ov_models / t2t /ko2en_int4 /openvino_encoder_model.xml
rippertnt's picture
Upload 257 files
6c33256 verified
<?xml version="1.0"?>
<net name="Model0" version="11">
<layers>
<layer id="1" name="input_ids" type="Parameter" version="opset1">
<data shape="?,?" element_type="i64" />
<output>
<port id="0" precision="I64" names="input_ids">
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="0" name="attention_mask" type="Parameter" version="opset1">
<data shape="?,?" element_type="i64" />
<output>
<port id="0" precision="I64" names="attention_mask">
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="2" name="self.embed_tokens.weight" type="Const" version="opset1">
<data element_type="u8" shape="30000, 768" offset="0" size="23040000" />
<output>
<port id="0" precision="U8">
<dim>30000</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="3" name="Convert_170423" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U8">
<dim>30000</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>30000</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="4" name="self.embed_tokens.weight/zero_point" type="Const" version="opset1">
<data element_type="u8" shape="30000, 1" offset="23040000" size="30000" />
<output>
<port id="0" precision="U8">
<dim>30000</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="5" name="Convert_170426" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U8">
<dim>30000</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>30000</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="6" name="self.embed_tokens.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>30000</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP16">
<dim>30000</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>30000</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="7" name="self.embed_tokens.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="30000, 1" offset="23070000" size="60000" />
<output>
<port id="0" precision="FP16">
<dim>30000</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="8" name="self.embed_tokens.weight/fq_weights_0" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>30000</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP16">
<dim>30000</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>30000</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="9" name="self.embed_tokens.weight/fq_weights_0/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<dim>30000</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="1" precision="FP32">
<dim>30000</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="10" name="Constant_25573" type="Const" version="opset1">
<data element_type="i64" shape="2" offset="23130000" size="16" />
<output>
<port id="0" precision="I64">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="11" name="aten::view/Reshape" type="Reshape" version="opset1">
<data special_zero="true" />
<input>
<port id="0" precision="I64">
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="I64">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="I64" names="24">
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="12" name="__module.embed_tokens/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="13" name="__module.embed_tokens/aten::embedding/Constant" type="Const" version="opset1">
<data element_type="i32" shape="" offset="23130016" size="4" />
<output>
<port id="0" precision="I32" />
</output>
</layer>
<layer id="14" name="__module.embed_tokens/aten::embedding/Gather" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="FP32">
<dim>30000</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="94,95,inputs_embeds">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="15" name="self.embed_positions.weight" type="Const" version="opset1">
<data element_type="u8" shape="1028, 768" offset="23130020" size="789504" />
<output>
<port id="0" precision="U8">
<dim>1028</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="16" name="Convert_170566" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U8">
<dim>1028</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>1028</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="17" name="self.embed_positions.weight/zero_point" type="Const" version="opset1">
<data element_type="u8" shape="1028, 1" offset="23919524" size="1028" />
<output>
<port id="0" precision="U8">
<dim>1028</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="18" name="Convert_170569" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U8">
<dim>1028</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>1028</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="19" name="self.embed_positions.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>1028</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP16">
<dim>1028</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>1028</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="20" name="self.embed_positions.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="1028, 1" offset="23920552" size="2056" />
<output>
<port id="0" precision="FP16">
<dim>1028</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="21" name="self.embed_positions.weight/fq_weights_0" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>1028</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP16">
<dim>1028</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>1028</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="22" name="self.embed_positions.weight/fq_weights_0/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<dim>1028</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="1" precision="FP32">
<dim>1028</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="23" name="104" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" names="104" />
</output>
</layer>
<layer id="24" name="ShapeOf_25486" 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_25487" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922616" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="26" name="Constant_25488" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="27" name="Gather_25489" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>2</dim>
</port>
<port id="1" precision="I64" />
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64" names="107,109,19,21" />
</output>
</layer>
<layer id="28" name="__module.embed_positions/aten::arange/Constant" type="Const" version="opset1">
<data element_type="i32" shape="" offset="23922624" size="4" />
<output>
<port id="0" precision="I32" />
</output>
</layer>
<layer id="29" name="__module.embed_positions/aten::arange/Range" type="Range" version="opset4">
<data output_type="i64" />
<input>
<port id="0" precision="I64" />
<port id="1" precision="I64" />
<port id="2" precision="I32" />
</input>
<output>
<port id="3" precision="I64" names="111">
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="30" name="Constant_24432" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922628" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="31" name="Add_23863" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="I64">
<dim>-1</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="I64">
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="32" name="Constant_23865" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="33" name="ShapeOf_23864" type="ShapeOf" version="opset3">
<data output_type="i64" />
<input>
<port id="0" precision="I64">
<dim>-1</dim>
</port>
</input>
<output>
<port id="1" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="34" name="Concat_23866" 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>
</input>
<output>
<port id="2" precision="I64">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="35" name="Constant_25495" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922608" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="36" name="Constant_25496" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="37" name="Gather_25497" 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="106">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="38" name="Constant_22106" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="39" name="__module.embed_positions/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>
</input>
<output>
<port id="2" precision="I64" names="112">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="40" name="Maximum_23867" type="Maximum" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="I64">
<dim>2</dim>
</port>
<port id="1" precision="I64">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="I64">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="41" name="__module.embed_positions/aten::add/Add_1" type="Broadcast" version="opset3">
<data mode="numpy" />
<input>
<port id="0" precision="I64">
<dim>-1</dim>
</port>
<port id="1" precision="I64">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="I64" names="114">
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="42" name="__module.embed_positions/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="43" name="__module.embed_positions/aten::embedding/Constant" type="Const" version="opset1">
<data element_type="i32" shape="" offset="23130016" size="4" />
<output>
<port id="0" precision="I32" />
</output>
</layer>
<layer id="44" name="__module.embed_positions/aten::embedding/Gather" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="FP32">
<dim>1028</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="115,34,embed_pos,embed_pos.1">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="45" name="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="36">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="46" name="__module.layernorm_embedding/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="47" name="__module.layernorm_embedding/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="48" name="Constant_24433" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="23922640" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="49" name="__module.layernorm_embedding/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="50" name="Constant_24434" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="23925712" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="51" name="__module.layernorm_embedding/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="122,input.1">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="52" name="Constant_24282" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="23928784" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="53" name="Convert_170598" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="54" name="Constant_24282/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="24223696" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="55" name="Convert_170600" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="56" name="Constant_24282/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="57" name="Constant_24282/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="24226000" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="58" name="Constant_24282/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="59" name="Constant_170605" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="60" name="Reshape_170606" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="61" name="Constant_24282/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="62" name="Multiply_24257" 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="63" name="Constant_24435" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="24235224" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="64" name="__module.layers.0.self_attn/aten::mul/Multiply" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="155,tensor.5">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="65" name="Constant_25574" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="66" name="__module.layers.0.self_attn/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="175">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="67" name="__module.layers.0.self_attn/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="68" name="__module.layers.0.self_attn/aten::transpose/Transpose" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="176">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="69" name="ShapeOf_25515" 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="70" name="Constant_25695" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922608" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="71" name="Constant_25517" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="72" name="Gather_25518" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>3</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64" names="149,251,353,455,557,659">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="73" name="Constant_25470" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="24238344" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="74" name="__module.layers.0.self_attn/aten::mul/Multiply_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="I64">
<dim>1</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="I64" names="170,171,172,173,190,191,272,273,274,275,292,293,374,375,376,377,394,395,476,477,478,479,496,497,578,579,580,581,598,599,680,681,682,683,700,701">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="75" name="__module.layers.0.self_attn/prim::ListConstruct/Reshape_0_1" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="24238352" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="76" name="__module.layers.0.self_attn/prim::ListConstruct/Reshape_2" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="24238360" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="77" name="__module.layers.0.self_attn/prim::ListConstruct/Concat_1" 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>
</input>
<output>
<port id="3" precision="I64" names="178,180,182,280,282,284,382,384,386,484,486,488,586,588,590,688,690,692">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="78" name="__module.layers.0.self_attn/aten::view/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="179">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="79" name="self.layers.0.self_attn.k_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="24238368" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="80" name="Convert_170587" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="81" name="self.layers.0.self_attn.k_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="24533280" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="82" name="Convert_170589" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="83" name="self.layers.0.self_attn.k_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="84" name="self.layers.0.self_attn.k_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="24535584" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="85" name="self.layers.0.self_attn.k_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="86" name="Constant_170594" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="87" name="Reshape_170595" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="88" name="self.layers.0.self_attn.k_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="89" name="__module.layers.0.self_attn.k_proj/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="90" name="Constant_24436" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="24544800" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="91" name="__module.layers.0.self_attn.k_proj/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="158,tensor.1">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="92" name="Constant_25575" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="93" name="__module.layers.0.self_attn/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="160">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="94" name="__module.layers.0.self_attn/aten::transpose/ScatterElementsUpdate_1" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="95" name="__module.layers.0.self_attn/aten::transpose/Transpose_1" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="161">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="96" name="__module.layers.0.self_attn/aten::reshape/Reshape" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="181,key_states.3">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="97" name="__module.layers.0.self_attn/aten::bmm/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>48</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="186,attn_weights.1">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="98" name="__module.layers.0.self_attn/prim::ListConstruct/Reshape_1" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="24238344" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="99" name="Constant_25524" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="100" name="Constant_25525" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="101" name="Gather_25526" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>3</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64" names="151,253,355,457,559,661">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="102" name="ShapeOf_25528" type="ShapeOf" version="opset3">
<data output_type="i64" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="103" name="Constant_25529" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="104" name="Constant_25530" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="105" name="Gather_25531" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>3</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64" names="184">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="106" name="__module.layers.0.self_attn/prim::ListConstruct/Concat_4" 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="187">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="107" name="__module.layers.0.self_attn/aten::view/Reshape_3" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<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="FP32" names="188">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="108" name="Constant_24438" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 1, 1" offset="24547872" 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="109" name="55" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922616" size="8" />
<output>
<port id="0" precision="I64" names="55" />
</output>
</layer>
<layer id="110" 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="56">
<dim>-1</dim>
<dim>1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="111" name="57" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922628" size="8" />
<output>
<port id="0" precision="I64" names="57" />
</output>
</layer>
<layer id="112" 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="58,63">
<dim>-1</dim>
<dim>1</dim>
<dim>1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="113" name="ShapeOf_25499" 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="114" name="Constant_25698" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922608" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="115" name="Constant_25501" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="116" name="Gather_25502" 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="42,44">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="117" name="Constant_22112" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="118" name="Constant_25703" type="Const" version="opset1">
<data element_type="i64" shape="2" offset="24547876" size="16" />
<output>
<port id="0" precision="I64">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="119" name="Constant_25704" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="120" name="Gather_25705" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>2</dim>
</port>
<port id="1" precision="I64">
<dim>2</dim>
</port>
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="121" name="prim::ListConstruct/Concat_1" type="Concat" version="opset1">
<data axis="0" />
<input>
<port id="0" precision="I64">
<dim>1</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
<port id="2" precision="I64">
<dim>2</dim>
</port>
</input>
<output>
<port id="3" precision="I64" names="65">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="122" 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="67">
<dim>-1</dim>
<dim>1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="123" name="aten::to/Convert_1" 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="72">
<dim>-1</dim>
<dim>1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="124" name="Constant_24437" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 1, 1" offset="24547872" 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="125" 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="126" 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="75,inverted_mask">
<dim>-1</dim>
<dim>1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="127" name="aten::to/Convert_2" 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="80">
<dim>-1</dim>
<dim>1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="128" name="aten::masked_fill/ConvertLike" type="Const" version="opset1">
<data element_type="f32" shape="" offset="24547892" size="4" />
<output>
<port id="0" precision="FP32" />
</output>
</layer>
<layer id="129" 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="82,attention_mask">
<dim>-1</dim>
<dim>1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="130" name="__module.layers.0.self_attn/aten::add/Add" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</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="189,attn_weights.3">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="131" name="__module.layers.0.self_attn/prim::ListConstruct/Concat_5" 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>
</input>
<output>
<port id="3" precision="I64" names="192">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="132" name="__module.layers.0.self_attn/aten::view/Reshape_4" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="193,input.3">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="133" name="__module.layers.0.self_attn/aten::softmax/Softmax" type="SoftMax" version="opset8">
<data axis="-1" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</input>
<output>
<port id="1" precision="FP32" names="194,input.5">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="134" name="self.layers.0.self_attn.v_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="24547896" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="135" name="Convert_170576" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="136" name="self.layers.0.self_attn.v_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="24842808" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="137" name="Convert_170578" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="138" name="self.layers.0.self_attn.v_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="139" name="self.layers.0.self_attn.v_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="24845112" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="140" name="self.layers.0.self_attn.v_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="141" name="Constant_170583" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="142" name="Reshape_170584" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="143" name="self.layers.0.self_attn.v_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="144" name="__module.layers.0.self_attn.v_proj/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="145" name="Constant_24439" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="24854328" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="146" name="__module.layers.0.self_attn.v_proj/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="165,tensor.3">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="147" name="Constant_25576" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="148" name="__module.layers.0.self_attn/aten::view/Reshape_5" type="Reshape" version="opset1">
<data special_zero="true" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I64">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="167">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="149" name="__module.layers.0.self_attn/aten::transpose/ScatterElementsUpdate_3" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="150" name="__module.layers.0.self_attn/aten::transpose/Transpose_3" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="168">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="151" name="__module.layers.0.self_attn/aten::reshape/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="183">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="152" name="__module.layers.0.self_attn/aten::bmm/MatMul_1" type="MatMul" version="opset1">
<data transpose_a="false" transpose_b="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="196,attn_output.1">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="153" name="__module.layers.0.self_attn/prim::ListConstruct/Concat_8" 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="197,299,401,503,605,707">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="154" name="__module.layers.0.self_attn/aten::view/Reshape_6" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="198,attn_output.3">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="155" name="__module.layers.0.self_attn/aten::transpose/ScatterElementsUpdate_4" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="156" name="__module.layers.0.self_attn/aten::transpose/Transpose_4" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="199,attn_output.5">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="157" name="Constant_25577" type="Const" version="opset1">
<data element_type="i64" shape="3" offset="24857400" size="24" />
<output>
<port id="0" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="158" name="__module.layers.0.self_attn/aten::reshape/Reshape_2" type="Reshape" version="opset1">
<data special_zero="true" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="201">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="159" name="self.layers.0.self_attn.out_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="24857424" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="160" name="Convert_170609" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="161" name="self.layers.0.self_attn.out_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="25152336" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="162" name="Convert_170611" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="163" name="self.layers.0.self_attn.out_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="164" name="self.layers.0.self_attn.out_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="25154640" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="165" name="self.layers.0.self_attn.out_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="166" name="Constant_170616" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="167" name="Reshape_170617" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="168" name="self.layers.0.self_attn.out_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="169" name="__module.layers.0.self_attn.out_proj/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="170" name="Constant_24440" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="25163856" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="171" name="__module.layers.0.self_attn.out_proj/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="204,input.7">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="172" name="__module.layers.0/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="206">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="173" name="__module.layers.0.self_attn_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="174" name="__module.layers.0.self_attn_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="175" name="Constant_24441" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="25166928" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="176" name="__module.layers.0.self_attn_layer_norm/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="177" name="Constant_24442" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="25170000" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="178" name="__module.layers.0.self_attn_layer_norm/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="210,residual.1">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="179" name="self.layers.0.fc1.weight" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 128" offset="25173072" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="180" name="Convert_170433" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="181" name="self.layers.0.fc1.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 1" offset="26352720" size="9216" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="182" name="Convert_170435" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="183" name="self.layers.0.fc1.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="184" name="self.layers.0.fc1.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="3072, 6, 1" offset="26361936" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="185" name="self.layers.0.fc1.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="186" name="Constant_170440" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="26398800" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="187" name="Reshape_170441" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="188" name="self.layers.0.fc1.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="189" name="__module.layers.0.fc1/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="190" name="Constant_24443" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 3072" offset="26398808" size="12288" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="191" name="__module.layers.0.fc1/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="213">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="192" name="__module.layers.0.activation_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="214,input.9">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="193" name="self.layers.0.fc2.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 128" offset="26411096" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="194" name="Convert_170444" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="195" name="self.layers.0.fc2.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 1" offset="27590744" size="9216" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="196" name="Convert_170446" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="197" name="self.layers.0.fc2.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="198" name="self.layers.0.fc2.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 24, 1" offset="27599960" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="199" name="self.layers.0.fc2.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="200" name="Constant_170451" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="27636824" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="201" name="Reshape_170452" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="202" name="self.layers.0.fc2.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="203" name="__module.layers.0.fc2/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="204" name="Constant_24444" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="27636832" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="205" name="__module.layers.0.fc2/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="218,input.11">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="206" name="__module.layers.0/aten::add/Add_1" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="220">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="207" name="__module.layers.0.final_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="208" name="__module.layers.0.final_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="209" name="Constant_24445" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="27639904" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="210" name="__module.layers.0.final_layer_norm/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="211" name="Constant_24446" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="27642976" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="212" name="__module.layers.0.final_layer_norm/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="224,hidden_states.7">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="213" name="Constant_24285" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="27646048" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="214" name="Convert_170642" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="215" name="Constant_24285/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="27940960" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="216" name="Convert_170644" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="217" name="Constant_24285/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="218" name="Constant_24285/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="27943264" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="219" name="Constant_24285/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="220" name="Constant_170649" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="221" name="Reshape_170650" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="222" name="Constant_24285/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="223" name="Multiply_24261" 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_24447" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="27952480" 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.layers.1.self_attn/aten::mul/Multiply" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="257,tensor.11">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="226" name="Constant_25578" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="227" name="__module.layers.1.self_attn/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="277">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="228" name="__module.layers.1.self_attn/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="229" name="__module.layers.1.self_attn/aten::transpose/Transpose" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="278">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="230" name="__module.layers.1.self_attn/aten::view/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="281">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="231" name="self.layers.1.self_attn.k_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="27955552" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="232" name="Convert_170631" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="233" name="self.layers.1.self_attn.k_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="28250464" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="234" name="Convert_170633" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="235" name="self.layers.1.self_attn.k_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="236" name="self.layers.1.self_attn.k_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="28252768" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="237" name="self.layers.1.self_attn.k_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="238" name="Constant_170638" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="239" name="Reshape_170639" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="240" name="self.layers.1.self_attn.k_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="241" name="__module.layers.1.self_attn.k_proj/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="242" name="Constant_24448" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="28261984" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="243" name="__module.layers.1.self_attn.k_proj/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="260,tensor.7">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="244" name="Constant_25579" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="245" name="__module.layers.1.self_attn/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="262">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="246" name="__module.layers.1.self_attn/aten::transpose/ScatterElementsUpdate_1" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="247" name="__module.layers.1.self_attn/aten::transpose/Transpose_1" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="263">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="248" name="__module.layers.1.self_attn/aten::reshape/Reshape" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="283,key_states.7">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="249" name="__module.layers.1.self_attn/aten::bmm/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>48</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="288,attn_weights.5">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="250" name="__module.layers.1.self_attn/prim::ListConstruct/Reshape_1" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="24238344" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="251" name="ShapeOf_25536" type="ShapeOf" version="opset3">
<data output_type="i64" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="252" name="Constant_25537" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="253" name="Constant_25538" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="254" name="Gather_25539" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>3</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64" names="286">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="255" name="__module.layers.1.self_attn/prim::ListConstruct/Concat_4" 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="289">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="256" name="__module.layers.1.self_attn/aten::view/Reshape_3" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<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="FP32" names="290">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="257" name="__module.layers.1.self_attn/aten::add/Add" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</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="291,attn_weights.7">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="258" name="__module.layers.1.self_attn/prim::ListConstruct/Concat_5" 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>
</input>
<output>
<port id="3" precision="I64" names="294">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="259" name="__module.layers.1.self_attn/aten::view/Reshape_4" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="295,input.13">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="260" name="__module.layers.1.self_attn/aten::softmax/Softmax" type="SoftMax" version="opset8">
<data axis="-1" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</input>
<output>
<port id="1" precision="FP32" names="296,input.15">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="261" name="self.layers.1.self_attn.v_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="28265056" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="262" name="Convert_170620" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="263" name="self.layers.1.self_attn.v_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="28559968" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="264" name="Convert_170622" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="265" name="self.layers.1.self_attn.v_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="266" name="self.layers.1.self_attn.v_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="28562272" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="267" name="self.layers.1.self_attn.v_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="268" name="Constant_170627" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="269" name="Reshape_170628" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="270" name="self.layers.1.self_attn.v_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="271" name="__module.layers.1.self_attn.v_proj/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="272" name="Constant_24449" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="28571488" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="273" name="__module.layers.1.self_attn.v_proj/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="267,tensor.9">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="274" name="Constant_25580" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="275" name="__module.layers.1.self_attn/aten::view/Reshape_5" 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="269">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="276" name="__module.layers.1.self_attn/aten::transpose/ScatterElementsUpdate_3" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="277" name="__module.layers.1.self_attn/aten::transpose/Transpose_3" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="270">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="278" name="__module.layers.1.self_attn/aten::reshape/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="285">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="279" name="__module.layers.1.self_attn/aten::bmm/MatMul_1" type="MatMul" version="opset1">
<data transpose_a="false" transpose_b="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="298,attn_output.7">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="280" name="__module.layers.1.self_attn/aten::view/Reshape_6" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="300,attn_output.9">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="281" name="__module.layers.1.self_attn/aten::transpose/ScatterElementsUpdate_4" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="282" name="__module.layers.1.self_attn/aten::transpose/Transpose_4" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="301,attn_output.11">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="283" name="Constant_25581" type="Const" version="opset1">
<data element_type="i64" shape="3" offset="24857400" size="24" />
<output>
<port id="0" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="284" name="__module.layers.1.self_attn/aten::reshape/Reshape_2" type="Reshape" version="opset1">
<data special_zero="true" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="303">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="285" name="self.layers.1.self_attn.out_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="28574560" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="286" name="Convert_170653" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="287" name="self.layers.1.self_attn.out_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="28869472" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="288" name="Convert_170655" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="289" name="self.layers.1.self_attn.out_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="290" name="self.layers.1.self_attn.out_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="28871776" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="291" name="self.layers.1.self_attn.out_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="292" name="Constant_170660" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="293" name="Reshape_170661" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="294" name="self.layers.1.self_attn.out_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="1" precision="FP32">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="295" name="__module.layers.1.self_attn.out_proj/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="296" name="Constant_24450" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="28880992" 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.layers.1.self_attn.out_proj/aten::linear/Add" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="306,input.17">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="298" name="__module.layers.1/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="308">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="299" name="__module.layers.1.self_attn_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="300" name="__module.layers.1.self_attn_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="301" name="Constant_24451" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="28884064" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="302" name="__module.layers.1.self_attn_layer_norm/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="303" name="Constant_24452" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="28887136" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="304" name="__module.layers.1.self_attn_layer_norm/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="312,residual.3">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="305" name="self.layers.1.fc1.weight" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 128" offset="28890208" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="306" name="Convert_170455" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="307" name="self.layers.1.fc1.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 1" offset="30069856" size="9216" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="308" name="Convert_170457" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="309" name="self.layers.1.fc1.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="310" name="self.layers.1.fc1.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="3072, 6, 1" offset="30079072" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="311" name="self.layers.1.fc1.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="312" name="Constant_170462" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="26398800" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="313" name="Reshape_170463" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="314" name="self.layers.1.fc1.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="315" name="__module.layers.1.fc1/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="316" name="Constant_24453" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 3072" offset="30115936" size="12288" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="317" name="__module.layers.1.fc1/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="315">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="318" name="__module.layers.1.activation_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="316,input.19">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="319" name="self.layers.1.fc2.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 128" offset="30128224" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="320" name="Convert_170466" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="321" name="self.layers.1.fc2.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 1" offset="31307872" size="9216" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="322" name="Convert_170468" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="323" name="self.layers.1.fc2.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="324" name="self.layers.1.fc2.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 24, 1" offset="31317088" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="325" name="self.layers.1.fc2.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="326" name="Constant_170473" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="27636824" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="327" name="Reshape_170474" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="328" name="self.layers.1.fc2.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="329" name="__module.layers.1.fc2/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="330" name="Constant_24454" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="31353952" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="331" name="__module.layers.1.fc2/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="320,input.21">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="332" name="__module.layers.1/aten::add/Add_1" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="322">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="333" name="__module.layers.1.final_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="334" name="__module.layers.1.final_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="335" name="Constant_24455" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="31357024" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="336" name="__module.layers.1.final_layer_norm/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="337" name="Constant_24456" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="31360096" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="338" name="__module.layers.1.final_layer_norm/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="326,hidden_states.13">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="339" name="Constant_24288" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="31363168" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="340" name="Convert_170686" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="341" name="Constant_24288/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="31658080" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="342" name="Convert_170688" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="343" name="Constant_24288/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="344" name="Constant_24288/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="31660384" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="345" name="Constant_24288/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="346" name="Constant_170693" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="347" name="Reshape_170694" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="348" name="Constant_24288/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="349" name="Multiply_24265" 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="350" name="Constant_24457" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="31669600" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="351" name="__module.layers.2.self_attn/aten::mul/Multiply" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="359,tensor.17">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="352" name="Constant_25582" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="353" name="__module.layers.2.self_attn/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="379">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="354" name="__module.layers.2.self_attn/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="355" name="__module.layers.2.self_attn/aten::transpose/Transpose" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="380">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="356" name="__module.layers.2.self_attn/aten::view/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="383">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="357" name="self.layers.2.self_attn.k_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="31672672" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="358" name="Convert_170675" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="359" name="self.layers.2.self_attn.k_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="31967584" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="360" name="Convert_170677" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="361" name="self.layers.2.self_attn.k_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="362" name="self.layers.2.self_attn.k_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="31969888" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="363" name="self.layers.2.self_attn.k_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="364" name="Constant_170682" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="365" name="Reshape_170683" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="366" name="self.layers.2.self_attn.k_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="367" name="__module.layers.2.self_attn.k_proj/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="368" name="Constant_24458" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="31979104" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="369" name="__module.layers.2.self_attn.k_proj/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="362,tensor.13">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="370" name="Constant_25583" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="371" name="__module.layers.2.self_attn/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="364">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="372" name="__module.layers.2.self_attn/aten::transpose/ScatterElementsUpdate_1" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="373" name="__module.layers.2.self_attn/aten::transpose/Transpose_1" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="365">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="374" name="__module.layers.2.self_attn/aten::reshape/Reshape" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="385,key_states.11">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="375" name="__module.layers.2.self_attn/aten::bmm/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>48</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="390,attn_weights.9">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="376" name="__module.layers.2.self_attn/prim::ListConstruct/Reshape_1" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="24238344" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="377" name="ShapeOf_25544" type="ShapeOf" version="opset3">
<data output_type="i64" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="378" name="Constant_25545" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="379" name="Constant_25546" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="380" name="Gather_25547" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>3</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64" names="388">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="381" name="__module.layers.2.self_attn/prim::ListConstruct/Concat_4" 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="391">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="382" name="__module.layers.2.self_attn/aten::view/Reshape_3" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<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="FP32" names="392">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="383" name="__module.layers.2.self_attn/aten::add/Add" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</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="393,attn_weights.11">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="384" name="__module.layers.2.self_attn/prim::ListConstruct/Concat_5" 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>
</input>
<output>
<port id="3" precision="I64" names="396">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="385" name="__module.layers.2.self_attn/aten::view/Reshape_4" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="397,input.23">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="386" name="__module.layers.2.self_attn/aten::softmax/Softmax" type="SoftMax" version="opset8">
<data axis="-1" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</input>
<output>
<port id="1" precision="FP32" names="398,input.25">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="387" name="self.layers.2.self_attn.v_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="31982176" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="388" name="Convert_170664" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="389" name="self.layers.2.self_attn.v_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="32277088" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="390" name="Convert_170666" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="391" name="self.layers.2.self_attn.v_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="392" name="self.layers.2.self_attn.v_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="32279392" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="393" name="self.layers.2.self_attn.v_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="394" name="Constant_170671" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="395" name="Reshape_170672" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="396" name="self.layers.2.self_attn.v_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="397" name="__module.layers.2.self_attn.v_proj/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="398" name="Constant_24459" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="32288608" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="399" name="__module.layers.2.self_attn.v_proj/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="369,tensor.15">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="400" name="Constant_25584" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="401" name="__module.layers.2.self_attn/aten::view/Reshape_5" 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="371">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="402" name="__module.layers.2.self_attn/aten::transpose/ScatterElementsUpdate_3" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="403" name="__module.layers.2.self_attn/aten::transpose/Transpose_3" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="372">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="404" name="__module.layers.2.self_attn/aten::reshape/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="387">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="405" name="__module.layers.2.self_attn/aten::bmm/MatMul_1" type="MatMul" version="opset1">
<data transpose_a="false" transpose_b="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="400,attn_output.13">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="406" name="__module.layers.2.self_attn/aten::view/Reshape_6" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="402,attn_output.15">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="407" name="__module.layers.2.self_attn/aten::transpose/ScatterElementsUpdate_4" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="408" name="__module.layers.2.self_attn/aten::transpose/Transpose_4" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="403,attn_output.17">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="409" name="Constant_25585" type="Const" version="opset1">
<data element_type="i64" shape="3" offset="24857400" size="24" />
<output>
<port id="0" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="410" name="__module.layers.2.self_attn/aten::reshape/Reshape_2" type="Reshape" version="opset1">
<data special_zero="true" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="405">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="411" name="self.layers.2.self_attn.out_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="32291680" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="412" name="Convert_170697" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="413" name="self.layers.2.self_attn.out_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="32586592" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="414" name="Convert_170699" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="415" name="self.layers.2.self_attn.out_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="416" name="self.layers.2.self_attn.out_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="32588896" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="417" name="self.layers.2.self_attn.out_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="418" name="Constant_170704" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="419" name="Reshape_170705" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="420" name="self.layers.2.self_attn.out_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="__module.layers.2.self_attn.out_proj/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="422" name="Constant_24460" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="32598112" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="423" name="__module.layers.2.self_attn.out_proj/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="408,input.27">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="424" name="__module.layers.2/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="410">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="425" name="__module.layers.2.self_attn_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="426" name="__module.layers.2.self_attn_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="427" name="Constant_24461" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="32601184" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="428" name="__module.layers.2.self_attn_layer_norm/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="429" name="Constant_24462" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="32604256" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="430" name="__module.layers.2.self_attn_layer_norm/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="414,residual.5">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="431" name="self.layers.2.fc1.weight" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 128" offset="32607328" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="432" name="Convert_170477" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="433" name="self.layers.2.fc1.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 1" offset="33786976" size="9216" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="434" name="Convert_170479" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="435" name="self.layers.2.fc1.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="436" name="self.layers.2.fc1.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="3072, 6, 1" offset="33796192" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="437" name="self.layers.2.fc1.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="438" name="Constant_170484" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="26398800" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="439" name="Reshape_170485" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="440" name="self.layers.2.fc1.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="441" name="__module.layers.2.fc1/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="442" name="Constant_24463" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 3072" offset="33833056" size="12288" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="443" name="__module.layers.2.fc1/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="417">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="444" name="__module.layers.2.activation_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="418,input.29">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="445" name="self.layers.2.fc2.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 128" offset="33845344" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="446" name="Convert_170488" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="447" name="self.layers.2.fc2.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 1" offset="35024992" size="9216" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="448" name="Convert_170490" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="449" name="self.layers.2.fc2.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="450" name="self.layers.2.fc2.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 24, 1" offset="35034208" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="451" name="self.layers.2.fc2.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="452" name="Constant_170495" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="27636824" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="453" name="Reshape_170496" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="454" name="self.layers.2.fc2.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="455" name="__module.layers.2.fc2/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="456" name="Constant_24464" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="35071072" 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.layers.2.fc2/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="422,input.31">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="458" name="__module.layers.2/aten::add/Add_1" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="424">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="459" name="__module.layers.2.final_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="460" name="__module.layers.2.final_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="461" name="Constant_24465" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="35074144" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="462" name="__module.layers.2.final_layer_norm/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="463" name="Constant_24466" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="35077216" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="464" name="__module.layers.2.final_layer_norm/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="428,hidden_states.19">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="465" name="Constant_24291" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="35080288" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="466" name="Convert_170730" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="467" name="Constant_24291/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="35375200" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="468" name="Convert_170732" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="469" name="Constant_24291/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="470" name="Constant_24291/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="35377504" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="471" name="Constant_24291/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="472" name="Constant_170737" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="473" name="Reshape_170738" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="474" name="Constant_24291/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="475" name="Multiply_24269" 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="476" name="Constant_24467" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="35386720" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="477" name="__module.layers.3.self_attn/aten::mul/Multiply" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</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,tensor.23">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="478" name="Constant_25586" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="479" name="__module.layers.3.self_attn/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="481">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="480" name="__module.layers.3.self_attn/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="481" name="__module.layers.3.self_attn/aten::transpose/Transpose" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="482">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="482" name="__module.layers.3.self_attn/aten::view/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="485">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="483" name="self.layers.3.self_attn.k_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="35389792" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="484" name="Convert_170719" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="485" name="self.layers.3.self_attn.k_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="35684704" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="486" name="Convert_170721" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="487" name="self.layers.3.self_attn.k_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="488" name="self.layers.3.self_attn.k_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="35687008" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="489" name="self.layers.3.self_attn.k_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="490" name="Constant_170726" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="491" name="Reshape_170727" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="492" name="self.layers.3.self_attn.k_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="493" name="__module.layers.3.self_attn.k_proj/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="494" name="Constant_24468" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="35696224" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="495" name="__module.layers.3.self_attn.k_proj/aten::linear/Add" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="464,tensor.19">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="496" name="Constant_25587" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="497" name="__module.layers.3.self_attn/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="466">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="498" name="__module.layers.3.self_attn/aten::transpose/ScatterElementsUpdate_1" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="499" name="__module.layers.3.self_attn/aten::transpose/Transpose_1" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="467">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="500" name="__module.layers.3.self_attn/aten::reshape/Reshape" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="487,key_states.15">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="501" name="__module.layers.3.self_attn/aten::bmm/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>48</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="492,attn_weights.13">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="502" name="__module.layers.3.self_attn/prim::ListConstruct/Reshape_1" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="24238344" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="503" name="ShapeOf_25552" type="ShapeOf" version="opset3">
<data output_type="i64" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="504" name="Constant_25553" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="505" name="Constant_25554" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="506" name="Gather_25555" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>3</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64" names="490">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="507" name="__module.layers.3.self_attn/prim::ListConstruct/Concat_4" 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="493">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="508" name="__module.layers.3.self_attn/aten::view/Reshape_3" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<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="FP32" names="494">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="509" name="__module.layers.3.self_attn/aten::add/Add" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</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="495,attn_weights.15">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="510" name="__module.layers.3.self_attn/prim::ListConstruct/Concat_5" 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>
</input>
<output>
<port id="3" precision="I64" names="498">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="511" name="__module.layers.3.self_attn/aten::view/Reshape_4" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="499,input.33">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="512" name="__module.layers.3.self_attn/aten::softmax/Softmax" type="SoftMax" version="opset8">
<data axis="-1" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</input>
<output>
<port id="1" precision="FP32" names="500,input.35">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="513" name="self.layers.3.self_attn.v_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="35699296" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="514" name="Convert_170708" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="515" name="self.layers.3.self_attn.v_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="35994208" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="516" name="Convert_170710" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="517" name="self.layers.3.self_attn.v_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="518" name="self.layers.3.self_attn.v_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="35996512" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="519" name="self.layers.3.self_attn.v_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="520" name="Constant_170715" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="521" name="Reshape_170716" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="522" name="self.layers.3.self_attn.v_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="523" name="__module.layers.3.self_attn.v_proj/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="524" name="Constant_24469" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="36005728" 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.layers.3.self_attn.v_proj/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="471,tensor.21">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="526" name="Constant_25588" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="527" name="__module.layers.3.self_attn/aten::view/Reshape_5" 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="473">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="528" name="__module.layers.3.self_attn/aten::transpose/ScatterElementsUpdate_3" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="529" name="__module.layers.3.self_attn/aten::transpose/Transpose_3" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="474">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="530" name="__module.layers.3.self_attn/aten::reshape/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="489">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="531" name="__module.layers.3.self_attn/aten::bmm/MatMul_1" type="MatMul" version="opset1">
<data transpose_a="false" transpose_b="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="502,attn_output.19">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="532" name="__module.layers.3.self_attn/aten::view/Reshape_6" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="504,attn_output.21">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="533" name="__module.layers.3.self_attn/aten::transpose/ScatterElementsUpdate_4" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="534" name="__module.layers.3.self_attn/aten::transpose/Transpose_4" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="505,attn_output.23">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="535" name="Constant_25589" type="Const" version="opset1">
<data element_type="i64" shape="3" offset="24857400" size="24" />
<output>
<port id="0" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="536" name="__module.layers.3.self_attn/aten::reshape/Reshape_2" type="Reshape" version="opset1">
<data special_zero="true" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="507">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="537" name="self.layers.3.self_attn.out_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="36008800" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="538" name="Convert_170741" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="539" name="self.layers.3.self_attn.out_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="36303712" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="540" name="Convert_170743" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="541" name="self.layers.3.self_attn.out_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="542" name="self.layers.3.self_attn.out_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="36306016" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="543" name="self.layers.3.self_attn.out_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="544" name="Constant_170748" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="545" name="Reshape_170749" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="546" name="self.layers.3.self_attn.out_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="547" name="__module.layers.3.self_attn.out_proj/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="548" name="Constant_24470" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="36315232" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="549" name="__module.layers.3.self_attn.out_proj/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="510,input.37">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="550" name="__module.layers.3/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="512">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="551" name="__module.layers.3.self_attn_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="552" name="__module.layers.3.self_attn_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="553" name="Constant_24471" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="36318304" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="554" name="__module.layers.3.self_attn_layer_norm/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="555" name="Constant_24472" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="36321376" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="556" name="__module.layers.3.self_attn_layer_norm/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="516,residual.7">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="557" name="self.layers.3.fc1.weight" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 128" offset="36324448" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="558" name="Convert_170499" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="559" name="self.layers.3.fc1.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 1" offset="37504096" size="9216" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="560" name="Convert_170501" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="561" name="self.layers.3.fc1.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="562" name="self.layers.3.fc1.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="3072, 6, 1" offset="37513312" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="563" name="self.layers.3.fc1.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="564" name="Constant_170506" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="26398800" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="565" name="Reshape_170507" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="566" name="self.layers.3.fc1.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="567" name="__module.layers.3.fc1/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="568" name="Constant_24473" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 3072" offset="37550176" size="12288" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="569" name="__module.layers.3.fc1/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="519">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="570" name="__module.layers.3.activation_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="520,input.39">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="571" name="self.layers.3.fc2.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 128" offset="37562464" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="572" name="Convert_170510" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="573" name="self.layers.3.fc2.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 1" offset="38742112" size="9216" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="574" name="Convert_170512" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="575" name="self.layers.3.fc2.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="576" name="self.layers.3.fc2.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 24, 1" offset="38751328" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="577" name="self.layers.3.fc2.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="578" name="Constant_170517" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="27636824" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="579" name="Reshape_170518" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="580" name="self.layers.3.fc2.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="581" name="__module.layers.3.fc2/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="582" name="Constant_24474" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="38788192" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="583" name="__module.layers.3.fc2/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="524,input.41">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="584" name="__module.layers.3/aten::add/Add_1" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="526">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="585" name="__module.layers.3.final_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="586" name="__module.layers.3.final_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="587" name="Constant_24475" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="38791264" 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.layers.3.final_layer_norm/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="589" name="Constant_24476" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="38794336" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="590" name="__module.layers.3.final_layer_norm/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="530,hidden_states.25">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="591" name="Constant_24294" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="38797408" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="592" name="Convert_170774" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="593" name="Constant_24294/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="39092320" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="594" name="Convert_170776" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="595" name="Constant_24294/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="596" name="Constant_24294/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="39094624" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="597" name="Constant_24294/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="598" name="Constant_170781" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="599" name="Reshape_170782" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="600" name="Constant_24294/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="601" name="Multiply_24273" 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="602" name="Constant_24477" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="39103840" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="603" name="__module.layers.4.self_attn/aten::mul/Multiply" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="563,tensor.29">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="604" name="Constant_25590" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="605" name="__module.layers.4.self_attn/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="583">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="606" name="__module.layers.4.self_attn/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="607" name="__module.layers.4.self_attn/aten::transpose/Transpose" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="584">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="608" name="__module.layers.4.self_attn/aten::view/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="587">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="609" name="self.layers.4.self_attn.k_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="39106912" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="610" name="Convert_170763" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="611" name="self.layers.4.self_attn.k_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="39401824" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="612" name="Convert_170765" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="613" name="self.layers.4.self_attn.k_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="614" name="self.layers.4.self_attn.k_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="39404128" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="615" name="self.layers.4.self_attn.k_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="616" name="Constant_170770" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="617" name="Reshape_170771" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="618" name="self.layers.4.self_attn.k_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="619" name="__module.layers.4.self_attn.k_proj/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="620" name="Constant_24478" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="39413344" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="621" name="__module.layers.4.self_attn.k_proj/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="566,tensor.25">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="622" name="Constant_25591" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="623" name="__module.layers.4.self_attn/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="568">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="624" name="__module.layers.4.self_attn/aten::transpose/ScatterElementsUpdate_1" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="625" name="__module.layers.4.self_attn/aten::transpose/Transpose_1" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="569">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="626" name="__module.layers.4.self_attn/aten::reshape/Reshape" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="589,key_states.19">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="627" name="__module.layers.4.self_attn/aten::bmm/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>48</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="594,attn_weights.17">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="628" name="__module.layers.4.self_attn/prim::ListConstruct/Reshape_1" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="24238344" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="629" name="ShapeOf_25560" type="ShapeOf" version="opset3">
<data output_type="i64" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="630" name="Constant_25561" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="631" name="Constant_25562" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="632" name="Gather_25563" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>3</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64" names="592">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="633" name="__module.layers.4.self_attn/prim::ListConstruct/Concat_4" 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="595">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="634" name="__module.layers.4.self_attn/aten::view/Reshape_3" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<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="FP32" names="596">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="635" name="__module.layers.4.self_attn/aten::add/Add" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</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="597,attn_weights.19">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="636" name="__module.layers.4.self_attn/prim::ListConstruct/Concat_5" 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>
</input>
<output>
<port id="3" precision="I64" names="600">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="637" name="__module.layers.4.self_attn/aten::view/Reshape_4" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="601,input.43">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="638" name="__module.layers.4.self_attn/aten::softmax/Softmax" type="SoftMax" version="opset8">
<data axis="-1" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</input>
<output>
<port id="1" precision="FP32" names="602,input.45">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="639" name="self.layers.4.self_attn.v_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="39416416" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="640" name="Convert_170752" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="641" name="self.layers.4.self_attn.v_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="39711328" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="642" name="Convert_170754" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="643" name="self.layers.4.self_attn.v_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="644" name="self.layers.4.self_attn.v_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="39713632" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="645" name="self.layers.4.self_attn.v_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="646" name="Constant_170759" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="647" name="Reshape_170760" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="648" name="self.layers.4.self_attn.v_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="649" name="__module.layers.4.self_attn.v_proj/aten::linear/MatMul" type="MatMul" version="opset1">
<data transpose_a="false" transpose_b="true" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>768</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="650" name="Constant_24479" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="39722848" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="651" name="__module.layers.4.self_attn.v_proj/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="573,tensor.27">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="652" name="Constant_25592" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="653" name="__module.layers.4.self_attn/aten::view/Reshape_5" 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="575">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="654" name="__module.layers.4.self_attn/aten::transpose/ScatterElementsUpdate_3" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="655" name="__module.layers.4.self_attn/aten::transpose/Transpose_3" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="576">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="656" name="__module.layers.4.self_attn/aten::reshape/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="591">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="657" name="__module.layers.4.self_attn/aten::bmm/MatMul_1" type="MatMul" version="opset1">
<data transpose_a="false" transpose_b="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="604,attn_output.25">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="658" name="__module.layers.4.self_attn/aten::view/Reshape_6" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="606,attn_output.27">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="659" name="__module.layers.4.self_attn/aten::transpose/ScatterElementsUpdate_4" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="660" name="__module.layers.4.self_attn/aten::transpose/Transpose_4" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="607,attn_output.29">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="661" name="Constant_25593" type="Const" version="opset1">
<data element_type="i64" shape="3" offset="24857400" size="24" />
<output>
<port id="0" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="662" name="__module.layers.4.self_attn/aten::reshape/Reshape_2" type="Reshape" version="opset1">
<data special_zero="true" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="609">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="663" name="self.layers.4.self_attn.out_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="39725920" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="664" name="Convert_170785" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="665" name="self.layers.4.self_attn.out_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="40020832" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="666" name="Convert_170787" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="667" name="self.layers.4.self_attn.out_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="668" name="self.layers.4.self_attn.out_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="40023136" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="669" name="self.layers.4.self_attn.out_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="670" name="Constant_170792" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="671" name="Reshape_170793" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="672" name="self.layers.4.self_attn.out_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="673" name="__module.layers.4.self_attn.out_proj/aten::linear/MatMul" type="MatMul" version="opset1">
<data transpose_a="false" transpose_b="true" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>768</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="674" name="Constant_24480" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="40032352" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="675" name="__module.layers.4.self_attn.out_proj/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="612,input.47">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="676" name="__module.layers.4/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="614">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="677" name="__module.layers.4.self_attn_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="678" name="__module.layers.4.self_attn_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="679" name="Constant_24481" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="40035424" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="680" name="__module.layers.4.self_attn_layer_norm/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="681" name="Constant_24482" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="40038496" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="682" name="__module.layers.4.self_attn_layer_norm/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="618,residual.9">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="683" name="self.layers.4.fc1.weight" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 128" offset="40041568" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="684" name="Convert_170521" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="685" name="self.layers.4.fc1.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 1" offset="41221216" size="9216" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="686" name="Convert_170523" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="687" name="self.layers.4.fc1.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="688" name="self.layers.4.fc1.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="3072, 6, 1" offset="41230432" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="689" name="self.layers.4.fc1.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="690" name="Constant_170528" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="26398800" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="691" name="Reshape_170529" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="692" name="self.layers.4.fc1.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="693" name="__module.layers.4.fc1/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="694" name="Constant_24483" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 3072" offset="41267296" size="12288" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="695" name="__module.layers.4.fc1/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="621">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="696" name="__module.layers.4.activation_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="622,input.49">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="697" name="self.layers.4.fc2.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 128" offset="41279584" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="698" name="Convert_170532" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="699" name="self.layers.4.fc2.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 24, 1" offset="42459232" size="9216" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="700" name="Convert_170534" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="701" name="self.layers.4.fc2.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="702" name="self.layers.4.fc2.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 24, 1" offset="42468448" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="703" name="self.layers.4.fc2.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="704" name="Constant_170539" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="27636824" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="705" name="Reshape_170540" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>24</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="706" name="self.layers.4.fc2.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="707" name="__module.layers.4.fc2/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="708" name="Constant_24484" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="42505312" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="709" name="__module.layers.4.fc2/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="626,input.51">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="710" name="__module.layers.4/aten::add/Add_1" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="628">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="711" name="__module.layers.4.final_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="712" name="__module.layers.4.final_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="713" name="Constant_24485" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="42508384" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="714" name="__module.layers.4.final_layer_norm/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="715" name="Constant_24486" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="42511456" 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.layers.4.final_layer_norm/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="632,hidden_states.31">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="717" name="Constant_24297" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="42514528" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="718" name="Convert_170818" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="719" name="Constant_24297/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="42809440" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="720" name="Convert_170820" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="721" name="Constant_24297/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="722" name="Constant_24297/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="42811744" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="723" name="Constant_24297/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="724" name="Constant_170825" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="725" name="Reshape_170826" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="726" name="Constant_24297/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="727" name="Multiply_24277" 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="728" name="Constant_24487" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="42820960" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="729" name="__module.layers.5.self_attn/aten::mul/Multiply" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="665,tensor">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="730" name="Constant_25594" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="731" name="__module.layers.5.self_attn/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="685">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="732" name="__module.layers.5.self_attn/aten::transpose/ScatterElementsUpdate" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="733" name="__module.layers.5.self_attn/aten::transpose/Transpose" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="686">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="734" name="__module.layers.5.self_attn/aten::view/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="689">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="735" name="self.layers.5.self_attn.k_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="42824032" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="736" name="Convert_170807" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="737" name="self.layers.5.self_attn.k_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="43118944" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="738" name="Convert_170809" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="739" name="self.layers.5.self_attn.k_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="740" name="self.layers.5.self_attn.k_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="43121248" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="741" name="self.layers.5.self_attn.k_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="742" name="Constant_170814" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="743" name="Reshape_170815" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="744" name="self.layers.5.self_attn.k_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="1" precision="FP32">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="745" name="__module.layers.5.self_attn.k_proj/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="746" name="Constant_24488" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="43130464" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="747" name="__module.layers.5.self_attn.k_proj/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="668,tensor.31">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="748" name="Constant_25595" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="749" name="__module.layers.5.self_attn/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="670">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="750" name="__module.layers.5.self_attn/aten::transpose/ScatterElementsUpdate_1" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="751" name="__module.layers.5.self_attn/aten::transpose/Transpose_1" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="671">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="752" name="__module.layers.5.self_attn/aten::reshape/Reshape" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="691,key_states">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="753" name="__module.layers.5.self_attn/aten::bmm/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>48</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="696,attn_weights.21">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="754" name="__module.layers.5.self_attn/prim::ListConstruct/Reshape_1" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="24238344" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="755" name="ShapeOf_25568" type="ShapeOf" version="opset3">
<data output_type="i64" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="756" name="Constant_25569" type="Const" version="opset1">
<data element_type="i64" shape="1" offset="23922616" size="8" />
<output>
<port id="0" precision="I64">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="757" name="Constant_25570" type="Const" version="opset1">
<data element_type="i64" shape="" offset="23922608" size="8" />
<output>
<port id="0" precision="I64" />
</output>
</layer>
<layer id="758" name="Gather_25571" type="Gather" version="opset8">
<data batch_dims="0" />
<input>
<port id="0" precision="I64">
<dim>3</dim>
</port>
<port id="1" precision="I64">
<dim>1</dim>
</port>
<port id="2" precision="I64" />
</input>
<output>
<port id="3" precision="I64" names="694">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="759" name="__module.layers.5.self_attn/prim::ListConstruct/Concat_4" 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="697">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="760" name="__module.layers.5.self_attn/aten::view/Reshape_3" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<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="FP32" names="698">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="761" name="__module.layers.5.self_attn/aten::add/Add" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</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="699,attn_weights">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="762" name="__module.layers.5.self_attn/prim::ListConstruct/Concat_5" 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>
</input>
<output>
<port id="3" precision="I64" names="702">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="763" name="__module.layers.5.self_attn/aten::view/Reshape_4" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="703,input.53">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="764" name="__module.layers.5.self_attn/aten::softmax/Softmax" type="SoftMax" version="opset8">
<data axis="-1" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</input>
<output>
<port id="1" precision="FP32" names="704,input.55">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
</output>
</layer>
<layer id="765" name="self.layers.5.self_attn.v_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="43133536" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="766" name="Convert_170796" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="767" name="self.layers.5.self_attn.v_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="43428448" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="768" name="Convert_170798" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="769" name="self.layers.5.self_attn.v_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="770" name="self.layers.5.self_attn.v_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="43430752" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="771" name="self.layers.5.self_attn.v_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="772" name="Constant_170803" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="773" name="Reshape_170804" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="774" name="self.layers.5.self_attn.v_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="775" name="__module.layers.5.self_attn.v_proj/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="776" name="Constant_24489" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="43439968" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="777" name="__module.layers.5.self_attn.v_proj/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="675,tensor.33">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="778" name="Constant_25596" type="Const" version="opset1">
<data element_type="i64" shape="4" offset="24238296" size="32" />
<output>
<port id="0" precision="I64">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="779" name="__module.layers.5.self_attn/aten::view/Reshape_5" 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="677">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="780" name="__module.layers.5.self_attn/aten::transpose/ScatterElementsUpdate_3" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="781" name="__module.layers.5.self_attn/aten::transpose/Transpose_3" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="678">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="782" name="__module.layers.5.self_attn/aten::reshape/Reshape_1" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="693">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="783" name="__module.layers.5.self_attn/aten::bmm/MatMul_1" type="MatMul" version="opset1">
<data transpose_a="false" transpose_b="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>-1</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="706,attn_output.31">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="784" name="__module.layers.5.self_attn/aten::view/Reshape_6" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="708,attn_output.33">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="785" name="__module.layers.5.self_attn/aten::transpose/ScatterElementsUpdate_4" type="Const" version="opset1">
<data element_type="i32" shape="4" offset="24238328" size="16" />
<output>
<port id="0" precision="I32">
<dim>4</dim>
</port>
</output>
</layer>
<layer id="786" name="__module.layers.5.self_attn/aten::transpose/Transpose_4" type="Transpose" version="opset1">
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>16</dim>
<dim>-1</dim>
<dim>48</dim>
</port>
<port id="1" precision="I32">
<dim>4</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="709,attn_output">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
</output>
</layer>
<layer id="787" name="Constant_25597" type="Const" version="opset1">
<data element_type="i64" shape="3" offset="24857400" size="24" />
<output>
<port id="0" precision="I64">
<dim>3</dim>
</port>
</output>
</layer>
<layer id="788" name="__module.layers.5.self_attn/aten::reshape/Reshape_2" type="Reshape" version="opset1">
<data special_zero="true" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>16</dim>
<dim>48</dim>
</port>
<port id="1" precision="I64">
<dim>3</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="711">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="789" name="self.layers.5.self_attn.out_proj.weight" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 128" offset="43443040" size="294912" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="790" name="Convert_170829" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="791" name="self.layers.5.self_attn.out_proj.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="768, 6, 1" offset="43737952" size="2304" />
<output>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="792" name="Convert_170831" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="793" name="self.layers.5.self_attn.out_proj.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="794" name="self.layers.5.self_attn.out_proj.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 6, 1" offset="43740256" size="9216" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="795" name="self.layers.5.self_attn.out_proj.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="796" name="Constant_170836" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="24235216" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="797" name="Reshape_170837" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="798" name="self.layers.5.self_attn.out_proj.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="799" name="__module.layers.5.self_attn.out_proj/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="800" name="Constant_24490" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="43749472" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="801" name="__module.layers.5.self_attn.out_proj/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="714,input.57">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="802" name="__module.layers.5/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="716">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="803" name="__module.layers.5.self_attn_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="804" name="__module.layers.5.self_attn_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="805" name="Constant_24491" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="43752544" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="806" name="__module.layers.5.self_attn_layer_norm/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="807" name="Constant_24492" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="43755616" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="808" name="__module.layers.5.self_attn_layer_norm/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="720,residual">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="809" name="self.layers.5.fc1.weight" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 128" offset="43758688" size="1179648" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="810" name="Convert_170543" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="811" name="self.layers.5.fc1.weight/zero_point" type="Const" version="opset1">
<data element_type="u4" shape="3072, 6, 1" offset="44938336" size="9216" />
<output>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="812" name="Convert_170545" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U4">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="813" name="self.layers.5.fc1.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="814" name="self.layers.5.fc1.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="3072, 6, 1" offset="44947552" size="36864" />
<output>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="815" name="self.layers.5.fc1.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
</output>
</layer>
<layer id="816" name="Constant_170550" type="Const" version="opset1">
<data element_type="i32" shape="2" offset="26398800" size="8" />
<output>
<port id="0" precision="I32">
<dim>2</dim>
</port>
</output>
</layer>
<layer id="817" name="Reshape_170551" type="Reshape" version="opset1">
<data special_zero="false" />
<input>
<port id="0" precision="FP16">
<dim>3072</dim>
<dim>6</dim>
<dim>128</dim>
</port>
<port id="1" precision="I32">
<dim>2</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>3072</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="818" name="self.layers.5.fc1.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="819" name="__module.layers.5.fc1/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="820" name="Constant_24493" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 3072" offset="44984416" size="12288" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="821" name="__module.layers.5.fc1/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="723">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="822" name="__module.layers.5.activation_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="724,input.59">
<dim>-1</dim>
<dim>-1</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="823" name="self.layers.5.fc2.weight" type="Const" version="opset1">
<data element_type="u8" shape="768, 3072" offset="44996704" size="2359296" />
<output>
<port id="0" precision="U8">
<dim>768</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="824" name="Convert_170555" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U8">
<dim>768</dim>
<dim>3072</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="825" name="self.layers.5.fc2.weight/zero_point" type="Const" version="opset1">
<data element_type="u8" shape="768, 1" offset="47356000" size="768" />
<output>
<port id="0" precision="U8">
<dim>768</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="826" name="Convert_170558" type="Convert" version="opset1">
<data destination_type="f16" />
<input>
<port id="0" precision="U8">
<dim>768</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="827" name="self.layers.5.fc2.weight/zero_point/subtract" type="Subtract" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="828" name="self.layers.5.fc2.weight/scale" type="Const" version="opset1">
<data element_type="f16" shape="768, 1" offset="47356768" size="1536" />
<output>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>1</dim>
</port>
</output>
</layer>
<layer id="829" name="self.layers.5.fc2.weight/fq_weights_1" type="Multiply" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
<port id="1" precision="FP16">
<dim>768</dim>
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP16">
<dim>768</dim>
<dim>3072</dim>
</port>
</output>
</layer>
<layer id="830" name="self.layers.5.fc2.weight/fq_weights_1/convert" type="Convert" version="opset1">
<data destination_type="f32" />
<input>
<port id="0" precision="FP16">
<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="831" name="__module.layers.5.fc2/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="832" name="Constant_24494" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="47358304" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="833" name="__module.layers.5.fc2/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="728,input">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="834" name="__module.layers.5/aten::add/Add_1" type="Add" version="opset1">
<data auto_broadcast="numpy" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</input>
<output>
<port id="2" precision="FP32" names="730">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="835" name="__module.layers.5.final_layer_norm/aten::layer_norm/Multiply" type="Const" version="opset1">
<data element_type="i32" shape="1" offset="23922636" size="4" />
<output>
<port id="0" precision="I32">
<dim>1</dim>
</port>
</output>
</layer>
<layer id="836" name="__module.layers.5.final_layer_norm/aten::layer_norm/MVN" type="MVN" version="opset6">
<data eps="9.9999997473787516e-06" normalize_variance="true" eps_mode="INSIDE_SQRT" />
<input>
<port id="0" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
<port id="1" precision="I32">
<dim>1</dim>
</port>
</input>
<output>
<port id="2" precision="FP32">
<dim>-1</dim>
<dim>-1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="837" name="Constant_24495" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="47361376" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="838" name="__module.layers.5.final_layer_norm/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="839" name="Constant_24496" type="Const" version="opset1">
<data element_type="f32" shape="1, 1, 768" offset="47364448" size="3072" />
<output>
<port id="0" precision="FP32">
<dim>1</dim>
<dim>1</dim>
<dim>768</dim>
</port>
</output>
</layer>
<layer id="840" name="__module.layers.5.final_layer_norm/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="841" name="Result_5251" 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="110" to-port="0" />
<edge from-layer="0" from-port="0" to-layer="113" to-port="0" />
<edge from-layer="1" from-port="0" to-layer="11" to-port="0" />
<edge from-layer="1" from-port="0" to-layer="24" to-port="0" />
<edge from-layer="2" from-port="0" to-layer="3" to-port="0" />
<edge from-layer="3" from-port="1" to-layer="6" to-port="0" />
<edge from-layer="4" from-port="0" to-layer="5" to-port="0" />
<edge from-layer="5" from-port="1" to-layer="6" to-port="1" />
<edge from-layer="6" from-port="2" to-layer="8" to-port="0" />
<edge from-layer="7" from-port="0" to-layer="8" to-port="1" />
<edge from-layer="8" from-port="2" to-layer="9" to-port="0" />
<edge from-layer="9" from-port="1" to-layer="14" to-port="0" />
<edge from-layer="10" from-port="0" to-layer="11" to-port="1" />
<edge from-layer="11" from-port="2" to-layer="12" to-port="0" />
<edge from-layer="12" from-port="1" to-layer="14" to-port="1" />
<edge from-layer="13" from-port="0" to-layer="14" to-port="2" />
<edge from-layer="14" from-port="3" to-layer="45" to-port="0" />
<edge from-layer="15" from-port="0" to-layer="16" to-port="0" />
<edge from-layer="16" from-port="1" to-layer="19" to-port="0" />
<edge from-layer="17" from-port="0" to-layer="18" to-port="0" />
<edge from-layer="18" from-port="1" to-layer="19" to-port="1" />
<edge from-layer="19" from-port="2" 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="22" to-port="0" />
<edge from-layer="22" from-port="1" to-layer="44" to-port="0" />
<edge from-layer="23" from-port="0" to-layer="29" to-port="0" />
<edge from-layer="24" from-port="1" to-layer="37" to-port="0" />
<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="29" to-port="1" />
<edge from-layer="28" from-port="0" to-layer="29" to-port="2" />
<edge from-layer="29" from-port="3" to-layer="31" to-port="0" />
<edge from-layer="30" from-port="0" to-layer="31" to-port="1" />
<edge from-layer="31" from-port="2" to-layer="33" to-port="0" />
<edge from-layer="31" from-port="2" to-layer="41" to-port="0" />
<edge from-layer="32" from-port="0" to-layer="34" to-port="0" />
<edge from-layer="33" from-port="1" to-layer="34" to-port="1" />
<edge from-layer="34" from-port="2" to-layer="40" to-port="0" />
<edge from-layer="35" from-port="0" to-layer="37" to-port="1" />
<edge from-layer="36" from-port="0" to-layer="37" to-port="2" />
<edge from-layer="37" from-port="3" to-layer="39" to-port="0" />
<edge from-layer="38" from-port="0" to-layer="39" to-port="1" />
<edge from-layer="39" from-port="2" to-layer="40" to-port="1" />
<edge from-layer="40" from-port="2" to-layer="41" to-port="1" />
<edge from-layer="41" from-port="2" to-layer="42" to-port="0" />
<edge from-layer="42" from-port="1" to-layer="44" to-port="1" />
<edge from-layer="43" from-port="0" to-layer="44" to-port="2" />
<edge from-layer="44" from-port="3" to-layer="45" to-port="1" />
<edge from-layer="45" from-port="2" to-layer="47" to-port="0" />
<edge from-layer="46" from-port="0" to-layer="47" to-port="1" />
<edge from-layer="47" from-port="2" to-layer="49" to-port="0" />
<edge from-layer="48" from-port="0" to-layer="49" to-port="1" />
<edge from-layer="49" from-port="2" to-layer="51" to-port="0" />
<edge from-layer="50" from-port="0" to-layer="51" to-port="1" />
<edge from-layer="51" from-port="2" to-layer="172" to-port="0" />
<edge from-layer="51" from-port="2" to-layer="62" to-port="0" />
<edge from-layer="51" from-port="2" to-layer="69" to-port="0" />
<edge from-layer="51" from-port="2" to-layer="89" to-port="0" />
<edge from-layer="51" from-port="2" to-layer="144" to-port="0" />
<edge from-layer="52" from-port="0" to-layer="53" to-port="0" />
<edge from-layer="53" from-port="1" to-layer="56" to-port="0" />
<edge from-layer="54" from-port="0" to-layer="55" to-port="0" />
<edge from-layer="55" from-port="1" to-layer="56" to-port="1" />
<edge from-layer="56" from-port="2" to-layer="58" to-port="0" />
<edge from-layer="57" from-port="0" to-layer="58" to-port="1" />
<edge from-layer="58" from-port="2" to-layer="60" to-port="0" />
<edge from-layer="59" from-port="0" to-layer="60" to-port="1" />
<edge from-layer="60" from-port="2" to-layer="61" to-port="0" />
<edge from-layer="61" from-port="1" to-layer="62" to-port="1" />
<edge from-layer="62" from-port="2" to-layer="64" to-port="0" />
<edge from-layer="63" from-port="0" to-layer="64" to-port="1" />
<edge from-layer="64" from-port="2" to-layer="66" to-port="0" />
<edge from-layer="65" from-port="0" to-layer="66" to-port="1" />
<edge from-layer="66" from-port="2" to-layer="68" to-port="0" />
<edge from-layer="67" from-port="0" to-layer="68" to-port="1" />
<edge from-layer="68" from-port="2" to-layer="78" to-port="0" />
<edge from-layer="69" from-port="1" to-layer="72" to-port="0" />
<edge from-layer="69" from-port="1" to-layer="101" to-port="0" />
<edge from-layer="70" from-port="0" to-layer="72" to-port="1" />
<edge from-layer="71" from-port="0" to-layer="72" to-port="2" />
<edge from-layer="72" from-port="3" to-layer="381" to-port="0" />
<edge from-layer="72" from-port="3" to-layer="74" to-port="0" />
<edge from-layer="72" from-port="3" to-layer="153" to-port="0" />
<edge from-layer="72" from-port="3" to-layer="255" to-port="0" />
<edge from-layer="72" from-port="3" to-layer="106" to-port="0" />
<edge from-layer="72" from-port="3" to-layer="507" to-port="0" />
<edge from-layer="72" from-port="3" to-layer="759" to-port="0" />
<edge from-layer="72" from-port="3" to-layer="633" to-port="0" />
<edge from-layer="73" from-port="0" to-layer="74" to-port="1" />
<edge from-layer="74" from-port="2" to-layer="131" to-port="0" />
<edge from-layer="74" from-port="2" to-layer="384" to-port="0" />
<edge from-layer="74" from-port="2" to-layer="510" to-port="0" />
<edge from-layer="74" from-port="2" to-layer="636" to-port="0" />
<edge from-layer="74" from-port="2" to-layer="762" to-port="0" />
<edge from-layer="74" from-port="2" to-layer="258" to-port="0" />
<edge from-layer="74" from-port="2" to-layer="77" to-port="0" />
<edge from-layer="75" from-port="0" to-layer="77" to-port="1" />
<edge from-layer="76" from-port="0" to-layer="77" to-port="2" />
<edge from-layer="76" from-port="0" to-layer="153" to-port="3" />
<edge from-layer="77" from-port="3" to-layer="78" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="752" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="734" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="656" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="626" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="608" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="530" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="500" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="482" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="404" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="96" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="248" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="230" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="278" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="374" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="356" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="151" to-port="1" />
<edge from-layer="77" from-port="3" to-layer="782" to-port="1" />
<edge from-layer="78" from-port="2" to-layer="97" to-port="0" />
<edge from-layer="79" from-port="0" to-layer="80" to-port="0" />
<edge from-layer="80" from-port="1" to-layer="83" to-port="0" />
<edge from-layer="81" from-port="0" to-layer="82" to-port="0" />
<edge from-layer="82" from-port="1" to-layer="83" to-port="1" />
<edge from-layer="83" from-port="2" to-layer="85" to-port="0" />
<edge from-layer="84" from-port="0" to-layer="85" to-port="1" />
<edge from-layer="85" from-port="2" to-layer="87" to-port="0" />
<edge from-layer="86" from-port="0" to-layer="87" to-port="1" />
<edge from-layer="87" from-port="2" to-layer="88" to-port="0" />
<edge from-layer="88" from-port="1" to-layer="89" to-port="1" />
<edge from-layer="89" from-port="2" to-layer="91" to-port="0" />
<edge from-layer="90" from-port="0" to-layer="91" to-port="1" />
<edge from-layer="91" from-port="2" to-layer="93" 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="96" to-port="0" />
<edge from-layer="96" from-port="2" to-layer="97" to-port="1" />
<edge from-layer="96" from-port="2" to-layer="102" to-port="0" />
<edge from-layer="97" from-port="2" to-layer="107" to-port="0" />
<edge from-layer="98" from-port="0" to-layer="106" to-port="1" />
<edge from-layer="98" from-port="0" to-layer="153" to-port="1" />
<edge from-layer="99" from-port="0" to-layer="101" to-port="1" />
<edge from-layer="100" from-port="0" to-layer="101" to-port="2" />
<edge from-layer="101" from-port="3" to-layer="384" to-port="1" />
<edge from-layer="101" from-port="3" to-layer="633" to-port="2" />
<edge from-layer="101" from-port="3" to-layer="131" to-port="1" />
<edge from-layer="101" from-port="3" to-layer="762" to-port="1" />
<edge from-layer="101" from-port="3" to-layer="153" to-port="2" />
<edge from-layer="101" from-port="3" to-layer="759" to-port="2" />
<edge from-layer="101" from-port="3" to-layer="510" to-port="1" />
<edge from-layer="101" from-port="3" to-layer="258" to-port="1" />
<edge from-layer="101" from-port="3" to-layer="507" to-port="2" />
<edge from-layer="101" from-port="3" to-layer="381" to-port="2" />
<edge from-layer="101" from-port="3" to-layer="255" to-port="2" />
<edge from-layer="101" from-port="3" to-layer="106" to-port="2" />
<edge from-layer="101" from-port="3" to-layer="636" to-port="1" />
<edge from-layer="102" from-port="1" to-layer="105" to-port="0" />
<edge from-layer="103" from-port="0" to-layer="105" to-port="1" />
<edge from-layer="104" from-port="0" to-layer="105" to-port="2" />
<edge from-layer="105" from-port="3" to-layer="106" to-port="3" />
<edge from-layer="105" from-port="3" to-layer="131" to-port="2" />
<edge from-layer="106" from-port="4" to-layer="107" to-port="1" />
<edge from-layer="107" from-port="2" to-layer="130" to-port="0" />
<edge from-layer="108" from-port="0" to-layer="126" to-port="0" />
<edge from-layer="109" from-port="0" to-layer="110" to-port="1" />
<edge from-layer="110" from-port="2" 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="2" to-layer="122" to-port="0" />
<edge from-layer="113" from-port="1" to-layer="120" to-port="0" />
<edge from-layer="113" from-port="1" to-layer="116" to-port="0" />
<edge from-layer="114" from-port="0" to-layer="116" to-port="1" />
<edge from-layer="115" from-port="0" to-layer="116" to-port="2" />
<edge from-layer="116" from-port="3" to-layer="121" to-port="0" />
<edge from-layer="117" from-port="0" to-layer="121" to-port="1" />
<edge from-layer="118" from-port="0" to-layer="120" to-port="1" />
<edge from-layer="119" from-port="0" to-layer="120" to-port="2" />
<edge from-layer="120" from-port="3" to-layer="121" to-port="2" />
<edge from-layer="121" from-port="3" to-layer="122" to-port="1" />
<edge from-layer="122" from-port="2" to-layer="123" to-port="0" />
<edge from-layer="123" from-port="1" to-layer="125" to-port="0" />
<edge from-layer="124" from-port="0" to-layer="125" to-port="1" />
<edge from-layer="125" from-port="2" to-layer="126" to-port="1" />
<edge from-layer="126" from-port="2" to-layer="129" to-port="2" />
<edge from-layer="126" from-port="2" to-layer="127" to-port="0" />
<edge from-layer="127" from-port="1" to-layer="129" to-port="0" />
<edge from-layer="128" from-port="0" to-layer="129" to-port="1" />
<edge from-layer="129" from-port="3" to-layer="761" to-port="1" />
<edge from-layer="129" from-port="3" to-layer="257" to-port="1" />
<edge from-layer="129" from-port="3" to-layer="130" to-port="1" />
<edge from-layer="129" from-port="3" to-layer="383" to-port="1" />
<edge from-layer="129" from-port="3" to-layer="635" to-port="1" />
<edge from-layer="129" from-port="3" to-layer="509" to-port="1" />
<edge from-layer="130" from-port="2" to-layer="132" to-port="0" />
<edge from-layer="131" from-port="3" to-layer="132" to-port="1" />
<edge from-layer="132" from-port="2" to-layer="133" to-port="0" />
<edge from-layer="133" from-port="1" to-layer="152" to-port="0" />
<edge from-layer="134" from-port="0" to-layer="135" to-port="0" />
<edge from-layer="135" from-port="1" to-layer="138" to-port="0" />
<edge from-layer="136" from-port="0" to-layer="137" to-port="0" />
<edge from-layer="137" from-port="1" to-layer="138" to-port="1" />
<edge from-layer="138" from-port="2" to-layer="140" to-port="0" />
<edge from-layer="139" from-port="0" to-layer="140" to-port="1" />
<edge from-layer="140" from-port="2" to-layer="142" to-port="0" />
<edge from-layer="141" from-port="0" to-layer="142" to-port="1" />
<edge from-layer="142" from-port="2" to-layer="143" to-port="0" />
<edge from-layer="143" from-port="1" to-layer="144" to-port="1" />
<edge from-layer="144" from-port="2" to-layer="146" to-port="0" />
<edge from-layer="145" from-port="0" to-layer="146" to-port="1" />
<edge from-layer="146" from-port="2" to-layer="148" to-port="0" />
<edge from-layer="147" from-port="0" to-layer="148" to-port="1" />
<edge from-layer="148" from-port="2" to-layer="150" to-port="0" />
<edge from-layer="149" from-port="0" to-layer="150" to-port="1" />
<edge from-layer="150" from-port="2" to-layer="151" to-port="0" />
<edge from-layer="151" from-port="2" to-layer="152" to-port="1" />
<edge from-layer="152" from-port="2" to-layer="154" to-port="0" />
<edge from-layer="153" from-port="4" to-layer="280" to-port="1" />
<edge from-layer="153" from-port="4" to-layer="784" to-port="1" />
<edge from-layer="153" from-port="4" to-layer="532" to-port="1" />
<edge from-layer="153" from-port="4" to-layer="154" to-port="1" />
<edge from-layer="153" from-port="4" to-layer="658" to-port="1" />
<edge from-layer="153" from-port="4" to-layer="406" to-port="1" />
<edge from-layer="154" from-port="2" to-layer="156" to-port="0" />
<edge from-layer="155" from-port="0" to-layer="156" to-port="1" />
<edge from-layer="156" from-port="2" to-layer="158" to-port="0" />
<edge from-layer="157" from-port="0" to-layer="158" to-port="1" />
<edge from-layer="158" from-port="2" to-layer="169" to-port="0" />
<edge from-layer="159" from-port="0" to-layer="160" to-port="0" />
<edge from-layer="160" from-port="1" to-layer="163" to-port="0" />
<edge from-layer="161" from-port="0" to-layer="162" to-port="0" />
<edge from-layer="162" from-port="1" to-layer="163" to-port="1" />
<edge from-layer="163" from-port="2" to-layer="165" to-port="0" />
<edge from-layer="164" from-port="0" to-layer="165" to-port="1" />
<edge from-layer="165" from-port="2" to-layer="167" to-port="0" />
<edge from-layer="166" from-port="0" to-layer="167" to-port="1" />
<edge from-layer="167" from-port="2" to-layer="168" to-port="0" />
<edge from-layer="168" from-port="1" to-layer="169" to-port="1" />
<edge from-layer="169" from-port="2" to-layer="171" to-port="0" />
<edge from-layer="170" from-port="0" to-layer="171" to-port="1" />
<edge from-layer="171" from-port="2" to-layer="172" to-port="1" />
<edge from-layer="172" from-port="2" to-layer="174" to-port="0" />
<edge from-layer="173" from-port="0" to-layer="174" to-port="1" />
<edge from-layer="174" from-port="2" to-layer="176" to-port="0" />
<edge from-layer="175" from-port="0" to-layer="176" to-port="1" />
<edge from-layer="176" from-port="2" to-layer="178" to-port="0" />
<edge from-layer="177" from-port="0" to-layer="178" to-port="1" />
<edge from-layer="178" from-port="2" to-layer="189" to-port="0" />
<edge from-layer="178" from-port="2" to-layer="206" to-port="0" />
<edge from-layer="179" from-port="0" to-layer="180" to-port="0" />
<edge from-layer="180" from-port="1" to-layer="183" to-port="0" />
<edge from-layer="181" from-port="0" to-layer="182" to-port="0" />
<edge from-layer="182" from-port="1" to-layer="183" to-port="1" />
<edge from-layer="183" from-port="2" to-layer="185" to-port="0" />
<edge from-layer="184" from-port="0" to-layer="185" to-port="1" />
<edge from-layer="185" from-port="2" to-layer="187" to-port="0" />
<edge from-layer="186" from-port="0" to-layer="187" to-port="1" />
<edge from-layer="187" from-port="2" to-layer="188" to-port="0" />
<edge from-layer="188" from-port="1" to-layer="189" to-port="1" />
<edge from-layer="189" from-port="2" to-layer="191" to-port="0" />
<edge from-layer="190" from-port="0" to-layer="191" to-port="1" />
<edge from-layer="191" from-port="2" to-layer="192" to-port="0" />
<edge from-layer="192" from-port="1" to-layer="203" to-port="0" />
<edge from-layer="193" from-port="0" to-layer="194" to-port="0" />
<edge from-layer="194" from-port="1" to-layer="197" to-port="0" />
<edge from-layer="195" from-port="0" to-layer="196" to-port="0" />
<edge from-layer="196" from-port="1" to-layer="197" to-port="1" />
<edge from-layer="197" from-port="2" to-layer="199" to-port="0" />
<edge from-layer="198" from-port="0" to-layer="199" to-port="1" />
<edge from-layer="199" from-port="2" to-layer="201" to-port="0" />
<edge from-layer="200" from-port="0" to-layer="201" to-port="1" />
<edge from-layer="201" from-port="2" to-layer="202" to-port="0" />
<edge from-layer="202" from-port="1" to-layer="203" to-port="1" />
<edge from-layer="203" from-port="2" to-layer="205" to-port="0" />
<edge from-layer="204" from-port="0" to-layer="205" to-port="1" />
<edge from-layer="205" from-port="2" to-layer="206" to-port="1" />
<edge from-layer="206" from-port="2" to-layer="208" to-port="0" />
<edge from-layer="207" from-port="0" to-layer="208" to-port="1" />
<edge from-layer="208" from-port="2" to-layer="210" to-port="0" />
<edge from-layer="209" from-port="0" to-layer="210" to-port="1" />
<edge from-layer="210" from-port="2" to-layer="212" to-port="0" />
<edge from-layer="211" from-port="0" to-layer="212" to-port="1" />
<edge from-layer="212" from-port="2" to-layer="241" to-port="0" />
<edge from-layer="212" from-port="2" to-layer="223" to-port="0" />
<edge from-layer="212" from-port="2" to-layer="298" to-port="0" />
<edge from-layer="212" from-port="2" to-layer="271" to-port="0" />
<edge from-layer="213" from-port="0" to-layer="214" to-port="0" />
<edge from-layer="214" from-port="1" to-layer="217" to-port="0" />
<edge from-layer="215" from-port="0" to-layer="216" to-port="0" />
<edge from-layer="216" from-port="1" to-layer="217" to-port="1" />
<edge from-layer="217" from-port="2" to-layer="219" to-port="0" />
<edge from-layer="218" from-port="0" to-layer="219" to-port="1" />
<edge from-layer="219" from-port="2" to-layer="221" to-port="0" />
<edge from-layer="220" from-port="0" to-layer="221" to-port="1" />
<edge from-layer="221" from-port="2" to-layer="222" to-port="0" />
<edge from-layer="222" from-port="1" 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="0" />
<edge from-layer="230" from-port="2" to-layer="249" to-port="0" />
<edge from-layer="231" from-port="0" to-layer="232" to-port="0" />
<edge from-layer="232" from-port="1" to-layer="235" to-port="0" />
<edge from-layer="233" from-port="0" to-layer="234" to-port="0" />
<edge from-layer="234" from-port="1" to-layer="235" to-port="1" />
<edge from-layer="235" from-port="2" to-layer="237" to-port="0" />
<edge from-layer="236" from-port="0" to-layer="237" to-port="1" />
<edge from-layer="237" from-port="2" to-layer="239" to-port="0" />
<edge from-layer="238" from-port="0" to-layer="239" to-port="1" />
<edge from-layer="239" from-port="2" to-layer="240" to-port="0" />
<edge from-layer="240" from-port="1" to-layer="241" to-port="1" />
<edge from-layer="241" from-port="2" to-layer="243" to-port="0" />
<edge from-layer="242" from-port="0" to-layer="243" to-port="1" />
<edge from-layer="243" from-port="2" to-layer="245" to-port="0" />
<edge from-layer="244" from-port="0" to-layer="245" to-port="1" />
<edge from-layer="245" from-port="2" to-layer="247" to-port="0" />
<edge from-layer="246" from-port="0" to-layer="247" to-port="1" />
<edge from-layer="247" from-port="2" to-layer="248" to-port="0" />
<edge from-layer="248" from-port="2" to-layer="249" to-port="1" />
<edge from-layer="248" from-port="2" to-layer="251" to-port="0" />
<edge from-layer="249" from-port="2" to-layer="256" to-port="0" />
<edge from-layer="250" from-port="0" to-layer="255" to-port="1" />
<edge from-layer="251" from-port="1" to-layer="254" to-port="0" />
<edge from-layer="252" from-port="0" to-layer="254" to-port="1" />
<edge from-layer="253" from-port="0" to-layer="254" to-port="2" />
<edge from-layer="254" from-port="3" to-layer="255" to-port="3" />
<edge from-layer="254" from-port="3" to-layer="258" to-port="2" />
<edge from-layer="255" from-port="4" to-layer="256" to-port="1" />
<edge from-layer="256" from-port="2" to-layer="257" to-port="0" />
<edge from-layer="257" from-port="2" to-layer="259" to-port="0" />
<edge from-layer="258" from-port="3" to-layer="259" to-port="1" />
<edge from-layer="259" from-port="2" to-layer="260" to-port="0" />
<edge from-layer="260" from-port="1" to-layer="279" to-port="0" />
<edge from-layer="261" from-port="0" to-layer="262" to-port="0" />
<edge from-layer="262" from-port="1" to-layer="265" to-port="0" />
<edge from-layer="263" from-port="0" to-layer="264" to-port="0" />
<edge from-layer="264" from-port="1" to-layer="265" to-port="1" />
<edge from-layer="265" from-port="2" to-layer="267" to-port="0" />
<edge from-layer="266" from-port="0" to-layer="267" to-port="1" />
<edge from-layer="267" from-port="2" to-layer="269" to-port="0" />
<edge from-layer="268" from-port="0" to-layer="269" to-port="1" />
<edge from-layer="269" from-port="2" to-layer="270" to-port="0" />
<edge from-layer="270" from-port="1" 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="275" to-port="0" />
<edge from-layer="274" from-port="0" to-layer="275" to-port="1" />
<edge from-layer="275" from-port="2" to-layer="277" to-port="0" />
<edge from-layer="276" from-port="0" to-layer="277" to-port="1" />
<edge from-layer="277" from-port="2" to-layer="278" to-port="0" />
<edge from-layer="278" from-port="2" to-layer="279" to-port="1" />
<edge from-layer="279" from-port="2" to-layer="280" to-port="0" />
<edge from-layer="280" from-port="2" to-layer="282" to-port="0" />
<edge from-layer="281" from-port="0" to-layer="282" to-port="1" />
<edge from-layer="282" from-port="2" to-layer="284" to-port="0" />
<edge from-layer="283" from-port="0" to-layer="284" to-port="1" />
<edge from-layer="284" from-port="2" to-layer="295" to-port="0" />
<edge from-layer="285" from-port="0" to-layer="286" to-port="0" />
<edge from-layer="286" from-port="1" to-layer="289" to-port="0" />
<edge from-layer="287" from-port="0" to-layer="288" to-port="0" />
<edge from-layer="288" from-port="1" to-layer="289" to-port="1" />
<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="294" to-port="0" />
<edge from-layer="294" from-port="1" to-layer="295" 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="298" to-port="1" />
<edge from-layer="298" from-port="2" to-layer="300" to-port="0" />
<edge from-layer="299" from-port="0" to-layer="300" to-port="1" />
<edge from-layer="300" from-port="2" to-layer="302" to-port="0" />
<edge from-layer="301" from-port="0" to-layer="302" to-port="1" />
<edge from-layer="302" from-port="2" to-layer="304" to-port="0" />
<edge from-layer="303" from-port="0" to-layer="304" to-port="1" />
<edge from-layer="304" from-port="2" to-layer="332" to-port="0" />
<edge from-layer="304" from-port="2" to-layer="315" to-port="0" />
<edge from-layer="305" from-port="0" to-layer="306" to-port="0" />
<edge from-layer="306" from-port="1" to-layer="309" to-port="0" />
<edge from-layer="307" from-port="0" to-layer="308" to-port="0" />
<edge from-layer="308" from-port="1" to-layer="309" to-port="1" />
<edge from-layer="309" from-port="2" to-layer="311" to-port="0" />
<edge from-layer="310" from-port="0" to-layer="311" to-port="1" />
<edge from-layer="311" from-port="2" to-layer="313" to-port="0" />
<edge from-layer="312" from-port="0" to-layer="313" to-port="1" />
<edge from-layer="313" from-port="2" to-layer="314" to-port="0" />
<edge from-layer="314" from-port="1" to-layer="315" to-port="1" />
<edge from-layer="315" from-port="2" to-layer="317" to-port="0" />
<edge from-layer="316" from-port="0" to-layer="317" to-port="1" />
<edge from-layer="317" from-port="2" to-layer="318" to-port="0" />
<edge from-layer="318" from-port="1" to-layer="329" 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="323" to-port="0" />
<edge from-layer="321" from-port="0" to-layer="322" to-port="0" />
<edge from-layer="322" from-port="1" 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="327" to-port="0" />
<edge from-layer="326" from-port="0" to-layer="327" to-port="1" />
<edge from-layer="327" from-port="2" to-layer="328" to-port="0" />
<edge from-layer="328" from-port="1" to-layer="329" to-port="1" />
<edge from-layer="329" from-port="2" to-layer="331" to-port="0" />
<edge from-layer="330" from-port="0" to-layer="331" to-port="1" />
<edge from-layer="331" from-port="2" to-layer="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="336" to-port="0" />
<edge from-layer="335" from-port="0" to-layer="336" to-port="1" />
<edge from-layer="336" from-port="2" to-layer="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="367" to-port="0" />
<edge from-layer="338" from-port="2" to-layer="424" to-port="0" />
<edge from-layer="338" from-port="2" to-layer="397" to-port="0" />
<edge from-layer="338" from-port="2" to-layer="349" to-port="0" />
<edge from-layer="339" from-port="0" to-layer="340" to-port="0" />
<edge from-layer="340" from-port="1" to-layer="343" to-port="0" />
<edge from-layer="341" from-port="0" to-layer="342" to-port="0" />
<edge from-layer="342" from-port="1" to-layer="343" to-port="1" />
<edge from-layer="343" from-port="2" to-layer="345" to-port="0" />
<edge from-layer="344" from-port="0" to-layer="345" to-port="1" />
<edge from-layer="345" from-port="2" to-layer="347" to-port="0" />
<edge from-layer="346" from-port="0" to-layer="347" to-port="1" />
<edge from-layer="347" from-port="2" to-layer="348" to-port="0" />
<edge from-layer="348" from-port="1" to-layer="349" to-port="1" />
<edge from-layer="349" from-port="2" to-layer="351" to-port="0" />
<edge from-layer="350" from-port="0" to-layer="351" to-port="1" />
<edge from-layer="351" from-port="2" to-layer="353" to-port="0" />
<edge from-layer="352" from-port="0" to-layer="353" to-port="1" />
<edge from-layer="353" from-port="2" to-layer="355" to-port="0" />
<edge from-layer="354" from-port="0" to-layer="355" to-port="1" />
<edge from-layer="355" from-port="2" to-layer="356" to-port="0" />
<edge from-layer="356" from-port="2" to-layer="375" to-port="0" />
<edge from-layer="357" from-port="0" to-layer="358" to-port="0" />
<edge from-layer="358" from-port="1" to-layer="361" to-port="0" />
<edge from-layer="359" from-port="0" to-layer="360" to-port="0" />
<edge from-layer="360" from-port="1" to-layer="361" to-port="1" />
<edge from-layer="361" from-port="2" to-layer="363" to-port="0" />
<edge from-layer="362" from-port="0" to-layer="363" to-port="1" />
<edge from-layer="363" from-port="2" 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="0" />
<edge from-layer="366" from-port="1" to-layer="367" to-port="1" />
<edge from-layer="367" from-port="2" to-layer="369" to-port="0" />
<edge from-layer="368" from-port="0" to-layer="369" to-port="1" />
<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="374" to-port="0" />
<edge from-layer="374" from-port="2" to-layer="375" to-port="1" />
<edge from-layer="374" from-port="2" to-layer="377" to-port="0" />
<edge from-layer="375" from-port="2" to-layer="382" to-port="0" />
<edge from-layer="376" from-port="0" to-layer="381" to-port="1" />
<edge from-layer="377" from-port="1" to-layer="380" to-port="0" />
<edge from-layer="378" from-port="0" to-layer="380" to-port="1" />
<edge from-layer="379" from-port="0" to-layer="380" to-port="2" />
<edge from-layer="380" from-port="3" to-layer="384" to-port="2" />
<edge from-layer="380" from-port="3" to-layer="381" to-port="3" />
<edge from-layer="381" from-port="4" to-layer="382" to-port="1" />
<edge from-layer="382" from-port="2" to-layer="383" to-port="0" />
<edge from-layer="383" from-port="2" to-layer="385" to-port="0" />
<edge from-layer="384" from-port="3" to-layer="385" to-port="1" />
<edge from-layer="385" from-port="2" to-layer="386" to-port="0" />
<edge from-layer="386" from-port="1" to-layer="405" to-port="0" />
<edge from-layer="387" from-port="0" to-layer="388" to-port="0" />
<edge from-layer="388" from-port="1" to-layer="391" to-port="0" />
<edge from-layer="389" from-port="0" to-layer="390" to-port="0" />
<edge from-layer="390" from-port="1" to-layer="391" to-port="1" />
<edge from-layer="391" from-port="2" to-layer="393" to-port="0" />
<edge from-layer="392" from-port="0" to-layer="393" to-port="1" />
<edge from-layer="393" from-port="2" to-layer="395" to-port="0" />
<edge from-layer="394" from-port="0" to-layer="395" to-port="1" />
<edge from-layer="395" from-port="2" to-layer="396" to-port="0" />
<edge from-layer="396" from-port="1" to-layer="397" to-port="1" />
<edge from-layer="397" from-port="2" to-layer="399" to-port="0" />
<edge from-layer="398" from-port="0" to-layer="399" to-port="1" />
<edge from-layer="399" from-port="2" 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="403" to-port="0" />
<edge from-layer="402" from-port="0" to-layer="403" to-port="1" />
<edge from-layer="403" from-port="2" to-layer="404" to-port="0" />
<edge from-layer="404" from-port="2" to-layer="405" to-port="1" />
<edge from-layer="405" from-port="2" to-layer="406" to-port="0" />
<edge from-layer="406" from-port="2" to-layer="408" to-port="0" />
<edge from-layer="407" from-port="0" to-layer="408" to-port="1" />
<edge from-layer="408" from-port="2" to-layer="410" to-port="0" />
<edge from-layer="409" from-port="0" to-layer="410" to-port="1" />
<edge from-layer="410" from-port="2" to-layer="421" to-port="0" />
<edge from-layer="411" from-port="0" to-layer="412" to-port="0" />
<edge from-layer="412" from-port="1" to-layer="415" to-port="0" />
<edge from-layer="413" from-port="0" to-layer="414" to-port="0" />
<edge from-layer="414" from-port="1" to-layer="415" to-port="1" />
<edge from-layer="415" from-port="2" to-layer="417" to-port="0" />
<edge from-layer="416" from-port="0" to-layer="417" to-port="1" />
<edge from-layer="417" from-port="2" to-layer="419" to-port="0" />
<edge from-layer="418" from-port="0" to-layer="419" to-port="1" />
<edge from-layer="419" from-port="2" to-layer="420" to-port="0" />
<edge from-layer="420" from-port="1" to-layer="421" to-port="1" />
<edge from-layer="421" from-port="2" to-layer="423" to-port="0" />
<edge from-layer="422" from-port="0" to-layer="423" to-port="1" />
<edge from-layer="423" from-port="2" to-layer="424" to-port="1" />
<edge from-layer="424" from-port="2" to-layer="426" to-port="0" />
<edge from-layer="425" from-port="0" to-layer="426" to-port="1" />
<edge from-layer="426" from-port="2" to-layer="428" to-port="0" />
<edge from-layer="427" from-port="0" to-layer="428" to-port="1" />
<edge from-layer="428" from-port="2" to-layer="430" to-port="0" />
<edge from-layer="429" from-port="0" to-layer="430" to-port="1" />
<edge from-layer="430" from-port="2" to-layer="441" to-port="0" />
<edge from-layer="430" from-port="2" to-layer="458" to-port="0" />
<edge from-layer="431" from-port="0" to-layer="432" to-port="0" />
<edge from-layer="432" from-port="1" to-layer="435" to-port="0" />
<edge from-layer="433" from-port="0" to-layer="434" to-port="0" />
<edge from-layer="434" from-port="1" to-layer="435" to-port="1" />
<edge from-layer="435" from-port="2" to-layer="437" to-port="0" />
<edge from-layer="436" from-port="0" to-layer="437" to-port="1" />
<edge from-layer="437" from-port="2" to-layer="439" to-port="0" />
<edge from-layer="438" from-port="0" to-layer="439" to-port="1" />
<edge from-layer="439" from-port="2" to-layer="440" to-port="0" />
<edge from-layer="440" from-port="1" to-layer="441" to-port="1" />
<edge from-layer="441" from-port="2" to-layer="443" to-port="0" />
<edge from-layer="442" from-port="0" to-layer="443" to-port="1" />
<edge from-layer="443" from-port="2" to-layer="444" to-port="0" />
<edge from-layer="444" from-port="1" to-layer="455" to-port="0" />
<edge from-layer="445" from-port="0" to-layer="446" to-port="0" />
<edge from-layer="446" from-port="1" to-layer="449" to-port="0" />
<edge from-layer="447" from-port="0" to-layer="448" to-port="0" />
<edge from-layer="448" from-port="1" to-layer="449" to-port="1" />
<edge from-layer="449" from-port="2" to-layer="451" to-port="0" />
<edge from-layer="450" from-port="0" to-layer="451" to-port="1" />
<edge from-layer="451" from-port="2" to-layer="453" to-port="0" />
<edge from-layer="452" from-port="0" to-layer="453" to-port="1" />
<edge from-layer="453" from-port="2" to-layer="454" to-port="0" />
<edge from-layer="454" from-port="1" 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="458" to-port="1" />
<edge from-layer="458" from-port="2" to-layer="460" to-port="0" />
<edge from-layer="459" from-port="0" to-layer="460" to-port="1" />
<edge from-layer="460" from-port="2" to-layer="462" to-port="0" />
<edge from-layer="461" from-port="0" to-layer="462" to-port="1" />
<edge from-layer="462" from-port="2" to-layer="464" to-port="0" />
<edge from-layer="463" from-port="0" to-layer="464" to-port="1" />
<edge from-layer="464" from-port="2" to-layer="475" to-port="0" />
<edge from-layer="464" from-port="2" to-layer="523" to-port="0" />
<edge from-layer="464" from-port="2" to-layer="550" to-port="0" />
<edge from-layer="464" from-port="2" to-layer="493" to-port="0" />
<edge from-layer="465" from-port="0" to-layer="466" to-port="0" />
<edge from-layer="466" from-port="1" to-layer="469" to-port="0" />
<edge from-layer="467" from-port="0" to-layer="468" to-port="0" />
<edge from-layer="468" from-port="1" to-layer="469" to-port="1" />
<edge from-layer="469" from-port="2" to-layer="471" to-port="0" />
<edge from-layer="470" from-port="0" to-layer="471" to-port="1" />
<edge from-layer="471" from-port="2" to-layer="473" to-port="0" />
<edge from-layer="472" from-port="0" to-layer="473" to-port="1" />
<edge from-layer="473" from-port="2" to-layer="474" to-port="0" />
<edge from-layer="474" from-port="1" to-layer="475" to-port="1" />
<edge from-layer="475" from-port="2" to-layer="477" to-port="0" />
<edge from-layer="476" from-port="0" to-layer="477" to-port="1" />
<edge from-layer="477" from-port="2" to-layer="479" to-port="0" />
<edge from-layer="478" from-port="0" to-layer="479" to-port="1" />
<edge from-layer="479" from-port="2" 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="0" />
<edge from-layer="482" from-port="2" to-layer="501" to-port="0" />
<edge from-layer="483" from-port="0" to-layer="484" to-port="0" />
<edge from-layer="484" from-port="1" to-layer="487" to-port="0" />
<edge from-layer="485" from-port="0" to-layer="486" to-port="0" />
<edge from-layer="486" from-port="1" 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="492" to-port="0" />
<edge from-layer="492" from-port="1" to-layer="493" to-port="1" />
<edge from-layer="493" from-port="2" to-layer="495" to-port="0" />
<edge from-layer="494" from-port="0" to-layer="495" to-port="1" />
<edge from-layer="495" from-port="2" to-layer="497" to-port="0" />
<edge from-layer="496" from-port="0" to-layer="497" to-port="1" />
<edge from-layer="497" from-port="2" to-layer="499" to-port="0" />
<edge from-layer="498" from-port="0" to-layer="499" to-port="1" />
<edge from-layer="499" from-port="2" to-layer="500" to-port="0" />
<edge from-layer="500" from-port="2" to-layer="501" to-port="1" />
<edge from-layer="500" from-port="2" to-layer="503" to-port="0" />
<edge from-layer="501" from-port="2" to-layer="508" to-port="0" />
<edge from-layer="502" from-port="0" to-layer="507" to-port="1" />
<edge from-layer="503" from-port="1" to-layer="506" to-port="0" />
<edge from-layer="504" from-port="0" to-layer="506" to-port="1" />
<edge from-layer="505" from-port="0" to-layer="506" to-port="2" />
<edge from-layer="506" from-port="3" to-layer="507" to-port="3" />
<edge from-layer="506" from-port="3" to-layer="510" to-port="2" />
<edge from-layer="507" from-port="4" to-layer="508" to-port="1" />
<edge from-layer="508" from-port="2" to-layer="509" to-port="0" />
<edge from-layer="509" from-port="2" to-layer="511" to-port="0" />
<edge from-layer="510" from-port="3" to-layer="511" to-port="1" />
<edge from-layer="511" from-port="2" to-layer="512" to-port="0" />
<edge from-layer="512" from-port="1" to-layer="531" to-port="0" />
<edge from-layer="513" from-port="0" to-layer="514" to-port="0" />
<edge from-layer="514" from-port="1" to-layer="517" to-port="0" />
<edge from-layer="515" from-port="0" to-layer="516" to-port="0" />
<edge from-layer="516" from-port="1" to-layer="517" to-port="1" />
<edge from-layer="517" from-port="2" to-layer="519" to-port="0" />
<edge from-layer="518" from-port="0" to-layer="519" to-port="1" />
<edge from-layer="519" from-port="2" to-layer="521" to-port="0" />
<edge from-layer="520" from-port="0" to-layer="521" to-port="1" />
<edge from-layer="521" from-port="2" to-layer="522" to-port="0" />
<edge from-layer="522" from-port="1" 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="528" from-port="0" to-layer="529" to-port="1" />
<edge from-layer="529" from-port="2" to-layer="530" to-port="0" />
<edge from-layer="530" from-port="2" to-layer="531" to-port="1" />
<edge from-layer="531" from-port="2" to-layer="532" to-port="0" />
<edge from-layer="532" from-port="2" to-layer="534" to-port="0" />
<edge from-layer="533" from-port="0" to-layer="534" to-port="1" />
<edge from-layer="534" from-port="2" to-layer="536" to-port="0" />
<edge from-layer="535" from-port="0" to-layer="536" to-port="1" />
<edge from-layer="536" from-port="2" to-layer="547" to-port="0" />
<edge from-layer="537" from-port="0" to-layer="538" to-port="0" />
<edge from-layer="538" from-port="1" to-layer="541" to-port="0" />
<edge from-layer="539" from-port="0" to-layer="540" to-port="0" />
<edge from-layer="540" from-port="1" to-layer="541" to-port="1" />
<edge from-layer="541" from-port="2" to-layer="543" to-port="0" />
<edge from-layer="542" from-port="0" to-layer="543" to-port="1" />
<edge from-layer="543" from-port="2" to-layer="545" to-port="0" />
<edge from-layer="544" from-port="0" to-layer="545" to-port="1" />
<edge from-layer="545" from-port="2" to-layer="546" to-port="0" />
<edge from-layer="546" from-port="1" to-layer="547" to-port="1" />
<edge from-layer="547" from-port="2" to-layer="549" to-port="0" />
<edge from-layer="548" from-port="0" to-layer="549" to-port="1" />
<edge from-layer="549" from-port="2" to-layer="550" to-port="1" />
<edge from-layer="550" from-port="2" to-layer="552" to-port="0" />
<edge from-layer="551" from-port="0" to-layer="552" to-port="1" />
<edge from-layer="552" from-port="2" to-layer="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="556" to-port="0" />
<edge from-layer="555" from-port="0" to-layer="556" to-port="1" />
<edge from-layer="556" from-port="2" to-layer="584" to-port="0" />
<edge from-layer="556" from-port="2" to-layer="567" to-port="0" />
<edge from-layer="557" from-port="0" to-layer="558" to-port="0" />
<edge from-layer="558" from-port="1" to-layer="561" to-port="0" />
<edge from-layer="559" from-port="0" to-layer="560" to-port="0" />
<edge from-layer="560" from-port="1" to-layer="561" to-port="1" />
<edge from-layer="561" from-port="2" to-layer="563" to-port="0" />
<edge from-layer="562" from-port="0" to-layer="563" to-port="1" />
<edge from-layer="563" from-port="2" to-layer="565" to-port="0" />
<edge from-layer="564" from-port="0" to-layer="565" to-port="1" />
<edge from-layer="565" from-port="2" to-layer="566" to-port="0" />
<edge from-layer="566" from-port="1" to-layer="567" to-port="1" />
<edge from-layer="567" from-port="2" to-layer="569" to-port="0" />
<edge from-layer="568" from-port="0" to-layer="569" to-port="1" />
<edge from-layer="569" from-port="2" to-layer="570" to-port="0" />
<edge from-layer="570" from-port="1" to-layer="581" to-port="0" />
<edge from-layer="571" from-port="0" to-layer="572" to-port="0" />
<edge from-layer="572" from-port="1" to-layer="575" to-port="0" />
<edge from-layer="573" from-port="0" to-layer="574" to-port="0" />
<edge from-layer="574" from-port="1" to-layer="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="579" to-port="0" />
<edge from-layer="578" from-port="0" to-layer="579" to-port="1" />
<edge from-layer="579" from-port="2" to-layer="580" to-port="0" />
<edge from-layer="580" from-port="1" to-layer="581" to-port="1" />
<edge from-layer="581" from-port="2" to-layer="583" to-port="0" />
<edge from-layer="582" from-port="0" to-layer="583" to-port="1" />
<edge from-layer="583" from-port="2" to-layer="584" to-port="1" />
<edge from-layer="584" from-port="2" to-layer="586" to-port="0" />
<edge from-layer="585" from-port="0" to-layer="586" to-port="1" />
<edge from-layer="586" from-port="2" to-layer="588" to-port="0" />
<edge from-layer="587" from-port="0" to-layer="588" to-port="1" />
<edge from-layer="588" from-port="2" to-layer="590" to-port="0" />
<edge from-layer="589" from-port="0" to-layer="590" to-port="1" />
<edge from-layer="590" from-port="2" to-layer="601" to-port="0" />
<edge from-layer="590" from-port="2" to-layer="619" to-port="0" />
<edge from-layer="590" from-port="2" to-layer="649" to-port="0" />
<edge from-layer="590" from-port="2" to-layer="676" to-port="0" />
<edge from-layer="591" from-port="0" to-layer="592" to-port="0" />
<edge from-layer="592" from-port="1" to-layer="595" to-port="0" />
<edge from-layer="593" from-port="0" to-layer="594" to-port="0" />
<edge from-layer="594" from-port="1" to-layer="595" to-port="1" />
<edge from-layer="595" from-port="2" 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="599" to-port="0" />
<edge from-layer="598" from-port="0" to-layer="599" to-port="1" />
<edge from-layer="599" from-port="2" to-layer="600" to-port="0" />
<edge from-layer="600" from-port="1" to-layer="601" to-port="1" />
<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="608" to-port="0" />
<edge from-layer="608" from-port="2" to-layer="627" to-port="0" />
<edge from-layer="609" from-port="0" to-layer="610" to-port="0" />
<edge from-layer="610" from-port="1" to-layer="613" to-port="0" />
<edge from-layer="611" from-port="0" to-layer="612" to-port="0" />
<edge from-layer="612" from-port="1" to-layer="613" to-port="1" />
<edge from-layer="613" from-port="2" to-layer="615" to-port="0" />
<edge from-layer="614" from-port="0" to-layer="615" to-port="1" />
<edge from-layer="615" from-port="2" to-layer="617" to-port="0" />
<edge from-layer="616" from-port="0" to-layer="617" to-port="1" />
<edge from-layer="617" from-port="2" to-layer="618" to-port="0" />
<edge from-layer="618" from-port="1" to-layer="619" to-port="1" />
<edge from-layer="619" from-port="2" to-layer="621" to-port="0" />
<edge from-layer="620" from-port="0" to-layer="621" to-port="1" />
<edge from-layer="621" from-port="2" to-layer="623" to-port="0" />
<edge from-layer="622" from-port="0" to-layer="623" to-port="1" />
<edge from-layer="623" from-port="2" to-layer="625" to-port="0" />
<edge from-layer="624" from-port="0" to-layer="625" to-port="1" />
<edge from-layer="625" from-port="2" to-layer="626" to-port="0" />
<edge from-layer="626" from-port="2" to-layer="629" to-port="0" />
<edge from-layer="626" from-port="2" to-layer="627" to-port="1" />
<edge from-layer="627" from-port="2" to-layer="634" to-port="0" />
<edge from-layer="628" from-port="0" to-layer="633" to-port="1" />
<edge from-layer="629" from-port="1" to-layer="632" to-port="0" />
<edge from-layer="630" from-port="0" to-layer="632" to-port="1" />
<edge from-layer="631" from-port="0" to-layer="632" to-port="2" />
<edge from-layer="632" from-port="3" to-layer="633" to-port="3" />
<edge from-layer="632" from-port="3" to-layer="636" to-port="2" />
<edge from-layer="633" from-port="4" to-layer="634" to-port="1" />
<edge from-layer="634" from-port="2" to-layer="635" to-port="0" />
<edge from-layer="635" from-port="2" to-layer="637" to-port="0" />
<edge from-layer="636" from-port="3" to-layer="637" to-port="1" />
<edge from-layer="637" from-port="2" to-layer="638" to-port="0" />
<edge from-layer="638" from-port="1" to-layer="657" to-port="0" />
<edge from-layer="639" from-port="0" to-layer="640" to-port="0" />
<edge from-layer="640" from-port="1" to-layer="643" to-port="0" />
<edge from-layer="641" from-port="0" to-layer="642" to-port="0" />
<edge from-layer="642" from-port="1" to-layer="643" to-port="1" />
<edge from-layer="643" from-port="2" to-layer="645" to-port="0" />
<edge from-layer="644" from-port="0" to-layer="645" to-port="1" />
<edge from-layer="645" from-port="2" to-layer="647" to-port="0" />
<edge from-layer="646" from-port="0" to-layer="647" to-port="1" />
<edge from-layer="647" from-port="2" to-layer="648" to-port="0" />
<edge from-layer="648" from-port="1" to-layer="649" to-port="1" />
<edge from-layer="649" from-port="2" to-layer="651" to-port="0" />
<edge from-layer="650" from-port="0" to-layer="651" to-port="1" />
<edge from-layer="651" from-port="2" to-layer="653" to-port="0" />
<edge from-layer="652" from-port="0" to-layer="653" to-port="1" />
<edge from-layer="653" from-port="2" to-layer="655" to-port="0" />
<edge from-layer="654" from-port="0" to-layer="655" to-port="1" />
<edge from-layer="655" from-port="2" to-layer="656" to-port="0" />
<edge from-layer="656" from-port="2" to-layer="657" to-port="1" />
<edge from-layer="657" from-port="2" to-layer="658" to-port="0" />
<edge from-layer="658" from-port="2" to-layer="660" to-port="0" />
<edge from-layer="659" from-port="0" to-layer="660" to-port="1" />
<edge from-layer="660" from-port="2" to-layer="662" to-port="0" />
<edge from-layer="661" from-port="0" to-layer="662" to-port="1" />
<edge from-layer="662" from-port="2" to-layer="673" to-port="0" />
<edge from-layer="663" from-port="0" to-layer="664" to-port="0" />
<edge from-layer="664" from-port="1" to-layer="667" to-port="0" />
<edge from-layer="665" from-port="0" to-layer="666" to-port="0" />
<edge from-layer="666" from-port="1" to-layer="667" to-port="1" />
<edge from-layer="667" from-port="2" to-layer="669" to-port="0" />
<edge from-layer="668" from-port="0" to-layer="669" to-port="1" />
<edge from-layer="669" from-port="2" to-layer="671" to-port="0" />
<edge from-layer="670" from-port="0" to-layer="671" to-port="1" />
<edge from-layer="671" from-port="2" to-layer="672" to-port="0" />
<edge from-layer="672" from-port="1" to-layer="673" to-port="1" />
<edge from-layer="673" from-port="2" to-layer="675" to-port="0" />
<edge from-layer="674" from-port="0" to-layer="675" to-port="1" />
<edge from-layer="675" from-port="2" to-layer="676" to-port="1" />
<edge from-layer="676" from-port="2" to-layer="678" to-port="0" />
<edge from-layer="677" from-port="0" to-layer="678" to-port="1" />
<edge from-layer="678" from-port="2" to-layer="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="710" to-port="0" />
<edge from-layer="682" from-port="2" to-layer="693" to-port="0" />
<edge from-layer="683" from-port="0" to-layer="684" to-port="0" />
<edge from-layer="684" from-port="1" to-layer="687" to-port="0" />
<edge from-layer="685" from-port="0" to-layer="686" to-port="0" />
<edge from-layer="686" from-port="1" 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="692" to-port="0" />
<edge from-layer="692" from-port="1" to-layer="693" to-port="1" />
<edge from-layer="693" from-port="2" to-layer="695" to-port="0" />
<edge from-layer="694" from-port="0" to-layer="695" to-port="1" />
<edge from-layer="695" from-port="2" to-layer="696" to-port="0" />
<edge from-layer="696" from-port="1" to-layer="707" to-port="0" />
<edge from-layer="697" from-port="0" to-layer="698" to-port="0" />
<edge from-layer="698" from-port="1" to-layer="701" to-port="0" />
<edge from-layer="699" from-port="0" to-layer="700" to-port="0" />
<edge from-layer="700" from-port="1" to-layer="701" to-port="1" />
<edge from-layer="701" from-port="2" to-layer="703" to-port="0" />
<edge from-layer="702" from-port="0" to-layer="703" to-port="1" />
<edge from-layer="703" from-port="2" to-layer="705" to-port="0" />
<edge from-layer="704" from-port="0" to-layer="705" to-port="1" />
<edge from-layer="705" from-port="2" to-layer="706" to-port="0" />
<edge from-layer="706" from-port="1" to-layer="707" to-port="1" />
<edge from-layer="707" from-port="2" to-layer="709" to-port="0" />
<edge from-layer="708" from-port="0" to-layer="709" to-port="1" />
<edge from-layer="709" from-port="2" to-layer="710" to-port="1" />
<edge from-layer="710" from-port="2" to-layer="712" to-port="0" />
<edge from-layer="711" from-port="0" to-layer="712" to-port="1" />
<edge from-layer="712" from-port="2" to-layer="714" to-port="0" />
<edge from-layer="713" from-port="0" to-layer="714" to-port="1" />
<edge from-layer="714" from-port="2" to-layer="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="727" to-port="0" />
<edge from-layer="716" from-port="2" to-layer="802" to-port="0" />
<edge from-layer="716" from-port="2" to-layer="745" to-port="0" />
<edge from-layer="716" from-port="2" to-layer="775" to-port="0" />
<edge from-layer="717" from-port="0" to-layer="718" to-port="0" />
<edge from-layer="718" from-port="1" to-layer="721" to-port="0" />
<edge from-layer="719" from-port="0" to-layer="720" to-port="0" />
<edge from-layer="720" from-port="1" 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="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="726" to-port="0" />
<edge from-layer="726" from-port="1" to-layer="727" to-port="1" />
<edge from-layer="727" from-port="2" to-layer="729" to-port="0" />
<edge from-layer="728" from-port="0" to-layer="729" to-port="1" />
<edge from-layer="729" from-port="2" to-layer="731" to-port="0" />
<edge from-layer="730" from-port="0" to-layer="731" to-port="1" />
<edge from-layer="731" from-port="2" to-layer="733" to-port="0" />
<edge from-layer="732" from-port="0" to-layer="733" to-port="1" />
<edge from-layer="733" from-port="2" to-layer="734" to-port="0" />
<edge from-layer="734" from-port="2" to-layer="753" to-port="0" />
<edge from-layer="735" from-port="0" to-layer="736" to-port="0" />
<edge from-layer="736" from-port="1" to-layer="739" to-port="0" />
<edge from-layer="737" from-port="0" to-layer="738" to-port="0" />
<edge from-layer="738" from-port="1" to-layer="739" to-port="1" />
<edge from-layer="739" from-port="2" to-layer="741" to-port="0" />
<edge from-layer="740" from-port="0" to-layer="741" to-port="1" />
<edge from-layer="741" from-port="2" to-layer="743" to-port="0" />
<edge from-layer="742" from-port="0" to-layer="743" to-port="1" />
<edge from-layer="743" from-port="2" to-layer="744" to-port="0" />
<edge from-layer="744" from-port="1" to-layer="745" to-port="1" />
<edge from-layer="745" from-port="2" to-layer="747" to-port="0" />
<edge from-layer="746" from-port="0" to-layer="747" to-port="1" />
<edge from-layer="747" from-port="2" to-layer="749" to-port="0" />
<edge from-layer="748" from-port="0" to-layer="749" to-port="1" />
<edge from-layer="749" from-port="2" to-layer="751" to-port="0" />
<edge from-layer="750" from-port="0" to-layer="751" to-port="1" />
<edge from-layer="751" from-port="2" to-layer="752" to-port="0" />
<edge from-layer="752" from-port="2" to-layer="755" to-port="0" />
<edge from-layer="752" from-port="2" to-layer="753" to-port="1" />
<edge from-layer="753" from-port="2" to-layer="760" to-port="0" />
<edge from-layer="754" from-port="0" to-layer="759" to-port="1" />
<edge from-layer="755" from-port="1" to-layer="758" to-port="0" />
<edge from-layer="756" from-port="0" to-layer="758" to-port="1" />
<edge from-layer="757" from-port="0" to-layer="758" to-port="2" />
<edge from-layer="758" from-port="3" to-layer="759" to-port="3" />
<edge from-layer="758" from-port="3" to-layer="762" to-port="2" />
<edge from-layer="759" from-port="4" to-layer="760" to-port="1" />
<edge from-layer="760" from-port="2" to-layer="761" to-port="0" />
<edge from-layer="761" from-port="2" to-layer="763" to-port="0" />
<edge from-layer="762" from-port="3" to-layer="763" to-port="1" />
<edge from-layer="763" from-port="2" to-layer="764" to-port="0" />
<edge from-layer="764" from-port="1" to-layer="783" to-port="0" />
<edge from-layer="765" from-port="0" to-layer="766" to-port="0" />
<edge from-layer="766" from-port="1" to-layer="769" 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="769" to-port="1" />
<edge from-layer="769" from-port="2" to-layer="771" to-port="0" />
<edge from-layer="770" from-port="0" 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="774" to-port="0" />
<edge from-layer="774" from-port="1" to-layer="775" to-port="1" />
<edge from-layer="775" from-port="2" to-layer="777" to-port="0" />
<edge from-layer="776" from-port="0" to-layer="777" to-port="1" />
<edge from-layer="777" from-port="2" to-layer="779" to-port="0" />
<edge from-layer="778" from-port="0" to-layer="779" to-port="1" />
<edge from-layer="779" from-port="2" to-layer="781" to-port="0" />
<edge from-layer="780" from-port="0" to-layer="781" to-port="1" />
<edge from-layer="781" from-port="2" to-layer="782" to-port="0" />
<edge from-layer="782" from-port="2" to-layer="783" to-port="1" />
<edge from-layer="783" from-port="2" to-layer="784" to-port="0" />
<edge from-layer="784" from-port="2" 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="788" to-port="0" />
<edge from-layer="787" from-port="0" to-layer="788" to-port="1" />
<edge from-layer="788" from-port="2" to-layer="799" to-port="0" />
<edge from-layer="789" from-port="0" to-layer="790" to-port="0" />
<edge from-layer="790" from-port="1" to-layer="793" to-port="0" />
<edge from-layer="791" from-port="0" to-layer="792" to-port="0" />
<edge from-layer="792" from-port="1" to-layer="793" to-port="1" />
<edge from-layer="793" from-port="2" to-layer="795" to-port="0" />
<edge from-layer="794" from-port="0" to-layer="795" to-port="1" />
<edge from-layer="795" from-port="2" to-layer="797" to-port="0" />
<edge from-layer="796" from-port="0" to-layer="797" to-port="1" />
<edge from-layer="797" from-port="2" to-layer="798" to-port="0" />
<edge from-layer="798" from-port="1" to-layer="799" to-port="1" />
<edge from-layer="799" from-port="2" to-layer="801" to-port="0" />
<edge from-layer="800" from-port="0" to-layer="801" to-port="1" />
<edge from-layer="801" from-port="2" to-layer="802" to-port="1" />
<edge from-layer="802" from-port="2" to-layer="804" to-port="0" />
<edge from-layer="803" from-port="0" to-layer="804" to-port="1" />
<edge from-layer="804" from-port="2" to-layer="806" to-port="0" />
<edge from-layer="805" from-port="0" to-layer="806" to-port="1" />
<edge from-layer="806" from-port="2" to-layer="808" to-port="0" />
<edge from-layer="807" from-port="0" to-layer="808" to-port="1" />
<edge from-layer="808" from-port="2" to-layer="819" to-port="0" />
<edge from-layer="808" from-port="2" to-layer="834" to-port="0" />
<edge from-layer="809" from-port="0" to-layer="810" to-port="0" />
<edge from-layer="810" from-port="1" to-layer="813" to-port="0" />
<edge from-layer="811" from-port="0" to-layer="812" to-port="0" />
<edge from-layer="812" from-port="1" to-layer="813" to-port="1" />
<edge from-layer="813" from-port="2" to-layer="815" to-port="0" />
<edge from-layer="814" from-port="0" to-layer="815" to-port="1" />
<edge from-layer="815" from-port="2" to-layer="817" to-port="0" />
<edge from-layer="816" from-port="0" to-layer="817" to-port="1" />
<edge from-layer="817" from-port="2" to-layer="818" to-port="0" />
<edge from-layer="818" from-port="1" to-layer="819" to-port="1" />
<edge from-layer="819" from-port="2" to-layer="821" to-port="0" />
<edge from-layer="820" from-port="0" to-layer="821" to-port="1" />
<edge from-layer="821" from-port="2" to-layer="822" to-port="0" />
<edge from-layer="822" from-port="1" to-layer="831" to-port="0" />
<edge from-layer="823" from-port="0" to-layer="824" to-port="0" />
<edge from-layer="824" from-port="1" to-layer="827" to-port="0" />
<edge from-layer="825" from-port="0" to-layer="826" to-port="0" />
<edge from-layer="826" from-port="1" to-layer="827" to-port="1" />
<edge from-layer="827" from-port="2" 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="0" />
<edge from-layer="830" from-port="1" to-layer="831" to-port="1" />
<edge from-layer="831" from-port="2" to-layer="833" to-port="0" />
<edge from-layer="832" from-port="0" to-layer="833" to-port="1" />
<edge from-layer="833" from-port="2" to-layer="834" to-port="1" />
<edge from-layer="834" from-port="2" to-layer="836" to-port="0" />
<edge from-layer="835" from-port="0" to-layer="836" to-port="1" />
<edge from-layer="836" from-port="2" to-layer="838" to-port="0" />
<edge from-layer="837" from-port="0" to-layer="838" to-port="1" />
<edge from-layer="838" from-port="2" to-layer="840" to-port="0" />
<edge from-layer="839" from-port="0" to-layer="840" to-port="1" />
<edge from-layer="840" from-port="2" to-layer="841" 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" />
<weight_compression>
<all_layers value="False" />
<awq value="False" />
<gptq value="False" />
<group_size value="128" />
<ignored_scope value="[]" />
<mode value="int4_asym" />
<ratio value="1.0" />
<scale_estimation value="False" />
<sensitivity_metric value="weight_quantization_error" />
</weight_compression>
</nncf>
<optimum>
<optimum_intel_version value="1.19.0.dev0+b5998f2" />
<optimum_version value="1.22.0.dev0" />
<pytorch_version value="2.3.1" />
<transformers_version value="4.43.4" />
</optimum>
</rt_info>
</net>