="169" from-port="2" to-layer="171" to-port="0" />
|
|
<edge from-layer="170" from-port="0" to-layer="171" to-port="1" />
|
|
<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="173" from-port="2" to-layer="243" to-port="1" />
|
|
<edge from-layer="174" from-port="0" to-layer="175" to-port="1" />
|
|
<edge from-layer="175" from-port="2" to-layer="180" to-port="0" />
|
|
<edge from-layer="176" from-port="0" to-layer="180" to-port="1" />
|
|
<edge from-layer="177" from-port="0" to-layer="180" to-port="2" />
|
|
<edge from-layer="178" from-port="0" to-layer="180" to-port="3" />
|
|
<edge from-layer="179" from-port="0" to-layer="180" to-port="4" />
|
|
<edge from-layer="180" from-port="5" to-layer="185" to-port="0" />
|
|
<edge from-layer="180" from-port="5" to-layer="201" to-port="0" />
|
|
<edge from-layer="180" from-port="5" to-layer="217" to-port="0" />
|
|
<edge from-layer="181" from-port="0" to-layer="182" to-port="0" />
|
|
<edge from-layer="182" from-port="1" to-layer="184" to-port="0" />
|
|
<edge from-layer="183" from-port="0" to-layer="184" to-port="1" />
|
|
<edge from-layer="184" from-port="2" to-layer="185" to-port="1" />
|
|
<edge from-layer="185" from-port="2" to-layer="187" to-port="0" />
|
|
<edge from-layer="186" from-port="0" to-layer="187" to-port="1" />
|
|
<edge from-layer="187" from-port="2" to-layer="192" to-port="0" />
|
|
<edge from-layer="188" from-port="0" to-layer="192" to-port="1" />
|
|
<edge from-layer="189" from-port="0" to-layer="192" to-port="2" />
|
|
<edge from-layer="190" from-port="0" to-layer="192" to-port="3" />
|
|
<edge from-layer="191" from-port="0" to-layer="192" to-port="4" />
|
|
<edge from-layer="192" from-port="5" to-layer="194" to-port="0" />
|
|
<edge from-layer="193" from-port="0" to-layer="194" to-port="1" />
|
|
<edge from-layer="194" from-port="2" to-layer="196" to-port="0" />
|
|
<edge from-layer="195" from-port="0" to-layer="196" to-port="1" />
|
|
<edge from-layer="196" from-port="2" to-layer="224" to-port="0" />
|
|
<edge from-layer="197" from-port="0" to-layer="198" to-port="0" />
|
|
<edge from-layer="198" from-port="1" 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="201" to-port="1" />
|
|
<edge from-layer="201" from-port="2" to-layer="203" to-port="0" />
|
|
<edge from-layer="202" from-port="0" to-layer="203" to-port="1" />
|
|
<edge from-layer="203" from-port="2" to-layer="208" to-port="0" />
|
|
<edge from-layer="204" from-port="0" to-layer="208" to-port="1" />
|
|
<edge from-layer="205" from-port="0" to-layer="208" to-port="2" />
|
|
<edge from-layer="206" from-port="0" to-layer="208" to-port="3" />
|
|
<edge from-layer="207" from-port="0" to-layer="208" to-port="4" />
|
|
<edge from-layer="208" from-port="5" to-layer="210" to-port="0" />
|
|
<edge from-layer="209" from-port="0" to-layer="210" to-port="1" />
|
|
<edge from-layer="210" from-port="2" to-layer="212" to-port="0" />
|
|
<edge from-layer="211" from-port="0" to-layer="212" to-port="1" />
|
|
<edge from-layer="212" from-port="2" to-layer="224" to-port="1" />
|
|
<edge from-layer="213" from-port="0" to-layer="214" to-port="0" />
|
|
<edge from-layer="214" from-port="1" 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="221" to-port="0" />
|
|
<edge from-layer="220" from-port="0" to-layer="221" to-port="1" />
|
|
<edge from-layer="221" from-port="2" to-layer="223" to-port="0" />
|
|
<edge from-layer="222" from-port="0" to-layer="223" to-port="1" />
|
|
<edge from-layer="223" from-port="2" to-layer="224" to-port="2" />
|
|
<edge from-layer="224" from-port="4" to-layer="226" to-port="0" />
|
|
<edge from-layer="225" from-port="0" 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="235" to-port="0" />
|
|
<edge from-layer="231" from-port="0" to-layer="235" to-port="1" />
|
|
<edge from-layer="232" from-port="0" to-layer="235" to-port="2" />
|
|
<edge from-layer="233" from-port="0" to-layer="235" to-port="3" />
|
|
<edge from-layer="234" from-port="0" to-layer="235" to-port="4" />
|
|
<edge from-layer="235" from-port="5" to-layer="240" to-port="0" />
|
|
<edge from-layer="236" from-port="0" to-layer="237" to-port="0" />
|
|
<edge from-layer="237" from-port="1" 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="240" to-port="1" />
|
|
<edge from-layer="240" from-port="2" to-layer="242" to-port="0" />
|
|
<edge from-layer="241" from-port="0" to-layer="242" to-port="1" />
|
|
<edge from-layer="242" from-port="2" to-layer="243" to-port="0" />
|
|
<edge from-layer="243" from-port="2" to-layer="245" to-port="0" />
|
|
<edge from-layer="244" from-port="0" to-layer="245" to-port="1" />
|
|
<edge from-layer="245" from-port="2" to-layer="247" to-port="0" />
|
|
<edge from-layer="246" from-port="0" to-layer="247" to-port="1" />
|
|
<edge from-layer="247" from-port="2" to-layer="249" to-port="0" />
|
|
<edge from-layer="248" from-port="0" to-layer="249" to-port="1" />
|
|
<edge from-layer="249" from-port="2" to-layer="251" to-port="0" />
|
|
<edge from-layer="249" from-port="2" to-layer="279" to-port="1" />
|
|
<edge from-layer="250" from-port="0" to-layer="251" to-port="1" />
|
|
<edge from-layer="251" from-port="2" to-layer="256" to-port="0" />
|
|
<edge from-layer="252" from-port="0" to-layer="256" to-port="1" />
|
|
<edge from-layer="253" from-port="0" to-layer="256" to-port="2" />
|
|
<edge from-layer="254" from-port="0" to-layer="256" to-port="3" />
|
|
<edge from-layer="255" from-port="0" to-layer="256" to-port="4" />
|
|
<edge from-layer="256" from-port="5" to-layer="261" to-port="0" />
|
|
<edge from-layer="257" from-port="0" to-layer="258" to-port="0" />
|
|
<edge from-layer="258" from-port="1" to-layer="260" to-port="0" />
|
|
<edge from-layer="259" from-port="0" to-layer="260" to-port="1" />
|
|
<edge from-layer="260" from-port="2" to-layer="261" to-port="1" />
|
|
<edge from-layer="261" from-port="2" to-layer="263" to-port="0" />
|
|
<edge from-layer="262" from-port="0" to-layer="263" to-port="1" />
|
|
<edge from-layer="263" from-port="2" to-layer="264" to-port="0" />
|
|
<edge from-layer="264" from-port="1" to-layer="266" to-port="0" />
|
|
<edge from-layer="265" from-port="0" to-layer="266" to-port="1" />
|
|
<edge from-layer="266" from-port="2" to-layer="271" to-port="0" />
|
|
<edge from-layer="267" from-port="0" to-layer="271" to-port="1" />
|
|
<edge from-layer="268" from-port="0" to-layer="271" to-port="2" />
|
|
<edge from-layer="269" from-port="0" to-layer="271" to-port="3" />
|
|
<edge from-layer="270" from-port="0" to-layer="271" to-port="4" />
|
|
<edge from-layer="271" from-port="5" to-layer="276" to-port="0" />
|
|
<edge from-layer="272" from-port="0" 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="276" to-port="1" />
|
|
<edge from-layer="276" from-port="2" to-layer="278" to-port="0" />
|
|
<edge from-layer="277" from-port="0" to-layer="278" to-port="1" />
|
|
<edge from-layer="278" from-port="2" to-layer="279" to-port="0" />
|
|
<edge from-layer="279" from-port="2" to-layer="281" to-port="0" />
|
|
<edge from-layer="280" from-port="0" to-layer="281" to-port="1" />
|
|
<edge from-layer="281" from-port="2" to-layer="283" to-port="0" />
|
|
<edge from-layer="282" from-port="0" to-layer="283" to-port="1" />
|
|
<edge from-layer="283" from-port="2" to-layer="285" to-port="0" />
|
|
<edge from-layer="284" from-port="0" to-layer="285" to-port="1" />
|
|
<edge from-layer="285" from-port="2" to-layer="355" 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="292" to-port="0" />
|
|
<edge from-layer="288" from-port="0" to-layer="292" to-port="1" />
|
|
<edge from-layer="289" from-port="0" to-layer="292" to-port="2" />
|
|
<edge from-layer="290" from-port="0" to-layer="292" to-port="3" />
|
|
<edge from-layer="291" from-port="0" to-layer="292" to-port="4" />
|
|
<edge from-layer="292" from-port="5" to-layer="297" to-port="0" />
|
|
<edge from-layer="292" from-port="5" to-layer="313" to-port="0" />
|
|
<edge from-layer="292" from-port="5" to-layer="329" to-port="0" />
|
|
<edge from-layer="293" from-port="0" to-layer="294" to-port="0" />
|
|
<edge from-layer="294" from-port="1" to-layer="296" to-port="0" />
|
|
<edge from-layer="295" from-port="0" to-layer="296" to-port="1" />
|
|
<edge from-layer="296" from-port="2" to-layer="297" to-port="1" />
|
|
<edge from-layer="297" from-port="2" to-layer="299" to-port="0" />
|
|
<edge from-layer="298" from-port="0" to-layer="299" to-port="1" />
|
|
<edge from-layer="299" from-port="2" to-layer="304" to-port="0" />
|
|
<edge from-layer="300" from-port="0" to-layer="304" to-port="1" />
|
|
<edge from-layer="301" from-port="0" to-layer="304" to-port="2" />
|
|
<edge from-layer="302" from-port="0" to-layer="304" to-port="3" />
|
|
<edge from-layer="303" from-port="0" to-layer="304" to-port="4" />
|
|
<edge from-layer="304" from-port="5" to-layer="306" to-port="0" />
|
|
<edge from-layer="305" from-port="0" 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="336" to-port="0" />
|
|
<edge from-layer="309" from-port="0" to-layer="310" to-port="0" />
|
|
<edge from-layer="310" from-port="1" to-layer="312" to-port="0" />
|
|
<edge from-layer="311" from-port="0" to-layer="312" to-port="1" />
|
|
<edge from-layer="312" from-port="2" to-layer="313" to-port="1" />
|
|
<edge from-layer="313" from-port="2" 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="320" to-port="0" />
|
|
<edge from-layer="316" from-port="0" to-layer="320" to-port="1" />
|
|
<edge from-layer="317" from-port="0" to-layer="320" to-port="2" />
|
|
<edge from-layer="318" from-port="0" to-layer="320" to-port="3" />
|
|
<edge from-layer="319" from-port="0" to-layer="320" to-port="4" />
|
|
<edge from-layer="320" from-port="5" to-layer="322" to-port="0" />
|
|
<edge from-layer="321" from-port="0" 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="336" to-port="1" />
|
|
<edge from-layer="325" from-port="0" to-layer="326" to-port="0" />
|
|
<edge from-layer="326" from-port="1" to-layer="328" to-port="0" />
|
|
<edge from-layer="327" from-port="0" to-layer="328" to-port="1" />
|
|
<edge from-layer="328" from-port="2" to-layer="329" to-port="1" />
|
|
<edge from-layer="329" from-port="2" 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="335" to-port="0" />
|
|
<edge from-layer="334" from-port="0" to-layer="335" to-port="1" />
|
|
<edge from-layer="335" from-port="2" to-layer="336" to-port="2" />
|
|
<edge from-layer="336" from-port="4" to-layer="338" to-port="0" />
|
|
<edge from-layer="337" from-port="0" to-layer="338" to-port="1" />
|
|
<edge from-layer="338" from-port="2" to-layer="340" to-port="0" />
|
|
<edge from-layer="339" from-port="0" to-layer="340" to-port="1" />
|
|
<edge from-layer="340" from-port="2" 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="347" to-port="0" />
|
|
<edge from-layer="343" from-port="0" to-layer="347" to-port="1" />
|
|
<edge from-layer="344" from-port="0" to-layer="347" to-port="2" />
|
|
<edge from-layer="345" from-port="0" to-layer="347" to-port="3" />
|
|
<edge from-layer="346" from-port="0" to-layer="347" to-port="4" />
|
|
<edge from-layer="347" from-port="5" to-layer="352" to-port="0" />
|
|
<edge from-layer="348" from-port="0" to-layer="349" to-port="0" />
|
|
<edge from-layer="349" from-port="1" to-layer="351" to-port="0" />
|
|
<edge from-layer="350" from-port="0" to-layer="351" to-port="1" />
|
|
<edge from-layer="351" from-port="2" 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="355" to-port="0" />
|
|
<edge from-layer="355" from-port="2" to-layer="357" to-port="0" />
|
|
<edge from-layer="356" from-port="0" to-layer="357" to-port="1" />
|
|
<edge from-layer="357" from-port="2" to-layer="359" to-port="0" />
|
|
<edge from-layer="358" from-port="0" to-layer="359" to-port="1" />
|
|
<edge from-layer="359" from-port="2" to-layer="361" to-port="0" />
|
|
<edge from-layer="360" from-port="0" to-layer="361" to-port="1" />
|
|
<edge from-layer="361" from-port="2" to-layer="363" to-port="0" />
|
|
<edge from-layer="361" from-port="2" to-layer="391" to-port="1" />
|
|
<edge from-layer="362" from-port="0" to-layer="363" to-port="1" />
|
|
<edge from-layer="363" from-port="2" to-layer="368" to-port="0" />
|
|
<edge from-layer="364" from-port="0" to-layer="368" to-port="1" />
|
|
<edge from-layer="365" from-port="0" to-layer="368" to-port="2" />
|
|
<edge from-layer="366" from-port="0" to-layer="368" to-port="3" />
|
|
<edge from-layer="367" from-port="0" to-layer="368" to-port="4" />
|
|
<edge from-layer="368" from-port="5" to-layer="373" to-port="0" />
|
|
<edge from-layer="369" from-port="0" to-layer="370" to-port="0" />
|
|
<edge from-layer="370" from-port="1" to-layer="372" to-port="0" />
|
|
<edge from-layer="371" from-port="0" to-layer="372" to-port="1" />
|
|
<edge from-layer="372" from-port="2" 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="376" to-port="0" />
|
|
<edge from-layer="376" from-port="1" to-layer="378" to-port="0" />
|
|
<edge from-layer="377" from-port="0" to-layer="378" to-port="1" />
|
|
<edge from-layer="378" from-port="2" to-layer="383" to-port="0" />
|
|
<edge from-layer="379" from-port="0" to-layer="383" to-port="1" />
|
|
<edge from-layer="380" from-port="0" to-layer="383" to-port="2" />
|
|
<edge from-layer="381" from-port="0" to-layer="383" to-port="3" />
|
|
<edge from-layer="382" from-port="0" to-layer="383" to-port="4" />
|
|
<edge from-layer="383" from-port="5" to-layer="388" to-port="0" />
|
|
<edge from-layer="384" from-port="0" to-layer="385" to-port="0" />
|
|
<edge from-layer="385" from-port="1" to-layer="387" to-port="0" />
|
|
<edge from-layer="386" from-port="0" to-layer="387" to-port="1" />
|
|
<edge from-layer="387" from-port="2" to-layer="388" to-port="1" />
|
|
<edge from-layer="388" from-port="2" to-layer="390" to-port="0" />
|
|
<edge from-layer="389" from-port="0" to-layer="390" to-port="1" />
|
|
<edge from-layer="390" from-port="2" to-layer="391" to-port="0" />
|
|
<edge from-layer="391" from-port="2" to-layer="393" to-port="0" />
|
|
<edge from-layer="392" from-port="0" to-layer="393" to-port="1" />
|
|
<edge from-layer="393" from-port="2" to-layer="395" to-port="0" />
|
|
<edge from-layer="394" from-port="0" to-layer="395" to-port="1" />
|
|
<edge from-layer="395" from-port="2" to-layer="397" to-port="0" />
|
|
<edge from-layer="396" from-port="0" to-layer="397" to-port="1" />
|
|
<edge from-layer="397" from-port="2" to-layer="399" to-port="0" />
|
|
<edge from-layer="397" from-port="2" to-layer="467" to-port="1" />
|
|
<edge from-layer="398" from-port="0" to-layer="399" to-port="1" />
|
|
<edge from-layer="399" from-port="2" to-layer="404" to-port="0" />
|
|
<edge from-layer="400" from-port="0" to-layer="404" to-port="1" />
|
|
<edge from-layer="401" from-port="0" to-layer="404" to-port="2" />
|
|
<edge from-layer="402" from-port="0" to-layer="404" to-port="3" />
|
|
<edge from-layer="403" from-port="0" to-layer="404" to-port="4" />
|
|
<edge from-layer="404" from-port="5" to-layer="409" to-port="0" />
|
|
<edge from-layer="404" from-port="5" to-layer="441" to-port="0" />
|
|
<edge from-layer="404" from-port="5" to-layer="425" to-port="0" />
|
|
<edge from-layer="405" from-port="0" to-layer="406" to-port="0" />
|
|
<edge from-layer="406" from-port="1" to-layer="408" to-port="0" />
|
|
<edge from-layer="407" from-port="0" to-layer="408" to-port="1" />
|
|
<edge from-layer="408" from-port="2" 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="416" to-port="0" />
|
|
<edge from-layer="412" from-port="0" to-layer="416" to-port="1" />
|
|
<edge from-layer="413" from-port="0" to-layer="416" to-port="2" />
|
|
<edge from-layer="414" from-port="0" to-layer="416" to-port="3" />
|
|
<edge from-layer="415" from-port="0" to-layer="416" to-port="4" />
|
|
<edge from-layer="416" from-port="5" to-layer="418" to-port="0" />
|
|
<edge from-layer="417" from-port="0" to-layer="418" to-port="1" />
|
|
<edge from-layer="418" from-port="2" to-layer="420" to-port="0" />
|
|
<edge from-layer="419" from-port="0" to-layer="420" to-port="1" />
|
|
<edge from-layer="420" from-port="2" to-layer="448" to-port="0" />
|
|
<edge from-layer="421" from-port="0" to-layer="422" to-port="0" />
|
|
<edge from-layer="422" from-port="1" to-layer="424" to-port="0" />
|
|
<edge from-layer="423" from-port="0" to-layer="424" to-port="1" />
|
|
<edge from-layer="424" from-port="2" to-layer="425" to-port="1" />
|
|
<edge from-layer="425" from-port="2" to-layer="427" to-port="0" />
|
|
<edge from-layer="426" from-port="0" to-layer="427" to-port="1" />
|
|
<edge from-layer="427" from-port="2" to-layer="432" to-port="0" />
|
|
<edge from-layer="428" from-port="0" to-layer="432" to-port="1" />
|
|
<edge from-layer="429" from-port="0" to-layer="432" to-port="2" />
|
|
<edge from-layer="430" from-port="0" to-layer="432" to-port="3" />
|
|
<edge from-layer="431" from-port="0" to-layer="432" to-port="4" />
|
|
<edge from-layer="432" from-port="5" 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="436" to-port="0" />
|
|
<edge from-layer="435" from-port="0" to-layer="436" to-port="1" />
|
|
<edge from-layer="436" from-port="2" to-layer="448" to-port="1" />
|
|
<edge from-layer="437" from-port="0" to-layer="438" to-port="0" />
|
|
<edge from-layer="438" from-port="1" to-layer="440" to-port="0" />
|
|
<edge from-layer="439" from-port="0" to-layer="440" to-port="1" />
|
|
<edge from-layer="440" from-port="2" to-layer="441" to-port="1" />
|
|
<edge from-layer="441" from-port="2" to-layer="443" to-port="0" />
|
|
<edge from-layer="442" from-port="0" to-layer="443" to-port="1" />
|
|
<edge from-layer="443" from-port="2" to-layer="445" to-port="0" />
|
|
<edge from-layer="444" from-port="0" to-layer="445" to-port="1" />
|
|
<edge from-layer="445" from-port="2" to-layer="447" to-port="0" />
|
|
<edge from-layer="446" from-port="0" to-layer="447" to-port="1" />
|
|
<edge from-layer="447" from-port="2" to-layer="448" to-port="2" />
|
|
<edge from-layer="448" from-port="4" 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="452" to-port="0" />
|
|
<edge from-layer="451" from-port="0" to-layer="452" to-port="1" />
|
|
<edge from-layer="452" from-port="2" to-layer="454" to-port="0" />
|
|
<edge from-layer="453" from-port="0" to-layer="454" to-port="1" />
|
|
<edge from-layer="454" from-port="2" to-layer="459" to-port="0" />
|
|
<edge from-layer="455" from-port="0" to-layer="459" to-port="1" />
|
|
<edge from-layer="456" from-port="0" to-layer="459" to-port="2" />
|
|
<edge from-layer="457" from-port="0" to-layer="459" to-port="3" />
|
|
<edge from-layer="458" from-port="0" to-layer="459" to-port="4" />
|
|
<edge from-layer="459" from-port="5" to-layer="464" to-port="0" />
|
|
<edge from-layer="460" from-port="0" to-layer="461" to-port="0" />
|
|
<edge from-layer="461" from-port="1" to-layer="463" to-port="0" />
|
|
<edge from-layer="462" from-port="0" to-layer="463" to-port="1" />
|
|
<edge from-layer="463" from-port="2" 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="0" />
|
|
<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="475" to-port="0" />
|
|
<edge from-layer="473" from-port="2" to-layer="503" to-port="1" />
|
|
<edge from-layer="474" from-port="0" to-layer="475" to-port="1" />
|
|
<edge from-layer="475" from-port="2" to-layer="480" to-port="0" />
|
|
<edge from-layer="476" from-port="0" to-layer="480" to-port="1" />
|
|
<edge from-layer="477" from-port="0" to-layer="480" to-port="2" />
|
|
<edge from-layer="478" from-port="0" to-layer="480" to-port="3" />
|
|
<edge from-layer="479" from-port="0" to-layer="480" to-port="4" />
|
|
<edge from-layer="480" from-port="5" to-layer="485" to-port="0" />
|
|
<edge from-layer="481" from-port="0" to-layer="482" to-port="0" />
|
|
<edge from-layer="482" from-port="1" to-layer="484" to-port="0" />
|
|
<edge from-layer="483" from-port="0" to-layer="484" to-port="1" />
|
|
<edge from-layer="484" from-port="2" to-layer="485" to-port="1" />
|
|
<edge from-layer="485" from-port="2" to-layer="487" to-port="0" />
|
|
<edge from-layer="486" from-port="0" to-layer="487" to-port="1" />
|
|
<edge from-layer="487" from-port="2" to-layer="488" to-port="0" />
|
|
<edge from-layer="488" from-port="1" 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="495" to-port="0" />
|
|
<edge from-layer="491" from-port="0" to-layer="495" to-port="1" />
|
|
<edge from-layer="492" from-port="0" to-layer="495" to-port="2" />
|
|
<edge from-layer="493" from-port="0" to-layer="495" to-port="3" />
|
|
<edge from-layer="494" from-port="0" to-layer="495" to-port="4" />
|
|
<edge from-layer="495" from-port="5" to-layer="500" to-port="0" />
|
|
<edge from-layer="496" from-port="0" to-layer="497" to-port="0" />
|
|
<edge from-layer="497" from-port="1" to-layer="499" to-port="0" />
|
|
<edge from-layer="498" from-port="0" to-layer="499" to-port="1" />
|
|
<edge from-layer="499" from-port="2" to-layer="500" to-port="1" />
|
|
<edge from-layer="500" from-port="2" to-layer="502" to-port="0" />
|
|
<edge from-layer="501" from-port="0" to-layer="502" to-port="1" />
|
|
<edge from-layer="502" from-port="2" to-layer="503" to-port="0" />
|
|
<edge from-layer="503" from-port="2" to-layer="505" to-port="0" />
|
|
<edge from-layer="504" from-port="0" to-layer="505" to-port="1" />
|
|
<edge from-layer="505" from-port="2" to-layer="507" to-port="0" />
|
|
<edge from-layer="506" from-port="0" to-layer="507" to-port="1" />
|
|
<edge from-layer="507" from-port="2" 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="511" to-port="0" />
|
|
<edge from-layer="509" from-port="2" to-layer="579" to-port="1" />
|
|
<edge from-layer="510" from-port="0" to-layer="511" to-port="1" />
|
|
<edge from-layer="511" from-port="2" to-layer="516" to-port="0" />
|
|
<edge from-layer="512" from-port="0" to-layer="516" to-port="1" />
|
|
<edge from-layer="513" from-port="0" to-layer="516" to-port="2" />
|
|
<edge from-layer="514" from-port="0" to-layer="516" to-port="3" />
|
|
<edge from-layer="515" from-port="0" to-layer="516" to-port="4" />
|
|
<edge from-layer="516" from-port="5" to-layer="521" to-port="0" />
|
|
<edge from-layer="516" from-port="5" to-layer="537" to-port="0" />
|
|
<edge from-layer="516" from-port="5" to-layer="553" to-port="0" />
|
|
<edge from-layer="517" from-port="0" to-layer="518" to-port="0" />
|
|
<edge from-layer="518" from-port="1" to-layer="520" to-port="0" />
|
|
<edge from-layer="519" from-port="0" to-layer="520" to-port="1" />
|
|
<edge from-layer="520" from-port="2" to-layer="521" to-port="1" />
|
|
<edge from-layer="521" from-port="2" to-layer="523" to-port="0" />
|
|
<edge from-layer="522" from-port="0" to-layer="523" to-port="1" />
|
|
<edge from-layer="523" from-port="2" to-layer="528" to-port="0" />
|
|
<edge from-layer="524" from-port="0" to-layer="528" to-port="1" />
|
|
<edge from-layer="525" from-port="0" to-layer="528" to-port="2" />
|
|
<edge from-layer="526" from-port="0" to-layer="528" to-port="3" />
|
|
<edge from-layer="527" from-port="0" to-layer="528" to-port="4" />
|
|
<edge from-layer="528" from-port="5" to-layer="530" to-port="0" />
|
|
<edge from-layer="529" from-port="0" to-layer="530" to-port="1" />
|
|
<edge from-layer="530" from-port="2" to-layer="532" to-port="0" />
|
|
<edge from-layer="531" from-port="0" to-layer="532" to-port="1" />
|
|
<edge from-layer="532" from-port="2" to-layer="560" to-port="0" />
|
|
<edge from-layer="533" from-port="0" to-layer="534" to-port="0" />
|
|
<edge from-layer="534" from-port="1" to-layer="536" to-port="0" />
|
|
<edge from-layer="535" from-port="0" to-layer="536" to-port="1" />
|
|
<edge from-layer="536" from-port="2" to-layer="537" to-port="1" />
|
|
<edge from-layer="537" from-port="2" 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="544" to-port="0" />
|
|
<edge from-layer="540" from-port="0" to-layer="544" to-port="1" />
|
|
<edge from-layer="541" from-port="0" to-layer="544" to-port="2" />
|
|
<edge from-layer="542" from-port="0" to-layer="544" to-port="3" />
|
|
<edge from-layer="543" from-port="0" to-layer="544" to-port="4" />
|
|
<edge from-layer="544" from-port="5" to-layer="546" to-port="0" />
|
|
<edge from-layer="545" from-port="0" to-layer="546" to-port="1" />
|
|
<edge from-layer="546" from-port="2" to-layer="548" to-port="0" />
|
|
<edge from-layer="547" from-port="0" to-layer="548" to-port="1" />
|
|
<edge from-layer="548" from-port="2" to-layer="560" to-port="1" />
|
|
<edge from-layer="549" from-port="0" to-layer="550" to-port="0" />
|
|
<edge from-layer="550" from-port="1" to-layer="552" to-port="0" />
|
|
<edge from-layer="551" from-port="0" to-layer="552" to-port="1" />
|
|
<edge from-layer="552" from-port="2" to-layer="553" to-port="1" />
|
|
<edge from-layer="553" from-port="2" 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="557" to-port="0" />
|
|
<edge from-layer="556" from-port="0" to-layer="557" to-port="1" />
|
|
<edge from-layer="557" from-port="2" to-layer="559" to-port="0" />
|
|
<edge from-layer="558" from-port="0" to-layer="559" to-port="1" />
|
|
<edge from-layer="559" from-port="2" to-layer="560" to-port="2" />
|
|
<edge from-layer="560" from-port="4" to-layer="562" to-port="0" />
|
|
<edge from-layer="561" from-port="0" to-layer="562" to-port="1" />
|
|
<edge from-layer="562" from-port="2" to-layer="564" to-port="0" />
|
|
<edge from-layer="563" from-port="0" to-layer="564" to-port="1" />
|
|
<edge from-layer="564" from-port="2" to-layer="566" to-port="0" />
|
|
<edge from-layer="565" from-port="0" to-layer="566" to-port="1" />
|
|
<edge from-layer="566" from-port="2" to-layer="571" to-port="0" />
|
|
<edge from-layer="567" from-port="0" to-layer="571" to-port="1" />
|
|
<edge from-layer="568" from-port="0" to-layer="571" to-port="2" />
|
|
<edge from-layer="569" from-port="0" to-layer="571" to-port="3" />
|
|
<edge from-layer="570" from-port="0" to-layer="571" to-port="4" />
|
|
<edge from-layer="571" from-port="5" to-layer="576" to-port="0" />
|
|
<edge from-layer="572" from-port="0" to-layer="573" to-port="0" />
|
|
<edge from-layer="573" from-port="1" to-layer="575" to-port="0" />
|
|
<edge from-layer="574" from-port="0" to-layer="575" to-port="1" />
|
|
<edge from-layer="575" from-port="2" to-layer="576" to-port="1" />
|
|
<edge from-layer="576" from-port="2" to-layer="578" to-port="0" />
|
|
<edge from-layer="577" from-port="0" to-layer="578" to-port="1" />
|
|
<edge from-layer="578" from-port="2" to-layer="579" to-port="0" />
|
|
<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="585" to-port="0" />
|
|
<edge from-layer="584" from-port="0" to-layer="585" to-port="1" />
|
|
<edge from-layer="585" from-port="2" to-layer="587" to-port="0" />
|
|
<edge from-layer="585" from-port="2" to-layer="615" to-port="1" />
|
|
<edge from-layer="586" from-port="0" to-layer="587" to-port="1" />
|
|
<edge from-layer="587" from-port="2" to-layer="592" to-port="0" />
|
|
<edge from-layer="588" from-port="0" to-layer="592" to-port="1" />
|
|
<edge from-layer="589" from-port="0" to-layer="592" to-port="2" />
|
|
<edge from-layer="590" from-port="0" to-layer="592" to-port="3" />
|
|
<edge from-layer="591" from-port="0" to-layer="592" to-port="4" />
|
|
<edge from-layer="592" from-port="5" to-layer="597" to-port="0" />
|
|
<edge from-layer="593" from-port="0" to-layer="594" to-port="0" />
|
|
<edge from-layer="594" from-port="1" to-layer="596" to-port="0" />
|
|
<edge from-layer="595" from-port="0" to-layer="596" to-port="1" />
|
|
<edge from-layer="596" from-port="2" to-layer="597" to-port="1" />
|
|
<edge from-layer="597" from-port="2" 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="0" />
|
|
<edge from-layer="600" from-port="1" 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="607" to-port="0" />
|
|
<edge from-layer="603" from-port="0" to-layer="607" to-port="1" />
|
|
<edge from-layer="604" from-port="0" to-layer="607" to-port="2" />
|
|
<edge from-layer="605" from-port="0" to-layer="607" to-port="3" />
|
|
<edge from-layer="606" from-port="0" to-layer="607" to-port="4" />
|
|
<edge from-layer="607" from-port="5" to-layer="612" to-port="0" />
|
|
<edge from-layer="608" from-port="0" to-layer="609" to-port="0" />
|
|
<edge from-layer="609" from-port="1" to-layer="611" to-port="0" />
|
|
<edge from-layer="610" from-port="0" to-layer="611" to-port="1" />
|
|
<edge from-layer="611" from-port="2" to-layer="612" to-port="1" />
|
|
<edge from-layer="612" from-port="2" to-layer="614" to-port="0" />
|
|
<edge from-layer="613" from-port="0" to-layer="614" to-port="1" />
|
|
<edge from-layer="614" from-port="2" to-layer="615" to-port="0" />
|
|
<edge from-layer="615" from-port="2" to-layer="617" to-port="0" />
|
|
<edge from-layer="616" from-port="0" to-layer="617" to-port="1" />
|
|
<edge from-layer="617" from-port="2" to-layer="619" to-port="0" />
|
|
<edge from-layer="618" from-port="0" to-layer="619" to-port="1" />
|
|
<edge from-layer="619" from-port="2" to-layer="621" to-port="0" />
|
|
<edge from-layer="620" from-port="0" to-layer="621" to-port="1" />
|
|
<edge from-layer="621" from-port="2" to-layer="623" to-port="0" />
|
|
<edge from-layer="621" from-port="2" to-layer="691" to-port="1" />
|
|
<edge from-layer="622" from-port="0" to-layer="623" to-port="1" />
|
|
<edge from-layer="623" from-port="2" to-layer="628" to-port="0" />
|
|
<edge from-layer="624" from-port="0" to-layer="628" to-port="1" />
|
|
<edge from-layer="625" from-port="0" to-layer="628" to-port="2" />
|
|
<edge from-layer="626" from-port="0" to-layer="628" to-port="3" />
|
|
<edge from-layer="627" from-port="0" to-layer="628" to-port="4" />
|
|
<edge from-layer="628" from-port="5" to-layer="633" to-port="0" />
|
|
<edge from-layer="628" from-port="5" to-layer="649" to-port="0" />
|
|
<edge from-layer="628" from-port="5" to-layer="665" to-port="0" />
|
|
<edge from-layer="629" from-port="0" to-layer="630" to-port="0" />
|
|
<edge from-layer="630" from-port="1" to-layer="632" to-port="0" />
|
|
<edge from-layer="631" from-port="0" to-layer="632" to-port="1" />
|
|
<edge from-layer="632" from-port="2" to-layer="633" to-port="1" />
|
|
<edge from-layer="633" from-port="2" 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="640" to-port="0" />
|
|
<edge from-layer="636" from-port="0" to-layer="640" to-port="1" />
|
|
<edge from-layer="637" from-port="0" to-layer="640" to-port="2" />
|
|
<edge from-layer="638" from-port="0" to-layer="640" to-port="3" />
|
|
<edge from-layer="639" from-port="0" to-layer="640" to-port="4" />
|
|
<edge from-layer="640" from-port="5" to-layer="642" to-port="0" />
|
|
<edge from-layer="641" from-port="0" to-layer="642" to-port="1" />
|
|
<edge from-layer="642" from-port="2" to-layer="644" to-port="0" />
|
|
<edge from-layer="643" from-port="0" to-layer="644" to-port="1" />
|
|
<edge from-layer="644" from-port="2" to-layer="672" to-port="0" />
|
|
<edge from-layer="645" from-port="0" to-layer="646" to-port="0" />
|
|
<edge from-layer="646" from-port="1" to-layer="648" to-port="0" />
|
|
<edge from-layer="647" from-port="0" to-layer="648" to-port="1" />
|
|
<edge from-layer="648" from-port="2" to-layer="649" to-port="1" />
|
|
<edge from-layer="649" from-port="2" to-layer="651" to-port="0" />
|
|
<edge from-layer="650" from-port="0" to-layer="651" to-port="1" />
|
|
<edge from-layer="651" from-port="2" to-layer="656" to-port="0" />
|
|
<edge from-layer="652" from-port="0" to-layer="656" to-port="1" />
|
|
<edge from-layer="653" from-port="0" to-layer="656" to-port="2" />
|
|
<edge from-layer="654" from-port="0" to-layer="656" to-port="3" />
|
|
<edge from-layer="655" from-port="0" to-layer="656" to-port="4" />
|
|
<edge from-layer="656" from-port="5" to-layer="658" to-port="0" />
|
|
<edge from-layer="657" from-port="0" to-layer="658" to-port="1" />
|
|
<edge from-layer="658" from-port="2" to-layer="660" to-port="0" />
|
|
<edge from-layer="659" from-port="0" to-layer="660" to-port="1" />
|
|
<edge from-layer="660" from-port="2" to-layer="672" to-port="1" />
|
|
<edge from-layer="661" from-port="0" to-layer="662" to-port="0" />
|
|
<edge from-layer="662" from-port="1" to-layer="664" to-port="0" />
|
|
<edge from-layer="663" from-port="0" to-layer="664" to-port="1" />
|
|
<edge from-layer="664" from-port="2" to-layer="665" to-port="1" />
|
|
<edge from-layer="665" from-port="2" to-layer="667" to-port="0" />
|
|
<edge from-layer="666" from-port="0" to-layer="667" to-port="1" />
|
|
<edge from-layer="667" from-port="2" to-layer="669" to-port="0" />
|
|
<edge from-layer="668" from-port="0" to-layer="669" to-port="1" />
|
|
<edge from-layer="669" from-port="2" to-layer="671" to-port="0" />
|
|
<edge from-layer="670" from-port="0" to-layer="671" to-port="1" />
|
|
<edge from-layer="671" from-port="2" to-layer="672" to-port="2" />
|
|
<edge from-layer="672" from-port="4" to-layer="674" to-port="0" />
|
|
<edge from-layer="673" from-port="0" to-layer="674" to-port="1" />
|
|
<edge from-layer="674" from-port="2" to-layer="676" to-port="0" />
|
|
<edge from-layer="675" from-port="0" to-layer="676" to-port="1" />
|
|
<edge from-layer="676" from-port="2" to-layer="678" to-port="0" />
|
|
<edge from-layer="677" from-port="0" to-layer="678" to-port="1" />
|
|
<edge from-layer="678" from-port="2" to-layer="683" to-port="0" />
|
|
<edge from-layer="679" from-port="0" to-layer="683" to-port="1" />
|
|
<edge from-layer="680" from-port="0" to-layer="683" to-port="2" />
|
|
<edge from-layer="681" from-port="0" to-layer="683" to-port="3" />
|
|
<edge from-layer="682" from-port="0" to-layer="683" to-port="4" />
|
|
<edge from-layer="683" from-port="5" to-layer="688" to-port="0" />
|
|
<edge from-layer="684" from-port="0" to-layer="685" to-port="0" />
|
|
<edge from-layer="685" from-port="1" 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="1" />
|
|
<edge from-layer="688" from-port="2" to-layer="690" to-port="0" />
|
|
<edge from-layer="689" from-port="0" to-layer="690" to-port="1" />
|
|
<edge from-layer="690" from-port="2" to-layer="691" to-port="0" />
|
|
<edge from-layer="691" from-port="2" 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="695" to-port="0" />
|
|
<edge from-layer="694" from-port="0" to-layer="695" to-port="1" />
|
|
<edge from-layer="695" from-port="2" to-layer="697" to-port="0" />
|
|
<edge from-layer="696" from-port="0" to-layer="697" to-port="1" />
|
|
<edge from-layer="697" from-port="2" to-layer="699" to-port="0" />
|
|
<edge from-layer="697" from-port="2" to-layer="727" to-port="1" />
|
|
<edge from-layer="698" from-port="0" to-layer="699" to-port="1" />
|
|
<edge from-layer="699" from-port="2" to-layer="704" to-port="0" />
|
|
<edge from-layer="700" from-port="0" to-layer="704" to-port="1" />
|
|
<edge from-layer="701" from-port="0" to-layer="704" to-port="2" />
|
|
<edge from-layer="702" from-port="0" to-layer="704" to-port="3" />
|
|
<edge from-layer="703" from-port="0" to-layer="704" to-port="4" />
|
|
<edge from-layer="704" from-port="5" to-layer="709" to-port="0" />
|
|
<edge from-layer="705" from-port="0" to-layer="706" to-port="0" />
|
|
<edge from-layer="706" from-port="1" to-layer="708" to-port="0" />
|
|
<edge from-layer="707" from-port="0" to-layer="708" to-port="1" />
|
|
<edge from-layer="708" from-port="2" to-layer="709" to-port="1" />
|
|
<edge from-layer="709" from-port="2" to-layer="711" to-port="0" />
|
|
<edge from-layer="710" from-port="0" to-layer="711" to-port="1" />
|
|
<edge from-layer="711" from-port="2" to-layer="712" to-port="0" />
|
|
<edge from-layer="712" from-port="1" to-layer="714" to-port="0" />
|
|
<edge from-layer="713" from-port="0" to-layer="714" to-port="1" />
|
|
<edge from-layer="714" from-port="2" to-layer="719" to-port="0" />
|
|
<edge from-layer="715" from-port="0" to-layer="719" to-port="1" />
|
|
<edge from-layer="716" from-port="0" to-layer="719" to-port="2" />
|
|
<edge from-layer="717" from-port="0" to-layer="719" to-port="3" />
|
|
<edge from-layer="718" from-port="0" to-layer="719" to-port="4" />
|
|
<edge from-layer="719" from-port="5" to-layer="724" to-port="0" />
|
|
<edge from-layer="720" from-port="0" to-layer="721" to-port="0" />
|
|
<edge from-layer="721" from-port="1" to-layer="723" to-port="0" />
|
|
<edge from-layer="722" from-port="0" to-layer="723" to-port="1" />
|
|
<edge from-layer="723" from-port="2" 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="727" to-port="0" />
|
|
<edge from-layer="727" from-port="2" to-layer="729" to-port="0" />
|
|
<edge from-layer="728" from-port="0" to-layer="729" to-port="1" />
|
|
<edge from-layer="729" from-port="2" to-layer="731" to-port="0" />
|
|
<edge from-layer="730" from-port="0" to-layer="731" to-port="1" />
|
|
<edge from-layer="731" from-port="2" to-layer="733" to-port="0" />
|
|
<edge from-layer="732" from-port="0" to-layer="733" to-port="1" />
|
|
<edge from-layer="733" from-port="2" to-layer="735" to-port="0" />
|
|
<edge from-layer="734" from-port="0" to-layer="735" to-port="1" />
|
|
<edge from-layer="735" from-port="3" 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="742" to-port="0" />
|
|
<edge from-layer="738" from-port="0" to-layer="742" to-port="1" />
|
|
<edge from-layer="739" from-port="0" to-layer="742" to-port="2" />
|
|
<edge from-layer="740" from-port="0" to-layer="742" to-port="3" />
|
|
<edge from-layer="741" from-port="0" to-layer="742" to-port="4" />
|
|
<edge from-layer="742" from-port="5" to-layer="747" to-port="0" />
|
|
<edge from-layer="743" from-port="0" to-layer="744" to-port="0" />
|
|
<edge from-layer="744" from-port="1" to-layer="746" to-port="0" />
|
|
<edge from-layer="745" from-port="0" to-layer="746" to-port="1" />
|
|
<edge from-layer="746" from-port="2" to-layer="747" to-port="1" />
|
|
<edge from-layer="747" from-port="2" to-layer="749" to-port="0" />
|
|
<edge from-layer="748" from-port="0" to-layer="749" to-port="1" />
|
|
<edge from-layer="749" from-port="2" 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="757" to-port="0" />
|
|
<edge from-layer="753" from-port="0" to-layer="757" to-port="1" />
|
|
<edge from-layer="754" from-port="0" to-layer="757" to-port="2" />
|
|
<edge from-layer="755" from-port="0" to-layer="757" to-port="3" />
|
|
<edge from-layer="756" from-port="0" to-layer="757" to-port="4" />
|
|
<edge from-layer="757" from-port="5" to-layer="762" to-port="0" />
|
|
<edge from-layer="758" from-port="0" to-layer="759" to-port="0" />
|
|
<edge from-layer="759" from-port="1" to-layer="761" to-port="0" />
|
|
<edge from-layer="760" from-port="0" to-layer="761" to-port="1" />
|
|
<edge from-layer="761" from-port="2" to-layer="762" to-port="1" />
|
|
<edge from-layer="762" from-port="2" to-layer="764" to-port="0" />
|
|
<edge from-layer="763" from-port="0" to-layer="764" to-port="1" />
|
|
<edge from-layer="764" from-port="2" to-layer="765" 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="1.20.1" />
|
|
<optimum_version value="1.24.0" />
|
|
<pytorch_version value="2.6.0+cu124" />
|
|
<transformers_version value="4.52.0.dev0" />
|
|
</optimum>
|
|
</rt_info>
|
|
</net>
|