0b18e2ded3
Model: sentence-transformers/facebook-dpr-ctx_encoder-multiset-base Source: Original Platform
21612 lines
692 KiB
XML
21612 lines
692 KiB
XML
<?xml version="1.0"?>
|
|
<net name="Model0" 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_43740286" 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_43740287" 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_43740288" 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.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.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.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.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="79,inputs_embeds">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="10" name="Constant_43740290" 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_43740291" 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_43740292" 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.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.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.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.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="81,token_type_embeddings.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="17" name="__module.embeddings/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="82_1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="18" name="Constant_43740294" 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_43740295" 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_43740296" 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.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.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="76">
|
|
<dim>1</dim>
|
|
<dim>512</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="23" name="__module.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_9038" 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_9159" 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_9040" 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_9041" 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="10,72,74,75,8">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="28" name="__module.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.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.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="77">
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="31" name="__module.embeddings.position_embeddings/aten::embedding/Convert" type="Convert" version="opset1">
|
|
<data destination_type="i32" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I32">
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="32" name="__module.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.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="84,position_embeddings.1">
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="34" name="__module.embeddings/aten::add_/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="82,embeddings.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="35" name="__module.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.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_8811" 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.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_8812" 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.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="89,input.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="41" name="__module.embeddings.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="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.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.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.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.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.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.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_43740298" 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_43740299" 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_43740300" 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.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.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_8813" 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.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="141,x.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="55" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="24569104" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="56" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="24569108" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="57" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="24569104" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="58" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="24569108" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="59" name="__module.encoder.layer.0.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="60" name="__module.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.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="145,x.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="62" name="Constant_259" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="146">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="63" name="__module.encoder.layer.0.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="147">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="64" name="Constant_43740302" 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_43740303" 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_43740304" 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.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.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_8814" 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.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="150,x.5">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="71" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25165144" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="72" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25165148" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="73" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25165144" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="74" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25165148" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="75" name="__module.encoder.layer.0.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="76" name="__module.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.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="154,x.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="78" name="Constant_284" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="155">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="79" name="__module.encoder.layer.0.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="156">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="80" name="Constant_43740306" 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_43740307" 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_43740308" 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.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.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_8815" 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.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="159,x.9">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="87" name="__module.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.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="163,x.11">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="89" name="Constant_309" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="164">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="90" name="__module.encoder.layer.0.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="165">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="91" name="Constant_8817" 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="25" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963900" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" names="25" />
|
|
</output>
|
|
</layer>
|
|
<layer id="93" name="aten::unsqueeze/Unsqueeze" type="Unsqueeze" version="opset1">
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I64" />
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="26">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="94" name="27" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="25761124" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" names="27" />
|
|
</output>
|
|
</layer>
|
|
<layer id="95" name="aten::unsqueeze/Unsqueeze_1" type="Unsqueeze" version="opset1">
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="I64" />
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="28,33">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="96" name="ShapeOf_9046" 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="97" name="Constant_9162" 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="98" name="Constant_9048" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="99" name="Gather_9049" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="2" precision="I64" />
|
|
</input>
|
|
<output>
|
|
<port id="3" precision="I64" names="13,15">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="100" name="Constant_7435" 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="101" name="Constant_9165" 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="102" name="Constant_9056" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="103" name="Gather_9057" 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="17,19">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="104" name="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>1</dim>
|
|
</port>
|
|
<port id="3" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="4" precision="I64" names="35">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="105" name="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="37">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="106" name="aten::to/Convert" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="42">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="107" name="Constant_8816" 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="108" name="aten::rsub/Multiply" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="109" name="aten::rsub/Subtract" type="Subtract" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="45,inverted_mask">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="110" name="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="50">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="111" name="aten::masked_fill/ConvertLike" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="25761132" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="112" name="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="52">
|
|
<dim>-1</dim>
|
|
<dim>1</dim>
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="113" name="__module.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="166,attn_output.1">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="114" name="__module.encoder.layer.0.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="115" name="__module.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="167,attn_output.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="116" name="__module.encoder.layer.0.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="117" name="Constant_7880" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="118" name="Constant_7881" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="119" name="Gather_7882" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="120" name="__module.encoder.layer.0.attention.self/prim::ListConstruct/Reshape_1_3" type="Const" version="opset1">
|
|
<data element_type="i64" shape="1" offset="25761168" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="121" name="__module.encoder.layer.0.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="168">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="122" name="__module.encoder.layer.0.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="169">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="123" name="__module.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="25761176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="124" name="__module.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="125" name="__module.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="25764248" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="126" name="__module.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="25764252" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="127" name="__module.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="25764248" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="128" name="__module.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="25764252" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="129" name="__module.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="130" name="Constant_43740310" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="25764256" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="131" name="Convert_43740311" 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="132" name="Constant_43740312" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="26354080" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="133" name="__module.encoder.layer.0.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="134" name="__module.encoder.layer.0.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="135" name="Constant_8818" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="26357152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="136" name="__module.encoder.layer.0.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="175,input.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="137" name="__module.encoder.layer.0.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="177">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="138" name="__module.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="139" name="__module.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="140" name="Constant_8819" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="26360224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="141" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/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="142" name="Constant_8820" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="26363296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="143" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add" 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="181,input_tensor.1">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="144" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="26366368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="145" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="146" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="26369440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="147" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="26369444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="148" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="26369440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="149" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="26369444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="150" name="__module.encoder.layer.0.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="151" name="Constant_43740314" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="26369448" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="152" name="Convert_43740315" 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="153" name="Constant_43740316" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="28728744" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="154" name="__module.encoder.layer.0.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="155" name="__module.encoder.layer.0.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="156" name="Constant_8821" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="28741032" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="157" name="__module.encoder.layer.0.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="186">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="158" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu" 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="187">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="159" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="28753320" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="160" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="161" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="28765608" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="162" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="28765612" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="163" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="28765608" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="164" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="28765612" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="165" name="__module.encoder.layer.0.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="166" name="Constant_43740318" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="28765616" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="167" name="Convert_43740319" 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="168" name="Constant_43740320" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="31124912" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="169" name="__module.encoder.layer.0.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="170" name="__module.encoder.layer.0.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="171" name="Constant_8822" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="31127984" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="172" name="__module.encoder.layer.0.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="193,input.5">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="173" name="__module.encoder.layer.0.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="195">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="174" name="__module.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="175" name="__module.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="176" name="Constant_8823" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="31131056" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="177" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/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="178" name="Constant_8824" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="31134128" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="179" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add" 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,hidden_states.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="180" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="31137200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="181" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="182" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31140272" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="183" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31140276" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="184" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31140272" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="185" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31140276" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="186" name="__module.encoder.layer.0.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="187" name="Constant_43740322" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="31140280" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="188" name="Convert_43740323" 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="189" name="Constant_43740324" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="31730104" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="190" name="__module.encoder.layer.1.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="191" name="__module.encoder.layer.1.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="192" name="Constant_8825" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="31733176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="193" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="212,x.13">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="194" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31736248" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="195" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31736252" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="196" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31736248" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="197" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="31736252" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="198" name="__module.encoder.layer.1.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="199" name="__module.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="200" name="__module.encoder.layer.1.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="216,x.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="201" name="Constant_467" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="217">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="202" name="__module.encoder.layer.1.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="218">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="203" name="Constant_43740326" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="31736256" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="204" name="Convert_43740327" 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="205" name="Constant_43740328" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="32326080" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="206" name="__module.encoder.layer.1.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="207" name="__module.encoder.layer.1.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="208" name="Constant_8826" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="32329152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="209" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="221,x.17">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="210" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32332224" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="211" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32332228" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="212" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32332224" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="213" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="32332228" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="214" name="__module.encoder.layer.1.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="215" name="__module.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="216" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="225,x.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="217" name="Constant_490" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="226">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="218" name="__module.encoder.layer.1.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="227">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="219" name="Constant_43740330" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="32332232" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="220" name="Convert_43740331" 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="221" name="Constant_43740332" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="32922056" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="222" name="__module.encoder.layer.1.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="223" name="__module.encoder.layer.1.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="224" name="Constant_8827" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="32925128" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="225" name="__module.encoder.layer.1.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="230,x.21">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="226" name="__module.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="227" name="__module.encoder.layer.1.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="234,x.23">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="228" name="Constant_513" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="235">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="229" name="__module.encoder.layer.1.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="236">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="230" name="__module.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="237,attn_output.5">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="231" name="__module.encoder.layer.1.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="232" name="__module.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="238,attn_output.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="233" name="__module.encoder.layer.1.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="234" name="Constant_7900" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="235" name="Constant_7901" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="236" name="Gather_7902" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="237" name="__module.encoder.layer.1.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="239">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="238" name="__module.encoder.layer.1.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="240">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="239" name="__module.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="32928200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="240" name="__module.encoder.layer.1.attention.self/aten::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="241" name="__module.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="32931272" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="242" name="__module.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="32931276" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="243" name="__module.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="32931272" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="244" name="__module.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="32931276" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="245" name="__module.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="246" name="Constant_43740334" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="32931280" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="247" name="Convert_43740335" 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="248" name="Constant_43740336" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="33521104" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="249" name="__module.encoder.layer.1.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="250" name="__module.encoder.layer.1.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="251" name="Constant_8828" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="33524176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="252" name="__module.encoder.layer.1.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="246,input.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="253" name="__module.encoder.layer.1.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="248">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="254" name="__module.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="255" name="__module.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="256" name="Constant_8829" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="33527248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="257" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="258" name="Constant_8830" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="33530320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="259" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add" 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="252,input_tensor.3">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="260" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="33533392" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="261" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" 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="262" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="33536464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="263" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="33536468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="264" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="33536464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="265" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="33536468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="266" name="__module.encoder.layer.1.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="267" name="Constant_43740338" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="33536472" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="268" name="Convert_43740339" 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="269" name="Constant_43740340" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="35895768" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="270" name="__module.encoder.layer.1.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="271" name="__module.encoder.layer.1.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="272" name="Constant_8831" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="35908056" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="273" name="__module.encoder.layer.1.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="257">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="274" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu" 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="258">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="275" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="35920344" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="276" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" 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="277" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="35932632" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="278" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="35932636" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="279" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="35932632" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="280" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="35932636" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="281" name="__module.encoder.layer.1.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="282" name="Constant_43740342" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="35932640" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="283" name="Convert_43740343" 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="284" name="Constant_43740344" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="38291936" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="285" name="__module.encoder.layer.1.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="286" name="__module.encoder.layer.1.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="287" name="Constant_8832" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="38295008" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="288" name="__module.encoder.layer.1.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="264,input.9">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="289" name="__module.encoder.layer.1.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="266">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="290" name="__module.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="291" name="__module.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="292" name="Constant_8833" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="38298080" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="293" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="294" name="Constant_8834" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="38301152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="295" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="270,hidden_states.13">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="296" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="38304224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="297" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" 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="298" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38307296" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="299" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38307300" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="300" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38307296" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="301" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38307300" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="302" name="__module.encoder.layer.1.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="303" name="Constant_43740346" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="38307304" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="304" name="Convert_43740347" 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="305" name="Constant_43740348" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="38897128" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="306" name="__module.encoder.layer.2.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="307" name="__module.encoder.layer.2.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="308" name="Constant_8835" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="38900200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="309" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add" 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,x.25">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="310" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38903272" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="311" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38903276" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="312" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38903272" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="313" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="38903276" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="314" name="__module.encoder.layer.2.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="315" name="__module.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="316" name="__module.encoder.layer.2.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="287,x.27">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="317" name="Constant_669" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="288">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="318" name="__module.encoder.layer.2.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="289">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="319" name="Constant_43740350" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="38903280" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="320" name="Convert_43740351" 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="321" name="Constant_43740352" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="39493104" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="322" name="__module.encoder.layer.2.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="323" name="__module.encoder.layer.2.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="324" name="Constant_8836" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="39496176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="325" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="292,x.29">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="326" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39499248" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="327" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39499252" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="328" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39499248" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="329" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="39499252" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="330" name="__module.encoder.layer.2.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="331" name="__module.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="332" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="296,x.31">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="333" name="Constant_692" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="297">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="334" name="__module.encoder.layer.2.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="298">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="335" name="Constant_43740354" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="39499256" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="336" name="Convert_43740355" 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="337" name="Constant_43740356" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="40089080" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="338" name="__module.encoder.layer.2.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="339" name="__module.encoder.layer.2.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="340" name="Constant_8837" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40092152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="341" name="__module.encoder.layer.2.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="301,x.33">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="342" name="__module.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="343" name="__module.encoder.layer.2.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="305,x.35">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="344" name="Constant_715" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="306">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="345" name="__module.encoder.layer.2.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="307">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="346" name="__module.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="308,attn_output.9">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="347" name="__module.encoder.layer.2.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="348" name="__module.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="309,attn_output.11">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="349" name="__module.encoder.layer.2.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="350" name="Constant_7920" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="351" name="Constant_7921" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="352" name="Gather_7922" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="353" name="__module.encoder.layer.2.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="310">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="354" name="__module.encoder.layer.2.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="311">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="355" name="__module.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="40095224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="356" name="__module.encoder.layer.2.attention.self/aten::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="357" name="__module.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="40098296" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="358" name="__module.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="40098300" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="359" name="__module.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="40098296" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="360" name="__module.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="40098300" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="361" name="__module.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="362" name="Constant_43740358" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="40098304" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="363" name="Convert_43740359" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="364" name="Constant_43740360" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="40688128" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="365" name="__module.encoder.layer.2.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="366" name="__module.encoder.layer.2.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="367" name="Constant_8838" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40691200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="368" name="__module.encoder.layer.2.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="317,input.11">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="369" name="__module.encoder.layer.2.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="319">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="370" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="371" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-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="372" name="Constant_8839" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40694272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="373" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="374" name="Constant_8840" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40697344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="375" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="323,input_tensor.5">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="376" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="40700416" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="377" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="378" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40703488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="379" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40703492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="380" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40703488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="381" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="40703492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="382" name="__module.encoder.layer.2.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="383" name="Constant_43740362" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="40703496" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="384" name="Convert_43740363" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="385" name="Constant_43740364" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="43062792" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="386" name="__module.encoder.layer.2.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="387" name="__module.encoder.layer.2.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="388" name="Constant_8841" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="43075080" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="389" name="__module.encoder.layer.2.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="328">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="390" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="329">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="391" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="43087368" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="392" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="393" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="43099656" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="394" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="43099660" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="395" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="43099656" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="396" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="43099660" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="397" name="__module.encoder.layer.2.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="398" name="Constant_43740366" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="43099664" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="399" name="Convert_43740367" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="400" name="Constant_43740368" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="45458960" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="401" name="__module.encoder.layer.2.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="402" name="__module.encoder.layer.2.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="403" name="Constant_8842" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="45462032" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="404" name="__module.encoder.layer.2.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="335,input.13">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="405" name="__module.encoder.layer.2.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="337">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="406" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
|
|
<data element_type="i32" shape="1" offset="23963908" size="4" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="407" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
|
|
<data eps="9.999999960041972e-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="408" name="Constant_8843" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="45465104" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="409" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="410" name="Constant_8844" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="45468176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="411" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="341,hidden_states.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="412" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="45471248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="413" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="414" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="45474320" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="415" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="45474324" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="416" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="45474320" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="417" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="45474324" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="418" name="__module.encoder.layer.2.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="419" name="Constant_43740370" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="45474328" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="420" name="Convert_43740371" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="421" name="Constant_43740372" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="46064152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="422" name="__module.encoder.layer.3.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="423" name="__module.encoder.layer.3.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="424" name="Constant_8845" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="46067224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="425" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="354,x.37">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="426" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46070296" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="427" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46070300" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="428" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46070296" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="429" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46070300" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="430" name="__module.encoder.layer.3.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="431" name="__module.encoder.layer.3.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="432" name="__module.encoder.layer.3.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="358,x.39">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="433" name="Constant_871" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="359">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="434" name="__module.encoder.layer.3.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="360">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="435" name="Constant_43740374" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="46070304" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="436" name="Convert_43740375" type="Convert" version="opset1">
|
|
<data destination_type="f32" />
|
|
<input>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="437" name="Constant_43740376" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="46660128" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="438" name="__module.encoder.layer.3.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="439" name="__module.encoder.layer.3.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="440" name="Constant_8846" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="46663200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="441" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="363,x.41">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="442" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46666272" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="443" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46666276" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="444" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46666272" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="445" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="46666276" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="446" name="__module.encoder.layer.3.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="447" name="__module.encoder.layer.3.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569112" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="448" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="367,x.43">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="449" name="Constant_894" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="368">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="450" name="__module.encoder.layer.3.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="369">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="451" name="Constant_43740378" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="46666280" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="452" name="Convert_43740379" 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="453" name="Constant_43740380" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="47256104" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="454" name="__module.encoder.layer.3.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="455" name="__module.encoder.layer.3.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="456" name="Constant_8847" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47259176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="457" name="__module.encoder.layer.3.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="372,x.45">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="458" name="__module.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="459" name="__module.encoder.layer.3.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="376,x.47">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="460" name="Constant_917" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="377">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="461" name="__module.encoder.layer.3.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="378">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="462" name="__module.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="379,attn_output.13">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="463" name="__module.encoder.layer.3.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="464" name="__module.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="380,attn_output.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="465" name="__module.encoder.layer.3.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="466" name="Constant_7940" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="467" name="Constant_7941" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="468" name="Gather_7942" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="469" name="__module.encoder.layer.3.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="381">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="470" name="__module.encoder.layer.3.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="382">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="471" name="__module.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="47262248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="472" name="__module.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="473" name="__module.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="47265320" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="474" name="__module.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="47265324" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="475" name="__module.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="47265320" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="476" name="__module.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="47265324" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="477" name="__module.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="478" name="Constant_43740382" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="47265328" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="479" name="Convert_43740383" 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="480" name="Constant_43740384" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="47855152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="481" name="__module.encoder.layer.3.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="482" name="__module.encoder.layer.3.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="483" name="Constant_8848" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47858224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="484" name="__module.encoder.layer.3.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="388,input.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="485" name="__module.encoder.layer.3.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="390">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="486" name="__module.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="487" name="__module.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="488" name="Constant_8849" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47861296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="489" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="490" name="Constant_8850" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47864368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="491" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="394,input_tensor.7">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="492" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="47867440" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="493" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" 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="494" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47870512" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="495" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47870516" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="496" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47870512" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="497" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="47870516" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="498" name="__module.encoder.layer.3.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" 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="499" name="Constant_43740386" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="47870520" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="500" name="Convert_43740387" 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="501" name="Constant_43740388" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="50229816" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="502" name="__module.encoder.layer.3.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="503" name="__module.encoder.layer.3.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="504" name="Constant_8851" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="50242104" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="505" name="__module.encoder.layer.3.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="399">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="506" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="400">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="507" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="50254392" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="508" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" 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="509" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="50266680" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="510" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="50266684" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="511" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="50266680" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="512" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="50266684" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="513" name="__module.encoder.layer.3.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" 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="514" name="Constant_43740390" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="50266688" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="515" name="Convert_43740391" 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="516" name="Constant_43740392" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="52625984" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="517" name="__module.encoder.layer.3.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="518" name="__module.encoder.layer.3.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="519" name="Constant_8852" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="52629056" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="520" name="__module.encoder.layer.3.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="406,input.17">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="521" name="__module.encoder.layer.3.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="408">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="522" name="__module.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="523" name="__module.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="524" name="Constant_8853" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="52632128" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="525" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="526" name="Constant_8854" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="52635200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="527" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="412,hidden_states.25">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="528" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="52638272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="529" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" 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="530" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="52641344" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="531" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="52641348" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="532" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="52641344" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="533" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="52641348" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="534" name="__module.encoder.layer.3.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" 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="535" name="Constant_43740394" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="52641352" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="536" name="Convert_43740395" 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="537" name="Constant_43740396" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="53231176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="538" name="__module.encoder.layer.4.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="539" name="__module.encoder.layer.4.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="540" name="Constant_8855" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="53234248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="541" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="425,x.49">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="542" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53237320" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="543" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53237324" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="544" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53237320" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="545" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53237324" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="546" name="__module.encoder.layer.4.attention.self.query/aten::linear/Add/fq_output_0" 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="547" name="__module.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="548" name="__module.encoder.layer.4.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="429,x.51">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="549" name="Constant_1073" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="430">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="550" name="__module.encoder.layer.4.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="431">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="551" name="Constant_43740398" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="53237328" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="552" name="Convert_43740399" 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="553" name="Constant_43740400" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="53827152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="554" name="__module.encoder.layer.4.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="555" name="__module.encoder.layer.4.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="556" name="Constant_8856" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="53830224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="557" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="434,x.53">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="558" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53833296" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="559" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53833300" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="560" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53833296" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="561" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="53833300" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="562" name="__module.encoder.layer.4.attention.self.key/aten::linear/Add/fq_output_0" 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="563" name="__module.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="564" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="438,x.55">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="565" name="Constant_1096" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="439">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="566" name="__module.encoder.layer.4.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="440">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="567" name="Constant_43740402" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="53833304" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="568" name="Convert_43740403" 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="569" name="Constant_43740404" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="54423128" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="570" name="__module.encoder.layer.4.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="571" name="__module.encoder.layer.4.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="572" name="Constant_8857" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="54426200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="573" name="__module.encoder.layer.4.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="443,x.57">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="574" name="__module.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="575" name="__module.encoder.layer.4.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="447,x.59">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="576" name="Constant_1119" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="448">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="577" name="__module.encoder.layer.4.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="449">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="578" name="__module.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="450,attn_output.17">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="579" name="__module.encoder.layer.4.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="580" name="__module.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="451,attn_output.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="581" name="__module.encoder.layer.4.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="582" name="Constant_7960" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="583" name="Constant_7961" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="584" name="Gather_7962" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="585" name="__module.encoder.layer.4.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="452">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="586" name="__module.encoder.layer.4.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="453">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="587" name="__module.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="54429272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="588" name="__module.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="589" name="__module.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="54432344" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="590" name="__module.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="54432348" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="591" name="__module.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="54432344" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="592" name="__module.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="54432348" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="593" name="__module.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="594" name="Constant_43740406" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="54432352" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="595" name="Convert_43740407" 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="596" name="Constant_43740408" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="55022176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="597" name="__module.encoder.layer.4.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="598" name="__module.encoder.layer.4.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="599" name="Constant_8858" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55025248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="600" name="__module.encoder.layer.4.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="459,input.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="601" name="__module.encoder.layer.4.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="461">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="602" name="__module.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="603" name="__module.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="604" name="Constant_8859" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55028320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="605" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="606" name="Constant_8860" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55031392" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="607" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/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="465,input_tensor.9">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="608" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="55034464" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="609" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_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="610" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55037536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="611" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55037540" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="612" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55037536" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="613" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="55037540" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="614" name="__module.encoder.layer.4.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" 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="615" name="Constant_43740410" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="55037544" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="616" name="Convert_43740411" 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="617" name="Constant_43740412" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="57396840" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="618" name="__module.encoder.layer.4.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="619" name="__module.encoder.layer.4.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="620" name="Constant_8861" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="57409128" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="621" name="__module.encoder.layer.4.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="470">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="622" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="471">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="623" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="57421416" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="624" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_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="625" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="57433704" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="626" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="57433708" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="627" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="57433704" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="628" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="57433708" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="629" name="__module.encoder.layer.4.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" 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="630" name="Constant_43740414" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="57433712" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="631" name="Convert_43740415" 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="632" name="Constant_43740416" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="59793008" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="633" name="__module.encoder.layer.4.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="634" name="__module.encoder.layer.4.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="635" name="Constant_8862" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="59796080" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="636" name="__module.encoder.layer.4.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="477,input.21">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="637" name="__module.encoder.layer.4.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="479">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="638" name="__module.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="639" name="__module.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="640" name="Constant_8863" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="59799152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="641" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="642" name="Constant_8864" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="59802224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="643" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/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="483,hidden_states.31">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="644" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="59805296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="645" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_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="646" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="59808368" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="647" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="59808372" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="648" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="59808368" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="649" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="59808372" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="650" name="__module.encoder.layer.4.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" 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="651" name="Constant_43740418" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="59808376" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="652" name="Convert_43740419" 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="653" name="Constant_43740420" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="60398200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="654" name="__module.encoder.layer.5.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="655" name="__module.encoder.layer.5.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="656" name="Constant_8865" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="60401272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="657" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="496,x.61">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="658" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60404344" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="659" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60404348" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="660" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60404344" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="661" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="60404348" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="662" name="__module.encoder.layer.5.attention.self.query/aten::linear/Add/fq_output_0" 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="663" name="__module.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="664" name="__module.encoder.layer.5.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="500,x.63">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="665" name="Constant_1275" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="501">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="666" name="__module.encoder.layer.5.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="502">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="667" name="Constant_43740422" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="60404352" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="668" name="Convert_43740423" 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="669" name="Constant_43740424" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="60994176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="670" name="__module.encoder.layer.5.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="671" name="__module.encoder.layer.5.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="672" name="Constant_8866" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="60997248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="673" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="505,x.65">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="674" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61000320" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="675" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61000324" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="676" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61000320" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="677" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="61000324" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="678" name="__module.encoder.layer.5.attention.self.key/aten::linear/Add/fq_output_0" 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="679" name="__module.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="680" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="509,x.67">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="681" name="Constant_1298" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="510">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="682" name="__module.encoder.layer.5.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="511">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="683" name="Constant_43740426" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="61000328" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="684" name="Convert_43740427" 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="685" name="Constant_43740428" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="61590152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="686" name="__module.encoder.layer.5.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="687" name="__module.encoder.layer.5.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="688" name="Constant_8867" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="61593224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="689" name="__module.encoder.layer.5.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="514,x.69">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="690" name="__module.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="691" name="__module.encoder.layer.5.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="518,x.71">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="692" name="Constant_1321" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="519">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="693" name="__module.encoder.layer.5.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="520">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="694" name="__module.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="521,attn_output.21">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="695" name="__module.encoder.layer.5.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="696" name="__module.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="522,attn_output.23">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="697" name="__module.encoder.layer.5.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="698" name="Constant_7980" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="699" name="Constant_7981" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="700" name="Gather_7982" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="701" name="__module.encoder.layer.5.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="523">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="702" name="__module.encoder.layer.5.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="524">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="703" name="__module.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="61596296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="704" name="__module.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="705" name="__module.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="61599368" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="706" name="__module.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="61599372" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="707" name="__module.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="61599368" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="708" name="__module.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="61599372" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="709" name="__module.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="710" name="Constant_43740430" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="61599376" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="711" name="Convert_43740431" 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="712" name="Constant_43740432" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="62189200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="713" name="__module.encoder.layer.5.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="714" name="__module.encoder.layer.5.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="715" name="Constant_8868" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="62192272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="716" name="__module.encoder.layer.5.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="530,input.23">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="717" name="__module.encoder.layer.5.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="532">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="718" name="__module.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="719" name="__module.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="720" name="Constant_8869" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="62195344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="721" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Multiply_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="722" name="Constant_8870" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="62198416" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="723" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="536,input_tensor.11">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="724" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="62201488" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="725" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="726" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="62204560" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="727" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="62204564" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="728" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="62204560" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="729" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="62204564" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="730" name="__module.encoder.layer.5.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" 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="731" name="Constant_43740434" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="62204568" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="732" name="Convert_43740435" 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="733" name="Constant_43740436" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="64563864" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="734" name="__module.encoder.layer.5.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="735" name="__module.encoder.layer.5.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="736" name="Constant_8871" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="64576152" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="737" name="__module.encoder.layer.5.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="541">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="738" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="542">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="739" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="64588440" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="740" name="__module.encoder.layer.5.intermediate.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="741" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="64600728" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="742" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="64600732" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="743" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="64600728" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="744" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="64600732" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="745" name="__module.encoder.layer.5.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" 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="746" name="Constant_43740438" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="64600736" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="747" name="Convert_43740439" 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="748" name="Constant_43740440" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="66960032" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="749" name="__module.encoder.layer.5.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="750" name="__module.encoder.layer.5.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="751" name="Constant_8872" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="66963104" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="752" name="__module.encoder.layer.5.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="548,input.25">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="753" name="__module.encoder.layer.5.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="550">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="754" name="__module.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="755" name="__module.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="756" name="Constant_8873" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="66966176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="757" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Multiply_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="758" name="Constant_8874" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="66969248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="759" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="554,hidden_states.37">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="760" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="66972320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="761" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="762" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="66975392" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="763" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="66975396" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="764" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="66975392" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="765" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="66975396" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="766" name="__module.encoder.layer.5.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" 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="767" name="Constant_43740442" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="66975400" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="768" name="Convert_43740443" 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="769" name="Constant_43740444" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="67565224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="770" name="__module.encoder.layer.6.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="771" name="__module.encoder.layer.6.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="772" name="Constant_8875" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="67568296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="773" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="567,x.73">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="774" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67571368" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="775" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67571372" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="776" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67571368" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="777" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="67571372" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="778" name="__module.encoder.layer.6.attention.self.query/aten::linear/Add/fq_output_0" 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="779" name="__module.encoder.layer.6.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="780" name="__module.encoder.layer.6.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="571,x.75">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="781" name="Constant_1477" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="572">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="782" name="__module.encoder.layer.6.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="573">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="783" name="Constant_43740446" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="67571376" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="784" name="Convert_43740447" 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="785" name="Constant_43740448" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="68161200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="786" name="__module.encoder.layer.6.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="787" name="__module.encoder.layer.6.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="788" name="Constant_8876" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="68164272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="789" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="576,x.77">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="790" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="68167344" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="791" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="68167348" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="792" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="68167344" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="793" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="68167348" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="794" name="__module.encoder.layer.6.attention.self.key/aten::linear/Add/fq_output_0" 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="795" name="__module.encoder.layer.6.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="796" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="580,x.79">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="797" name="Constant_1500" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="581">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="798" name="__module.encoder.layer.6.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="582">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="799" name="Constant_43740450" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="68167352" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="800" name="Convert_43740451" 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="801" name="Constant_43740452" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="68757176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="802" name="__module.encoder.layer.6.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="803" name="__module.encoder.layer.6.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="804" name="Constant_8877" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="68760248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="805" name="__module.encoder.layer.6.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="585,x.81">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="806" name="__module.encoder.layer.6.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="807" name="__module.encoder.layer.6.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="589,x.83">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="808" name="Constant_1523" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="590">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="809" name="__module.encoder.layer.6.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="591">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="810" name="__module.encoder.layer.6.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="592,attn_output.25">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="811" name="__module.encoder.layer.6.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="812" name="__module.encoder.layer.6.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="593,attn_output.27">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="813" name="__module.encoder.layer.6.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="814" name="Constant_8000" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="815" name="Constant_8001" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="816" name="Gather_8002" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="817" name="__module.encoder.layer.6.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="594">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="818" name="__module.encoder.layer.6.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="595">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="819" name="__module.encoder.layer.6.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="68763320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="820" name="__module.encoder.layer.6.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="821" name="__module.encoder.layer.6.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="68766392" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="822" name="__module.encoder.layer.6.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="68766396" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="823" name="__module.encoder.layer.6.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="68766392" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="824" name="__module.encoder.layer.6.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="68766396" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="825" name="__module.encoder.layer.6.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="826" name="Constant_43740454" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="68766400" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="827" name="Convert_43740455" 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="828" name="Constant_43740456" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="69356224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="829" name="__module.encoder.layer.6.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="830" name="__module.encoder.layer.6.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="831" name="Constant_8878" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="69359296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="832" name="__module.encoder.layer.6.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="601,input.27">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="833" name="__module.encoder.layer.6.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="603">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="834" name="__module.encoder.layer.6.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="835" name="__module.encoder.layer.6.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="836" name="Constant_8879" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="69362368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="837" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="838" name="Constant_8880" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="69365440" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="839" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="607,input_tensor.13">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="840" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="69368512" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="841" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/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="842" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69371584" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="843" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69371588" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="844" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69371584" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="845" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="69371588" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="846" name="__module.encoder.layer.6.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" 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="847" name="Constant_43740458" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="69371592" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="848" name="Convert_43740459" 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="849" name="Constant_43740460" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="71730888" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="850" name="__module.encoder.layer.6.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="851" name="__module.encoder.layer.6.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="852" name="Constant_8881" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="71743176" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="853" name="__module.encoder.layer.6.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="612">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="854" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="613">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="855" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="71755464" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="856" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="857" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="71767752" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="858" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="71767756" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="859" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="71767752" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="860" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="71767756" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="861" name="__module.encoder.layer.6.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" 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="862" name="Constant_43740462" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="71767760" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="863" name="Convert_43740463" 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="864" name="Constant_43740464" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="74127056" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="865" name="__module.encoder.layer.6.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="866" name="__module.encoder.layer.6.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="867" name="Constant_8882" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="74130128" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="868" name="__module.encoder.layer.6.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="619,input.29">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="869" name="__module.encoder.layer.6.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="621">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="870" name="__module.encoder.layer.6.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="871" name="__module.encoder.layer.6.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="872" name="Constant_8883" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="74133200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="873" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="874" name="Constant_8884" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="74136272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="875" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="625,hidden_states.43">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="876" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="74139344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="877" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/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="878" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="74142416" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="879" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="74142420" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="880" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="74142416" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="881" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="74142420" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="882" name="__module.encoder.layer.6.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="883" name="Constant_43740466" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="74142424" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="884" name="Convert_43740467" 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="885" name="Constant_43740468" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="74732248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="886" name="__module.encoder.layer.7.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="887" name="__module.encoder.layer.7.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="888" name="Constant_8885" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="74735320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="889" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="638,x.85">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="890" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="74738392" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="891" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="74738396" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="892" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="74738392" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="893" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="74738396" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="894" name="__module.encoder.layer.7.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="895" name="__module.encoder.layer.7.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="896" name="__module.encoder.layer.7.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="642,x.87">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="897" name="Constant_1679" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="643">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="898" name="__module.encoder.layer.7.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="644">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="899" name="Constant_43740470" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="74738400" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="900" name="Convert_43740471" 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="901" name="Constant_43740472" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="75328224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="902" name="__module.encoder.layer.7.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="903" name="__module.encoder.layer.7.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="904" name="Constant_8886" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="75331296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="905" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="647,x.89">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="906" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75334368" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="907" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75334372" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="908" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75334368" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="909" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="75334372" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="910" name="__module.encoder.layer.7.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="911" name="__module.encoder.layer.7.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="912" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="651,x.91">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="913" name="Constant_1702" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="652">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="914" name="__module.encoder.layer.7.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="653">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="915" name="Constant_43740474" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="75334376" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="916" name="Convert_43740475" 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="917" name="Constant_43740476" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="75924200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="918" name="__module.encoder.layer.7.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="919" name="__module.encoder.layer.7.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="920" name="Constant_8887" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="75927272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="921" name="__module.encoder.layer.7.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="656,x.93">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="922" name="__module.encoder.layer.7.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="923" name="__module.encoder.layer.7.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="660,x.95">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="924" name="Constant_1725" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="661">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="925" name="__module.encoder.layer.7.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="662">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="926" name="__module.encoder.layer.7.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="663,attn_output.29">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="927" name="__module.encoder.layer.7.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="928" name="__module.encoder.layer.7.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="664,attn_output.31">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="929" name="__module.encoder.layer.7.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="930" name="Constant_8020" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="931" name="Constant_8021" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="932" name="Gather_8022" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="933" name="__module.encoder.layer.7.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="665">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="934" name="__module.encoder.layer.7.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="666">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="935" name="__module.encoder.layer.7.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="75930344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="936" name="__module.encoder.layer.7.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="937" name="__module.encoder.layer.7.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="75933416" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="938" name="__module.encoder.layer.7.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="75933420" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="939" name="__module.encoder.layer.7.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="75933416" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="940" name="__module.encoder.layer.7.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="75933420" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="941" name="__module.encoder.layer.7.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="942" name="Constant_43740478" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="75933424" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="943" name="Convert_43740479" 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="944" name="Constant_43740480" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="76523248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="945" name="__module.encoder.layer.7.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="946" name="__module.encoder.layer.7.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="947" name="Constant_8888" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="76526320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="948" name="__module.encoder.layer.7.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="672,input.31">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="949" name="__module.encoder.layer.7.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="674">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="950" name="__module.encoder.layer.7.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="951" name="__module.encoder.layer.7.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="952" name="Constant_8889" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="76529392" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="953" name="__module.encoder.layer.7.attention.output.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="954" name="Constant_8890" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="76532464" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="955" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/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="678,input_tensor.15">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="956" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="76535536" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="957" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="958" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76538608" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="959" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76538612" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="960" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76538608" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="961" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="76538612" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="962" name="__module.encoder.layer.7.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="963" name="Constant_43740482" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="76538616" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="964" name="Convert_43740483" 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="965" name="Constant_43740484" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="78897912" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="966" name="__module.encoder.layer.7.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="967" name="__module.encoder.layer.7.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="968" name="Constant_8891" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="78910200" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="969" name="__module.encoder.layer.7.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="683">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="970" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="684">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="971" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="78922488" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="972" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="973" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="78934776" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="974" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="78934780" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="975" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="78934776" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="976" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="78934780" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="977" name="__module.encoder.layer.7.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="978" name="Constant_43740486" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="78934784" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="979" name="Convert_43740487" 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="980" name="Constant_43740488" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="81294080" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="981" name="__module.encoder.layer.7.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="982" name="__module.encoder.layer.7.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="983" name="Constant_8892" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="81297152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="984" name="__module.encoder.layer.7.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="690,input.33">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="985" name="__module.encoder.layer.7.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="692">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="986" name="__module.encoder.layer.7.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="987" name="__module.encoder.layer.7.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="988" name="Constant_8893" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="81300224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="989" name="__module.encoder.layer.7.output.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="990" name="Constant_8894" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="81303296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="991" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/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="696,hidden_states.49">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="992" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="81306368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="993" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="994" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="81309440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="995" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="81309444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="996" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="81309440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="997" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="81309444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="998" name="__module.encoder.layer.7.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="999" name="Constant_43740490" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="81309448" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1000" name="Convert_43740491" 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="1001" name="Constant_43740492" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="81899272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1002" name="__module.encoder.layer.8.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1003" name="__module.encoder.layer.8.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1004" name="Constant_8895" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="81902344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1005" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="709,x.97">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1006" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="81905416" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1007" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="81905420" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1008" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="81905416" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1009" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="81905420" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1010" name="__module.encoder.layer.8.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1011" name="__module.encoder.layer.8.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="1012" name="__module.encoder.layer.8.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="713,x.99">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1013" name="Constant_1881" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="714">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1014" name="__module.encoder.layer.8.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="715">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1015" name="Constant_43740494" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="81905424" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1016" name="Convert_43740495" 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="1017" name="Constant_43740496" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="82495248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1018" name="__module.encoder.layer.8.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1019" name="__module.encoder.layer.8.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1020" name="Constant_8896" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="82498320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1021" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="718,x.101">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1022" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82501392" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1023" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82501396" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1024" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82501392" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1025" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="82501396" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1026" name="__module.encoder.layer.8.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1027" name="__module.encoder.layer.8.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="1028" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="722,x.103">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1029" name="Constant_1904" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="723">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1030" name="__module.encoder.layer.8.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="724">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1031" name="Constant_43740498" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="82501400" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1032" name="Convert_43740499" 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="1033" name="Constant_43740500" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="83091224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1034" name="__module.encoder.layer.8.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1035" name="__module.encoder.layer.8.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1036" name="Constant_8897" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="83094296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1037" name="__module.encoder.layer.8.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="727,x.105">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1038" name="__module.encoder.layer.8.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="1039" name="__module.encoder.layer.8.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="731,x.107">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1040" name="Constant_1927" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="732">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1041" name="__module.encoder.layer.8.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="733">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1042" name="__module.encoder.layer.8.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="734,attn_output.33">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1043" name="__module.encoder.layer.8.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1044" name="__module.encoder.layer.8.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="735,attn_output.35">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1045" name="__module.encoder.layer.8.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1046" name="Constant_8040" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1047" name="Constant_8041" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1048" name="Gather_8042" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="1049" name="__module.encoder.layer.8.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="736">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1050" name="__module.encoder.layer.8.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="737">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1051" name="__module.encoder.layer.8.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="83097368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1052" name="__module.encoder.layer.8.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="1053" name="__module.encoder.layer.8.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="83100440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1054" name="__module.encoder.layer.8.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="83100444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1055" name="__module.encoder.layer.8.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="83100440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1056" name="__module.encoder.layer.8.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="83100444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1057" name="__module.encoder.layer.8.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="1058" name="Constant_43740502" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="83100448" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1059" name="Convert_43740503" 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="1060" name="Constant_43740504" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="83690272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1061" name="__module.encoder.layer.8.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1062" name="__module.encoder.layer.8.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1063" name="Constant_8898" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="83693344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1064" name="__module.encoder.layer.8.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="743,input.35">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1065" name="__module.encoder.layer.8.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="745">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1066" name="__module.encoder.layer.8.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="1067" name="__module.encoder.layer.8.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="1068" name="Constant_8899" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="83696416" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1069" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1070" name="Constant_8900" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="83699488" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1071" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="749,input_tensor.17">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1072" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="83702560" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1073" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1074" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="83705632" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1075" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="83705636" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1076" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="83705632" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1077" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="83705636" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1078" name="__module.encoder.layer.8.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1079" name="Constant_43740506" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="83705640" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1080" name="Convert_43740507" 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="1081" name="Constant_43740508" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="86064936" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1082" name="__module.encoder.layer.8.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1083" name="__module.encoder.layer.8.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1084" name="Constant_8901" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="86077224" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1085" name="__module.encoder.layer.8.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="754">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1086" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="755">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1087" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="86089512" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1088" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1089" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="86101800" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1090" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="86101804" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1091" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="86101800" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1092" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="86101804" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1093" name="__module.encoder.layer.8.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1094" name="Constant_43740510" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="86101808" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1095" name="Convert_43740511" 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="1096" name="Constant_43740512" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="88461104" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1097" name="__module.encoder.layer.8.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1098" name="__module.encoder.layer.8.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1099" name="Constant_8902" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="88464176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1100" name="__module.encoder.layer.8.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="761,input.37">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1101" name="__module.encoder.layer.8.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="763">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1102" name="__module.encoder.layer.8.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="1103" name="__module.encoder.layer.8.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="1104" name="Constant_8903" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="88467248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1105" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1106" name="Constant_8904" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="88470320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1107" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="767,hidden_states.55">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1108" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="88473392" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1109" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1110" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="88476464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1111" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="88476468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1112" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="88476464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1113" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="88476468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1114" name="__module.encoder.layer.8.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1115" name="Constant_43740514" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="88476472" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1116" name="Convert_43740515" 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="1117" name="Constant_43740516" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="89066296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1118" name="__module.encoder.layer.9.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1119" name="__module.encoder.layer.9.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1120" name="Constant_8905" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="89069368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1121" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="780,x.109">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1122" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89072440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1123" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89072444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1124" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89072440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1125" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89072444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1126" name="__module.encoder.layer.9.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1127" name="__module.encoder.layer.9.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="1128" name="__module.encoder.layer.9.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="784,x.111">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1129" name="Constant_2083" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="785">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1130" name="__module.encoder.layer.9.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="786">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1131" name="Constant_43740518" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="89072448" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1132" name="Convert_43740519" 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="1133" name="Constant_43740520" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="89662272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1134" name="__module.encoder.layer.9.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1135" name="__module.encoder.layer.9.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1136" name="Constant_8906" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="89665344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1137" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="789,x.113">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1138" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89668416" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1139" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89668420" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1140" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89668416" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1141" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="89668420" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1142" name="__module.encoder.layer.9.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1143" name="__module.encoder.layer.9.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="1144" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="793,x.115">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1145" name="Constant_2106" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="794">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1146" name="__module.encoder.layer.9.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="795">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1147" name="Constant_43740522" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="89668424" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1148" name="Convert_43740523" 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="1149" name="Constant_43740524" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="90258248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1150" name="__module.encoder.layer.9.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1151" name="__module.encoder.layer.9.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1152" name="Constant_8907" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="90261320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1153" name="__module.encoder.layer.9.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="798,x.117">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1154" name="__module.encoder.layer.9.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="1155" name="__module.encoder.layer.9.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="802,x.119">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1156" name="Constant_2129" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="803">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1157" name="__module.encoder.layer.9.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="804">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1158" name="__module.encoder.layer.9.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="805,attn_output.37">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1159" name="__module.encoder.layer.9.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1160" name="__module.encoder.layer.9.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="806,attn_output.39">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1161" name="__module.encoder.layer.9.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1162" name="Constant_8060" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1163" name="Constant_8061" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1164" name="Gather_8062" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="1165" name="__module.encoder.layer.9.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="807">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1166" name="__module.encoder.layer.9.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="808">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1167" name="__module.encoder.layer.9.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="90264392" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1168" name="__module.encoder.layer.9.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="1169" name="__module.encoder.layer.9.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="90267464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1170" name="__module.encoder.layer.9.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="90267468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1171" name="__module.encoder.layer.9.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="90267464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1172" name="__module.encoder.layer.9.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="90267468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1173" name="__module.encoder.layer.9.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="1174" name="Constant_43740526" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="90267472" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1175" name="Convert_43740527" 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="1176" name="Constant_43740528" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="90857296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1177" name="__module.encoder.layer.9.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1178" name="__module.encoder.layer.9.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1179" name="Constant_8908" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="90860368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1180" name="__module.encoder.layer.9.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="814,input.39">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1181" name="__module.encoder.layer.9.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="816">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1182" name="__module.encoder.layer.9.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="1183" name="__module.encoder.layer.9.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="1184" name="Constant_8909" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="90863440" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1185" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1186" name="Constant_8910" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="90866512" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1187" name="__module.encoder.layer.9.attention.output.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="820,input_tensor.19">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1188" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="90869584" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1189" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/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="1190" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="90872656" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1191" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="90872660" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1192" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="90872656" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1193" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="90872660" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1194" name="__module.encoder.layer.9.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1195" name="Constant_43740530" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="90872664" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1196" name="Convert_43740531" 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="1197" name="Constant_43740532" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="93231960" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1198" name="__module.encoder.layer.9.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1199" name="__module.encoder.layer.9.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1200" name="Constant_8911" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="93244248" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1201" name="__module.encoder.layer.9.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="825">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1202" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="826">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1203" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="93256536" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1204" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1205" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="93268824" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1206" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="93268828" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1207" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="93268824" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1208" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="93268828" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1209" name="__module.encoder.layer.9.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1210" name="Constant_43740534" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="93268832" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1211" name="Convert_43740535" 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="1212" name="Constant_43740536" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="95628128" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1213" name="__module.encoder.layer.9.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1214" name="__module.encoder.layer.9.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1215" name="Constant_8912" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="95631200" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1216" name="__module.encoder.layer.9.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="832,input.41">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1217" name="__module.encoder.layer.9.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="834">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1218" name="__module.encoder.layer.9.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="1219" name="__module.encoder.layer.9.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="1220" name="Constant_8913" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="95634272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1221" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1222" name="Constant_8914" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="95637344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1223" name="__module.encoder.layer.9.output.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="838,hidden_states.61">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1224" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="95640416" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1225" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/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="1226" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="95643488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1227" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="95643492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1228" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="95643488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1229" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="95643492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1230" name="__module.encoder.layer.9.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1231" name="Constant_43740538" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="95643496" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1232" name="Convert_43740539" 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="1233" name="Constant_43740540" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="96233320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1234" name="__module.encoder.layer.10.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1235" name="__module.encoder.layer.10.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1236" name="Constant_8915" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="96236392" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1237" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="851,x.121">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1238" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96239464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1239" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96239468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1240" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96239464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1241" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96239468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1242" name="__module.encoder.layer.10.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1243" name="__module.encoder.layer.10.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="1244" name="__module.encoder.layer.10.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="855,x.123">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1245" name="Constant_2285" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="856">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1246" name="__module.encoder.layer.10.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="857">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1247" name="Constant_43740542" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="96239472" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1248" name="Convert_43740543" 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="1249" name="Constant_43740544" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="96829296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1250" name="__module.encoder.layer.10.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1251" name="__module.encoder.layer.10.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1252" name="Constant_8916" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="96832368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1253" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="860,x.125">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1254" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96835440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1255" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96835444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1256" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96835440" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1257" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="96835444" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1258" name="__module.encoder.layer.10.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1259" name="__module.encoder.layer.10.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="1260" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="864,x.127">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1261" name="Constant_2308" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="865">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1262" name="__module.encoder.layer.10.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="866">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1263" name="Constant_43740546" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="96835448" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1264" name="Convert_43740547" 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="1265" name="Constant_43740548" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="97425272" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1266" name="__module.encoder.layer.10.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1267" name="__module.encoder.layer.10.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1268" name="Constant_8917" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="97428344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1269" name="__module.encoder.layer.10.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="869,x.129">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1270" name="__module.encoder.layer.10.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="1271" name="__module.encoder.layer.10.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="873,x.131">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1272" name="Constant_2331" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="874">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1273" name="__module.encoder.layer.10.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="875">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1274" name="__module.encoder.layer.10.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="876,attn_output.41">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1275" name="__module.encoder.layer.10.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1276" name="__module.encoder.layer.10.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="877,attn_output.43">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1277" name="__module.encoder.layer.10.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1278" name="Constant_8080" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1279" name="Constant_8081" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1280" name="Gather_8082" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="1281" name="__module.encoder.layer.10.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="878">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1282" name="__module.encoder.layer.10.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="879">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1283" name="__module.encoder.layer.10.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="97431416" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1284" name="__module.encoder.layer.10.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="1285" name="__module.encoder.layer.10.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="97434488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1286" name="__module.encoder.layer.10.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="97434492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1287" name="__module.encoder.layer.10.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="97434488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1288" name="__module.encoder.layer.10.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="97434492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1289" name="__module.encoder.layer.10.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="1290" name="Constant_43740550" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="97434496" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1291" name="Convert_43740551" 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="1292" name="Constant_43740552" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="98024320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1293" name="__module.encoder.layer.10.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1294" name="__module.encoder.layer.10.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1295" name="Constant_8918" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98027392" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1296" name="__module.encoder.layer.10.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="885,input.43">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1297" name="__module.encoder.layer.10.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="887">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1298" name="__module.encoder.layer.10.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="1299" name="__module.encoder.layer.10.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="1300" name="Constant_8919" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98030464" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1301" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1302" name="Constant_8920" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98033536" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1303" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="891,input_tensor.21">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1304" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="98036608" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1305" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1306" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98039680" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1307" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98039684" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1308" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98039680" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1309" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="98039684" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1310" name="__module.encoder.layer.10.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1311" name="Constant_43740554" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="98039688" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1312" name="Convert_43740555" 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="1313" name="Constant_43740556" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="100398984" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1314" name="__module.encoder.layer.10.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1315" name="__module.encoder.layer.10.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1316" name="Constant_8921" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="100411272" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1317" name="__module.encoder.layer.10.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="896">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1318" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="897">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1319" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="100423560" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1320" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1321" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="100435848" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1322" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="100435852" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1323" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="100435848" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1324" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="100435852" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1325" name="__module.encoder.layer.10.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1326" name="Constant_43740558" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="100435856" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1327" name="Convert_43740559" 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="1328" name="Constant_43740560" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="102795152" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1329" name="__module.encoder.layer.10.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1330" name="__module.encoder.layer.10.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1331" name="Constant_8922" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="102798224" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1332" name="__module.encoder.layer.10.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="903,input.45">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1333" name="__module.encoder.layer.10.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="905">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1334" name="__module.encoder.layer.10.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="1335" name="__module.encoder.layer.10.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="1336" name="Constant_8923" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="102801296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1337" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1338" name="Constant_8924" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="102804368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1339" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="909,hidden_states.67">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1340" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="102807440" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1341" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1342" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="102810512" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1343" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="102810516" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1344" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="102810512" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1345" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="102810516" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1346" name="__module.encoder.layer.10.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1347" name="Constant_43740562" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="102810520" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1348" name="Convert_43740563" 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="1349" name="Constant_43740564" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="103400344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1350" name="__module.encoder.layer.11.attention.self.query/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1351" name="__module.encoder.layer.11.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1352" name="Constant_8925" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="103403416" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1353" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="922,x.133">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1354" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="103406488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1355" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="103406492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1356" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="103406488" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1357" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="103406492" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1358" name="__module.encoder.layer.11.attention.self.query/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1359" name="__module.encoder.layer.11.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="1360" name="__module.encoder.layer.11.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="926,x.135">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1361" name="Constant_2487" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="927">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1362" name="__module.encoder.layer.11.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="928">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1363" name="Constant_43740566" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="103406496" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1364" name="Convert_43740567" 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="1365" name="Constant_43740568" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="103996320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1366" name="__module.encoder.layer.11.attention.self.key/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1367" name="__module.encoder.layer.11.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1368" name="Constant_8926" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="103999392" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1369" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="931,x.137">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1370" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104002464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1371" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104002468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1372" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104002464" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1373" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="104002468" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1374" name="__module.encoder.layer.11.attention.self.key/aten::linear/Add/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1375" name="__module.encoder.layer.11.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="1376" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="935,x.139">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1377" name="Constant_2510" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="936">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1378" name="__module.encoder.layer.11.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="937">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1379" name="Constant_43740570" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="104002472" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1380" name="Convert_43740571" 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="1381" name="Constant_43740572" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="104592296" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1382" name="__module.encoder.layer.11.attention.self.value/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1383" name="__module.encoder.layer.11.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1384" name="Constant_8927" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="104595368" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1385" name="__module.encoder.layer.11.attention.self.value/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="940,x.141">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1386" name="__module.encoder.layer.11.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="1387" name="__module.encoder.layer.11.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
|
|
<data special_zero="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="944,x">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1388" name="Constant_2533" type="Const" version="opset1">
|
|
<data element_type="i64" shape="4" offset="24569144" size="32" />
|
|
<output>
|
|
<port id="0" precision="I64" names="945">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1389" name="__module.encoder.layer.11.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>4</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="946">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1390" name="__module.encoder.layer.11.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="947,attn_output.45">
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>-1</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1391" name="__module.encoder.layer.11.attention.self/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
|
|
<data element_type="i32" shape="4" offset="25761136" size="16" />
|
|
<output>
|
|
<port id="0" precision="I32">
|
|
<dim>4</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1392" name="__module.encoder.layer.11.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="948,attn_output">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>12</dim>
|
|
<dim>64</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1393" name="__module.encoder.layer.11.attention.self/aten::size/ShapeOf_6" type="ShapeOf" version="opset3">
|
|
<data output_type="i64" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="I64">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1394" name="Constant_8100" type="Const" version="opset1">
|
|
<data element_type="i64" shape="2" offset="25761152" size="16" />
|
|
<output>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1395" name="Constant_8101" type="Const" version="opset1">
|
|
<data element_type="i64" shape="" offset="23963892" size="8" />
|
|
<output>
|
|
<port id="0" precision="I64" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1396" name="Gather_8102" type="Gather" version="opset8">
|
|
<data batch_dims="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>3</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="1397" name="__module.encoder.layer.11.attention.self/prim::ListConstruct/Concat_3" type="Concat" version="opset1">
|
|
<data axis="0" />
|
|
<input>
|
|
<port id="0" precision="I64">
|
|
<dim>2</dim>
|
|
</port>
|
|
<port id="1" precision="I64">
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="I64" names="949">
|
|
<dim>3</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1398" name="__module.encoder.layer.11.attention.self/aten::reshape/Reshape" type="Reshape" version="opset1">
|
|
<data special_zero="false" />
|
|
<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="950">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1399" name="__module.encoder.layer.11.attention.self/aten::reshape/Reshape_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="104598440" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1400" name="__module.encoder.layer.11.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="1401" name="__module.encoder.layer.11.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="104601512" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1402" name="__module.encoder.layer.11.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="104601516" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1403" name="__module.encoder.layer.11.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="104601512" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1404" name="__module.encoder.layer.11.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="104601516" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1405" name="__module.encoder.layer.11.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="1406" name="Constant_43740574" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 768" offset="104601520" size="589824" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1407" name="Convert_43740575" 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="1408" name="Constant_43740576" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="105191344" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1409" name="__module.encoder.layer.11.attention.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1410" name="__module.encoder.layer.11.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1411" name="Constant_8928" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="105194416" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1412" name="__module.encoder.layer.11.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="956,input.47">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1413" name="__module.encoder.layer.11.attention.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="958">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1414" name="__module.encoder.layer.11.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="1415" name="__module.encoder.layer.11.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="1416" name="Constant_8929" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="105197488" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1417" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1418" name="Constant_8930" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="105200560" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1419" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="962,input_tensor">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1420" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="105203632" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1421" name="__module.encoder.layer.11.attention.output.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="1422" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="105206704" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1423" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="105206708" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1424" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="105206704" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1425" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="105206708" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1426" name="__module.encoder.layer.11.attention.output.LayerNorm/aten::layer_norm/Add_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1427" name="Constant_43740578" type="Const" version="opset1">
|
|
<data element_type="i8" shape="3072, 768" offset="105206712" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1428" name="Convert_43740579" 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="1429" name="Constant_43740580" type="Const" version="opset1">
|
|
<data element_type="f32" shape="3072, 1" offset="107566008" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1430" name="__module.encoder.layer.11.intermediate.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1431" name="__module.encoder.layer.11.intermediate.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>3072</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1432" name="Constant_8931" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="107578296" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1433" name="__module.encoder.layer.11.intermediate.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="967">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1434" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu" type="Gelu" version="opset7">
|
|
<data approximation_mode="ERF" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="1" precision="FP32" names="968">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1435" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/scale" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 3072" offset="107590584" size="12288" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1436" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1437" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="107602872" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1438" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/input_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="107602876" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1439" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_low" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="107602872" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1440" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0/output_high" type="Const" version="opset1">
|
|
<data element_type="f32" shape="" offset="107602876" size="4" />
|
|
<output>
|
|
<port id="0" precision="FP32" />
|
|
</output>
|
|
</layer>
|
|
<layer id="1441" name="__module.encoder.layer.11.intermediate.intermediate_act_fn/aten::gelu/Gelu_0_0/nncf_smooth_quant/fq_output_0" type="FakeQuantize" version="opset1">
|
|
<data levels="256" auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32" />
|
|
<port id="2" precision="FP32" />
|
|
<port id="3" precision="FP32" />
|
|
<port id="4" precision="FP32" />
|
|
</input>
|
|
<output>
|
|
<port id="5" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1442" name="Constant_43740582" type="Const" version="opset1">
|
|
<data element_type="i8" shape="768, 3072" offset="107602880" size="2359296" />
|
|
<output>
|
|
<port id="0" precision="I8">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1443" name="Convert_43740583" 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="1444" name="Constant_43740584" type="Const" version="opset1">
|
|
<data element_type="f32" shape="768, 1" offset="109962176" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1445" name="__module.encoder.layer.11.output.dense/aten::linear/MatMul/fq_weights_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>1</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1446" name="__module.encoder.layer.11.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
|
|
<data transpose_a="false" transpose_b="true" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>768</dim>
|
|
<dim>3072</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1447" name="Constant_8932" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="109965248" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1448" name="__module.encoder.layer.11.output.dense/aten::linear/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="974,input">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1449" name="__module.encoder.layer.11.output/aten::add/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="976">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1450" name="__module.encoder.layer.11.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="1451" name="__module.encoder.layer.11.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="1452" name="Constant_8933" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="109968320" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1453" name="__module.encoder.layer.11.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1454" name="Constant_8934" type="Const" version="opset1">
|
|
<data element_type="f32" shape="1, 1, 768" offset="109971392" size="3072" />
|
|
<output>
|
|
<port id="0" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1455" name="__module.encoder.layer.11.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
|
|
<data auto_broadcast="numpy" />
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
<port id="1" precision="FP32">
|
|
<dim>1</dim>
|
|
<dim>1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
<output>
|
|
<port id="2" precision="FP32" names="last_hidden_state">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</output>
|
|
</layer>
|
|
<layer id="1456" name="Result_4120" type="Result" version="opset1">
|
|
<input>
|
|
<port id="0" precision="FP32">
|
|
<dim>-1</dim>
|
|
<dim>-1</dim>
|
|
<dim>768</dim>
|
|
</port>
|
|
</input>
|
|
</layer>
|
|
</layers>
|
|
<edges>
|
|
<edge from-layer="0" from-port="0" to-layer="14" to-port="0" />
|
|
<edge from-layer="1" from-port="0" to-layer="93" to-port="0" />
|
|
<edge from-layer="1" from-port="0" to-layer="96" 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="27" 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="27" from-port="3" to-layer="104" 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="116" to-port="0" />
|
|
<edge from-layer="40" from-port="2" to-layer="137" 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="68" to-port="0" />
|
|
<edge from-layer="47" from-port="5" to-layer="84" to-port="0" />
|
|
<edge from-layer="47" from-port="5" to-layer="52" 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="113" 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="113" 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="113" to-port="2" />
|
|
<edge from-layer="91" from-port="0" to-layer="109" to-port="0" />
|
|
<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="105" to-port="0" />
|
|
<edge from-layer="96" from-port="1" to-layer="99" to-port="0" />
|
|
<edge from-layer="96" from-port="1" to-layer="103" to-port="0" />
|
|
<edge from-layer="97" from-port="0" to-layer="99" to-port="1" />
|
|
<edge from-layer="98" from-port="0" to-layer="99" to-port="2" />
|
|
<edge from-layer="99" from-port="3" to-layer="104" to-port="0" />
|
|
<edge from-layer="100" from-port="0" to-layer="104" to-port="1" />
|
|
<edge from-layer="101" from-port="0" to-layer="103" to-port="1" />
|
|
<edge from-layer="102" from-port="0" to-layer="103" to-port="2" />
|
|
<edge from-layer="103" from-port="3" to-layer="104" to-port="3" />
|
|
<edge from-layer="104" from-port="4" to-layer="105" to-port="1" />
|
|
<edge from-layer="105" from-port="2" to-layer="106" to-port="0" />
|
|
<edge from-layer="106" from-port="1" to-layer="108" to-port="0" />
|
|
<edge from-layer="107" from-port="0" to-layer="108" to-port="1" />
|
|
<edge from-layer="108" from-port="2" to-layer="109" to-port="1" />
|
|
<edge from-layer="109" from-port="2" to-layer="112" to-port="2" />
|
|
<edge from-layer="109" from-port="2" to-layer="110" to-port="0" />
|
|
<edge from-layer="110" from-port="1" to-layer="112" to-port="0" />
|
|
<edge from-layer="111" from-port="0" to-layer="112" to-port="1" />
|
|
<edge from-layer="112" from-port="3" to-layer="113" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="578" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="462" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="346" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="1390" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="1274" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="230" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="1158" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="1042" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="926" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="810" to-port="3" />
|
|
<edge from-layer="112" from-port="3" to-layer="694" to-port="3" />
|
|
<edge from-layer="113" from-port="4" to-layer="115" to-port="0" />
|
|
<edge from-layer="114" from-port="0" to-layer="115" to-port="1" />
|
|
<edge from-layer="115" from-port="2" to-layer="122" to-port="0" />
|
|
<edge from-layer="116" from-port="1" to-layer="119" to-port="0" />
|
|
<edge from-layer="117" from-port="0" to-layer="119" to-port="1" />
|
|
<edge from-layer="118" from-port="0" to-layer="119" to-port="2" />
|
|
<edge from-layer="119" from-port="3" to-layer="121" to-port="0" />
|
|
<edge from-layer="120" from-port="0" to-layer="121" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="585" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="817" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="933" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="1049" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="1165" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="701" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="237" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="1281" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="1397" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="353" to-port="1" />
|
|
<edge from-layer="120" from-port="0" to-layer="469" to-port="1" />
|
|
<edge from-layer="121" from-port="2" to-layer="122" to-port="1" />
|
|
<edge from-layer="122" from-port="2" to-layer="124" to-port="0" />
|
|
<edge from-layer="123" from-port="0" to-layer="124" to-port="1" />
|
|
<edge from-layer="124" from-port="2" to-layer="129" to-port="0" />
|
|
<edge from-layer="125" from-port="0" to-layer="129" to-port="1" />
|
|
<edge from-layer="126" from-port="0" to-layer="129" to-port="2" />
|
|
<edge from-layer="127" from-port="0" to-layer="129" to-port="3" />
|
|
<edge from-layer="128" from-port="0" to-layer="129" to-port="4" />
|
|
<edge from-layer="129" from-port="5" to-layer="134" to-port="0" />
|
|
<edge from-layer="130" from-port="0" to-layer="131" to-port="0" />
|
|
<edge from-layer="131" from-port="1" 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="134" to-port="1" />
|
|
<edge from-layer="134" from-port="2" to-layer="136" to-port="0" />
|
|
<edge from-layer="135" from-port="0" to-layer="136" to-port="1" />
|
|
<edge from-layer="136" from-port="2" to-layer="137" to-port="0" />
|
|
<edge from-layer="137" from-port="2" to-layer="139" to-port="0" />
|
|
<edge from-layer="138" from-port="0" to-layer="139" to-port="1" />
|
|
<edge from-layer="139" from-port="2" to-layer="141" to-port="0" />
|
|
<edge from-layer="140" from-port="0" to-layer="141" to-port="1" />
|
|
<edge from-layer="141" from-port="2" to-layer="143" to-port="0" />
|
|
<edge from-layer="142" from-port="0" to-layer="143" to-port="1" />
|
|
<edge from-layer="143" from-port="2" to-layer="173" to-port="1" />
|
|
<edge from-layer="143" from-port="2" to-layer="145" to-port="0" />
|
|
<edge from-layer="144" from-port="0" to-layer="145" to-port="1" />
|
|
<edge from-layer="145" from-port="2" to-layer="150" to-port="0" />
|
|
<edge from-layer="146" from-port="0" to-layer="150" to-port="1" />
|
|
<edge from-layer="147" from-port="0" to-layer="150" to-port="2" />
|
|
<edge from-layer="148" from-port="0" to-layer="150" to-port="3" />
|
|
<edge from-layer="149" from-port="0" to-layer="150" to-port="4" />
|
|
<edge from-layer="150" from-port="5" to-layer="155" to-port="0" />
|
|
<edge from-layer="151" from-port="0" 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="155" to-port="1" />
|
|
<edge from-layer="155" from-port="2" to-layer="157" to-port="0" />
|
|
<edge from-layer="156" from-port="0" to-layer="157" to-port="1" />
|
|
<edge from-layer="157" from-port="2" to-layer="158" to-port="0" />
|
|
<edge from-layer="158" from-port="1" to-layer="160" to-port="0" />
|
|
<edge from-layer="159" from-port="0" to-layer="160" to-port="1" />
|
|
<edge from-layer="160" from-port="2" to-layer="165" to-port="0" />
|
|
<edge from-layer="161" from-port="0" to-layer="165" to-port="1" />
|
|
<edge from-layer="162" from-port="0" to-layer="165" to-port="2" />
|
|
<edge from-layer="163" from-port="0" to-layer="165" to-port="3" />
|
|
<edge from-layer="164" from-port="0" to-layer="165" to-port="4" />
|
|
<edge from-layer="165" from-port="5" to-layer="170" to-port="0" />
|
|
<edge from-layer="166" from-port="0" to-layer="167" to-port="0" />
|
|
<edge from-layer="167" from-port="1" 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="170" to-port="1" />
|
|
<edge from-layer="170" from-port="2" to-layer="172" to-port="0" />
|
|
<edge from-layer="171" from-port="0" to-layer="172" to-port="1" />
|
|
<edge from-layer="172" from-port="2" to-layer="173" to-port="0" />
|
|
<edge from-layer="173" from-port="2" to-layer="175" to-port="0" />
|
|
<edge from-layer="174" from-port="0" to-layer="175" to-port="1" />
|
|
<edge from-layer="175" from-port="2" to-layer="177" to-port="0" />
|
|
<edge from-layer="176" from-port="0" to-layer="177" to-port="1" />
|
|
<edge from-layer="177" from-port="2" to-layer="179" to-port="0" />
|
|
<edge from-layer="178" from-port="0" to-layer="179" to-port="1" />
|
|
<edge from-layer="179" from-port="2" to-layer="233" to-port="0" />
|
|
<edge from-layer="179" from-port="2" to-layer="253" to-port="1" />
|
|
<edge from-layer="179" from-port="2" to-layer="181" to-port="0" />
|
|
<edge from-layer="180" from-port="0" to-layer="181" to-port="1" />
|
|
<edge from-layer="181" from-port="2" to-layer="186" to-port="0" />
|
|
<edge from-layer="182" from-port="0" to-layer="186" to-port="1" />
|
|
<edge from-layer="183" from-port="0" to-layer="186" to-port="2" />
|
|
<edge from-layer="184" from-port="0" to-layer="186" to-port="3" />
|
|
<edge from-layer="185" from-port="0" to-layer="186" to-port="4" />
|
|
<edge from-layer="186" from-port="5" to-layer="191" to-port="0" />
|
|
<edge from-layer="186" from-port="5" to-layer="207" to-port="0" />
|
|
<edge from-layer="186" from-port="5" to-layer="223" to-port="0" />
|
|
<edge from-layer="187" from-port="0" to-layer="188" to-port="0" />
|
|
<edge from-layer="188" from-port="1" to-layer="190" to-port="0" />
|
|
<edge from-layer="189" from-port="0" to-layer="190" to-port="1" />
|
|
<edge from-layer="190" from-port="2" to-layer="191" to-port="1" />
|
|
<edge from-layer="191" from-port="2" to-layer="193" to-port="0" />
|
|
<edge from-layer="192" from-port="0" to-layer="193" to-port="1" />
|
|
<edge from-layer="193" from-port="2" to-layer="198" to-port="0" />
|
|
<edge from-layer="194" from-port="0" to-layer="198" to-port="1" />
|
|
<edge from-layer="195" from-port="0" to-layer="198" to-port="2" />
|
|
<edge from-layer="196" from-port="0" to-layer="198" to-port="3" />
|
|
<edge from-layer="197" from-port="0" to-layer="198" to-port="4" />
|
|
<edge from-layer="198" from-port="5" 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="202" to-port="0" />
|
|
<edge from-layer="201" from-port="0" to-layer="202" to-port="1" />
|
|
<edge from-layer="202" from-port="2" to-layer="230" to-port="0" />
|
|
<edge from-layer="203" from-port="0" to-layer="204" to-port="0" />
|
|
<edge from-layer="204" from-port="1" to-layer="206" to-port="0" />
|
|
<edge from-layer="205" from-port="0" to-layer="206" to-port="1" />
|
|
<edge from-layer="206" from-port="2" to-layer="207" to-port="1" />
|
|
<edge from-layer="207" from-port="2" to-layer="209" to-port="0" />
|
|
<edge from-layer="208" from-port="0" to-layer="209" to-port="1" />
|
|
<edge from-layer="209" from-port="2" to-layer="214" to-port="0" />
|
|
<edge from-layer="210" from-port="0" to-layer="214" to-port="1" />
|
|
<edge from-layer="211" from-port="0" to-layer="214" to-port="2" />
|
|
<edge from-layer="212" from-port="0" to-layer="214" to-port="3" />
|
|
<edge from-layer="213" from-port="0" to-layer="214" to-port="4" />
|
|
<edge from-layer="214" from-port="5" 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="218" to-port="0" />
|
|
<edge from-layer="217" from-port="0" to-layer="218" to-port="1" />
|
|
<edge from-layer="218" from-port="2" to-layer="230" to-port="1" />
|
|
<edge from-layer="219" from-port="0" to-layer="220" to-port="0" />
|
|
<edge from-layer="220" from-port="1" to-layer="222" to-port="0" />
|
|
<edge from-layer="221" from-port="0" to-layer="222" to-port="1" />
|
|
<edge from-layer="222" from-port="2" to-layer="223" to-port="1" />
|
|
<edge from-layer="223" from-port="2" to-layer="225" to-port="0" />
|
|
<edge from-layer="224" from-port="0" to-layer="225" to-port="1" />
|
|
<edge from-layer="225" from-port="2" to-layer="227" to-port="0" />
|
|
<edge from-layer="226" from-port="0" to-layer="227" to-port="1" />
|
|
<edge from-layer="227" from-port="2" to-layer="229" to-port="0" />
|
|
<edge from-layer="228" from-port="0" to-layer="229" to-port="1" />
|
|
<edge from-layer="229" from-port="2" to-layer="230" to-port="2" />
|
|
<edge from-layer="230" from-port="4" to-layer="232" to-port="0" />
|
|
<edge from-layer="231" from-port="0" to-layer="232" to-port="1" />
|
|
<edge from-layer="232" from-port="2" to-layer="238" to-port="0" />
|
|
<edge from-layer="233" from-port="1" to-layer="236" to-port="0" />
|
|
<edge from-layer="234" from-port="0" to-layer="236" to-port="1" />
|
|
<edge from-layer="235" from-port="0" to-layer="236" to-port="2" />
|
|
<edge from-layer="236" from-port="3" to-layer="237" to-port="0" />
|
|
<edge from-layer="237" from-port="2" to-layer="238" to-port="1" />
|
|
<edge from-layer="238" from-port="2" to-layer="240" to-port="0" />
|
|
<edge from-layer="239" from-port="0" to-layer="240" to-port="1" />
|
|
<edge from-layer="240" from-port="2" to-layer="245" to-port="0" />
|
|
<edge from-layer="241" from-port="0" to-layer="245" to-port="1" />
|
|
<edge from-layer="242" from-port="0" to-layer="245" to-port="2" />
|
|
<edge from-layer="243" from-port="0" to-layer="245" to-port="3" />
|
|
<edge from-layer="244" from-port="0" to-layer="245" to-port="4" />
|
|
<edge from-layer="245" from-port="5" to-layer="250" to-port="0" />
|
|
<edge from-layer="246" from-port="0" to-layer="247" to-port="0" />
|
|
<edge from-layer="247" from-port="1" 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="250" to-port="1" />
|
|
<edge from-layer="250" from-port="2" to-layer="252" to-port="0" />
|
|
<edge from-layer="251" from-port="0" to-layer="252" to-port="1" />
|
|
<edge from-layer="252" from-port="2" to-layer="253" to-port="0" />
|
|
<edge from-layer="253" from-port="2" to-layer="255" to-port="0" />
|
|
<edge from-layer="254" from-port="0" to-layer="255" to-port="1" />
|
|
<edge from-layer="255" from-port="2" to-layer="257" to-port="0" />
|
|
<edge from-layer="256" from-port="0" to-layer="257" to-port="1" />
|
|
<edge from-layer="257" from-port="2" to-layer="259" to-port="0" />
|
|
<edge from-layer="258" from-port="0" to-layer="259" to-port="1" />
|
|
<edge from-layer="259" from-port="2" to-layer="289" to-port="1" />
|
|
<edge from-layer="259" from-port="2" to-layer="261" to-port="0" />
|
|
<edge from-layer="260" from-port="0" to-layer="261" to-port="1" />
|
|
<edge from-layer="261" from-port="2" to-layer="266" to-port="0" />
|
|
<edge from-layer="262" from-port="0" to-layer="266" to-port="1" />
|
|
<edge from-layer="263" from-port="0" to-layer="266" to-port="2" />
|
|
<edge from-layer="264" from-port="0" to-layer="266" to-port="3" />
|
|
<edge from-layer="265" from-port="0" to-layer="266" to-port="4" />
|
|
<edge from-layer="266" from-port="5" to-layer="271" to-port="0" />
|
|
<edge from-layer="267" from-port="0" to-layer="268" to-port="0" />
|
|
<edge from-layer="268" from-port="1" to-layer="270" to-port="0" />
|
|
<edge from-layer="269" from-port="0" to-layer="270" to-port="1" />
|
|
<edge from-layer="270" from-port="2" to-layer="271" to-port="1" />
|
|
<edge from-layer="271" from-port="2" to-layer="273" to-port="0" />
|
|
<edge from-layer="272" from-port="0" to-layer="273" to-port="1" />
|
|
<edge from-layer="273" from-port="2" to-layer="274" to-port="0" />
|
|
<edge from-layer="274" from-port="1" to-layer="276" to-port="0" />
|
|
<edge from-layer="275" from-port="0" to-layer="276" to-port="1" />
|
|
<edge from-layer="276" from-port="2" to-layer="281" to-port="0" />
|
|
<edge from-layer="277" from-port="0" to-layer="281" to-port="1" />
|
|
<edge from-layer="278" from-port="0" to-layer="281" to-port="2" />
|
|
<edge from-layer="279" from-port="0" to-layer="281" to-port="3" />
|
|
<edge from-layer="280" from-port="0" to-layer="281" to-port="4" />
|
|
<edge from-layer="281" from-port="5" to-layer="286" to-port="0" />
|
|
<edge from-layer="282" from-port="0" to-layer="283" to-port="0" />
|
|
<edge from-layer="283" from-port="1" 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="286" to-port="1" />
|
|
<edge from-layer="286" from-port="2" to-layer="288" to-port="0" />
|
|
<edge from-layer="287" from-port="0" to-layer="288" to-port="1" />
|
|
<edge from-layer="288" from-port="2" to-layer="289" to-port="0" />
|
|
<edge from-layer="289" from-port="2" to-layer="291" to-port="0" />
|
|
<edge from-layer="290" from-port="0" to-layer="291" to-port="1" />
|
|
<edge from-layer="291" from-port="2" to-layer="293" to-port="0" />
|
|
<edge from-layer="292" from-port="0" to-layer="293" to-port="1" />
|
|
<edge from-layer="293" from-port="2" to-layer="295" to-port="0" />
|
|
<edge from-layer="294" from-port="0" to-layer="295" to-port="1" />
|
|
<edge from-layer="295" from-port="2" to-layer="349" to-port="0" />
|
|
<edge from-layer="295" from-port="2" to-layer="369" to-port="1" />
|
|
<edge from-layer="295" from-port="2" to-layer="297" to-port="0" />
|
|
<edge from-layer="296" from-port="0" to-layer="297" to-port="1" />
|
|
<edge from-layer="297" from-port="2" to-layer="302" to-port="0" />
|
|
<edge from-layer="298" from-port="0" to-layer="302" to-port="1" />
|
|
<edge from-layer="299" from-port="0" to-layer="302" to-port="2" />
|
|
<edge from-layer="300" from-port="0" to-layer="302" to-port="3" />
|
|
<edge from-layer="301" from-port="0" to-layer="302" to-port="4" />
|
|
<edge from-layer="302" from-port="5" to-layer="339" to-port="0" />
|
|
<edge from-layer="302" from-port="5" to-layer="323" to-port="0" />
|
|
<edge from-layer="302" from-port="5" to-layer="307" to-port="0" />
|
|
<edge from-layer="303" from-port="0" to-layer="304" to-port="0" />
|
|
<edge from-layer="304" from-port="1" 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="307" to-port="1" />
|
|
<edge from-layer="307" from-port="2" to-layer="309" to-port="0" />
|
|
<edge from-layer="308" from-port="0" to-layer="309" to-port="1" />
|
|
<edge from-layer="309" from-port="2" to-layer="314" to-port="0" />
|
|
<edge from-layer="310" from-port="0" to-layer="314" to-port="1" />
|
|
<edge from-layer="311" from-port="0" to-layer="314" to-port="2" />
|
|
<edge from-layer="312" from-port="0" to-layer="314" to-port="3" />
|
|
<edge from-layer="313" from-port="0" to-layer="314" to-port="4" />
|
|
<edge from-layer="314" from-port="5" to-layer="316" to-port="0" />
|
|
<edge from-layer="315" from-port="0" to-layer="316" to-port="1" />
|
|
<edge from-layer="316" from-port="2" to-layer="318" to-port="0" />
|
|
<edge from-layer="317" from-port="0" to-layer="318" to-port="1" />
|
|
<edge from-layer="318" from-port="2" to-layer="346" to-port="0" />
|
|
<edge from-layer="319" from-port="0" to-layer="320" to-port="0" />
|
|
<edge from-layer="320" from-port="1" 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="323" to-port="1" />
|
|
<edge from-layer="323" from-port="2" to-layer="325" to-port="0" />
|
|
<edge from-layer="324" from-port="0" to-layer="325" to-port="1" />
|
|
<edge from-layer="325" from-port="2" to-layer="330" to-port="0" />
|
|
<edge from-layer="326" from-port="0" to-layer="330" to-port="1" />
|
|
<edge from-layer="327" from-port="0" to-layer="330" to-port="2" />
|
|
<edge from-layer="328" from-port="0" to-layer="330" to-port="3" />
|
|
<edge from-layer="329" from-port="0" to-layer="330" to-port="4" />
|
|
<edge from-layer="330" from-port="5" to-layer="332" to-port="0" />
|
|
<edge from-layer="331" from-port="0" to-layer="332" to-port="1" />
|
|
<edge from-layer="332" from-port="2" to-layer="334" to-port="0" />
|
|
<edge from-layer="333" from-port="0" to-layer="334" to-port="1" />
|
|
<edge from-layer="334" from-port="2" to-layer="346" to-port="1" />
|
|
<edge from-layer="335" from-port="0" to-layer="336" to-port="0" />
|
|
<edge from-layer="336" from-port="1" 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="339" to-port="1" />
|
|
<edge from-layer="339" from-port="2" to-layer="341" to-port="0" />
|
|
<edge from-layer="340" from-port="0" to-layer="341" to-port="1" />
|
|
<edge from-layer="341" from-port="2" to-layer="343" to-port="0" />
|
|
<edge from-layer="342" from-port="0" to-layer="343" to-port="1" />
|
|
<edge from-layer="343" from-port="2" to-layer="345" to-port="0" />
|
|
<edge from-layer="344" from-port="0" to-layer="345" to-port="1" />
|
|
<edge from-layer="345" from-port="2" to-layer="346" to-port="2" />
|
|
<edge from-layer="346" from-port="4" to-layer="348" to-port="0" />
|
|
<edge from-layer="347" from-port="0" to-layer="348" to-port="1" />
|
|
<edge from-layer="348" from-port="2" to-layer="354" to-port="0" />
|
|
<edge from-layer="349" from-port="1" to-layer="352" to-port="0" />
|
|
<edge from-layer="350" from-port="0" to-layer="352" to-port="1" />
|
|
<edge from-layer="351" from-port="0" to-layer="352" to-port="2" />
|
|
<edge from-layer="352" from-port="3" to-layer="353" to-port="0" />
|
|
<edge from-layer="353" from-port="2" to-layer="354" to-port="1" />
|
|
<edge from-layer="354" from-port="2" to-layer="356" to-port="0" />
|
|
<edge from-layer="355" from-port="0" to-layer="356" to-port="1" />
|
|
<edge from-layer="356" from-port="2" to-layer="361" to-port="0" />
|
|
<edge from-layer="357" from-port="0" to-layer="361" to-port="1" />
|
|
<edge from-layer="358" from-port="0" to-layer="361" to-port="2" />
|
|
<edge from-layer="359" from-port="0" to-layer="361" to-port="3" />
|
|
<edge from-layer="360" from-port="0" to-layer="361" to-port="4" />
|
|
<edge from-layer="361" from-port="5" to-layer="366" to-port="0" />
|
|
<edge from-layer="362" from-port="0" to-layer="363" to-port="0" />
|
|
<edge from-layer="363" from-port="1" to-layer="365" to-port="0" />
|
|
<edge from-layer="364" from-port="0" to-layer="365" to-port="1" />
|
|
<edge from-layer="365" from-port="2" to-layer="366" to-port="1" />
|
|
<edge from-layer="366" from-port="2" to-layer="368" to-port="0" />
|
|
<edge from-layer="367" from-port="0" to-layer="368" to-port="1" />
|
|
<edge from-layer="368" from-port="2" to-layer="369" to-port="0" />
|
|
<edge from-layer="369" from-port="2" to-layer="371" to-port="0" />
|
|
<edge from-layer="370" from-port="0" to-layer="371" to-port="1" />
|
|
<edge from-layer="371" from-port="2" to-layer="373" to-port="0" />
|
|
<edge from-layer="372" from-port="0" to-layer="373" to-port="1" />
|
|
<edge from-layer="373" from-port="2" to-layer="375" to-port="0" />
|
|
<edge from-layer="374" from-port="0" to-layer="375" to-port="1" />
|
|
<edge from-layer="375" from-port="2" to-layer="405" to-port="1" />
|
|
<edge from-layer="375" from-port="2" to-layer="377" to-port="0" />
|
|
<edge from-layer="376" from-port="0" to-layer="377" to-port="1" />
|
|
<edge from-layer="377" from-port="2" to-layer="382" to-port="0" />
|
|
<edge from-layer="378" from-port="0" to-layer="382" to-port="1" />
|
|
<edge from-layer="379" from-port="0" to-layer="382" to-port="2" />
|
|
<edge from-layer="380" from-port="0" to-layer="382" to-port="3" />
|
|
<edge from-layer="381" from-port="0" to-layer="382" to-port="4" />
|
|
<edge from-layer="382" from-port="5" to-layer="387" to-port="0" />
|
|
<edge from-layer="383" from-port="0" to-layer="384" to-port="0" />
|
|
<edge from-layer="384" from-port="1" to-layer="386" to-port="0" />
|
|
<edge from-layer="385" from-port="0" to-layer="386" to-port="1" />
|
|
<edge from-layer="386" from-port="2" to-layer="387" to-port="1" />
|
|
<edge from-layer="387" from-port="2" to-layer="389" to-port="0" />
|
|
<edge from-layer="388" from-port="0" to-layer="389" to-port="1" />
|
|
<edge from-layer="389" from-port="2" to-layer="390" to-port="0" />
|
|
<edge from-layer="390" from-port="1" to-layer="392" to-port="0" />
|
|
<edge from-layer="391" from-port="0" to-layer="392" to-port="1" />
|
|
<edge from-layer="392" from-port="2" to-layer="397" to-port="0" />
|
|
<edge from-layer="393" from-port="0" to-layer="397" to-port="1" />
|
|
<edge from-layer="394" from-port="0" to-layer="397" to-port="2" />
|
|
<edge from-layer="395" from-port="0" to-layer="397" to-port="3" />
|
|
<edge from-layer="396" from-port="0" to-layer="397" to-port="4" />
|
|
<edge from-layer="397" from-port="5" to-layer="402" to-port="0" />
|
|
<edge from-layer="398" from-port="0" to-layer="399" to-port="0" />
|
|
<edge from-layer="399" from-port="1" to-layer="401" to-port="0" />
|
|
<edge from-layer="400" from-port="0" to-layer="401" to-port="1" />
|
|
<edge from-layer="401" from-port="2" to-layer="402" to-port="1" />
|
|
<edge from-layer="402" from-port="2" to-layer="404" to-port="0" />
|
|
<edge from-layer="403" from-port="0" to-layer="404" to-port="1" />
|
|
<edge from-layer="404" from-port="2" to-layer="405" to-port="0" />
|
|
<edge from-layer="405" from-port="2" to-layer="407" to-port="0" />
|
|
<edge from-layer="406" from-port="0" to-layer="407" to-port="1" />
|
|
<edge from-layer="407" from-port="2" to-layer="409" to-port="0" />
|
|
<edge from-layer="408" from-port="0" to-layer="409" to-port="1" />
|
|
<edge from-layer="409" from-port="2" to-layer="411" to-port="0" />
|
|
<edge from-layer="410" from-port="0" to-layer="411" to-port="1" />
|
|
<edge from-layer="411" from-port="2" to-layer="413" to-port="0" />
|
|
<edge from-layer="411" from-port="2" to-layer="465" to-port="0" />
|
|
<edge from-layer="411" from-port="2" to-layer="485" to-port="1" />
|
|
<edge from-layer="412" from-port="0" to-layer="413" to-port="1" />
|
|
<edge from-layer="413" from-port="2" to-layer="418" to-port="0" />
|
|
<edge from-layer="414" from-port="0" to-layer="418" to-port="1" />
|
|
<edge from-layer="415" from-port="0" to-layer="418" to-port="2" />
|
|
<edge from-layer="416" from-port="0" to-layer="418" to-port="3" />
|
|
<edge from-layer="417" from-port="0" to-layer="418" to-port="4" />
|
|
<edge from-layer="418" from-port="5" to-layer="439" to-port="0" />
|
|
<edge from-layer="418" from-port="5" to-layer="455" to-port="0" />
|
|
<edge from-layer="418" from-port="5" to-layer="423" to-port="0" />
|
|
<edge from-layer="419" from-port="0" to-layer="420" to-port="0" />
|
|
<edge from-layer="420" from-port="1" to-layer="422" to-port="0" />
|
|
<edge from-layer="421" from-port="0" to-layer="422" to-port="1" />
|
|
<edge from-layer="422" from-port="2" to-layer="423" to-port="1" />
|
|
<edge from-layer="423" from-port="2" to-layer="425" to-port="0" />
|
|
<edge from-layer="424" from-port="0" to-layer="425" to-port="1" />
|
|
<edge from-layer="425" from-port="2" to-layer="430" to-port="0" />
|
|
<edge from-layer="426" from-port="0" to-layer="430" to-port="1" />
|
|
<edge from-layer="427" from-port="0" to-layer="430" to-port="2" />
|
|
<edge from-layer="428" from-port="0" to-layer="430" to-port="3" />
|
|
<edge from-layer="429" from-port="0" to-layer="430" to-port="4" />
|
|
<edge from-layer="430" from-port="5" to-layer="432" to-port="0" />
|
|
<edge from-layer="431" from-port="0" to-layer="432" to-port="1" />
|
|
<edge from-layer="432" from-port="2" to-layer="434" to-port="0" />
|
|
<edge from-layer="433" from-port="0" to-layer="434" to-port="1" />
|
|
<edge from-layer="434" from-port="2" to-layer="462" to-port="0" />
|
|
<edge from-layer="435" from-port="0" to-layer="436" to-port="0" />
|
|
<edge from-layer="436" from-port="1" to-layer="438" to-port="0" />
|
|
<edge from-layer="437" from-port="0" to-layer="438" to-port="1" />
|
|
<edge from-layer="438" from-port="2" to-layer="439" to-port="1" />
|
|
<edge from-layer="439" from-port="2" to-layer="441" to-port="0" />
|
|
<edge from-layer="440" from-port="0" to-layer="441" to-port="1" />
|
|
<edge from-layer="441" from-port="2" to-layer="446" to-port="0" />
|
|
<edge from-layer="442" from-port="0" to-layer="446" to-port="1" />
|
|
<edge from-layer="443" from-port="0" to-layer="446" to-port="2" />
|
|
<edge from-layer="444" from-port="0" to-layer="446" to-port="3" />
|
|
<edge from-layer="445" from-port="0" to-layer="446" to-port="4" />
|
|
<edge from-layer="446" from-port="5" to-layer="448" to-port="0" />
|
|
<edge from-layer="447" from-port="0" to-layer="448" to-port="1" />
|
|
<edge from-layer="448" from-port="2" to-layer="450" to-port="0" />
|
|
<edge from-layer="449" from-port="0" to-layer="450" to-port="1" />
|
|
<edge from-layer="450" from-port="2" to-layer="462" to-port="1" />
|
|
<edge from-layer="451" from-port="0" to-layer="452" to-port="0" />
|
|
<edge from-layer="452" from-port="1" 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="455" to-port="1" />
|
|
<edge from-layer="455" from-port="2" to-layer="457" to-port="0" />
|
|
<edge from-layer="456" from-port="0" to-layer="457" to-port="1" />
|
|
<edge from-layer="457" from-port="2" to-layer="459" to-port="0" />
|
|
<edge from-layer="458" from-port="0" to-layer="459" to-port="1" />
|
|
<edge from-layer="459" from-port="2" to-layer="461" to-port="0" />
|
|
<edge from-layer="460" from-port="0" to-layer="461" to-port="1" />
|
|
<edge from-layer="461" from-port="2" to-layer="462" to-port="2" />
|
|
<edge from-layer="462" from-port="4" to-layer="464" to-port="0" />
|
|
<edge from-layer="463" from-port="0" to-layer="464" to-port="1" />
|
|
<edge from-layer="464" from-port="2" to-layer="470" to-port="0" />
|
|
<edge from-layer="465" from-port="1" to-layer="468" to-port="0" />
|
|
<edge from-layer="466" from-port="0" to-layer="468" to-port="1" />
|
|
<edge from-layer="467" from-port="0" to-layer="468" to-port="2" />
|
|
<edge from-layer="468" from-port="3" to-layer="469" to-port="0" />
|
|
<edge from-layer="469" from-port="2" to-layer="470" to-port="1" />
|
|
<edge from-layer="470" from-port="2" to-layer="472" to-port="0" />
|
|
<edge from-layer="471" from-port="0" to-layer="472" to-port="1" />
|
|
<edge from-layer="472" from-port="2" to-layer="477" to-port="0" />
|
|
<edge from-layer="473" from-port="0" to-layer="477" to-port="1" />
|
|
<edge from-layer="474" from-port="0" to-layer="477" to-port="2" />
|
|
<edge from-layer="475" from-port="0" to-layer="477" to-port="3" />
|
|
<edge from-layer="476" from-port="0" to-layer="477" to-port="4" />
|
|
<edge from-layer="477" from-port="5" to-layer="482" to-port="0" />
|
|
<edge from-layer="478" from-port="0" to-layer="479" to-port="0" />
|
|
<edge from-layer="479" from-port="1" to-layer="481" to-port="0" />
|
|
<edge from-layer="480" from-port="0" to-layer="481" to-port="1" />
|
|
<edge from-layer="481" from-port="2" to-layer="482" to-port="1" />
|
|
<edge from-layer="482" from-port="2" 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="0" />
|
|
<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="489" to-port="0" />
|
|
<edge from-layer="488" from-port="0" to-layer="489" to-port="1" />
|
|
<edge from-layer="489" from-port="2" to-layer="491" to-port="0" />
|
|
<edge from-layer="490" from-port="0" to-layer="491" to-port="1" />
|
|
<edge from-layer="491" from-port="2" to-layer="493" to-port="0" />
|
|
<edge from-layer="491" from-port="2" to-layer="521" to-port="1" />
|
|
<edge from-layer="492" from-port="0" to-layer="493" to-port="1" />
|
|
<edge from-layer="493" from-port="2" to-layer="498" to-port="0" />
|
|
<edge from-layer="494" from-port="0" to-layer="498" to-port="1" />
|
|
<edge from-layer="495" from-port="0" to-layer="498" to-port="2" />
|
|
<edge from-layer="496" from-port="0" to-layer="498" to-port="3" />
|
|
<edge from-layer="497" from-port="0" to-layer="498" to-port="4" />
|
|
<edge from-layer="498" from-port="5" to-layer="503" to-port="0" />
|
|
<edge from-layer="499" from-port="0" to-layer="500" to-port="0" />
|
|
<edge from-layer="500" from-port="1" 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="1" />
|
|
<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="506" to-port="0" />
|
|
<edge from-layer="506" from-port="1" to-layer="508" to-port="0" />
|
|
<edge from-layer="507" from-port="0" to-layer="508" to-port="1" />
|
|
<edge from-layer="508" from-port="2" to-layer="513" to-port="0" />
|
|
<edge from-layer="509" from-port="0" to-layer="513" to-port="1" />
|
|
<edge from-layer="510" from-port="0" to-layer="513" to-port="2" />
|
|
<edge from-layer="511" from-port="0" to-layer="513" to-port="3" />
|
|
<edge from-layer="512" from-port="0" to-layer="513" to-port="4" />
|
|
<edge from-layer="513" from-port="5" to-layer="518" to-port="0" />
|
|
<edge from-layer="514" from-port="0" to-layer="515" to-port="0" />
|
|
<edge from-layer="515" from-port="1" to-layer="517" to-port="0" />
|
|
<edge from-layer="516" from-port="0" to-layer="517" to-port="1" />
|
|
<edge from-layer="517" from-port="2" to-layer="518" to-port="1" />
|
|
<edge from-layer="518" from-port="2" 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="0" />
|
|
<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="525" to-port="0" />
|
|
<edge from-layer="524" from-port="0" to-layer="525" to-port="1" />
|
|
<edge from-layer="525" from-port="2" to-layer="527" to-port="0" />
|
|
<edge from-layer="526" from-port="0" to-layer="527" to-port="1" />
|
|
<edge from-layer="527" from-port="2" to-layer="529" to-port="0" />
|
|
<edge from-layer="527" from-port="2" to-layer="581" to-port="0" />
|
|
<edge from-layer="527" from-port="2" to-layer="601" to-port="1" />
|
|
<edge from-layer="528" from-port="0" to-layer="529" to-port="1" />
|
|
<edge from-layer="529" from-port="2" to-layer="534" to-port="0" />
|
|
<edge from-layer="530" from-port="0" to-layer="534" to-port="1" />
|
|
<edge from-layer="531" from-port="0" to-layer="534" to-port="2" />
|
|
<edge from-layer="532" from-port="0" to-layer="534" to-port="3" />
|
|
<edge from-layer="533" from-port="0" to-layer="534" to-port="4" />
|
|
<edge from-layer="534" from-port="5" to-layer="555" to-port="0" />
|
|
<edge from-layer="534" from-port="5" to-layer="539" to-port="0" />
|
|
<edge from-layer="534" from-port="5" to-layer="571" to-port="0" />
|
|
<edge from-layer="535" from-port="0" to-layer="536" to-port="0" />
|
|
<edge from-layer="536" from-port="1" to-layer="538" to-port="0" />
|
|
<edge from-layer="537" from-port="0" to-layer="538" to-port="1" />
|
|
<edge from-layer="538" from-port="2" to-layer="539" to-port="1" />
|
|
<edge from-layer="539" from-port="2" to-layer="541" to-port="0" />
|
|
<edge from-layer="540" from-port="0" to-layer="541" to-port="1" />
|
|
<edge from-layer="541" from-port="2" to-layer="546" to-port="0" />
|
|
<edge from-layer="542" from-port="0" to-layer="546" to-port="1" />
|
|
<edge from-layer="543" from-port="0" to-layer="546" to-port="2" />
|
|
<edge from-layer="544" from-port="0" to-layer="546" to-port="3" />
|
|
<edge from-layer="545" from-port="0" to-layer="546" to-port="4" />
|
|
<edge from-layer="546" from-port="5" 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="550" to-port="0" />
|
|
<edge from-layer="549" from-port="0" to-layer="550" to-port="1" />
|
|
<edge from-layer="550" from-port="2" to-layer="578" to-port="0" />
|
|
<edge from-layer="551" from-port="0" to-layer="552" to-port="0" />
|
|
<edge from-layer="552" from-port="1" to-layer="554" to-port="0" />
|
|
<edge from-layer="553" from-port="0" to-layer="554" to-port="1" />
|
|
<edge from-layer="554" from-port="2" 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="562" to-port="0" />
|
|
<edge from-layer="558" from-port="0" to-layer="562" to-port="1" />
|
|
<edge from-layer="559" from-port="0" to-layer="562" to-port="2" />
|
|
<edge from-layer="560" from-port="0" to-layer="562" to-port="3" />
|
|
<edge from-layer="561" from-port="0" to-layer="562" to-port="4" />
|
|
<edge from-layer="562" from-port="5" 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="578" to-port="1" />
|
|
<edge from-layer="567" from-port="0" to-layer="568" to-port="0" />
|
|
<edge from-layer="568" from-port="1" to-layer="570" to-port="0" />
|
|
<edge from-layer="569" from-port="0" to-layer="570" to-port="1" />
|
|
<edge from-layer="570" from-port="2" to-layer="571" to-port="1" />
|
|
<edge from-layer="571" from-port="2" to-layer="573" to-port="0" />
|
|
<edge from-layer="572" from-port="0" to-layer="573" to-port="1" />
|
|
<edge from-layer="573" from-port="2" 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="577" to-port="0" />
|
|
<edge from-layer="576" from-port="0" to-layer="577" to-port="1" />
|
|
<edge from-layer="577" from-port="2" to-layer="578" to-port="2" />
|
|
<edge from-layer="578" from-port="4" to-layer="580" to-port="0" />
|
|
<edge from-layer="579" from-port="0" to-layer="580" to-port="1" />
|
|
<edge from-layer="580" from-port="2" to-layer="586" to-port="0" />
|
|
<edge from-layer="581" from-port="1" to-layer="584" to-port="0" />
|
|
<edge from-layer="582" from-port="0" to-layer="584" to-port="1" />
|
|
<edge from-layer="583" from-port="0" to-layer="584" to-port="2" />
|
|
<edge from-layer="584" from-port="3" to-layer="585" to-port="0" />
|
|
<edge from-layer="585" from-port="2" to-layer="586" to-port="1" />
|
|
<edge from-layer="586" from-port="2" to-layer="588" to-port="0" />
|
|
<edge from-layer="587" from-port="0" to-layer="588" to-port="1" />
|
|
<edge from-layer="588" from-port="2" to-layer="593" to-port="0" />
|
|
<edge from-layer="589" from-port="0" to-layer="593" to-port="1" />
|
|
<edge from-layer="590" from-port="0" to-layer="593" to-port="2" />
|
|
<edge from-layer="591" from-port="0" to-layer="593" to-port="3" />
|
|
<edge from-layer="592" from-port="0" to-layer="593" to-port="4" />
|
|
<edge from-layer="593" from-port="5" to-layer="598" to-port="0" />
|
|
<edge from-layer="594" from-port="0" to-layer="595" to-port="0" />
|
|
<edge from-layer="595" from-port="1" to-layer="597" to-port="0" />
|
|
<edge from-layer="596" from-port="0" to-layer="597" to-port="1" />
|
|
<edge from-layer="597" from-port="2" to-layer="598" to-port="1" />
|
|
<edge from-layer="598" from-port="2" to-layer="600" to-port="0" />
|
|
<edge from-layer="599" from-port="0" to-layer="600" to-port="1" />
|
|
<edge from-layer="600" from-port="2" to-layer="601" to-port="0" />
|
|
<edge from-layer="601" from-port="2" to-layer="603" to-port="0" />
|
|
<edge from-layer="602" from-port="0" to-layer="603" to-port="1" />
|
|
<edge from-layer="603" from-port="2" to-layer="605" to-port="0" />
|
|
<edge from-layer="604" from-port="0" to-layer="605" to-port="1" />
|
|
<edge from-layer="605" from-port="2" to-layer="607" to-port="0" />
|
|
<edge from-layer="606" from-port="0" to-layer="607" to-port="1" />
|
|
<edge from-layer="607" from-port="2" to-layer="637" to-port="1" />
|
|
<edge from-layer="607" from-port="2" to-layer="609" to-port="0" />
|
|
<edge from-layer="608" from-port="0" to-layer="609" to-port="1" />
|
|
<edge from-layer="609" from-port="2" to-layer="614" to-port="0" />
|
|
<edge from-layer="610" from-port="0" to-layer="614" to-port="1" />
|
|
<edge from-layer="611" from-port="0" to-layer="614" to-port="2" />
|
|
<edge from-layer="612" from-port="0" to-layer="614" to-port="3" />
|
|
<edge from-layer="613" from-port="0" to-layer="614" to-port="4" />
|
|
<edge from-layer="614" from-port="5" to-layer="619" to-port="0" />
|
|
<edge from-layer="615" from-port="0" to-layer="616" to-port="0" />
|
|
<edge from-layer="616" from-port="1" to-layer="618" to-port="0" />
|
|
<edge from-layer="617" from-port="0" to-layer="618" to-port="1" />
|
|
<edge from-layer="618" from-port="2" 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="622" to-port="0" />
|
|
<edge from-layer="622" from-port="1" to-layer="624" to-port="0" />
|
|
<edge from-layer="623" from-port="0" to-layer="624" to-port="1" />
|
|
<edge from-layer="624" from-port="2" to-layer="629" to-port="0" />
|
|
<edge from-layer="625" from-port="0" to-layer="629" to-port="1" />
|
|
<edge from-layer="626" from-port="0" to-layer="629" to-port="2" />
|
|
<edge from-layer="627" from-port="0" to-layer="629" to-port="3" />
|
|
<edge from-layer="628" from-port="0" to-layer="629" to-port="4" />
|
|
<edge from-layer="629" from-port="5" to-layer="634" to-port="0" />
|
|
<edge from-layer="630" from-port="0" to-layer="631" to-port="0" />
|
|
<edge from-layer="631" from-port="1" to-layer="633" to-port="0" />
|
|
<edge from-layer="632" from-port="0" to-layer="633" to-port="1" />
|
|
<edge from-layer="633" from-port="2" to-layer="634" to-port="1" />
|
|
<edge from-layer="634" from-port="2" to-layer="636" to-port="0" />
|
|
<edge from-layer="635" from-port="0" to-layer="636" to-port="1" />
|
|
<edge from-layer="636" from-port="2" to-layer="637" to-port="0" />
|
|
<edge from-layer="637" from-port="2" to-layer="639" to-port="0" />
|
|
<edge from-layer="638" from-port="0" to-layer="639" to-port="1" />
|
|
<edge from-layer="639" from-port="2" to-layer="641" to-port="0" />
|
|
<edge from-layer="640" from-port="0" to-layer="641" to-port="1" />
|
|
<edge from-layer="641" from-port="2" to-layer="643" to-port="0" />
|
|
<edge from-layer="642" from-port="0" to-layer="643" to-port="1" />
|
|
<edge from-layer="643" from-port="2" to-layer="697" to-port="0" />
|
|
<edge from-layer="643" from-port="2" to-layer="717" to-port="1" />
|
|
<edge from-layer="643" from-port="2" to-layer="645" to-port="0" />
|
|
<edge from-layer="644" from-port="0" to-layer="645" to-port="1" />
|
|
<edge from-layer="645" from-port="2" to-layer="650" to-port="0" />
|
|
<edge from-layer="646" from-port="0" to-layer="650" to-port="1" />
|
|
<edge from-layer="647" from-port="0" to-layer="650" to-port="2" />
|
|
<edge from-layer="648" from-port="0" to-layer="650" to-port="3" />
|
|
<edge from-layer="649" from-port="0" to-layer="650" to-port="4" />
|
|
<edge from-layer="650" from-port="5" to-layer="687" to-port="0" />
|
|
<edge from-layer="650" from-port="5" to-layer="671" to-port="0" />
|
|
<edge from-layer="650" from-port="5" to-layer="655" to-port="0" />
|
|
<edge from-layer="651" from-port="0" to-layer="652" to-port="0" />
|
|
<edge from-layer="652" from-port="1" to-layer="654" to-port="0" />
|
|
<edge from-layer="653" from-port="0" to-layer="654" to-port="1" />
|
|
<edge from-layer="654" from-port="2" to-layer="655" to-port="1" />
|
|
<edge from-layer="655" from-port="2" to-layer="657" to-port="0" />
|
|
<edge from-layer="656" from-port="0" to-layer="657" to-port="1" />
|
|
<edge from-layer="657" from-port="2" to-layer="662" to-port="0" />
|
|
<edge from-layer="658" from-port="0" to-layer="662" to-port="1" />
|
|
<edge from-layer="659" from-port="0" to-layer="662" to-port="2" />
|
|
<edge from-layer="660" from-port="0" to-layer="662" to-port="3" />
|
|
<edge from-layer="661" from-port="0" to-layer="662" to-port="4" />
|
|
<edge from-layer="662" from-port="5" 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="666" to-port="0" />
|
|
<edge from-layer="665" from-port="0" to-layer="666" to-port="1" />
|
|
<edge from-layer="666" from-port="2" to-layer="694" to-port="0" />
|
|
<edge from-layer="667" from-port="0" to-layer="668" to-port="0" />
|
|
<edge from-layer="668" from-port="1" to-layer="670" to-port="0" />
|
|
<edge from-layer="669" from-port="0" to-layer="670" to-port="1" />
|
|
<edge from-layer="670" from-port="2" to-layer="671" to-port="1" />
|
|
<edge from-layer="671" from-port="2" to-layer="673" to-port="0" />
|
|
<edge from-layer="672" from-port="0" to-layer="673" to-port="1" />
|
|
<edge from-layer="673" from-port="2" to-layer="678" to-port="0" />
|
|
<edge from-layer="674" from-port="0" to-layer="678" to-port="1" />
|
|
<edge from-layer="675" from-port="0" to-layer="678" to-port="2" />
|
|
<edge from-layer="676" from-port="0" to-layer="678" to-port="3" />
|
|
<edge from-layer="677" from-port="0" to-layer="678" to-port="4" />
|
|
<edge from-layer="678" from-port="5" to-layer="680" to-port="0" />
|
|
<edge from-layer="679" from-port="0" to-layer="680" to-port="1" />
|
|
<edge from-layer="680" from-port="2" to-layer="682" to-port="0" />
|
|
<edge from-layer="681" from-port="0" to-layer="682" to-port="1" />
|
|
<edge from-layer="682" from-port="2" to-layer="694" to-port="1" />
|
|
<edge from-layer="683" from-port="0" to-layer="684" to-port="0" />
|
|
<edge from-layer="684" from-port="1" to-layer="686" to-port="0" />
|
|
<edge from-layer="685" from-port="0" to-layer="686" to-port="1" />
|
|
<edge from-layer="686" from-port="2" to-layer="687" to-port="1" />
|
|
<edge from-layer="687" from-port="2" to-layer="689" to-port="0" />
|
|
<edge from-layer="688" from-port="0" to-layer="689" to-port="1" />
|
|
<edge from-layer="689" from-port="2" to-layer="691" to-port="0" />
|
|
<edge from-layer="690" from-port="0" to-layer="691" to-port="1" />
|
|
<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="694" to-port="2" />
|
|
<edge from-layer="694" from-port="4" to-layer="696" to-port="0" />
|
|
<edge from-layer="695" from-port="0" to-layer="696" to-port="1" />
|
|
<edge from-layer="696" from-port="2" to-layer="702" to-port="0" />
|
|
<edge from-layer="697" from-port="1" to-layer="700" to-port="0" />
|
|
<edge from-layer="698" from-port="0" to-layer="700" to-port="1" />
|
|
<edge from-layer="699" from-port="0" to-layer="700" to-port="2" />
|
|
<edge from-layer="700" from-port="3" to-layer="701" to-port="0" />
|
|
<edge from-layer="701" from-port="2" to-layer="702" to-port="1" />
|
|
<edge from-layer="702" from-port="2" to-layer="704" to-port="0" />
|
|
<edge from-layer="703" from-port="0" to-layer="704" to-port="1" />
|
|
<edge from-layer="704" from-port="2" to-layer="709" to-port="0" />
|
|
<edge from-layer="705" from-port="0" to-layer="709" to-port="1" />
|
|
<edge from-layer="706" from-port="0" to-layer="709" to-port="2" />
|
|
<edge from-layer="707" from-port="0" to-layer="709" to-port="3" />
|
|
<edge from-layer="708" from-port="0" to-layer="709" to-port="4" />
|
|
<edge from-layer="709" from-port="5" to-layer="714" to-port="0" />
|
|
<edge from-layer="710" from-port="0" to-layer="711" to-port="0" />
|
|
<edge from-layer="711" from-port="1" to-layer="713" to-port="0" />
|
|
<edge from-layer="712" from-port="0" to-layer="713" to-port="1" />
|
|
<edge from-layer="713" from-port="2" to-layer="714" to-port="1" />
|
|
<edge from-layer="714" from-port="2" to-layer="716" to-port="0" />
|
|
<edge from-layer="715" from-port="0" to-layer="716" to-port="1" />
|
|
<edge from-layer="716" from-port="2" to-layer="717" to-port="0" />
|
|
<edge from-layer="717" from-port="2" to-layer="719" to-port="0" />
|
|
<edge from-layer="718" from-port="0" to-layer="719" to-port="1" />
|
|
<edge from-layer="719" from-port="2" to-layer="721" to-port="0" />
|
|
<edge from-layer="720" from-port="0" to-layer="721" to-port="1" />
|
|
<edge from-layer="721" from-port="2" 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="753" to-port="1" />
|
|
<edge from-layer="723" from-port="2" to-layer="725" to-port="0" />
|
|
<edge from-layer="724" from-port="0" to-layer="725" to-port="1" />
|
|
<edge from-layer="725" from-port="2" to-layer="730" to-port="0" />
|
|
<edge from-layer="726" from-port="0" to-layer="730" to-port="1" />
|
|
<edge from-layer="727" from-port="0" to-layer="730" to-port="2" />
|
|
<edge from-layer="728" from-port="0" to-layer="730" to-port="3" />
|
|
<edge from-layer="729" from-port="0" to-layer="730" to-port="4" />
|
|
<edge from-layer="730" from-port="5" to-layer="735" to-port="0" />
|
|
<edge from-layer="731" from-port="0" to-layer="732" to-port="0" />
|
|
<edge from-layer="732" from-port="1" to-layer="734" to-port="0" />
|
|
<edge from-layer="733" from-port="0" to-layer="734" to-port="1" />
|
|
<edge from-layer="734" from-port="2" to-layer="735" to-port="1" />
|
|
<edge from-layer="735" from-port="2" to-layer="737" to-port="0" />
|
|
<edge from-layer="736" from-port="0" to-layer="737" to-port="1" />
|
|
<edge from-layer="737" from-port="2" to-layer="738" to-port="0" />
|
|
<edge from-layer="738" from-port="1" to-layer="740" to-port="0" />
|
|
<edge from-layer="739" from-port="0" to-layer="740" to-port="1" />
|
|
<edge from-layer="740" from-port="2" to-layer="745" to-port="0" />
|
|
<edge from-layer="741" from-port="0" to-layer="745" to-port="1" />
|
|
<edge from-layer="742" from-port="0" to-layer="745" to-port="2" />
|
|
<edge from-layer="743" from-port="0" to-layer="745" to-port="3" />
|
|
<edge from-layer="744" from-port="0" to-layer="745" to-port="4" />
|
|
<edge from-layer="745" from-port="5" to-layer="750" to-port="0" />
|
|
<edge from-layer="746" from-port="0" to-layer="747" to-port="0" />
|
|
<edge from-layer="747" from-port="1" 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="1" />
|
|
<edge from-layer="750" from-port="2" to-layer="752" to-port="0" />
|
|
<edge from-layer="751" from-port="0" to-layer="752" to-port="1" />
|
|
<edge from-layer="752" from-port="2" to-layer="753" to-port="0" />
|
|
<edge from-layer="753" from-port="2" to-layer="755" to-port="0" />
|
|
<edge from-layer="754" from-port="0" to-layer="755" to-port="1" />
|
|
<edge from-layer="755" from-port="2" to-layer="757" to-port="0" />
|
|
<edge from-layer="756" from-port="0" to-layer="757" to-port="1" />
|
|
<edge from-layer="757" from-port="2" to-layer="759" to-port="0" />
|
|
<edge from-layer="758" from-port="0" to-layer="759" to-port="1" />
|
|
<edge from-layer="759" from-port="2" to-layer="761" to-port="0" />
|
|
<edge from-layer="759" from-port="2" to-layer="813" to-port="0" />
|
|
<edge from-layer="759" from-port="2" to-layer="833" to-port="1" />
|
|
<edge from-layer="760" from-port="0" to-layer="761" to-port="1" />
|
|
<edge from-layer="761" from-port="2" to-layer="766" to-port="0" />
|
|
<edge from-layer="762" from-port="0" to-layer="766" to-port="1" />
|
|
<edge from-layer="763" from-port="0" to-layer="766" to-port="2" />
|
|
<edge from-layer="764" from-port="0" to-layer="766" to-port="3" />
|
|
<edge from-layer="765" from-port="0" to-layer="766" to-port="4" />
|
|
<edge from-layer="766" from-port="5" to-layer="771" to-port="0" />
|
|
<edge from-layer="766" from-port="5" to-layer="803" to-port="0" />
|
|
<edge from-layer="766" from-port="5" to-layer="787" to-port="0" />
|
|
<edge from-layer="767" from-port="0" to-layer="768" to-port="0" />
|
|
<edge from-layer="768" from-port="1" to-layer="770" to-port="0" />
|
|
<edge from-layer="769" from-port="0" to-layer="770" to-port="1" />
|
|
<edge from-layer="770" from-port="2" to-layer="771" to-port="1" />
|
|
<edge from-layer="771" from-port="2" to-layer="773" to-port="0" />
|
|
<edge from-layer="772" from-port="0" to-layer="773" to-port="1" />
|
|
<edge from-layer="773" from-port="2" to-layer="778" to-port="0" />
|
|
<edge from-layer="774" from-port="0" to-layer="778" to-port="1" />
|
|
<edge from-layer="775" from-port="0" to-layer="778" to-port="2" />
|
|
<edge from-layer="776" from-port="0" to-layer="778" to-port="3" />
|
|
<edge from-layer="777" from-port="0" to-layer="778" to-port="4" />
|
|
<edge from-layer="778" from-port="5" to-layer="780" to-port="0" />
|
|
<edge from-layer="779" from-port="0" to-layer="780" to-port="1" />
|
|
<edge from-layer="780" from-port="2" to-layer="782" to-port="0" />
|
|
<edge from-layer="781" from-port="0" to-layer="782" to-port="1" />
|
|
<edge from-layer="782" from-port="2" to-layer="810" to-port="0" />
|
|
<edge from-layer="783" from-port="0" to-layer="784" to-port="0" />
|
|
<edge from-layer="784" from-port="1" to-layer="786" to-port="0" />
|
|
<edge from-layer="785" from-port="0" to-layer="786" to-port="1" />
|
|
<edge from-layer="786" from-port="2" to-layer="787" to-port="1" />
|
|
<edge from-layer="787" from-port="2" to-layer="789" to-port="0" />
|
|
<edge from-layer="788" from-port="0" to-layer="789" to-port="1" />
|
|
<edge from-layer="789" from-port="2" to-layer="794" to-port="0" />
|
|
<edge from-layer="790" from-port="0" to-layer="794" to-port="1" />
|
|
<edge from-layer="791" from-port="0" to-layer="794" to-port="2" />
|
|
<edge from-layer="792" from-port="0" to-layer="794" to-port="3" />
|
|
<edge from-layer="793" from-port="0" to-layer="794" to-port="4" />
|
|
<edge from-layer="794" from-port="5" to-layer="796" to-port="0" />
|
|
<edge from-layer="795" from-port="0" to-layer="796" to-port="1" />
|
|
<edge from-layer="796" from-port="2" to-layer="798" to-port="0" />
|
|
<edge from-layer="797" from-port="0" to-layer="798" to-port="1" />
|
|
<edge from-layer="798" from-port="2" to-layer="810" to-port="1" />
|
|
<edge from-layer="799" from-port="0" to-layer="800" to-port="0" />
|
|
<edge from-layer="800" from-port="1" to-layer="802" to-port="0" />
|
|
<edge from-layer="801" from-port="0" to-layer="802" to-port="1" />
|
|
<edge from-layer="802" from-port="2" to-layer="803" to-port="1" />
|
|
<edge from-layer="803" from-port="2" to-layer="805" to-port="0" />
|
|
<edge from-layer="804" from-port="0" to-layer="805" to-port="1" />
|
|
<edge from-layer="805" from-port="2" to-layer="807" to-port="0" />
|
|
<edge from-layer="806" from-port="0" to-layer="807" to-port="1" />
|
|
<edge from-layer="807" from-port="2" to-layer="809" to-port="0" />
|
|
<edge from-layer="808" from-port="0" to-layer="809" to-port="1" />
|
|
<edge from-layer="809" from-port="2" to-layer="810" to-port="2" />
|
|
<edge from-layer="810" from-port="4" to-layer="812" to-port="0" />
|
|
<edge from-layer="811" from-port="0" to-layer="812" to-port="1" />
|
|
<edge from-layer="812" from-port="2" to-layer="818" to-port="0" />
|
|
<edge from-layer="813" from-port="1" to-layer="816" to-port="0" />
|
|
<edge from-layer="814" from-port="0" to-layer="816" to-port="1" />
|
|
<edge from-layer="815" from-port="0" to-layer="816" to-port="2" />
|
|
<edge from-layer="816" from-port="3" to-layer="817" to-port="0" />
|
|
<edge from-layer="817" from-port="2" to-layer="818" to-port="1" />
|
|
<edge from-layer="818" from-port="2" to-layer="820" to-port="0" />
|
|
<edge from-layer="819" from-port="0" to-layer="820" to-port="1" />
|
|
<edge from-layer="820" from-port="2" to-layer="825" to-port="0" />
|
|
<edge from-layer="821" from-port="0" to-layer="825" to-port="1" />
|
|
<edge from-layer="822" from-port="0" to-layer="825" to-port="2" />
|
|
<edge from-layer="823" from-port="0" to-layer="825" to-port="3" />
|
|
<edge from-layer="824" from-port="0" to-layer="825" to-port="4" />
|
|
<edge from-layer="825" from-port="5" to-layer="830" to-port="0" />
|
|
<edge from-layer="826" from-port="0" to-layer="827" to-port="0" />
|
|
<edge from-layer="827" from-port="1" to-layer="829" to-port="0" />
|
|
<edge from-layer="828" from-port="0" to-layer="829" to-port="1" />
|
|
<edge from-layer="829" from-port="2" to-layer="830" to-port="1" />
|
|
<edge from-layer="830" from-port="2" to-layer="832" to-port="0" />
|
|
<edge from-layer="831" from-port="0" to-layer="832" to-port="1" />
|
|
<edge from-layer="832" from-port="2" to-layer="833" to-port="0" />
|
|
<edge from-layer="833" from-port="2" to-layer="835" to-port="0" />
|
|
<edge from-layer="834" from-port="0" to-layer="835" to-port="1" />
|
|
<edge from-layer="835" from-port="2" to-layer="837" to-port="0" />
|
|
<edge from-layer="836" from-port="0" to-layer="837" to-port="1" />
|
|
<edge from-layer="837" from-port="2" to-layer="839" to-port="0" />
|
|
<edge from-layer="838" from-port="0" to-layer="839" to-port="1" />
|
|
<edge from-layer="839" from-port="2" to-layer="841" to-port="0" />
|
|
<edge from-layer="839" from-port="2" to-layer="869" to-port="1" />
|
|
<edge from-layer="840" from-port="0" to-layer="841" to-port="1" />
|
|
<edge from-layer="841" from-port="2" to-layer="846" to-port="0" />
|
|
<edge from-layer="842" from-port="0" to-layer="846" to-port="1" />
|
|
<edge from-layer="843" from-port="0" to-layer="846" to-port="2" />
|
|
<edge from-layer="844" from-port="0" to-layer="846" to-port="3" />
|
|
<edge from-layer="845" from-port="0" to-layer="846" to-port="4" />
|
|
<edge from-layer="846" from-port="5" to-layer="851" to-port="0" />
|
|
<edge from-layer="847" from-port="0" to-layer="848" to-port="0" />
|
|
<edge from-layer="848" from-port="1" to-layer="850" to-port="0" />
|
|
<edge from-layer="849" from-port="0" to-layer="850" to-port="1" />
|
|
<edge from-layer="850" from-port="2" to-layer="851" to-port="1" />
|
|
<edge from-layer="851" from-port="2" to-layer="853" to-port="0" />
|
|
<edge from-layer="852" from-port="0" to-layer="853" to-port="1" />
|
|
<edge from-layer="853" from-port="2" to-layer="854" to-port="0" />
|
|
<edge from-layer="854" from-port="1" to-layer="856" to-port="0" />
|
|
<edge from-layer="855" from-port="0" to-layer="856" to-port="1" />
|
|
<edge from-layer="856" from-port="2" to-layer="861" to-port="0" />
|
|
<edge from-layer="857" from-port="0" to-layer="861" to-port="1" />
|
|
<edge from-layer="858" from-port="0" to-layer="861" to-port="2" />
|
|
<edge from-layer="859" from-port="0" to-layer="861" to-port="3" />
|
|
<edge from-layer="860" from-port="0" to-layer="861" to-port="4" />
|
|
<edge from-layer="861" from-port="5" to-layer="866" to-port="0" />
|
|
<edge from-layer="862" from-port="0" to-layer="863" to-port="0" />
|
|
<edge from-layer="863" from-port="1" to-layer="865" to-port="0" />
|
|
<edge from-layer="864" from-port="0" to-layer="865" to-port="1" />
|
|
<edge from-layer="865" from-port="2" to-layer="866" to-port="1" />
|
|
<edge from-layer="866" from-port="2" to-layer="868" to-port="0" />
|
|
<edge from-layer="867" from-port="0" to-layer="868" to-port="1" />
|
|
<edge from-layer="868" from-port="2" to-layer="869" to-port="0" />
|
|
<edge from-layer="869" from-port="2" to-layer="871" to-port="0" />
|
|
<edge from-layer="870" from-port="0" to-layer="871" to-port="1" />
|
|
<edge from-layer="871" from-port="2" to-layer="873" to-port="0" />
|
|
<edge from-layer="872" from-port="0" to-layer="873" to-port="1" />
|
|
<edge from-layer="873" from-port="2" to-layer="875" to-port="0" />
|
|
<edge from-layer="874" from-port="0" to-layer="875" to-port="1" />
|
|
<edge from-layer="875" from-port="2" to-layer="877" to-port="0" />
|
|
<edge from-layer="875" from-port="2" to-layer="929" to-port="0" />
|
|
<edge from-layer="875" from-port="2" to-layer="949" to-port="1" />
|
|
<edge from-layer="876" from-port="0" to-layer="877" to-port="1" />
|
|
<edge from-layer="877" from-port="2" to-layer="882" to-port="0" />
|
|
<edge from-layer="878" from-port="0" to-layer="882" to-port="1" />
|
|
<edge from-layer="879" from-port="0" to-layer="882" to-port="2" />
|
|
<edge from-layer="880" from-port="0" to-layer="882" to-port="3" />
|
|
<edge from-layer="881" from-port="0" to-layer="882" to-port="4" />
|
|
<edge from-layer="882" from-port="5" to-layer="887" to-port="0" />
|
|
<edge from-layer="882" from-port="5" to-layer="903" to-port="0" />
|
|
<edge from-layer="882" from-port="5" to-layer="919" to-port="0" />
|
|
<edge from-layer="883" from-port="0" to-layer="884" to-port="0" />
|
|
<edge from-layer="884" from-port="1" to-layer="886" to-port="0" />
|
|
<edge from-layer="885" from-port="0" to-layer="886" to-port="1" />
|
|
<edge from-layer="886" from-port="2" to-layer="887" to-port="1" />
|
|
<edge from-layer="887" from-port="2" to-layer="889" to-port="0" />
|
|
<edge from-layer="888" from-port="0" to-layer="889" to-port="1" />
|
|
<edge from-layer="889" from-port="2" to-layer="894" to-port="0" />
|
|
<edge from-layer="890" from-port="0" to-layer="894" to-port="1" />
|
|
<edge from-layer="891" from-port="0" to-layer="894" to-port="2" />
|
|
<edge from-layer="892" from-port="0" to-layer="894" to-port="3" />
|
|
<edge from-layer="893" from-port="0" to-layer="894" to-port="4" />
|
|
<edge from-layer="894" from-port="5" to-layer="896" to-port="0" />
|
|
<edge from-layer="895" from-port="0" to-layer="896" to-port="1" />
|
|
<edge from-layer="896" from-port="2" to-layer="898" to-port="0" />
|
|
<edge from-layer="897" from-port="0" to-layer="898" to-port="1" />
|
|
<edge from-layer="898" from-port="2" to-layer="926" to-port="0" />
|
|
<edge from-layer="899" from-port="0" to-layer="900" to-port="0" />
|
|
<edge from-layer="900" from-port="1" to-layer="902" to-port="0" />
|
|
<edge from-layer="901" from-port="0" to-layer="902" to-port="1" />
|
|
<edge from-layer="902" from-port="2" to-layer="903" to-port="1" />
|
|
<edge from-layer="903" from-port="2" to-layer="905" to-port="0" />
|
|
<edge from-layer="904" from-port="0" to-layer="905" to-port="1" />
|
|
<edge from-layer="905" from-port="2" to-layer="910" to-port="0" />
|
|
<edge from-layer="906" from-port="0" to-layer="910" to-port="1" />
|
|
<edge from-layer="907" from-port="0" to-layer="910" to-port="2" />
|
|
<edge from-layer="908" from-port="0" to-layer="910" to-port="3" />
|
|
<edge from-layer="909" from-port="0" to-layer="910" to-port="4" />
|
|
<edge from-layer="910" from-port="5" to-layer="912" to-port="0" />
|
|
<edge from-layer="911" from-port="0" to-layer="912" to-port="1" />
|
|
<edge from-layer="912" from-port="2" to-layer="914" to-port="0" />
|
|
<edge from-layer="913" from-port="0" to-layer="914" to-port="1" />
|
|
<edge from-layer="914" from-port="2" to-layer="926" to-port="1" />
|
|
<edge from-layer="915" from-port="0" to-layer="916" to-port="0" />
|
|
<edge from-layer="916" from-port="1" to-layer="918" to-port="0" />
|
|
<edge from-layer="917" from-port="0" to-layer="918" to-port="1" />
|
|
<edge from-layer="918" from-port="2" to-layer="919" to-port="1" />
|
|
<edge from-layer="919" from-port="2" to-layer="921" to-port="0" />
|
|
<edge from-layer="920" from-port="0" to-layer="921" to-port="1" />
|
|
<edge from-layer="921" from-port="2" to-layer="923" to-port="0" />
|
|
<edge from-layer="922" from-port="0" to-layer="923" to-port="1" />
|
|
<edge from-layer="923" from-port="2" to-layer="925" to-port="0" />
|
|
<edge from-layer="924" from-port="0" to-layer="925" to-port="1" />
|
|
<edge from-layer="925" from-port="2" to-layer="926" to-port="2" />
|
|
<edge from-layer="926" from-port="4" to-layer="928" to-port="0" />
|
|
<edge from-layer="927" from-port="0" to-layer="928" to-port="1" />
|
|
<edge from-layer="928" from-port="2" to-layer="934" to-port="0" />
|
|
<edge from-layer="929" from-port="1" to-layer="932" to-port="0" />
|
|
<edge from-layer="930" from-port="0" to-layer="932" to-port="1" />
|
|
<edge from-layer="931" from-port="0" to-layer="932" to-port="2" />
|
|
<edge from-layer="932" from-port="3" to-layer="933" to-port="0" />
|
|
<edge from-layer="933" from-port="2" to-layer="934" to-port="1" />
|
|
<edge from-layer="934" from-port="2" to-layer="936" to-port="0" />
|
|
<edge from-layer="935" from-port="0" to-layer="936" to-port="1" />
|
|
<edge from-layer="936" from-port="2" to-layer="941" to-port="0" />
|
|
<edge from-layer="937" from-port="0" to-layer="941" to-port="1" />
|
|
<edge from-layer="938" from-port="0" to-layer="941" to-port="2" />
|
|
<edge from-layer="939" from-port="0" to-layer="941" to-port="3" />
|
|
<edge from-layer="940" from-port="0" to-layer="941" to-port="4" />
|
|
<edge from-layer="941" from-port="5" to-layer="946" to-port="0" />
|
|
<edge from-layer="942" from-port="0" to-layer="943" to-port="0" />
|
|
<edge from-layer="943" from-port="1" to-layer="945" to-port="0" />
|
|
<edge from-layer="944" from-port="0" to-layer="945" to-port="1" />
|
|
<edge from-layer="945" from-port="2" to-layer="946" to-port="1" />
|
|
<edge from-layer="946" from-port="2" to-layer="948" to-port="0" />
|
|
<edge from-layer="947" from-port="0" to-layer="948" to-port="1" />
|
|
<edge from-layer="948" from-port="2" to-layer="949" to-port="0" />
|
|
<edge from-layer="949" from-port="2" to-layer="951" to-port="0" />
|
|
<edge from-layer="950" from-port="0" to-layer="951" to-port="1" />
|
|
<edge from-layer="951" from-port="2" to-layer="953" to-port="0" />
|
|
<edge from-layer="952" from-port="0" to-layer="953" to-port="1" />
|
|
<edge from-layer="953" from-port="2" to-layer="955" to-port="0" />
|
|
<edge from-layer="954" from-port="0" to-layer="955" to-port="1" />
|
|
<edge from-layer="955" from-port="2" to-layer="957" to-port="0" />
|
|
<edge from-layer="955" from-port="2" to-layer="985" to-port="1" />
|
|
<edge from-layer="956" from-port="0" to-layer="957" to-port="1" />
|
|
<edge from-layer="957" from-port="2" to-layer="962" to-port="0" />
|
|
<edge from-layer="958" from-port="0" to-layer="962" to-port="1" />
|
|
<edge from-layer="959" from-port="0" to-layer="962" to-port="2" />
|
|
<edge from-layer="960" from-port="0" to-layer="962" to-port="3" />
|
|
<edge from-layer="961" from-port="0" to-layer="962" to-port="4" />
|
|
<edge from-layer="962" from-port="5" to-layer="967" to-port="0" />
|
|
<edge from-layer="963" from-port="0" to-layer="964" to-port="0" />
|
|
<edge from-layer="964" from-port="1" to-layer="966" to-port="0" />
|
|
<edge from-layer="965" from-port="0" to-layer="966" to-port="1" />
|
|
<edge from-layer="966" from-port="2" to-layer="967" to-port="1" />
|
|
<edge from-layer="967" from-port="2" to-layer="969" to-port="0" />
|
|
<edge from-layer="968" from-port="0" to-layer="969" to-port="1" />
|
|
<edge from-layer="969" from-port="2" to-layer="970" to-port="0" />
|
|
<edge from-layer="970" from-port="1" to-layer="972" to-port="0" />
|
|
<edge from-layer="971" from-port="0" to-layer="972" to-port="1" />
|
|
<edge from-layer="972" from-port="2" to-layer="977" to-port="0" />
|
|
<edge from-layer="973" from-port="0" to-layer="977" to-port="1" />
|
|
<edge from-layer="974" from-port="0" to-layer="977" to-port="2" />
|
|
<edge from-layer="975" from-port="0" to-layer="977" to-port="3" />
|
|
<edge from-layer="976" from-port="0" to-layer="977" to-port="4" />
|
|
<edge from-layer="977" from-port="5" to-layer="982" to-port="0" />
|
|
<edge from-layer="978" from-port="0" to-layer="979" to-port="0" />
|
|
<edge from-layer="979" from-port="1" to-layer="981" to-port="0" />
|
|
<edge from-layer="980" from-port="0" to-layer="981" to-port="1" />
|
|
<edge from-layer="981" from-port="2" to-layer="982" to-port="1" />
|
|
<edge from-layer="982" from-port="2" to-layer="984" to-port="0" />
|
|
<edge from-layer="983" from-port="0" to-layer="984" to-port="1" />
|
|
<edge from-layer="984" from-port="2" to-layer="985" to-port="0" />
|
|
<edge from-layer="985" from-port="2" to-layer="987" to-port="0" />
|
|
<edge from-layer="986" from-port="0" to-layer="987" to-port="1" />
|
|
<edge from-layer="987" from-port="2" to-layer="989" to-port="0" />
|
|
<edge from-layer="988" from-port="0" to-layer="989" to-port="1" />
|
|
<edge from-layer="989" from-port="2" to-layer="991" to-port="0" />
|
|
<edge from-layer="990" from-port="0" to-layer="991" to-port="1" />
|
|
<edge from-layer="991" from-port="2" to-layer="1045" to-port="0" />
|
|
<edge from-layer="991" from-port="2" to-layer="1065" to-port="1" />
|
|
<edge from-layer="991" from-port="2" to-layer="993" to-port="0" />
|
|
<edge from-layer="992" from-port="0" to-layer="993" to-port="1" />
|
|
<edge from-layer="993" from-port="2" to-layer="998" to-port="0" />
|
|
<edge from-layer="994" from-port="0" to-layer="998" to-port="1" />
|
|
<edge from-layer="995" from-port="0" to-layer="998" to-port="2" />
|
|
<edge from-layer="996" from-port="0" to-layer="998" to-port="3" />
|
|
<edge from-layer="997" from-port="0" to-layer="998" to-port="4" />
|
|
<edge from-layer="998" from-port="5" to-layer="1003" to-port="0" />
|
|
<edge from-layer="998" from-port="5" to-layer="1035" to-port="0" />
|
|
<edge from-layer="998" from-port="5" to-layer="1019" to-port="0" />
|
|
<edge from-layer="999" from-port="0" to-layer="1000" to-port="0" />
|
|
<edge from-layer="1000" from-port="1" to-layer="1002" to-port="0" />
|
|
<edge from-layer="1001" from-port="0" to-layer="1002" to-port="1" />
|
|
<edge from-layer="1002" from-port="2" to-layer="1003" to-port="1" />
|
|
<edge from-layer="1003" from-port="2" to-layer="1005" to-port="0" />
|
|
<edge from-layer="1004" from-port="0" to-layer="1005" to-port="1" />
|
|
<edge from-layer="1005" from-port="2" to-layer="1010" to-port="0" />
|
|
<edge from-layer="1006" from-port="0" to-layer="1010" to-port="1" />
|
|
<edge from-layer="1007" from-port="0" to-layer="1010" to-port="2" />
|
|
<edge from-layer="1008" from-port="0" to-layer="1010" to-port="3" />
|
|
<edge from-layer="1009" from-port="0" to-layer="1010" to-port="4" />
|
|
<edge from-layer="1010" from-port="5" to-layer="1012" to-port="0" />
|
|
<edge from-layer="1011" from-port="0" to-layer="1012" to-port="1" />
|
|
<edge from-layer="1012" from-port="2" to-layer="1014" to-port="0" />
|
|
<edge from-layer="1013" from-port="0" to-layer="1014" to-port="1" />
|
|
<edge from-layer="1014" from-port="2" to-layer="1042" to-port="0" />
|
|
<edge from-layer="1015" from-port="0" to-layer="1016" to-port="0" />
|
|
<edge from-layer="1016" from-port="1" to-layer="1018" to-port="0" />
|
|
<edge from-layer="1017" from-port="0" to-layer="1018" to-port="1" />
|
|
<edge from-layer="1018" from-port="2" to-layer="1019" to-port="1" />
|
|
<edge from-layer="1019" from-port="2" to-layer="1021" to-port="0" />
|
|
<edge from-layer="1020" from-port="0" to-layer="1021" to-port="1" />
|
|
<edge from-layer="1021" from-port="2" to-layer="1026" to-port="0" />
|
|
<edge from-layer="1022" from-port="0" to-layer="1026" to-port="1" />
|
|
<edge from-layer="1023" from-port="0" to-layer="1026" to-port="2" />
|
|
<edge from-layer="1024" from-port="0" to-layer="1026" to-port="3" />
|
|
<edge from-layer="1025" from-port="0" to-layer="1026" to-port="4" />
|
|
<edge from-layer="1026" from-port="5" to-layer="1028" to-port="0" />
|
|
<edge from-layer="1027" from-port="0" to-layer="1028" to-port="1" />
|
|
<edge from-layer="1028" from-port="2" to-layer="1030" to-port="0" />
|
|
<edge from-layer="1029" from-port="0" to-layer="1030" to-port="1" />
|
|
<edge from-layer="1030" from-port="2" to-layer="1042" to-port="1" />
|
|
<edge from-layer="1031" from-port="0" to-layer="1032" to-port="0" />
|
|
<edge from-layer="1032" from-port="1" to-layer="1034" to-port="0" />
|
|
<edge from-layer="1033" from-port="0" to-layer="1034" to-port="1" />
|
|
<edge from-layer="1034" from-port="2" to-layer="1035" to-port="1" />
|
|
<edge from-layer="1035" from-port="2" to-layer="1037" to-port="0" />
|
|
<edge from-layer="1036" from-port="0" to-layer="1037" to-port="1" />
|
|
<edge from-layer="1037" from-port="2" to-layer="1039" to-port="0" />
|
|
<edge from-layer="1038" from-port="0" to-layer="1039" to-port="1" />
|
|
<edge from-layer="1039" from-port="2" to-layer="1041" to-port="0" />
|
|
<edge from-layer="1040" from-port="0" to-layer="1041" to-port="1" />
|
|
<edge from-layer="1041" from-port="2" to-layer="1042" to-port="2" />
|
|
<edge from-layer="1042" from-port="4" to-layer="1044" to-port="0" />
|
|
<edge from-layer="1043" from-port="0" to-layer="1044" to-port="1" />
|
|
<edge from-layer="1044" from-port="2" to-layer="1050" to-port="0" />
|
|
<edge from-layer="1045" from-port="1" to-layer="1048" to-port="0" />
|
|
<edge from-layer="1046" from-port="0" to-layer="1048" to-port="1" />
|
|
<edge from-layer="1047" from-port="0" to-layer="1048" to-port="2" />
|
|
<edge from-layer="1048" from-port="3" to-layer="1049" to-port="0" />
|
|
<edge from-layer="1049" from-port="2" to-layer="1050" to-port="1" />
|
|
<edge from-layer="1050" from-port="2" to-layer="1052" to-port="0" />
|
|
<edge from-layer="1051" from-port="0" to-layer="1052" to-port="1" />
|
|
<edge from-layer="1052" from-port="2" to-layer="1057" to-port="0" />
|
|
<edge from-layer="1053" from-port="0" to-layer="1057" to-port="1" />
|
|
<edge from-layer="1054" from-port="0" to-layer="1057" to-port="2" />
|
|
<edge from-layer="1055" from-port="0" to-layer="1057" to-port="3" />
|
|
<edge from-layer="1056" from-port="0" to-layer="1057" to-port="4" />
|
|
<edge from-layer="1057" from-port="5" to-layer="1062" to-port="0" />
|
|
<edge from-layer="1058" from-port="0" to-layer="1059" to-port="0" />
|
|
<edge from-layer="1059" from-port="1" to-layer="1061" to-port="0" />
|
|
<edge from-layer="1060" from-port="0" to-layer="1061" to-port="1" />
|
|
<edge from-layer="1061" from-port="2" to-layer="1062" to-port="1" />
|
|
<edge from-layer="1062" from-port="2" to-layer="1064" to-port="0" />
|
|
<edge from-layer="1063" from-port="0" to-layer="1064" to-port="1" />
|
|
<edge from-layer="1064" from-port="2" to-layer="1065" to-port="0" />
|
|
<edge from-layer="1065" from-port="2" to-layer="1067" to-port="0" />
|
|
<edge from-layer="1066" from-port="0" to-layer="1067" to-port="1" />
|
|
<edge from-layer="1067" from-port="2" to-layer="1069" to-port="0" />
|
|
<edge from-layer="1068" from-port="0" to-layer="1069" to-port="1" />
|
|
<edge from-layer="1069" from-port="2" to-layer="1071" to-port="0" />
|
|
<edge from-layer="1070" from-port="0" to-layer="1071" to-port="1" />
|
|
<edge from-layer="1071" from-port="2" to-layer="1073" to-port="0" />
|
|
<edge from-layer="1071" from-port="2" to-layer="1101" to-port="1" />
|
|
<edge from-layer="1072" from-port="0" to-layer="1073" to-port="1" />
|
|
<edge from-layer="1073" from-port="2" to-layer="1078" to-port="0" />
|
|
<edge from-layer="1074" from-port="0" to-layer="1078" to-port="1" />
|
|
<edge from-layer="1075" from-port="0" to-layer="1078" to-port="2" />
|
|
<edge from-layer="1076" from-port="0" to-layer="1078" to-port="3" />
|
|
<edge from-layer="1077" from-port="0" to-layer="1078" to-port="4" />
|
|
<edge from-layer="1078" from-port="5" to-layer="1083" to-port="0" />
|
|
<edge from-layer="1079" from-port="0" to-layer="1080" to-port="0" />
|
|
<edge from-layer="1080" from-port="1" to-layer="1082" to-port="0" />
|
|
<edge from-layer="1081" from-port="0" to-layer="1082" to-port="1" />
|
|
<edge from-layer="1082" from-port="2" to-layer="1083" to-port="1" />
|
|
<edge from-layer="1083" from-port="2" to-layer="1085" to-port="0" />
|
|
<edge from-layer="1084" from-port="0" to-layer="1085" to-port="1" />
|
|
<edge from-layer="1085" from-port="2" to-layer="1086" to-port="0" />
|
|
<edge from-layer="1086" from-port="1" to-layer="1088" to-port="0" />
|
|
<edge from-layer="1087" from-port="0" to-layer="1088" to-port="1" />
|
|
<edge from-layer="1088" from-port="2" to-layer="1093" to-port="0" />
|
|
<edge from-layer="1089" from-port="0" to-layer="1093" to-port="1" />
|
|
<edge from-layer="1090" from-port="0" to-layer="1093" to-port="2" />
|
|
<edge from-layer="1091" from-port="0" to-layer="1093" to-port="3" />
|
|
<edge from-layer="1092" from-port="0" to-layer="1093" to-port="4" />
|
|
<edge from-layer="1093" from-port="5" to-layer="1098" to-port="0" />
|
|
<edge from-layer="1094" from-port="0" to-layer="1095" to-port="0" />
|
|
<edge from-layer="1095" from-port="1" to-layer="1097" to-port="0" />
|
|
<edge from-layer="1096" from-port="0" to-layer="1097" to-port="1" />
|
|
<edge from-layer="1097" from-port="2" to-layer="1098" to-port="1" />
|
|
<edge from-layer="1098" from-port="2" to-layer="1100" to-port="0" />
|
|
<edge from-layer="1099" from-port="0" to-layer="1100" to-port="1" />
|
|
<edge from-layer="1100" from-port="2" to-layer="1101" to-port="0" />
|
|
<edge from-layer="1101" from-port="2" to-layer="1103" to-port="0" />
|
|
<edge from-layer="1102" from-port="0" to-layer="1103" to-port="1" />
|
|
<edge from-layer="1103" from-port="2" to-layer="1105" to-port="0" />
|
|
<edge from-layer="1104" from-port="0" to-layer="1105" to-port="1" />
|
|
<edge from-layer="1105" from-port="2" to-layer="1107" to-port="0" />
|
|
<edge from-layer="1106" from-port="0" to-layer="1107" to-port="1" />
|
|
<edge from-layer="1107" from-port="2" to-layer="1109" to-port="0" />
|
|
<edge from-layer="1107" from-port="2" to-layer="1161" to-port="0" />
|
|
<edge from-layer="1107" from-port="2" to-layer="1181" to-port="1" />
|
|
<edge from-layer="1108" from-port="0" to-layer="1109" to-port="1" />
|
|
<edge from-layer="1109" from-port="2" to-layer="1114" to-port="0" />
|
|
<edge from-layer="1110" from-port="0" to-layer="1114" to-port="1" />
|
|
<edge from-layer="1111" from-port="0" to-layer="1114" to-port="2" />
|
|
<edge from-layer="1112" from-port="0" to-layer="1114" to-port="3" />
|
|
<edge from-layer="1113" from-port="0" to-layer="1114" to-port="4" />
|
|
<edge from-layer="1114" from-port="5" to-layer="1119" to-port="0" />
|
|
<edge from-layer="1114" from-port="5" to-layer="1135" to-port="0" />
|
|
<edge from-layer="1114" from-port="5" to-layer="1151" to-port="0" />
|
|
<edge from-layer="1115" from-port="0" to-layer="1116" to-port="0" />
|
|
<edge from-layer="1116" from-port="1" to-layer="1118" to-port="0" />
|
|
<edge from-layer="1117" from-port="0" to-layer="1118" to-port="1" />
|
|
<edge from-layer="1118" from-port="2" to-layer="1119" to-port="1" />
|
|
<edge from-layer="1119" from-port="2" to-layer="1121" to-port="0" />
|
|
<edge from-layer="1120" from-port="0" to-layer="1121" to-port="1" />
|
|
<edge from-layer="1121" from-port="2" to-layer="1126" to-port="0" />
|
|
<edge from-layer="1122" from-port="0" to-layer="1126" to-port="1" />
|
|
<edge from-layer="1123" from-port="0" to-layer="1126" to-port="2" />
|
|
<edge from-layer="1124" from-port="0" to-layer="1126" to-port="3" />
|
|
<edge from-layer="1125" from-port="0" to-layer="1126" to-port="4" />
|
|
<edge from-layer="1126" from-port="5" to-layer="1128" to-port="0" />
|
|
<edge from-layer="1127" from-port="0" to-layer="1128" to-port="1" />
|
|
<edge from-layer="1128" from-port="2" to-layer="1130" to-port="0" />
|
|
<edge from-layer="1129" from-port="0" to-layer="1130" to-port="1" />
|
|
<edge from-layer="1130" from-port="2" to-layer="1158" to-port="0" />
|
|
<edge from-layer="1131" from-port="0" to-layer="1132" to-port="0" />
|
|
<edge from-layer="1132" from-port="1" to-layer="1134" to-port="0" />
|
|
<edge from-layer="1133" from-port="0" to-layer="1134" to-port="1" />
|
|
<edge from-layer="1134" from-port="2" to-layer="1135" to-port="1" />
|
|
<edge from-layer="1135" from-port="2" to-layer="1137" to-port="0" />
|
|
<edge from-layer="1136" from-port="0" to-layer="1137" to-port="1" />
|
|
<edge from-layer="1137" from-port="2" to-layer="1142" to-port="0" />
|
|
<edge from-layer="1138" from-port="0" to-layer="1142" to-port="1" />
|
|
<edge from-layer="1139" from-port="0" to-layer="1142" to-port="2" />
|
|
<edge from-layer="1140" from-port="0" to-layer="1142" to-port="3" />
|
|
<edge from-layer="1141" from-port="0" to-layer="1142" to-port="4" />
|
|
<edge from-layer="1142" from-port="5" to-layer="1144" to-port="0" />
|
|
<edge from-layer="1143" from-port="0" to-layer="1144" to-port="1" />
|
|
<edge from-layer="1144" from-port="2" to-layer="1146" to-port="0" />
|
|
<edge from-layer="1145" from-port="0" to-layer="1146" to-port="1" />
|
|
<edge from-layer="1146" from-port="2" to-layer="1158" to-port="1" />
|
|
<edge from-layer="1147" from-port="0" to-layer="1148" to-port="0" />
|
|
<edge from-layer="1148" from-port="1" to-layer="1150" to-port="0" />
|
|
<edge from-layer="1149" from-port="0" to-layer="1150" to-port="1" />
|
|
<edge from-layer="1150" from-port="2" to-layer="1151" to-port="1" />
|
|
<edge from-layer="1151" from-port="2" to-layer="1153" to-port="0" />
|
|
<edge from-layer="1152" from-port="0" to-layer="1153" to-port="1" />
|
|
<edge from-layer="1153" from-port="2" to-layer="1155" to-port="0" />
|
|
<edge from-layer="1154" from-port="0" to-layer="1155" to-port="1" />
|
|
<edge from-layer="1155" from-port="2" to-layer="1157" to-port="0" />
|
|
<edge from-layer="1156" from-port="0" to-layer="1157" to-port="1" />
|
|
<edge from-layer="1157" from-port="2" to-layer="1158" to-port="2" />
|
|
<edge from-layer="1158" from-port="4" to-layer="1160" to-port="0" />
|
|
<edge from-layer="1159" from-port="0" to-layer="1160" to-port="1" />
|
|
<edge from-layer="1160" from-port="2" to-layer="1166" to-port="0" />
|
|
<edge from-layer="1161" from-port="1" to-layer="1164" to-port="0" />
|
|
<edge from-layer="1162" from-port="0" to-layer="1164" to-port="1" />
|
|
<edge from-layer="1163" from-port="0" to-layer="1164" to-port="2" />
|
|
<edge from-layer="1164" from-port="3" to-layer="1165" to-port="0" />
|
|
<edge from-layer="1165" from-port="2" to-layer="1166" to-port="1" />
|
|
<edge from-layer="1166" from-port="2" to-layer="1168" to-port="0" />
|
|
<edge from-layer="1167" from-port="0" to-layer="1168" to-port="1" />
|
|
<edge from-layer="1168" from-port="2" to-layer="1173" to-port="0" />
|
|
<edge from-layer="1169" from-port="0" to-layer="1173" to-port="1" />
|
|
<edge from-layer="1170" from-port="0" to-layer="1173" to-port="2" />
|
|
<edge from-layer="1171" from-port="0" to-layer="1173" to-port="3" />
|
|
<edge from-layer="1172" from-port="0" to-layer="1173" to-port="4" />
|
|
<edge from-layer="1173" from-port="5" to-layer="1178" to-port="0" />
|
|
<edge from-layer="1174" from-port="0" to-layer="1175" to-port="0" />
|
|
<edge from-layer="1175" from-port="1" to-layer="1177" to-port="0" />
|
|
<edge from-layer="1176" from-port="0" to-layer="1177" to-port="1" />
|
|
<edge from-layer="1177" from-port="2" to-layer="1178" to-port="1" />
|
|
<edge from-layer="1178" from-port="2" to-layer="1180" to-port="0" />
|
|
<edge from-layer="1179" from-port="0" to-layer="1180" to-port="1" />
|
|
<edge from-layer="1180" from-port="2" to-layer="1181" to-port="0" />
|
|
<edge from-layer="1181" from-port="2" to-layer="1183" to-port="0" />
|
|
<edge from-layer="1182" from-port="0" to-layer="1183" to-port="1" />
|
|
<edge from-layer="1183" from-port="2" to-layer="1185" to-port="0" />
|
|
<edge from-layer="1184" from-port="0" to-layer="1185" to-port="1" />
|
|
<edge from-layer="1185" from-port="2" to-layer="1187" to-port="0" />
|
|
<edge from-layer="1186" from-port="0" to-layer="1187" to-port="1" />
|
|
<edge from-layer="1187" from-port="2" to-layer="1189" to-port="0" />
|
|
<edge from-layer="1187" from-port="2" to-layer="1217" to-port="1" />
|
|
<edge from-layer="1188" from-port="0" to-layer="1189" to-port="1" />
|
|
<edge from-layer="1189" from-port="2" to-layer="1194" to-port="0" />
|
|
<edge from-layer="1190" from-port="0" to-layer="1194" to-port="1" />
|
|
<edge from-layer="1191" from-port="0" to-layer="1194" to-port="2" />
|
|
<edge from-layer="1192" from-port="0" to-layer="1194" to-port="3" />
|
|
<edge from-layer="1193" from-port="0" to-layer="1194" to-port="4" />
|
|
<edge from-layer="1194" from-port="5" to-layer="1199" to-port="0" />
|
|
<edge from-layer="1195" from-port="0" to-layer="1196" to-port="0" />
|
|
<edge from-layer="1196" from-port="1" to-layer="1198" to-port="0" />
|
|
<edge from-layer="1197" from-port="0" to-layer="1198" to-port="1" />
|
|
<edge from-layer="1198" from-port="2" to-layer="1199" to-port="1" />
|
|
<edge from-layer="1199" from-port="2" to-layer="1201" to-port="0" />
|
|
<edge from-layer="1200" from-port="0" to-layer="1201" to-port="1" />
|
|
<edge from-layer="1201" from-port="2" to-layer="1202" to-port="0" />
|
|
<edge from-layer="1202" from-port="1" to-layer="1204" to-port="0" />
|
|
<edge from-layer="1203" from-port="0" to-layer="1204" to-port="1" />
|
|
<edge from-layer="1204" from-port="2" to-layer="1209" to-port="0" />
|
|
<edge from-layer="1205" from-port="0" to-layer="1209" to-port="1" />
|
|
<edge from-layer="1206" from-port="0" to-layer="1209" to-port="2" />
|
|
<edge from-layer="1207" from-port="0" to-layer="1209" to-port="3" />
|
|
<edge from-layer="1208" from-port="0" to-layer="1209" to-port="4" />
|
|
<edge from-layer="1209" from-port="5" to-layer="1214" to-port="0" />
|
|
<edge from-layer="1210" from-port="0" to-layer="1211" to-port="0" />
|
|
<edge from-layer="1211" from-port="1" to-layer="1213" to-port="0" />
|
|
<edge from-layer="1212" from-port="0" to-layer="1213" to-port="1" />
|
|
<edge from-layer="1213" from-port="2" to-layer="1214" to-port="1" />
|
|
<edge from-layer="1214" from-port="2" to-layer="1216" to-port="0" />
|
|
<edge from-layer="1215" from-port="0" to-layer="1216" to-port="1" />
|
|
<edge from-layer="1216" from-port="2" to-layer="1217" to-port="0" />
|
|
<edge from-layer="1217" from-port="2" to-layer="1219" to-port="0" />
|
|
<edge from-layer="1218" from-port="0" to-layer="1219" to-port="1" />
|
|
<edge from-layer="1219" from-port="2" to-layer="1221" to-port="0" />
|
|
<edge from-layer="1220" from-port="0" to-layer="1221" to-port="1" />
|
|
<edge from-layer="1221" from-port="2" to-layer="1223" to-port="0" />
|
|
<edge from-layer="1222" from-port="0" to-layer="1223" to-port="1" />
|
|
<edge from-layer="1223" from-port="2" to-layer="1225" to-port="0" />
|
|
<edge from-layer="1223" from-port="2" to-layer="1297" to-port="1" />
|
|
<edge from-layer="1223" from-port="2" to-layer="1277" to-port="0" />
|
|
<edge from-layer="1224" from-port="0" to-layer="1225" to-port="1" />
|
|
<edge from-layer="1225" from-port="2" to-layer="1230" to-port="0" />
|
|
<edge from-layer="1226" from-port="0" to-layer="1230" to-port="1" />
|
|
<edge from-layer="1227" from-port="0" to-layer="1230" to-port="2" />
|
|
<edge from-layer="1228" from-port="0" to-layer="1230" to-port="3" />
|
|
<edge from-layer="1229" from-port="0" to-layer="1230" to-port="4" />
|
|
<edge from-layer="1230" from-port="5" to-layer="1235" to-port="0" />
|
|
<edge from-layer="1230" from-port="5" to-layer="1267" to-port="0" />
|
|
<edge from-layer="1230" from-port="5" to-layer="1251" to-port="0" />
|
|
<edge from-layer="1231" from-port="0" to-layer="1232" to-port="0" />
|
|
<edge from-layer="1232" from-port="1" to-layer="1234" to-port="0" />
|
|
<edge from-layer="1233" from-port="0" to-layer="1234" to-port="1" />
|
|
<edge from-layer="1234" from-port="2" to-layer="1235" to-port="1" />
|
|
<edge from-layer="1235" from-port="2" to-layer="1237" to-port="0" />
|
|
<edge from-layer="1236" from-port="0" to-layer="1237" to-port="1" />
|
|
<edge from-layer="1237" from-port="2" to-layer="1242" to-port="0" />
|
|
<edge from-layer="1238" from-port="0" to-layer="1242" to-port="1" />
|
|
<edge from-layer="1239" from-port="0" to-layer="1242" to-port="2" />
|
|
<edge from-layer="1240" from-port="0" to-layer="1242" to-port="3" />
|
|
<edge from-layer="1241" from-port="0" to-layer="1242" to-port="4" />
|
|
<edge from-layer="1242" from-port="5" to-layer="1244" to-port="0" />
|
|
<edge from-layer="1243" from-port="0" to-layer="1244" to-port="1" />
|
|
<edge from-layer="1244" from-port="2" to-layer="1246" to-port="0" />
|
|
<edge from-layer="1245" from-port="0" to-layer="1246" to-port="1" />
|
|
<edge from-layer="1246" from-port="2" to-layer="1274" to-port="0" />
|
|
<edge from-layer="1247" from-port="0" to-layer="1248" to-port="0" />
|
|
<edge from-layer="1248" from-port="1" to-layer="1250" to-port="0" />
|
|
<edge from-layer="1249" from-port="0" to-layer="1250" to-port="1" />
|
|
<edge from-layer="1250" from-port="2" to-layer="1251" to-port="1" />
|
|
<edge from-layer="1251" from-port="2" to-layer="1253" to-port="0" />
|
|
<edge from-layer="1252" from-port="0" to-layer="1253" to-port="1" />
|
|
<edge from-layer="1253" from-port="2" to-layer="1258" to-port="0" />
|
|
<edge from-layer="1254" from-port="0" to-layer="1258" to-port="1" />
|
|
<edge from-layer="1255" from-port="0" to-layer="1258" to-port="2" />
|
|
<edge from-layer="1256" from-port="0" to-layer="1258" to-port="3" />
|
|
<edge from-layer="1257" from-port="0" to-layer="1258" to-port="4" />
|
|
<edge from-layer="1258" from-port="5" to-layer="1260" to-port="0" />
|
|
<edge from-layer="1259" from-port="0" to-layer="1260" to-port="1" />
|
|
<edge from-layer="1260" from-port="2" to-layer="1262" to-port="0" />
|
|
<edge from-layer="1261" from-port="0" to-layer="1262" to-port="1" />
|
|
<edge from-layer="1262" from-port="2" to-layer="1274" to-port="1" />
|
|
<edge from-layer="1263" from-port="0" to-layer="1264" to-port="0" />
|
|
<edge from-layer="1264" from-port="1" to-layer="1266" to-port="0" />
|
|
<edge from-layer="1265" from-port="0" to-layer="1266" to-port="1" />
|
|
<edge from-layer="1266" from-port="2" to-layer="1267" to-port="1" />
|
|
<edge from-layer="1267" from-port="2" to-layer="1269" to-port="0" />
|
|
<edge from-layer="1268" from-port="0" to-layer="1269" to-port="1" />
|
|
<edge from-layer="1269" from-port="2" to-layer="1271" to-port="0" />
|
|
<edge from-layer="1270" from-port="0" to-layer="1271" to-port="1" />
|
|
<edge from-layer="1271" from-port="2" to-layer="1273" to-port="0" />
|
|
<edge from-layer="1272" from-port="0" to-layer="1273" to-port="1" />
|
|
<edge from-layer="1273" from-port="2" to-layer="1274" to-port="2" />
|
|
<edge from-layer="1274" from-port="4" to-layer="1276" to-port="0" />
|
|
<edge from-layer="1275" from-port="0" to-layer="1276" to-port="1" />
|
|
<edge from-layer="1276" from-port="2" to-layer="1282" to-port="0" />
|
|
<edge from-layer="1277" from-port="1" to-layer="1280" to-port="0" />
|
|
<edge from-layer="1278" from-port="0" to-layer="1280" to-port="1" />
|
|
<edge from-layer="1279" from-port="0" to-layer="1280" to-port="2" />
|
|
<edge from-layer="1280" from-port="3" to-layer="1281" to-port="0" />
|
|
<edge from-layer="1281" from-port="2" to-layer="1282" to-port="1" />
|
|
<edge from-layer="1282" from-port="2" to-layer="1284" to-port="0" />
|
|
<edge from-layer="1283" from-port="0" to-layer="1284" to-port="1" />
|
|
<edge from-layer="1284" from-port="2" to-layer="1289" to-port="0" />
|
|
<edge from-layer="1285" from-port="0" to-layer="1289" to-port="1" />
|
|
<edge from-layer="1286" from-port="0" to-layer="1289" to-port="2" />
|
|
<edge from-layer="1287" from-port="0" to-layer="1289" to-port="3" />
|
|
<edge from-layer="1288" from-port="0" to-layer="1289" to-port="4" />
|
|
<edge from-layer="1289" from-port="5" to-layer="1294" to-port="0" />
|
|
<edge from-layer="1290" from-port="0" to-layer="1291" to-port="0" />
|
|
<edge from-layer="1291" from-port="1" to-layer="1293" to-port="0" />
|
|
<edge from-layer="1292" from-port="0" to-layer="1293" to-port="1" />
|
|
<edge from-layer="1293" from-port="2" to-layer="1294" to-port="1" />
|
|
<edge from-layer="1294" from-port="2" to-layer="1296" to-port="0" />
|
|
<edge from-layer="1295" from-port="0" to-layer="1296" to-port="1" />
|
|
<edge from-layer="1296" from-port="2" to-layer="1297" to-port="0" />
|
|
<edge from-layer="1297" from-port="2" to-layer="1299" to-port="0" />
|
|
<edge from-layer="1298" from-port="0" to-layer="1299" to-port="1" />
|
|
<edge from-layer="1299" from-port="2" to-layer="1301" to-port="0" />
|
|
<edge from-layer="1300" from-port="0" to-layer="1301" to-port="1" />
|
|
<edge from-layer="1301" from-port="2" to-layer="1303" to-port="0" />
|
|
<edge from-layer="1302" from-port="0" to-layer="1303" to-port="1" />
|
|
<edge from-layer="1303" from-port="2" to-layer="1305" to-port="0" />
|
|
<edge from-layer="1303" from-port="2" to-layer="1333" to-port="1" />
|
|
<edge from-layer="1304" from-port="0" to-layer="1305" to-port="1" />
|
|
<edge from-layer="1305" from-port="2" to-layer="1310" to-port="0" />
|
|
<edge from-layer="1306" from-port="0" to-layer="1310" to-port="1" />
|
|
<edge from-layer="1307" from-port="0" to-layer="1310" to-port="2" />
|
|
<edge from-layer="1308" from-port="0" to-layer="1310" to-port="3" />
|
|
<edge from-layer="1309" from-port="0" to-layer="1310" to-port="4" />
|
|
<edge from-layer="1310" from-port="5" to-layer="1315" to-port="0" />
|
|
<edge from-layer="1311" from-port="0" to-layer="1312" to-port="0" />
|
|
<edge from-layer="1312" from-port="1" to-layer="1314" to-port="0" />
|
|
<edge from-layer="1313" from-port="0" to-layer="1314" to-port="1" />
|
|
<edge from-layer="1314" from-port="2" to-layer="1315" to-port="1" />
|
|
<edge from-layer="1315" from-port="2" to-layer="1317" to-port="0" />
|
|
<edge from-layer="1316" from-port="0" to-layer="1317" to-port="1" />
|
|
<edge from-layer="1317" from-port="2" to-layer="1318" to-port="0" />
|
|
<edge from-layer="1318" from-port="1" to-layer="1320" to-port="0" />
|
|
<edge from-layer="1319" from-port="0" to-layer="1320" to-port="1" />
|
|
<edge from-layer="1320" from-port="2" to-layer="1325" to-port="0" />
|
|
<edge from-layer="1321" from-port="0" to-layer="1325" to-port="1" />
|
|
<edge from-layer="1322" from-port="0" to-layer="1325" to-port="2" />
|
|
<edge from-layer="1323" from-port="0" to-layer="1325" to-port="3" />
|
|
<edge from-layer="1324" from-port="0" to-layer="1325" to-port="4" />
|
|
<edge from-layer="1325" from-port="5" to-layer="1330" to-port="0" />
|
|
<edge from-layer="1326" from-port="0" to-layer="1327" to-port="0" />
|
|
<edge from-layer="1327" from-port="1" to-layer="1329" to-port="0" />
|
|
<edge from-layer="1328" from-port="0" to-layer="1329" to-port="1" />
|
|
<edge from-layer="1329" from-port="2" to-layer="1330" to-port="1" />
|
|
<edge from-layer="1330" from-port="2" to-layer="1332" to-port="0" />
|
|
<edge from-layer="1331" from-port="0" to-layer="1332" to-port="1" />
|
|
<edge from-layer="1332" from-port="2" to-layer="1333" to-port="0" />
|
|
<edge from-layer="1333" from-port="2" to-layer="1335" to-port="0" />
|
|
<edge from-layer="1334" from-port="0" to-layer="1335" to-port="1" />
|
|
<edge from-layer="1335" from-port="2" to-layer="1337" to-port="0" />
|
|
<edge from-layer="1336" from-port="0" to-layer="1337" to-port="1" />
|
|
<edge from-layer="1337" from-port="2" to-layer="1339" to-port="0" />
|
|
<edge from-layer="1338" from-port="0" to-layer="1339" to-port="1" />
|
|
<edge from-layer="1339" from-port="2" to-layer="1341" to-port="0" />
|
|
<edge from-layer="1339" from-port="2" to-layer="1413" to-port="1" />
|
|
<edge from-layer="1339" from-port="2" to-layer="1393" to-port="0" />
|
|
<edge from-layer="1340" from-port="0" to-layer="1341" to-port="1" />
|
|
<edge from-layer="1341" from-port="2" to-layer="1346" to-port="0" />
|
|
<edge from-layer="1342" from-port="0" to-layer="1346" to-port="1" />
|
|
<edge from-layer="1343" from-port="0" to-layer="1346" to-port="2" />
|
|
<edge from-layer="1344" from-port="0" to-layer="1346" to-port="3" />
|
|
<edge from-layer="1345" from-port="0" to-layer="1346" to-port="4" />
|
|
<edge from-layer="1346" from-port="5" to-layer="1367" to-port="0" />
|
|
<edge from-layer="1346" from-port="5" to-layer="1383" to-port="0" />
|
|
<edge from-layer="1346" from-port="5" to-layer="1351" to-port="0" />
|
|
<edge from-layer="1347" from-port="0" to-layer="1348" to-port="0" />
|
|
<edge from-layer="1348" from-port="1" to-layer="1350" to-port="0" />
|
|
<edge from-layer="1349" from-port="0" to-layer="1350" to-port="1" />
|
|
<edge from-layer="1350" from-port="2" to-layer="1351" to-port="1" />
|
|
<edge from-layer="1351" from-port="2" to-layer="1353" to-port="0" />
|
|
<edge from-layer="1352" from-port="0" to-layer="1353" to-port="1" />
|
|
<edge from-layer="1353" from-port="2" to-layer="1358" to-port="0" />
|
|
<edge from-layer="1354" from-port="0" to-layer="1358" to-port="1" />
|
|
<edge from-layer="1355" from-port="0" to-layer="1358" to-port="2" />
|
|
<edge from-layer="1356" from-port="0" to-layer="1358" to-port="3" />
|
|
<edge from-layer="1357" from-port="0" to-layer="1358" to-port="4" />
|
|
<edge from-layer="1358" from-port="5" to-layer="1360" to-port="0" />
|
|
<edge from-layer="1359" from-port="0" to-layer="1360" to-port="1" />
|
|
<edge from-layer="1360" from-port="2" to-layer="1362" to-port="0" />
|
|
<edge from-layer="1361" from-port="0" to-layer="1362" to-port="1" />
|
|
<edge from-layer="1362" from-port="2" to-layer="1390" to-port="0" />
|
|
<edge from-layer="1363" from-port="0" to-layer="1364" to-port="0" />
|
|
<edge from-layer="1364" from-port="1" to-layer="1366" to-port="0" />
|
|
<edge from-layer="1365" from-port="0" to-layer="1366" to-port="1" />
|
|
<edge from-layer="1366" from-port="2" to-layer="1367" to-port="1" />
|
|
<edge from-layer="1367" from-port="2" to-layer="1369" to-port="0" />
|
|
<edge from-layer="1368" from-port="0" to-layer="1369" to-port="1" />
|
|
<edge from-layer="1369" from-port="2" to-layer="1374" to-port="0" />
|
|
<edge from-layer="1370" from-port="0" to-layer="1374" to-port="1" />
|
|
<edge from-layer="1371" from-port="0" to-layer="1374" to-port="2" />
|
|
<edge from-layer="1372" from-port="0" to-layer="1374" to-port="3" />
|
|
<edge from-layer="1373" from-port="0" to-layer="1374" to-port="4" />
|
|
<edge from-layer="1374" from-port="5" to-layer="1376" to-port="0" />
|
|
<edge from-layer="1375" from-port="0" to-layer="1376" to-port="1" />
|
|
<edge from-layer="1376" from-port="2" to-layer="1378" to-port="0" />
|
|
<edge from-layer="1377" from-port="0" to-layer="1378" to-port="1" />
|
|
<edge from-layer="1378" from-port="2" to-layer="1390" to-port="1" />
|
|
<edge from-layer="1379" from-port="0" to-layer="1380" to-port="0" />
|
|
<edge from-layer="1380" from-port="1" to-layer="1382" to-port="0" />
|
|
<edge from-layer="1381" from-port="0" to-layer="1382" to-port="1" />
|
|
<edge from-layer="1382" from-port="2" to-layer="1383" to-port="1" />
|
|
<edge from-layer="1383" from-port="2" to-layer="1385" to-port="0" />
|
|
<edge from-layer="1384" from-port="0" to-layer="1385" to-port="1" />
|
|
<edge from-layer="1385" from-port="2" to-layer="1387" to-port="0" />
|
|
<edge from-layer="1386" from-port="0" to-layer="1387" to-port="1" />
|
|
<edge from-layer="1387" from-port="2" to-layer="1389" to-port="0" />
|
|
<edge from-layer="1388" from-port="0" to-layer="1389" to-port="1" />
|
|
<edge from-layer="1389" from-port="2" to-layer="1390" to-port="2" />
|
|
<edge from-layer="1390" from-port="4" to-layer="1392" to-port="0" />
|
|
<edge from-layer="1391" from-port="0" to-layer="1392" to-port="1" />
|
|
<edge from-layer="1392" from-port="2" to-layer="1398" to-port="0" />
|
|
<edge from-layer="1393" from-port="1" to-layer="1396" to-port="0" />
|
|
<edge from-layer="1394" from-port="0" to-layer="1396" to-port="1" />
|
|
<edge from-layer="1395" from-port="0" to-layer="1396" to-port="2" />
|
|
<edge from-layer="1396" from-port="3" to-layer="1397" to-port="0" />
|
|
<edge from-layer="1397" from-port="2" to-layer="1398" to-port="1" />
|
|
<edge from-layer="1398" from-port="2" to-layer="1400" to-port="0" />
|
|
<edge from-layer="1399" from-port="0" to-layer="1400" to-port="1" />
|
|
<edge from-layer="1400" from-port="2" to-layer="1405" to-port="0" />
|
|
<edge from-layer="1401" from-port="0" to-layer="1405" to-port="1" />
|
|
<edge from-layer="1402" from-port="0" to-layer="1405" to-port="2" />
|
|
<edge from-layer="1403" from-port="0" to-layer="1405" to-port="3" />
|
|
<edge from-layer="1404" from-port="0" to-layer="1405" to-port="4" />
|
|
<edge from-layer="1405" from-port="5" to-layer="1410" to-port="0" />
|
|
<edge from-layer="1406" from-port="0" to-layer="1407" to-port="0" />
|
|
<edge from-layer="1407" from-port="1" to-layer="1409" to-port="0" />
|
|
<edge from-layer="1408" from-port="0" to-layer="1409" to-port="1" />
|
|
<edge from-layer="1409" from-port="2" to-layer="1410" to-port="1" />
|
|
<edge from-layer="1410" from-port="2" to-layer="1412" to-port="0" />
|
|
<edge from-layer="1411" from-port="0" to-layer="1412" to-port="1" />
|
|
<edge from-layer="1412" from-port="2" to-layer="1413" to-port="0" />
|
|
<edge from-layer="1413" from-port="2" to-layer="1415" to-port="0" />
|
|
<edge from-layer="1414" from-port="0" to-layer="1415" to-port="1" />
|
|
<edge from-layer="1415" from-port="2" to-layer="1417" to-port="0" />
|
|
<edge from-layer="1416" from-port="0" to-layer="1417" to-port="1" />
|
|
<edge from-layer="1417" from-port="2" to-layer="1419" to-port="0" />
|
|
<edge from-layer="1418" from-port="0" to-layer="1419" to-port="1" />
|
|
<edge from-layer="1419" from-port="2" to-layer="1421" to-port="0" />
|
|
<edge from-layer="1419" from-port="2" to-layer="1449" to-port="1" />
|
|
<edge from-layer="1420" from-port="0" to-layer="1421" to-port="1" />
|
|
<edge from-layer="1421" from-port="2" to-layer="1426" to-port="0" />
|
|
<edge from-layer="1422" from-port="0" to-layer="1426" to-port="1" />
|
|
<edge from-layer="1423" from-port="0" to-layer="1426" to-port="2" />
|
|
<edge from-layer="1424" from-port="0" to-layer="1426" to-port="3" />
|
|
<edge from-layer="1425" from-port="0" to-layer="1426" to-port="4" />
|
|
<edge from-layer="1426" from-port="5" to-layer="1431" to-port="0" />
|
|
<edge from-layer="1427" from-port="0" to-layer="1428" to-port="0" />
|
|
<edge from-layer="1428" from-port="1" to-layer="1430" to-port="0" />
|
|
<edge from-layer="1429" from-port="0" to-layer="1430" to-port="1" />
|
|
<edge from-layer="1430" from-port="2" to-layer="1431" to-port="1" />
|
|
<edge from-layer="1431" from-port="2" to-layer="1433" to-port="0" />
|
|
<edge from-layer="1432" from-port="0" to-layer="1433" to-port="1" />
|
|
<edge from-layer="1433" from-port="2" to-layer="1434" to-port="0" />
|
|
<edge from-layer="1434" from-port="1" to-layer="1436" to-port="0" />
|
|
<edge from-layer="1435" from-port="0" to-layer="1436" to-port="1" />
|
|
<edge from-layer="1436" from-port="2" to-layer="1441" to-port="0" />
|
|
<edge from-layer="1437" from-port="0" to-layer="1441" to-port="1" />
|
|
<edge from-layer="1438" from-port="0" to-layer="1441" to-port="2" />
|
|
<edge from-layer="1439" from-port="0" to-layer="1441" to-port="3" />
|
|
<edge from-layer="1440" from-port="0" to-layer="1441" to-port="4" />
|
|
<edge from-layer="1441" from-port="5" to-layer="1446" to-port="0" />
|
|
<edge from-layer="1442" from-port="0" to-layer="1443" to-port="0" />
|
|
<edge from-layer="1443" from-port="1" to-layer="1445" to-port="0" />
|
|
<edge from-layer="1444" from-port="0" to-layer="1445" to-port="1" />
|
|
<edge from-layer="1445" from-port="2" to-layer="1446" to-port="1" />
|
|
<edge from-layer="1446" from-port="2" to-layer="1448" to-port="0" />
|
|
<edge from-layer="1447" from-port="0" to-layer="1448" to-port="1" />
|
|
<edge from-layer="1448" from-port="2" to-layer="1449" to-port="0" />
|
|
<edge from-layer="1449" from-port="2" to-layer="1451" to-port="0" />
|
|
<edge from-layer="1450" from-port="0" to-layer="1451" to-port="1" />
|
|
<edge from-layer="1451" from-port="2" to-layer="1453" to-port="0" />
|
|
<edge from-layer="1452" from-port="0" to-layer="1453" to-port="1" />
|
|
<edge from-layer="1453" from-port="2" to-layer="1455" to-port="0" />
|
|
<edge from-layer="1454" from-port="0" to-layer="1455" to-port="1" />
|
|
<edge from-layer="1455" from-port="2" to-layer="1456" to-port="0" />
|
|
</edges>
|
|
<rt_info>
|
|
<Runtime_version value="2024.3.0-16041-1e3b88e4e3f-releases/2024/3" />
|
|
<conversion_parameters>
|
|
<framework value="pytorch" />
|
|
<is_python_object value="True" />
|
|
</conversion_parameters>
|
|
<nncf>
|
|
<friendly_names_were_updated value="True" />
|
|
<quantization>
|
|
<advanced_parameters value="{'overflow_fix': 'disable', 'quantize_outputs': False, 'inplace_statistics': True, 'disable_channel_alignment': True, 'disable_bias_correction': False, 'batchwise_statistics': None, 'activations_quantization_params': None, 'weights_quantization_params': None, 'activations_range_estimator_params': {'min': {'statistics_type': None, 'aggregator_type': None, 'clipping_value': None, 'quantile_outlier_prob': 0.0001}, 'max': {'statistics_type': None, 'aggregator_type': None, 'clipping_value': None, 'quantile_outlier_prob': 0.0001}}, 'weights_range_estimator_params': {'min': {'statistics_type': None, 'aggregator_type': None, 'clipping_value': None, 'quantile_outlier_prob': 0.0001}, 'max': {'statistics_type': None, 'aggregator_type': None, 'clipping_value': None, 'quantile_outlier_prob': 0.0001}}, 'bias_correction_params': {'apply_for_all_nodes': False, 'threshold': None}, 'smooth_quant_alphas': {'convolution': -1, 'matmul': 0.95}, 'smooth_quant_alpha': None, 'backend_params': {}}" />
|
|
<fast_bias_correction value="True" />
|
|
<ignored_scope>
|
|
<types value="['GroupNormalization']" />
|
|
</ignored_scope>
|
|
<model_type value="transformer" />
|
|
<preset value="mixed" />
|
|
<subset_size value="300" />
|
|
<target_device value="ANY" />
|
|
</quantization>
|
|
</nncf>
|
|
<optimum>
|
|
<optimum_intel_version value="1.20.0.dev0+95a80f0" />
|
|
<optimum_version value="1.24.0.dev0" />
|
|
<pytorch_version value="2.5.0.dev20240807+cu121" />
|
|
<transformers_version value="4.43.4" />
|
|
</optimum>
|
|
</rt_info>
|
|
</net>
|