openvino_model.xml

<?xml version="1.0"?>
<net name="Model0" version="11">
	<layers>
		<layer id="2" name="input_ids" type="Parameter" version="opset1">
			<data shape="?,?" element_type="i64" />
			<output>
				<port id="0" precision="I64" names="input_ids">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="1" name="attention_mask" type="Parameter" version="opset1">
			<data shape="?,?" element_type="i64" />
			<output>
				<port id="0" precision="I64" names="39,attention_mask">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="0" name="token_type_ids" type="Parameter" version="opset1">
			<data shape="?,?" element_type="i64" />
			<output>
				<port id="0" precision="I64" names="token_type_ids">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="3" name="self.bert.embeddings.word_embeddings.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="61056, 384" offset="0" size="46891008" />
			<output>
				<port id="0" precision="FP16" names="self.bert.embeddings.word_embeddings.weight">
					<dim>61056</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="4" name="self.bert.embeddings.word_embeddings.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>61056</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>61056</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="5" name="__module.bert.embeddings.word_embeddings/aten::embedding/Convert" type="Convert" version="opset1">
			<data destination_type="i32" />
			<input>
				<port id="0" precision="I64">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="I32">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="6" name="__module.bert.embeddings.word_embeddings/aten::embedding/Constant" type="Const" version="opset1">
			<data element_type="i32" shape="" offset="46891008" size="4" />
			<output>
				<port id="0" precision="I32" />
			</output>
		</layer>
		<layer id="7" name="__module.bert.embeddings.word_embeddings/aten::embedding/Gather" type="Gather" version="opset8">
			<data batch_dims="0" />
			<input>
				<port id="0" precision="FP32">
					<dim>61056</dim>
					<dim>384</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="51,inputs_embeds">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="8" name="self.bert.embeddings.token_type_embeddings.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="2, 384" offset="46891012" size="1536" />
			<output>
				<port id="0" precision="FP16" names="self.bert.embeddings.token_type_embeddings.weight">
					<dim>2</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="9" name="self.bert.embeddings.token_type_embeddings.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>2</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>2</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="10" name="__module.bert.embeddings.token_type_embeddings/aten::embedding/Convert" type="Convert" version="opset1">
			<data destination_type="i32" />
			<input>
				<port id="0" precision="I64">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="I32">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="11" name="__module.bert.embeddings.token_type_embeddings/aten::embedding/Constant" type="Const" version="opset1">
			<data element_type="i32" shape="" offset="46891008" size="4" />
			<output>
				<port id="0" precision="I32" />
			</output>
		</layer>
		<layer id="12" name="__module.bert.embeddings.token_type_embeddings/aten::embedding/Gather" type="Gather" version="opset8">
			<data batch_dims="0" />
			<input>
				<port id="0" precision="FP32">
					<dim>2</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I32">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="2" precision="I32" />
			</input>
			<output>
				<port id="3" precision="FP32" names="53,token_type_embeddings.1">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="13" name="__module.bert.embeddings/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="54,input.1">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="14" name="__module.bert.embeddings.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="46892548" size="4" />
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="15" name="__module.bert.embeddings.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
			<data eps="9.999999960041972e-13" normalize_variance="true" eps_mode="INSIDE_SQRT" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</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>384</dim>
				</port>
			</output>
		</layer>
		<layer id="16" name="Constant_4723_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="46892552" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="17" name="Constant_4723" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="18" name="__module.bert.embeddings.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="19" name="Constant_4724_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="46893320" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="20" name="Constant_4724" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="21" name="__module.bert.embeddings.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="58,input.3">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="22" name="self.bert.encoder.model.layer.0.attention.self.query.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="46894088" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.0.attention.self.query.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="23" name="self.bert.encoder.model.layer.0.attention.self.query.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="24" name="__module.bert.encoder.model.layer.0.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="25" name="Constant_4725_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="47189000" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="26" name="Constant_4725" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="27" name="__module.bert.encoder.model.layer.0.attention.self.query/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="104,x.9">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="28" name="__module.bert.encoder.model.layer.0.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="29" name="__module.bert.encoder.model.layer.0.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="126,x.11">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="30" name="Constant_302" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="127">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="31" name="__module.bert.encoder.model.layer.0.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="128,query_layer.1">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="32" name="self.bert.encoder.model.layer.0.attention.self.key.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="47189832" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.0.attention.self.key.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="33" name="self.bert.encoder.model.layer.0.attention.self.key.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="34" name="__module.bert.encoder.model.layer.0.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="35" name="Constant_4726_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="47484744" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="36" name="Constant_4726" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="37" name="__module.bert.encoder.model.layer.0.attention.self.key/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="107,x.1">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="38" name="__module.bert.encoder.model.layer.0.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="39" name="__module.bert.encoder.model.layer.0.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="111,x.3">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="40" name="Constant_258" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="112">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="41" name="__module.bert.encoder.model.layer.0.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="113,key_layer.1">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="42" name="__module.bert.encoder.model.layer.0.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="130,attention_scores.1">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="43" name="Constant_4727_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 1, 1" offset="47485512" size="2" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="44" name="Constant_4727" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="45" name="__module.bert.encoder.model.layer.0.attention.self/aten::div/Divide" type="Divide" version="opset1">
			<data auto_broadcast="numpy" m_pythondiv="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="131,attention_scores.3">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="46" name="Constant_4729_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 1, 1" offset="47485514" size="2" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="47" name="Constant_4729" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="48" name="33" type="Const" version="opset1">
			<data element_type="i64" shape="" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64" names="33" />
			</output>
		</layer>
		<layer id="49" name="__module.bert/aten::unsqueeze/Unsqueeze" type="Unsqueeze" version="opset1">
			<input>
				<port id="0" precision="I64">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="I64" />
			</input>
			<output>
				<port id="2" precision="I64" names="40">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="50" name="32" type="Const" version="opset1">
			<data element_type="i64" shape="" offset="47485524" size="8" />
			<output>
				<port id="0" precision="I64" names="32" />
			</output>
		</layer>
		<layer id="51" name="__module.bert/aten::unsqueeze/Unsqueeze_1" type="Unsqueeze" version="opset1">
			<input>
				<port id="0" precision="I64">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="I64" />
			</input>
			<output>
				<port id="2" precision="I64" names="41,42,extended_attention_mask">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="52" name="__module.bert/aten::to/Convert" type="Convert" version="opset1">
			<data destination_type="f32" />
			<input>
				<port id="0" precision="I64">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="43">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="53" name="Constant_4728_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 1, 1" offset="47485514" size="2" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="54" name="Constant_4728" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="55" name="__module.bert/aten::rsub/Multiply" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="56" name="__module.bert/aten::rsub/Subtract" type="Subtract" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="44">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="57" name="Constant_4730" type="Const" version="opset1">
			<data element_type="f32" shape="1, 1, 1, 1" offset="47485532" 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="58" name="__module.bert/aten::mul/Multiply" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="45,attention_mask">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="59" name="__module.bert.encoder.model.layer.0.attention.self/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="132,attention_scores.5">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="60" name="__module.bert.encoder/aten::unsqueeze/Unsqueeze_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="12, 1, 1" offset="47485536" size="24" />
			<output>
				<port id="0" precision="FP16" names="87">
					<dim>12</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="61" name="__module.bert.encoder/aten::unsqueeze/Unsqueeze" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>12</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>12</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="62" name="__module.bert.encoder/aten::arange/Constant_2" type="Const" version="opset1">
			<data element_type="i32" shape="" offset="46891008" size="4" />
			<output>
				<port id="0" precision="I32" />
			</output>
		</layer>
		<layer id="63" name="ShapeOf_4811" type="ShapeOf" version="opset3">
			<data output_type="i64" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="I64">
					<dim>3</dim>
				</port>
			</output>
		</layer>
		<layer id="64" name="Constant_4812" type="Const" version="opset1">
			<data element_type="i64" shape="" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64" />
			</output>
		</layer>
		<layer id="65" name="Constant_4813" type="Const" version="opset1">
			<data element_type="i64" shape="" offset="47485560" size="8" />
			<output>
				<port id="0" precision="I64" />
			</output>
		</layer>
		<layer id="66" name="Gather_4814" type="Gather" version="opset8">
			<data batch_dims="0" />
			<input>
				<port id="0" precision="I64">
					<dim>3</dim>
				</port>
				<port id="1" precision="I64" />
				<port id="2" precision="I64" />
			</input>
			<output>
				<port id="3" precision="I64" names="72,90" />
			</output>
		</layer>
		<layer id="67" name="__module.bert.encoder/aten::arange/Constant_3" type="Const" version="opset1">
			<data element_type="i32" shape="" offset="47485568" size="4" />
			<output>
				<port id="0" precision="I32" />
			</output>
		</layer>
		<layer id="68" name="__module.bert.encoder/aten::arange/Range_1" type="Range" version="opset4">
			<data output_type="f32" />
			<input>
				<port id="0" precision="I32" />
				<port id="1" precision="I64" />
				<port id="2" precision="I32" />
			</input>
			<output>
				<port id="3" precision="FP32">
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="69" name="__module.bert.encoder/aten::arange/ConvertLike" type="Convert" version="opset1">
			<data destination_type="i64" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="I64" names="73,74,76">
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="70" name="35" type="Const" version="opset1">
			<data element_type="i64" shape="" offset="47485560" size="8" />
			<output>
				<port id="0" precision="I64" names="35" />
			</output>
		</layer>
		<layer id="71" name="__module.bert.encoder/aten::unsqueeze/Unsqueeze_1" type="Unsqueeze" version="opset1">
			<input>
				<port id="0" precision="I64">
					<dim>-1</dim>
				</port>
				<port id="1" precision="I64" />
			</input>
			<output>
				<port id="2" precision="I64" names="77,78,memory_position">
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="72" name="__module.bert.encoder/aten::unsqueeze/Unsqueeze_2" type="Unsqueeze" version="opset1">
			<input>
				<port id="0" precision="I64">
					<dim>-1</dim>
				</port>
				<port id="1" precision="I64" />
			</input>
			<output>
				<port id="2" precision="I64" names="75,context_position">
					<dim>-1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="73" name="__module.bert.encoder/aten::sub/Subtract" type="Subtract" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="I64">
					<dim>1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="I64">
					<dim>-1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="I64" names="79">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="74" name="__module.bert.encoder/aten::abs/Abs" type="Abs" version="opset1">
			<input>
				<port id="0" precision="I64">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="I64" names="80,relative_position.1">
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="75" name="__module.bert.encoder/aten::unsqueeze/Unsqueeze_3" 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="81">
					<dim>1</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="76" name="__module.bert.encoder/aten::expand/Abs" type="Const" version="opset1">
			<data element_type="i64" shape="3" offset="47485572" size="24" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>3</dim>
				</port>
			</output>
		</layer>
		<layer id="77" name="__module.bert.encoder/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>
				</port>
				<port id="1" precision="I64">
					<dim>3</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="I64" names="83,relative_position">
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="78" name="__module.bert.encoder/aten::mul/Convert" type="Convert" version="opset1">
			<data destination_type="f32" />
			<input>
				<port id="0" precision="I64">
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="79" name="__module.bert.encoder/aten::mul/Multiply" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>12</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="88,alibi.1">
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="80" name="__module.bert.encoder/aten::unsqueeze/Unsqueeze_4" type="Unsqueeze" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="I64" />
			</input>
			<output>
				<port id="2" precision="FP32" names="89,91,92,alibi">
					<dim>1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="81" name="Constant_4502" type="Const" version="opset1">
			<data element_type="i64" shape="2" offset="47485596" size="16" />
			<output>
				<port id="0" precision="I64">
					<dim>2</dim>
				</port>
			</output>
		</layer>
		<layer id="82" name="Constant_4816" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485516" size="8" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="83" name="Reshape_4817" type="Reshape" version="opset1">
			<data special_zero="false" />
			<input>
				<port id="0" precision="I64" />
				<port id="1" precision="I64">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="84" name="Constant_4505" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="47485612" size="4" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="85" name="Concat_4497" type="Broadcast" version="opset3">
			<data mode="numpy" />
			<input>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
				<port id="1" precision="I32">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="I64">
					<dim>2</dim>
				</port>
			</output>
		</layer>
		<layer id="86" name="Constant_4501" type="Const" version="opset1">
			<data element_type="i64" shape="2" offset="47485616" size="16" />
			<output>
				<port id="0" precision="I64">
					<dim>2</dim>
				</port>
			</output>
		</layer>
		<layer id="87" name="Constant_4500" type="Const" version="opset1">
			<data element_type="i64" shape="2" offset="47485632" size="16" />
			<output>
				<port id="0" precision="I64">
					<dim>2</dim>
				</port>
			</output>
		</layer>
		<layer id="88" name="__module.bert.encoder/aten::slice/Slice_5" type="Slice" version="opset8">
			<input>
				<port id="0" precision="FP32">
					<dim>1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="I64">
					<dim>2</dim>
				</port>
				<port id="2" precision="I64">
					<dim>2</dim>
				</port>
				<port id="3" precision="I64">
					<dim>2</dim>
				</port>
				<port id="4" precision="I64">
					<dim>2</dim>
				</port>
			</input>
			<output>
				<port id="5" precision="FP32" names="94,bias.3">
					<dim>1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="89" name="__module.bert.encoder.model.layer.0.attention.self/aten::add/Add_1" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="133,input.5">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="90" name="__module.bert.encoder.model.layer.0.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
			<data axis="-1" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="134,input.7">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="91" name="self.bert.encoder.model.layer.0.attention.self.value.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="47485648" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.0.attention.self.value.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="92" name="self.bert.encoder.model.layer.0.attention.self.value.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="93" name="__module.bert.encoder.model.layer.0.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="94" name="Constant_4731_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="47780560" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="95" name="Constant_4731" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="96" name="__module.bert.encoder.model.layer.0.attention.self.value/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="116,x.5">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="97" name="__module.bert.encoder.model.layer.0.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="98" name="__module.bert.encoder.model.layer.0.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="120,x.7">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="99" name="Constant_283" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="121">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="100" name="__module.bert.encoder.model.layer.0.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="122,value_layer.1">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="101" name="__module.bert.encoder.model.layer.0.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="false" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="136,context_layer.1">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="102" name="Constant_365" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="137">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="103" name="__module.bert.encoder.model.layer.0.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="138">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="104" name="__module.bert.encoder.model.layer.0.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
			<data element_type="i64" shape="3" offset="47781328" size="24" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>3</dim>
				</port>
			</output>
		</layer>
		<layer id="105" name="__module.bert.encoder.model.layer.0.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>3</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="143,input.9">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="106" name="self.bert.encoder.model.layer.0.attention.output.dense.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="47781352" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.0.attention.output.dense.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="107" name="self.bert.encoder.model.layer.0.attention.output.dense.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="108" name="__module.bert.encoder.model.layer.0.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="109" name="Constant_4732_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="48076264" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="110" name="Constant_4732" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="111" name="__module.bert.encoder.model.layer.0.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="148,input.11">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="112" name="__module.bert.encoder.model.layer.0.attention.output/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="150,input.13">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="113" name="__module.bert.encoder.model.layer.0.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="46892548" size="4" />
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="114" name="__module.bert.encoder.model.layer.0.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
			<data eps="9.999999960041972e-13" normalize_variance="true" eps_mode="INSIDE_SQRT" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</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>384</dim>
				</port>
			</output>
		</layer>
		<layer id="115" name="Constant_4733_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="48077032" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="116" name="Constant_4733" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="117" name="__module.bert.encoder.model.layer.0.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="118" name="Constant_4734_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="48077800" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="119" name="Constant_4734" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="120" name="__module.bert.encoder.model.layer.0.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="154,input.15">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="121" name="self.bert.encoder.model.layer.0.mlp.gated_layers.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="3072, 384" offset="48078568" size="2359296" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.0.mlp.gated_layers.weight">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="122" name="self.bert.encoder.model.layer.0.mlp.gated_layers.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="123" name="__module.bert.encoder.model.layer.0.mlp.gated_layers/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>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="159,160,161,163,164,hidden_states.5">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
			</output>
		</layer>
		<layer id="124" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Reshape_8" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485560" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="125" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Reshape_9" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437864" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="126" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Reshape_10" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="127" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Reshape_11" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485524" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="128" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Slice_2" type="Slice" version="opset8">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
				<port id="1" precision="I64">
					<dim>1</dim>
				</port>
				<port id="2" precision="I64">
					<dim>1</dim>
				</port>
				<port id="3" precision="I64">
					<dim>1</dim>
				</port>
				<port id="4" precision="I64">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="5" precision="FP32" names="162,input.17">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="129" name="__module.bert.encoder.model.layer.0.mlp.act/aten::gelu/Gelu" type="Gelu" version="opset7">
			<data approximation_mode="ERF" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="166">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="130" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Reshape_20" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437864" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="131" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Reshape_21" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437872" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="132" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Reshape_22" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="133" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Reshape_23" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485524" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="134" name="__module.bert.encoder.model.layer.0.mlp/aten::slice/Slice_5" type="Slice" version="opset8">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
				<port id="1" precision="I64">
					<dim>1</dim>
				</port>
				<port id="2" precision="I64">
					<dim>1</dim>
				</port>
				<port id="3" precision="I64">
					<dim>1</dim>
				</port>
				<port id="4" precision="I64">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="5" precision="FP32" names="165,non_gated.1">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="135" name="__module.bert.encoder.model.layer.0.mlp/aten::mul/Multiply" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="167,input.19">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="136" name="self.bert.encoder.model.layer.0.mlp.wo.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 1536" offset="50437880" size="1179648" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.0.mlp.wo.weight">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="137" name="self.bert.encoder.model.layer.0.mlp.wo.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="138" name="__module.bert.encoder.model.layer.0.mlp.wo/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>1536</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="139" name="Constant_4735_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="51617528" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="140" name="Constant_4735" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="141" name="__module.bert.encoder.model.layer.0.mlp.wo/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="171,hidden_states.7">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="142" name="__module.bert.encoder.model.layer.0.mlp/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="172,input.23">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="143" name="__module.bert.encoder.model.layer.0.mlp.layernorm/aten::layer_norm/Multiply" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="46892548" size="4" />
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="144" name="__module.bert.encoder.model.layer.0.mlp.layernorm/aten::layer_norm/MVN" type="MVN" version="opset6">
			<data eps="9.999999960041972e-13" normalize_variance="true" eps_mode="INSIDE_SQRT" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</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>384</dim>
				</port>
			</output>
		</layer>
		<layer id="145" name="Constant_4736_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="51618296" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="146" name="Constant_4736" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="147" name="__module.bert.encoder.model.layer.0.mlp.layernorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="148" name="Constant_4737_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="51619064" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="149" name="Constant_4737" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="150" name="__module.bert.encoder.model.layer.0.mlp.layernorm/aten::layer_norm/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="176,input.25">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="151" name="self.bert.encoder.model.layer.1.attention.self.query.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="51619832" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.1.attention.self.query.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="152" name="self.bert.encoder.model.layer.1.attention.self.query.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="153" name="__module.bert.encoder.model.layer.1.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="154" name="Constant_4738_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="51914744" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="155" name="Constant_4738" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="156" name="__module.bert.encoder.model.layer.1.attention.self.query/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="186,x.21">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="157" name="__module.bert.encoder.model.layer.1.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="158" name="__module.bert.encoder.model.layer.1.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="208,x.23">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="159" name="Constant_565" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="209">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="160" name="__module.bert.encoder.model.layer.1.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="210,query_layer.3">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="161" name="self.bert.encoder.model.layer.1.attention.self.key.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="51915512" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.1.attention.self.key.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="162" name="self.bert.encoder.model.layer.1.attention.self.key.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="163" name="__module.bert.encoder.model.layer.1.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="164" name="Constant_4739_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="52210424" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="165" name="Constant_4739" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="166" name="__module.bert.encoder.model.layer.1.attention.self.key/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="189,x.13">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="167" name="__module.bert.encoder.model.layer.1.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="168" name="__module.bert.encoder.model.layer.1.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="193,x.15">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="169" name="Constant_525" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="194">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="170" name="__module.bert.encoder.model.layer.1.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="195,key_layer.3">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="171" name="__module.bert.encoder.model.layer.1.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="212,attention_scores.7">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="172" name="Constant_4740_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 1, 1" offset="47485512" size="2" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="173" name="Constant_4740" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="174" name="__module.bert.encoder.model.layer.1.attention.self/aten::div/Divide" type="Divide" version="opset1">
			<data auto_broadcast="numpy" m_pythondiv="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="213,attention_scores.9">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="175" name="__module.bert.encoder.model.layer.1.attention.self/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="214,attention_scores.11">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="176" name="__module.bert.encoder.model.layer.1.attention.self/aten::add/Add_1" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="215,input.27">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="177" name="__module.bert.encoder.model.layer.1.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
			<data axis="-1" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="216,input.29">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="178" name="self.bert.encoder.model.layer.1.attention.self.value.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="52211192" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.1.attention.self.value.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="179" name="self.bert.encoder.model.layer.1.attention.self.value.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="180" name="__module.bert.encoder.model.layer.1.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="181" name="Constant_4741_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="52506104" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="182" name="Constant_4741" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="183" name="__module.bert.encoder.model.layer.1.attention.self.value/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="198,x.17">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="184" name="__module.bert.encoder.model.layer.1.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="185" name="__module.bert.encoder.model.layer.1.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="202,x.19">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="186" name="Constant_548" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="203">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="187" name="__module.bert.encoder.model.layer.1.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="204,value_layer.3">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="188" name="__module.bert.encoder.model.layer.1.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="false" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="218,context_layer.5">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="189" name="Constant_628" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="219">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="190" name="__module.bert.encoder.model.layer.1.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="220">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="191" name="__module.bert.encoder.model.layer.1.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
			<data element_type="i64" shape="3" offset="47781328" size="24" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>3</dim>
				</port>
			</output>
		</layer>
		<layer id="192" name="__module.bert.encoder.model.layer.1.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>3</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="225,input.31">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="193" name="self.bert.encoder.model.layer.1.attention.output.dense.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="52506872" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.1.attention.output.dense.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="194" name="self.bert.encoder.model.layer.1.attention.output.dense.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="195" name="__module.bert.encoder.model.layer.1.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="196" name="Constant_4742_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="52801784" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="197" name="Constant_4742" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="198" name="__module.bert.encoder.model.layer.1.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="230,input.33">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="199" name="__module.bert.encoder.model.layer.1.attention.output/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="232,input.35">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="200" name="__module.bert.encoder.model.layer.1.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="46892548" size="4" />
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="201" name="__module.bert.encoder.model.layer.1.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
			<data eps="9.999999960041972e-13" normalize_variance="true" eps_mode="INSIDE_SQRT" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</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>384</dim>
				</port>
			</output>
		</layer>
		<layer id="202" name="Constant_4743_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="52802552" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="203" name="Constant_4743" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="204" name="__module.bert.encoder.model.layer.1.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="205" name="Constant_4744_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="52803320" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="206" name="Constant_4744" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="207" name="__module.bert.encoder.model.layer.1.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="236,input.37">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="208" name="self.bert.encoder.model.layer.1.mlp.gated_layers.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="3072, 384" offset="52804088" size="2359296" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.1.mlp.gated_layers.weight">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="209" name="self.bert.encoder.model.layer.1.mlp.gated_layers.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="210" name="__module.bert.encoder.model.layer.1.mlp.gated_layers/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>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="241,242,243,245,246,hidden_states.11">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
			</output>
		</layer>
		<layer id="211" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Reshape_8" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485560" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="212" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Reshape_9" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437864" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="213" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Reshape_10" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="214" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Reshape_11" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485524" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="215" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Slice_2" type="Slice" version="opset8">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
				<port id="1" precision="I64">
					<dim>1</dim>
				</port>
				<port id="2" precision="I64">
					<dim>1</dim>
				</port>
				<port id="3" precision="I64">
					<dim>1</dim>
				</port>
				<port id="4" precision="I64">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="5" precision="FP32" names="244,input.39">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="216" name="__module.bert.encoder.model.layer.1.mlp.act/aten::gelu/Gelu" type="Gelu" version="opset7">
			<data approximation_mode="ERF" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="248">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="217" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Reshape_20" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437864" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="218" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Reshape_21" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437872" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="219" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Reshape_22" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="220" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Reshape_23" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485524" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="221" name="__module.bert.encoder.model.layer.1.mlp/aten::slice/Slice_5" type="Slice" version="opset8">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
				<port id="1" precision="I64">
					<dim>1</dim>
				</port>
				<port id="2" precision="I64">
					<dim>1</dim>
				</port>
				<port id="3" precision="I64">
					<dim>1</dim>
				</port>
				<port id="4" precision="I64">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="5" precision="FP32" names="247,non_gated.3">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="222" name="__module.bert.encoder.model.layer.1.mlp/aten::mul/Multiply" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="249,input.41">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="223" name="self.bert.encoder.model.layer.1.mlp.wo.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 1536" offset="55163384" size="1179648" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.1.mlp.wo.weight">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="224" name="self.bert.encoder.model.layer.1.mlp.wo.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="225" name="__module.bert.encoder.model.layer.1.mlp.wo/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>1536</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="226" name="Constant_4745_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="56343032" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="227" name="Constant_4745" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="228" name="__module.bert.encoder.model.layer.1.mlp.wo/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="253,hidden_states.13">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="229" name="__module.bert.encoder.model.layer.1.mlp/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="254,input.45">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="230" name="__module.bert.encoder.model.layer.1.mlp.layernorm/aten::layer_norm/Multiply" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="46892548" size="4" />
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="231" name="__module.bert.encoder.model.layer.1.mlp.layernorm/aten::layer_norm/MVN" type="MVN" version="opset6">
			<data eps="9.999999960041972e-13" normalize_variance="true" eps_mode="INSIDE_SQRT" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</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>384</dim>
				</port>
			</output>
		</layer>
		<layer id="232" name="Constant_4746_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="56343800" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="233" name="Constant_4746" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="234" name="__module.bert.encoder.model.layer.1.mlp.layernorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="235" name="Constant_4747_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="56344568" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="236" name="Constant_4747" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="237" name="__module.bert.encoder.model.layer.1.mlp.layernorm/aten::layer_norm/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="258,input.47">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="238" name="self.bert.encoder.model.layer.2.attention.self.query.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="56345336" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.2.attention.self.query.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="239" name="self.bert.encoder.model.layer.2.attention.self.query.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="240" name="__module.bert.encoder.model.layer.2.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="241" name="Constant_4748_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="56640248" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="242" name="Constant_4748" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="243" name="__module.bert.encoder.model.layer.2.attention.self.query/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="268,x.33">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="244" name="__module.bert.encoder.model.layer.2.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="245" name="__module.bert.encoder.model.layer.2.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="290,x.35">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="246" name="Constant_828" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="291">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="247" name="__module.bert.encoder.model.layer.2.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="292,query_layer.5">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="248" name="self.bert.encoder.model.layer.2.attention.self.key.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="56641016" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.2.attention.self.key.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="249" name="self.bert.encoder.model.layer.2.attention.self.key.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="250" name="__module.bert.encoder.model.layer.2.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="251" name="Constant_4749_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="56935928" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="252" name="Constant_4749" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="253" name="__module.bert.encoder.model.layer.2.attention.self.key/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="271,x.25">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="254" name="__module.bert.encoder.model.layer.2.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="255" name="__module.bert.encoder.model.layer.2.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="275,x.27">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="256" name="Constant_788" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="276">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="257" name="__module.bert.encoder.model.layer.2.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="277,key_layer.5">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="258" name="__module.bert.encoder.model.layer.2.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="294,attention_scores.13">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="259" name="Constant_4750_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 1, 1" offset="47485512" size="2" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="260" name="Constant_4750" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="261" name="__module.bert.encoder.model.layer.2.attention.self/aten::div/Divide" type="Divide" version="opset1">
			<data auto_broadcast="numpy" m_pythondiv="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="295,attention_scores.15">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="262" name="__module.bert.encoder.model.layer.2.attention.self/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="296,attention_scores.17">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="263" name="__module.bert.encoder.model.layer.2.attention.self/aten::add/Add_1" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="297,input.49">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="264" name="__module.bert.encoder.model.layer.2.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
			<data axis="-1" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="298,input.51">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="265" name="self.bert.encoder.model.layer.2.attention.self.value.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="56936696" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.2.attention.self.value.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="266" name="self.bert.encoder.model.layer.2.attention.self.value.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="267" name="__module.bert.encoder.model.layer.2.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="268" name="Constant_4751_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="57231608" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="269" name="Constant_4751" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="270" name="__module.bert.encoder.model.layer.2.attention.self.value/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="280,x.29">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="271" name="__module.bert.encoder.model.layer.2.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="272" name="__module.bert.encoder.model.layer.2.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="284,x.31">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="273" name="Constant_811" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="285">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="274" name="__module.bert.encoder.model.layer.2.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="286,value_layer.5">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="275" name="__module.bert.encoder.model.layer.2.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="false" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="300,context_layer.9">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="276" name="Constant_891" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="301">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="277" name="__module.bert.encoder.model.layer.2.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="302">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="278" name="__module.bert.encoder.model.layer.2.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
			<data element_type="i64" shape="3" offset="47781328" size="24" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>3</dim>
				</port>
			</output>
		</layer>
		<layer id="279" name="__module.bert.encoder.model.layer.2.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>3</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="307,input.53">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="280" name="self.bert.encoder.model.layer.2.attention.output.dense.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="57232376" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.2.attention.output.dense.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="281" name="self.bert.encoder.model.layer.2.attention.output.dense.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="282" name="__module.bert.encoder.model.layer.2.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="283" name="Constant_4752_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="57527288" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="284" name="Constant_4752" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="285" name="__module.bert.encoder.model.layer.2.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="312,input.55">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="286" name="__module.bert.encoder.model.layer.2.attention.output/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="314,input.57">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="287" name="__module.bert.encoder.model.layer.2.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="46892548" size="4" />
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="288" name="__module.bert.encoder.model.layer.2.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
			<data eps="9.999999960041972e-13" normalize_variance="true" eps_mode="INSIDE_SQRT" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</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>384</dim>
				</port>
			</output>
		</layer>
		<layer id="289" name="Constant_4753_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="57528056" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="290" name="Constant_4753" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="291" name="__module.bert.encoder.model.layer.2.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="292" name="Constant_4754_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="57528824" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="293" name="Constant_4754" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="294" name="__module.bert.encoder.model.layer.2.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="318,input.59">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="295" name="self.bert.encoder.model.layer.2.mlp.gated_layers.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="3072, 384" offset="57529592" size="2359296" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.2.mlp.gated_layers.weight">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="296" name="self.bert.encoder.model.layer.2.mlp.gated_layers.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="297" name="__module.bert.encoder.model.layer.2.mlp.gated_layers/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>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="323,324,325,327,328,hidden_states.17">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
			</output>
		</layer>
		<layer id="298" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Reshape_8" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485560" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="299" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Reshape_9" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437864" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="300" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Reshape_10" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="301" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Reshape_11" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485524" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="302" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Slice_2" type="Slice" version="opset8">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
				<port id="1" precision="I64">
					<dim>1</dim>
				</port>
				<port id="2" precision="I64">
					<dim>1</dim>
				</port>
				<port id="3" precision="I64">
					<dim>1</dim>
				</port>
				<port id="4" precision="I64">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="5" precision="FP32" names="326,input.61">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="303" name="__module.bert.encoder.model.layer.2.mlp.act/aten::gelu/Gelu" type="Gelu" version="opset7">
			<data approximation_mode="ERF" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="330">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="304" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Reshape_20" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437864" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="305" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Reshape_21" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437872" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="306" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Reshape_22" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="307" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Reshape_23" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485524" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="308" name="__module.bert.encoder.model.layer.2.mlp/aten::slice/Slice_5" type="Slice" version="opset8">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
				<port id="1" precision="I64">
					<dim>1</dim>
				</port>
				<port id="2" precision="I64">
					<dim>1</dim>
				</port>
				<port id="3" precision="I64">
					<dim>1</dim>
				</port>
				<port id="4" precision="I64">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="5" precision="FP32" names="329,non_gated.5">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="309" name="__module.bert.encoder.model.layer.2.mlp/aten::mul/Multiply" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="331,input.63">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="310" name="self.bert.encoder.model.layer.2.mlp.wo.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 1536" offset="59888888" size="1179648" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.2.mlp.wo.weight">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="311" name="self.bert.encoder.model.layer.2.mlp.wo.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="312" name="__module.bert.encoder.model.layer.2.mlp.wo/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>1536</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="313" name="Constant_4755_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="61068536" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="314" name="Constant_4755" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="315" name="__module.bert.encoder.model.layer.2.mlp.wo/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="335,hidden_states.19">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="316" name="__module.bert.encoder.model.layer.2.mlp/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="336,input.67">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="317" name="__module.bert.encoder.model.layer.2.mlp.layernorm/aten::layer_norm/Multiply" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="46892548" size="4" />
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="318" name="__module.bert.encoder.model.layer.2.mlp.layernorm/aten::layer_norm/MVN" type="MVN" version="opset6">
			<data eps="9.999999960041972e-13" normalize_variance="true" eps_mode="INSIDE_SQRT" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</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>384</dim>
				</port>
			</output>
		</layer>
		<layer id="319" name="Constant_4756_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="61069304" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="320" name="Constant_4756" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="321" name="__module.bert.encoder.model.layer.2.mlp.layernorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="322" name="Constant_4757_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="61070072" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="323" name="Constant_4757" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="324" name="__module.bert.encoder.model.layer.2.mlp.layernorm/aten::layer_norm/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="340,input.69">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="325" name="self.bert.encoder.model.layer.3.attention.self.query.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="61070840" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.3.attention.self.query.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="326" name="self.bert.encoder.model.layer.3.attention.self.query.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="327" name="__module.bert.encoder.model.layer.3.attention.self.query/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="328" name="Constant_4758_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="61365752" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="329" name="Constant_4758" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="330" name="__module.bert.encoder.model.layer.3.attention.self.query/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="350,x.45">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="331" name="__module.bert.encoder.model.layer.3.attention.self/prim::ListConstruct/Concat" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="332" name="__module.bert.encoder.model.layer.3.attention.self/aten::view/Reshape" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="372,x">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="333" name="Constant_1091" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="373">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="334" name="__module.bert.encoder.model.layer.3.attention.self/aten::permute/Transpose" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="374,query_layer">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="335" name="self.bert.encoder.model.layer.3.attention.self.key.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="61366520" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.3.attention.self.key.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="336" name="self.bert.encoder.model.layer.3.attention.self.key.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="337" name="__module.bert.encoder.model.layer.3.attention.self.key/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="338" name="Constant_4759_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="61661432" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="339" name="Constant_4759" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="340" name="__module.bert.encoder.model.layer.3.attention.self.key/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="353,x.37">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="341" name="__module.bert.encoder.model.layer.3.attention.self/prim::ListConstruct/Concat_1" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="342" name="__module.bert.encoder.model.layer.3.attention.self/aten::view/Reshape_1" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="357,x.39">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="343" name="Constant_1051" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="358">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="344" name="__module.bert.encoder.model.layer.3.attention.self/aten::permute/Transpose_1" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="359,key_layer">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="345" name="__module.bert.encoder.model.layer.3.attention.self/aten::matmul/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="376,attention_scores.19">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="346" name="Constant_4760_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 1, 1" offset="47485512" size="2" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="347" name="Constant_4760" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="348" name="__module.bert.encoder.model.layer.3.attention.self/aten::div/Divide" type="Divide" version="opset1">
			<data auto_broadcast="numpy" m_pythondiv="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="377,attention_scores.21">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="349" name="__module.bert.encoder.model.layer.3.attention.self/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
					<dim>1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="378,attention_scores">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="350" name="__module.bert.encoder.model.layer.3.attention.self/aten::add/Add_1" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="379,input.71">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="351" name="__module.bert.encoder.model.layer.3.attention.self/aten::softmax/Softmax" type="SoftMax" version="opset8">
			<data axis="-1" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="380,input.73">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
			</output>
		</layer>
		<layer id="352" name="self.bert.encoder.model.layer.3.attention.self.value.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="61662200" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.3.attention.self.value.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="353" name="self.bert.encoder.model.layer.3.attention.self.value.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="354" name="__module.bert.encoder.model.layer.3.attention.self.value/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="355" name="Constant_4761_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="61957112" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="356" name="Constant_4761" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="357" name="__module.bert.encoder.model.layer.3.attention.self.value/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="362,x.41">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="358" name="__module.bert.encoder.model.layer.3.attention.self/prim::ListConstruct/Concat_2" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189768" size="32" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="359" name="__module.bert.encoder.model.layer.3.attention.self/aten::view/Reshape_2" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="366,x.43">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="360" name="Constant_1074" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="367">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="361" name="__module.bert.encoder.model.layer.3.attention.self/aten::permute/Transpose_2" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="368,value_layer">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="362" name="__module.bert.encoder.model.layer.3.attention.self/aten::matmul/MatMul_1" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="false" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>-1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="382,context_layer.13">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="363" name="Constant_1154" type="Const" version="opset1">
			<data element_type="i64" shape="4" offset="47189800" size="32" />
			<output>
				<port id="0" precision="I64" names="383">
					<dim>4</dim>
				</port>
			</output>
		</layer>
		<layer id="364" name="__module.bert.encoder.model.layer.3.attention.self/aten::permute/Transpose_3" type="Transpose" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>12</dim>
					<dim>-1</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>4</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="384">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
			</output>
		</layer>
		<layer id="365" name="__module.bert.encoder.model.layer.3.attention.self/prim::ListConstruct/Concat_3" type="Const" version="opset1">
			<data element_type="i64" shape="3" offset="47781328" size="24" />
			<rt_info>
				<attribute name="precise" version="0" />
			</rt_info>
			<output>
				<port id="0" precision="I64">
					<dim>3</dim>
				</port>
			</output>
		</layer>
		<layer id="366" name="__module.bert.encoder.model.layer.3.attention.self/aten::view/Reshape_3" type="Reshape" version="opset1">
			<data special_zero="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>12</dim>
					<dim>32</dim>
				</port>
				<port id="1" precision="I64">
					<dim>3</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="389,input.75">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="367" name="self.bert.encoder.model.layer.3.attention.output.dense.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="61957880" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.3.attention.output.dense.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="368" name="self.bert.encoder.model.layer.3.attention.output.dense.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="369" name="__module.bert.encoder.model.layer.3.attention.output.dense/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="370" name="Constant_4762_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="62252792" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="371" name="Constant_4762" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="372" name="__module.bert.encoder.model.layer.3.attention.output.dense/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="394,input.77">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="373" name="__module.bert.encoder.model.layer.3.attention.output/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="396,input.79">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="374" name="__module.bert.encoder.model.layer.3.attention.output.LayerNorm/aten::layer_norm/Multiply" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="46892548" size="4" />
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="375" name="__module.bert.encoder.model.layer.3.attention.output.LayerNorm/aten::layer_norm/MVN" type="MVN" version="opset6">
			<data eps="9.999999960041972e-13" normalize_variance="true" eps_mode="INSIDE_SQRT" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</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>384</dim>
				</port>
			</output>
		</layer>
		<layer id="376" name="Constant_4763_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="62253560" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="377" name="Constant_4763" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="378" name="__module.bert.encoder.model.layer.3.attention.output.LayerNorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="379" name="Constant_4764_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="62254328" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="380" name="Constant_4764" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="381" name="__module.bert.encoder.model.layer.3.attention.output.LayerNorm/aten::layer_norm/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="400,input.81">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="382" name="self.bert.encoder.model.layer.3.mlp.gated_layers.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="3072, 384" offset="62255096" size="2359296" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.3.mlp.gated_layers.weight">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="383" name="self.bert.encoder.model.layer.3.mlp.gated_layers.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="384" name="__module.bert.encoder.model.layer.3.mlp.gated_layers/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>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>3072</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="405,406,407,409,410,hidden_states.23">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
			</output>
		</layer>
		<layer id="385" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Reshape_8" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485560" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="386" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Reshape_9" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437864" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="387" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Reshape_10" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="388" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Reshape_11" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485524" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="389" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Slice_2" type="Slice" version="opset8">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
				<port id="1" precision="I64">
					<dim>1</dim>
				</port>
				<port id="2" precision="I64">
					<dim>1</dim>
				</port>
				<port id="3" precision="I64">
					<dim>1</dim>
				</port>
				<port id="4" precision="I64">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="5" precision="FP32" names="408,input.83">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="390" name="__module.bert.encoder.model.layer.3.mlp.act/aten::gelu/Gelu" type="Gelu" version="opset7">
			<data approximation_mode="ERF" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="412">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="391" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Reshape_20" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437864" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="392" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Reshape_21" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="50437872" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="393" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Reshape_22" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485516" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="394" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Reshape_23" type="Const" version="opset1">
			<data element_type="i64" shape="1" offset="47485524" size="8" />
			<output>
				<port id="0" precision="I64">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="395" name="__module.bert.encoder.model.layer.3.mlp/aten::slice/Slice_5" type="Slice" version="opset8">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>3072</dim>
				</port>
				<port id="1" precision="I64">
					<dim>1</dim>
				</port>
				<port id="2" precision="I64">
					<dim>1</dim>
				</port>
				<port id="3" precision="I64">
					<dim>1</dim>
				</port>
				<port id="4" precision="I64">
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="5" precision="FP32" names="411,non_gated">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="396" name="__module.bert.encoder.model.layer.3.mlp/aten::mul/Multiply" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="413,input.85">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="397" name="self.bert.encoder.model.layer.3.mlp.wo.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 1536" offset="64614392" size="1179648" />
			<output>
				<port id="0" precision="FP16" names="self.bert.encoder.model.layer.3.mlp.wo.weight">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="398" name="self.bert.encoder.model.layer.3.mlp.wo.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</output>
		</layer>
		<layer id="399" name="__module.bert.encoder.model.layer.3.mlp.wo/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>1536</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>1536</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="400" name="Constant_4765_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="65794040" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="401" name="Constant_4765" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="402" name="__module.bert.encoder.model.layer.3.mlp.wo/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="417,hidden_states.25">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="403" name="__module.bert.encoder.model.layer.3.mlp/aten::add/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="418,input.89">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="404" name="__module.bert.encoder.model.layer.3.mlp.layernorm/aten::layer_norm/Multiply" type="Const" version="opset1">
			<data element_type="i32" shape="1" offset="46892548" size="4" />
			<output>
				<port id="0" precision="I32">
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="405" name="__module.bert.encoder.model.layer.3.mlp.layernorm/aten::layer_norm/MVN" type="MVN" version="opset6">
			<data eps="9.999999960041972e-13" normalize_variance="true" eps_mode="INSIDE_SQRT" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</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>384</dim>
				</port>
			</output>
		</layer>
		<layer id="406" name="Constant_4766_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="65794808" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="407" name="Constant_4766" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="408" name="__module.bert.encoder.model.layer.3.mlp.layernorm/aten::layer_norm/Multiply_1" type="Multiply" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="409" name="Constant_4767_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1, 384" offset="65795576" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="410" name="Constant_4767" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="411" name="__module.bert.encoder.model.layer.3.mlp.layernorm/aten::layer_norm/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="422,425,hidden_states">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="412" name="__module.bert.pooler/aten::select/Gather" type="Gather" version="opset8">
			<data batch_dims="0" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="I64" />
				<port id="2" precision="I64" />
			</input>
			<output>
				<port id="3" precision="FP32" names="426,input.91">
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="413" name="self.bert.pooler.dense.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="384, 384" offset="65796344" size="294912" />
			<output>
				<port id="0" precision="FP16" names="self.bert.pooler.dense.weight">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="414" name="self.bert.pooler.dense.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="415" name="__module.bert.pooler.dense/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>384</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="416" name="Constant_4768_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 384" offset="66091256" size="768" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="417" name="Constant_4768" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="418" name="__module.bert.pooler.dense/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="429,input.93">
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="419" name="__module.bert.pooler.activation/aten::tanh/Tanh" type="Tanh" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32" names="430,input.95">
					<dim>-1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="420" name="self.classifier.weight_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 384" offset="66092024" size="768" />
			<output>
				<port id="0" precision="FP16" names="self.classifier.weight">
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="421" name="self.classifier.weight" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</output>
		</layer>
		<layer id="422" name="__module.classifier/aten::linear/MatMul" type="MatMul" version="opset1">
			<data transpose_a="false" transpose_b="true" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>384</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>384</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="423" name="Constant_4769_compressed" type="Const" version="opset1">
			<data element_type="f16" shape="1, 1" offset="66092792" size="2" />
			<output>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="424" name="Constant_4769" type="Convert" version="opset1">
			<data destination_type="f32" />
			<rt_info>
				<attribute name="decompression" version="0" />
			</rt_info>
			<input>
				<port id="0" precision="FP16">
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="425" name="__module.classifier/aten::linear/Add" type="Add" version="opset1">
			<data auto_broadcast="numpy" />
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
				</port>
				<port id="1" precision="FP32">
					<dim>1</dim>
					<dim>1</dim>
				</port>
			</input>
			<output>
				<port id="2" precision="FP32" names="logits">
					<dim>-1</dim>
					<dim>1</dim>
				</port>
			</output>
		</layer>
		<layer id="426" name="Result_2243" type="Result" version="opset1">
			<input>
				<port id="0" precision="FP32">
					<dim>-1</dim>
					<dim>1</dim>
				</port>
			</input>
		</layer>
	</layers>
	<edges>
		<edge from-layer="0" from-port="0" to-layer="10" to-port="0" />
		<edge from-layer="1" from-port="0" to-layer="49" to-port="0" />
		<edge from-layer="2" from-port="0" to-layer="5" to-port="0" />
		<edge from-layer="3" from-port="0" to-layer="4" to-port="0" />
		<edge from-layer="4" from-port="1" to-layer="7" to-port="0" />
		<edge from-layer="5" from-port="1" to-layer="7" to-port="1" />
		<edge from-layer="6" from-port="0" to-layer="7" to-port="2" />
		<edge from-layer="7" from-port="3" to-layer="13" to-port="0" />
		<edge from-layer="8" from-port="0" to-layer="9" to-port="0" />
		<edge from-layer="9" from-port="1" to-layer="12" to-port="0" />
		<edge from-layer="10" from-port="1" to-layer="12" to-port="1" />
		<edge from-layer="11" from-port="0" to-layer="12" to-port="2" />
		<edge from-layer="12" from-port="3" to-layer="13" to-port="1" />
		<edge from-layer="13" from-port="2" to-layer="15" to-port="0" />
		<edge from-layer="14" from-port="0" to-layer="15" to-port="1" />
		<edge from-layer="15" from-port="2" to-layer="18" to-port="0" />
		<edge from-layer="16" from-port="0" to-layer="17" to-port="0" />
		<edge from-layer="17" from-port="1" to-layer="18" to-port="1" />
		<edge from-layer="18" from-port="2" to-layer="21" to-port="0" />
		<edge from-layer="19" from-port="0" to-layer="20" to-port="0" />
		<edge from-layer="20" from-port="1" to-layer="21" to-port="1" />
		<edge from-layer="21" from-port="2" to-layer="24" to-port="0" />
		<edge from-layer="21" from-port="2" to-layer="63" to-port="0" />
		<edge from-layer="21" from-port="2" to-layer="112" to-port="1" />
		<edge from-layer="21" from-port="2" to-layer="34" to-port="0" />
		<edge from-layer="21" from-port="2" to-layer="93" to-port="0" />
		<edge from-layer="22" from-port="0" to-layer="23" to-port="0" />
		<edge from-layer="23" from-port="1" to-layer="24" to-port="1" />
		<edge from-layer="24" from-port="2" to-layer="27" to-port="0" />
		<edge from-layer="25" from-port="0" to-layer="26" to-port="0" />
		<edge from-layer="26" from-port="1" to-layer="27" to-port="1" />
		<edge from-layer="27" from-port="2" to-layer="29" to-port="0" />
		<edge from-layer="28" from-port="0" to-layer="29" to-port="1" />
		<edge from-layer="29" from-port="2" 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="42" to-port="0" />
		<edge from-layer="32" from-port="0" to-layer="33" 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="37" to-port="0" />
		<edge from-layer="35" from-port="0" to-layer="36" to-port="0" />
		<edge from-layer="36" from-port="1" to-layer="37" to-port="1" />
		<edge from-layer="37" from-port="2" 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="41" to-port="0" />
		<edge from-layer="40" from-port="0" to-layer="41" to-port="1" />
		<edge from-layer="41" from-port="2" to-layer="42" to-port="1" />
		<edge from-layer="42" from-port="2" to-layer="45" to-port="0" />
		<edge from-layer="43" from-port="0" to-layer="44" to-port="0" />
		<edge from-layer="44" from-port="1" to-layer="45" to-port="1" />
		<edge from-layer="45" from-port="2" to-layer="59" to-port="0" />
		<edge from-layer="46" from-port="0" to-layer="47" to-port="0" />
		<edge from-layer="47" from-port="1" to-layer="56" to-port="0" />
		<edge from-layer="48" from-port="0" to-layer="49" to-port="1" />
		<edge from-layer="48" from-port="0" to-layer="72" to-port="1" />
		<edge from-layer="48" from-port="0" to-layer="412" to-port="2" />
		<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="52" to-port="0" />
		<edge from-layer="52" from-port="1" to-layer="55" to-port="0" />
		<edge from-layer="53" from-port="0" to-layer="54" to-port="0" />
		<edge from-layer="54" from-port="1" to-layer="55" to-port="1" />
		<edge from-layer="55" from-port="2" 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="59" to-port="1" />
		<edge from-layer="58" from-port="2" to-layer="349" to-port="1" />
		<edge from-layer="58" from-port="2" to-layer="175" to-port="1" />
		<edge from-layer="58" from-port="2" to-layer="262" to-port="1" />
		<edge from-layer="59" from-port="2" to-layer="89" to-port="0" />
		<edge from-layer="60" from-port="0" to-layer="61" to-port="0" />
		<edge from-layer="61" from-port="1" to-layer="79" to-port="0" />
		<edge from-layer="62" from-port="0" to-layer="68" to-port="0" />
		<edge from-layer="63" from-port="1" to-layer="66" to-port="0" />
		<edge from-layer="64" from-port="0" to-layer="66" to-port="1" />
		<edge from-layer="65" from-port="0" to-layer="66" to-port="2" />
		<edge from-layer="66" from-port="3" to-layer="83" to-port="0" />
		<edge from-layer="66" from-port="3" to-layer="68" to-port="1" />
		<edge from-layer="67" from-port="0" to-layer="68" to-port="2" />
		<edge from-layer="68" from-port="3" to-layer="69" to-port="0" />
		<edge from-layer="69" from-port="1" to-layer="71" to-port="0" />
		<edge from-layer="69" from-port="1" to-layer="72" to-port="0" />
		<edge from-layer="70" from-port="0" to-layer="75" to-port="1" />
		<edge from-layer="70" from-port="0" to-layer="80" to-port="1" />
		<edge from-layer="70" from-port="0" to-layer="412" to-port="1" />
		<edge from-layer="70" from-port="0" to-layer="71" to-port="1" />
		<edge from-layer="71" from-port="2" to-layer="73" to-port="0" />
		<edge from-layer="72" from-port="2" to-layer="73" to-port="1" />
		<edge from-layer="73" from-port="2" to-layer="74" to-port="0" />
		<edge from-layer="74" from-port="1" to-layer="75" to-port="0" />
		<edge from-layer="75" from-port="2" to-layer="77" to-port="0" />
		<edge from-layer="76" from-port="0" to-layer="77" to-port="1" />
		<edge from-layer="77" from-port="2" to-layer="78" to-port="0" />
		<edge from-layer="78" from-port="1" to-layer="79" to-port="1" />
		<edge from-layer="79" from-port="2" to-layer="80" to-port="0" />
		<edge from-layer="80" from-port="2" to-layer="88" to-port="0" />
		<edge from-layer="81" from-port="0" to-layer="88" to-port="1" />
		<edge from-layer="82" from-port="0" 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="88" to-port="2" />
		<edge from-layer="86" from-port="0" to-layer="88" to-port="3" />
		<edge from-layer="87" from-port="0" to-layer="88" to-port="4" />
		<edge from-layer="88" from-port="5" to-layer="89" to-port="1" />
		<edge from-layer="88" from-port="5" to-layer="176" to-port="1" />
		<edge from-layer="88" from-port="5" to-layer="263" to-port="1" />
		<edge from-layer="88" from-port="5" to-layer="350" to-port="1" />
		<edge from-layer="89" from-port="2" to-layer="90" to-port="0" />
		<edge from-layer="90" from-port="1" to-layer="101" to-port="0" />
		<edge from-layer="91" from-port="0" to-layer="92" to-port="0" />
		<edge from-layer="92" from-port="1" to-layer="93" to-port="1" />
		<edge from-layer="93" from-port="2" to-layer="96" to-port="0" />
		<edge from-layer="94" from-port="0" to-layer="95" to-port="0" />
		<edge from-layer="95" from-port="1" to-layer="96" to-port="1" />
		<edge from-layer="96" from-port="2" to-layer="98" to-port="0" />
		<edge from-layer="97" from-port="0" to-layer="98" to-port="1" />
		<edge from-layer="98" from-port="2" to-layer="100" to-port="0" />
		<edge from-layer="99" from-port="0" to-layer="100" to-port="1" />
		<edge from-layer="100" from-port="2" to-layer="101" to-port="1" />
		<edge from-layer="101" from-port="2" to-layer="103" to-port="0" />
		<edge from-layer="102" from-port="0" to-layer="103" to-port="1" />
		<edge from-layer="103" from-port="2" to-layer="105" to-port="0" />
		<edge from-layer="104" from-port="0" to-layer="105" to-port="1" />
		<edge from-layer="105" from-port="2" to-layer="108" to-port="0" />
		<edge from-layer="106" from-port="0" to-layer="107" to-port="0" />
		<edge from-layer="107" from-port="1" to-layer="108" to-port="1" />
		<edge from-layer="108" from-port="2" to-layer="111" to-port="0" />
		<edge from-layer="109" from-port="0" to-layer="110" to-port="0" />
		<edge from-layer="110" from-port="1" to-layer="111" to-port="1" />
		<edge from-layer="111" from-port="2" to-layer="112" to-port="0" />
		<edge from-layer="112" from-port="2" to-layer="114" to-port="0" />
		<edge from-layer="113" from-port="0" to-layer="114" to-port="1" />
		<edge from-layer="114" from-port="2" to-layer="117" to-port="0" />
		<edge from-layer="115" from-port="0" to-layer="116" to-port="0" />
		<edge from-layer="116" from-port="1" to-layer="117" to-port="1" />
		<edge from-layer="117" from-port="2" to-layer="120" to-port="0" />
		<edge from-layer="118" from-port="0" to-layer="119" to-port="0" />
		<edge from-layer="119" from-port="1" to-layer="120" to-port="1" />
		<edge from-layer="120" from-port="2" to-layer="123" to-port="0" />
		<edge from-layer="120" from-port="2" to-layer="142" to-port="1" />
		<edge from-layer="121" from-port="0" to-layer="122" to-port="0" />
		<edge from-layer="122" from-port="1" to-layer="123" to-port="1" />
		<edge from-layer="123" from-port="2" to-layer="128" to-port="0" />
		<edge from-layer="123" from-port="2" to-layer="134" to-port="0" />
		<edge from-layer="124" from-port="0" to-layer="128" to-port="1" />
		<edge from-layer="125" from-port="0" to-layer="128" to-port="2" />
		<edge from-layer="126" from-port="0" to-layer="128" to-port="3" />
		<edge from-layer="127" from-port="0" to-layer="128" to-port="4" />
		<edge from-layer="128" from-port="5" to-layer="129" to-port="0" />
		<edge from-layer="129" from-port="1" to-layer="135" to-port="0" />
		<edge from-layer="130" from-port="0" to-layer="134" to-port="1" />
		<edge from-layer="131" from-port="0" to-layer="134" to-port="2" />
		<edge from-layer="132" from-port="0" to-layer="134" to-port="3" />
		<edge from-layer="133" from-port="0" to-layer="134" to-port="4" />
		<edge from-layer="134" from-port="5" to-layer="135" to-port="1" />
		<edge from-layer="135" from-port="2" 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="141" to-port="0" />
		<edge from-layer="139" from-port="0" to-layer="140" to-port="0" />
		<edge from-layer="140" from-port="1" to-layer="141" to-port="1" />
		<edge from-layer="141" from-port="2" to-layer="142" to-port="0" />
		<edge from-layer="142" from-port="2" to-layer="144" to-port="0" />
		<edge from-layer="143" from-port="0" to-layer="144" to-port="1" />
		<edge from-layer="144" from-port="2" to-layer="147" to-port="0" />
		<edge from-layer="145" from-port="0" to-layer="146" to-port="0" />
		<edge from-layer="146" from-port="1" to-layer="147" to-port="1" />
		<edge from-layer="147" from-port="2" to-layer="150" to-port="0" />
		<edge from-layer="148" from-port="0" to-layer="149" to-port="0" />
		<edge from-layer="149" from-port="1" to-layer="150" to-port="1" />
		<edge from-layer="150" from-port="2" to-layer="153" to-port="0" />
		<edge from-layer="150" from-port="2" to-layer="163" to-port="0" />
		<edge from-layer="150" from-port="2" to-layer="199" to-port="1" />
		<edge from-layer="150" from-port="2" to-layer="180" to-port="0" />
		<edge from-layer="151" from-port="0" to-layer="152" to-port="0" />
		<edge from-layer="152" from-port="1" to-layer="153" to-port="1" />
		<edge from-layer="153" from-port="2" to-layer="156" to-port="0" />
		<edge from-layer="154" from-port="0" to-layer="155" to-port="0" />
		<edge from-layer="155" from-port="1" 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="160" to-port="0" />
		<edge from-layer="159" from-port="0" to-layer="160" to-port="1" />
		<edge from-layer="160" from-port="2" to-layer="171" 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="166" to-port="0" />
		<edge from-layer="164" from-port="0" to-layer="165" to-port="0" />
		<edge from-layer="165" from-port="1" to-layer="166" to-port="1" />
		<edge from-layer="166" from-port="2" to-layer="168" to-port="0" />
		<edge from-layer="167" from-port="0" to-layer="168" to-port="1" />
		<edge from-layer="168" from-port="2" to-layer="170" to-port="0" />
		<edge from-layer="169" from-port="0" to-layer="170" to-port="1" />
		<edge from-layer="170" from-port="2" to-layer="171" to-port="1" />
		<edge from-layer="171" from-port="2" to-layer="174" to-port="0" />
		<edge from-layer="172" from-port="0" to-layer="173" to-port="0" />
		<edge from-layer="173" from-port="1" to-layer="174" to-port="1" />
		<edge from-layer="174" from-port="2" to-layer="175" to-port="0" />
		<edge from-layer="175" from-port="2" to-layer="176" to-port="0" />
		<edge from-layer="176" from-port="2" to-layer="177" to-port="0" />
		<edge from-layer="177" from-port="1" to-layer="188" to-port="0" />
		<edge from-layer="178" from-port="0" to-layer="179" to-port="0" />
		<edge from-layer="179" from-port="1" to-layer="180" to-port="1" />
		<edge from-layer="180" from-port="2" 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="1" />
		<edge from-layer="188" from-port="2" to-layer="190" to-port="0" />
		<edge from-layer="189" from-port="0" to-layer="190" to-port="1" />
		<edge from-layer="190" from-port="2" to-layer="192" to-port="0" />
		<edge from-layer="191" from-port="0" to-layer="192" to-port="1" />
		<edge from-layer="192" from-port="2" to-layer="195" 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="195" to-port="1" />
		<edge from-layer="195" from-port="2" to-layer="198" to-port="0" />
		<edge from-layer="196" from-port="0" to-layer="197" to-port="0" />
		<edge from-layer="197" from-port="1" to-layer="198" to-port="1" />
		<edge from-layer="198" from-port="2" to-layer="199" to-port="0" />
		<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="204" to-port="0" />
		<edge from-layer="202" from-port="0" to-layer="203" to-port="0" />
		<edge from-layer="203" from-port="1" to-layer="204" to-port="1" />
		<edge from-layer="204" from-port="2" to-layer="207" to-port="0" />
		<edge from-layer="205" from-port="0" to-layer="206" to-port="0" />
		<edge from-layer="206" from-port="1" to-layer="207" to-port="1" />
		<edge from-layer="207" from-port="2" to-layer="229" to-port="1" />
		<edge from-layer="207" from-port="2" to-layer="210" to-port="0" />
		<edge from-layer="208" from-port="0" to-layer="209" to-port="0" />
		<edge from-layer="209" from-port="1" to-layer="210" to-port="1" />
		<edge from-layer="210" from-port="2" to-layer="215" to-port="0" />
		<edge from-layer="210" from-port="2" to-layer="221" to-port="0" />
		<edge from-layer="211" from-port="0" to-layer="215" to-port="1" />
		<edge from-layer="212" from-port="0" to-layer="215" to-port="2" />
		<edge from-layer="213" from-port="0" to-layer="215" to-port="3" />
		<edge from-layer="214" from-port="0" to-layer="215" to-port="4" />
		<edge from-layer="215" from-port="5" to-layer="216" to-port="0" />
		<edge from-layer="216" from-port="1" to-layer="222" to-port="0" />
		<edge from-layer="217" from-port="0" to-layer="221" to-port="1" />
		<edge from-layer="218" from-port="0" to-layer="221" to-port="2" />
		<edge from-layer="219" from-port="0" to-layer="221" to-port="3" />
		<edge from-layer="220" from-port="0" to-layer="221" to-port="4" />
		<edge from-layer="221" from-port="5" to-layer="222" to-port="1" />
		<edge from-layer="222" from-port="2" to-layer="225" to-port="0" />
		<edge from-layer="223" from-port="0" to-layer="224" to-port="0" />
		<edge from-layer="224" from-port="1" to-layer="225" to-port="1" />
		<edge from-layer="225" from-port="2" to-layer="228" to-port="0" />
		<edge from-layer="226" from-port="0" to-layer="227" to-port="0" />
		<edge from-layer="227" from-port="1" to-layer="228" to-port="1" />
		<edge from-layer="228" from-port="2" to-layer="229" to-port="0" />
		<edge from-layer="229" from-port="2" to-layer="231" to-port="0" />
		<edge from-layer="230" from-port="0" to-layer="231" to-port="1" />
		<edge from-layer="231" from-port="2" to-layer="234" to-port="0" />
		<edge from-layer="232" from-port="0" to-layer="233" to-port="0" />
		<edge from-layer="233" from-port="1" to-layer="234" to-port="1" />
		<edge from-layer="234" from-port="2" to-layer="237" to-port="0" />
		<edge from-layer="235" from-port="0" to-layer="236" to-port="0" />
		<edge from-layer="236" from-port="1" to-layer="237" to-port="1" />
		<edge from-layer="237" from-port="2" to-layer="267" to-port="0" />
		<edge from-layer="237" from-port="2" to-layer="240" to-port="0" />
		<edge from-layer="237" from-port="2" to-layer="286" to-port="1" />
		<edge from-layer="237" from-port="2" to-layer="250" to-port="0" />
		<edge from-layer="238" from-port="0" to-layer="239" to-port="0" />
		<edge from-layer="239" from-port="1" to-layer="240" to-port="1" />
		<edge from-layer="240" from-port="2" to-layer="243" to-port="0" />
		<edge from-layer="241" from-port="0" to-layer="242" to-port="0" />
		<edge from-layer="242" from-port="1" 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="258" to-port="0" />
		<edge from-layer="248" from-port="0" to-layer="249" to-port="0" />
		<edge from-layer="249" from-port="1" to-layer="250" to-port="1" />
		<edge from-layer="250" from-port="2" to-layer="253" to-port="0" />
		<edge from-layer="251" from-port="0" to-layer="252" to-port="0" />
		<edge from-layer="252" from-port="1" to-layer="253" to-port="1" />
		<edge from-layer="253" from-port="2" to-layer="255" to-port="0" />
		<edge from-layer="254" from-port="0" to-layer="255" to-port="1" />
		<edge from-layer="255" from-port="2" to-layer="257" to-port="0" />
		<edge from-layer="256" from-port="0" to-layer="257" to-port="1" />
		<edge from-layer="257" from-port="2" to-layer="258" to-port="1" />
		<edge from-layer="258" from-port="2" to-layer="261" to-port="0" />
		<edge from-layer="259" from-port="0" to-layer="260" to-port="0" />
		<edge from-layer="260" from-port="1" to-layer="261" to-port="1" />
		<edge from-layer="261" from-port="2" to-layer="262" to-port="0" />
		<edge from-layer="262" from-port="2" to-layer="263" to-port="0" />
		<edge from-layer="263" from-port="2" to-layer="264" to-port="0" />
		<edge from-layer="264" from-port="1" to-layer="275" to-port="0" />
		<edge from-layer="265" from-port="0" to-layer="266" to-port="0" />
		<edge from-layer="266" from-port="1" to-layer="267" to-port="1" />
		<edge from-layer="267" from-port="2" to-layer="270" to-port="0" />
		<edge from-layer="268" from-port="0" to-layer="269" to-port="0" />
		<edge from-layer="269" from-port="1" to-layer="270" to-port="1" />
		<edge from-layer="270" from-port="2" to-layer="272" to-port="0" />
		<edge from-layer="271" from-port="0" to-layer="272" to-port="1" />
		<edge from-layer="272" from-port="2" to-layer="274" to-port="0" />
		<edge from-layer="273" from-port="0" to-layer="274" to-port="1" />
		<edge from-layer="274" from-port="2" 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="279" to-port="0" />
		<edge from-layer="278" from-port="0" to-layer="279" to-port="1" />
		<edge from-layer="279" from-port="2" to-layer="282" to-port="0" />
		<edge from-layer="280" from-port="0" to-layer="281" to-port="0" />
		<edge from-layer="281" from-port="1" to-layer="282" to-port="1" />
		<edge from-layer="282" from-port="2" to-layer="285" to-port="0" />
		<edge from-layer="283" from-port="0" to-layer="284" to-port="0" />
		<edge from-layer="284" from-port="1" to-layer="285" to-port="1" />
		<edge from-layer="285" from-port="2" to-layer="286" to-port="0" />
		<edge from-layer="286" from-port="2" to-layer="288" to-port="0" />
		<edge from-layer="287" from-port="0" to-layer="288" to-port="1" />
		<edge from-layer="288" from-port="2" to-layer="291" to-port="0" />
		<edge from-layer="289" from-port="0" to-layer="290" to-port="0" />
		<edge from-layer="290" from-port="1" to-layer="291" to-port="1" />
		<edge from-layer="291" from-port="2" to-layer="294" to-port="0" />
		<edge from-layer="292" from-port="0" to-layer="293" to-port="0" />
		<edge from-layer="293" from-port="1" to-layer="294" to-port="1" />
		<edge from-layer="294" from-port="2" to-layer="297" to-port="0" />
		<edge from-layer="294" from-port="2" to-layer="316" to-port="1" />
		<edge from-layer="295" from-port="0" to-layer="296" to-port="0" />
		<edge from-layer="296" from-port="1" to-layer="297" to-port="1" />
		<edge from-layer="297" from-port="2" to-layer="302" to-port="0" />
		<edge from-layer="297" from-port="2" to-layer="308" to-port="0" />
		<edge from-layer="298" from-port="0" to-layer="302" to-port="1" />
		<edge from-layer="299" from-port="0" to-layer="302" to-port="2" />
		<edge from-layer="300" from-port="0" to-layer="302" to-port="3" />
		<edge from-layer="301" from-port="0" to-layer="302" to-port="4" />
		<edge from-layer="302" from-port="5" to-layer="303" to-port="0" />
		<edge from-layer="303" from-port="1" to-layer="309" to-port="0" />
		<edge from-layer="304" from-port="0" to-layer="308" to-port="1" />
		<edge from-layer="305" from-port="0" to-layer="308" to-port="2" />
		<edge from-layer="306" from-port="0" to-layer="308" to-port="3" />
		<edge from-layer="307" from-port="0" to-layer="308" to-port="4" />
		<edge from-layer="308" from-port="5" to-layer="309" to-port="1" />
		<edge from-layer="309" from-port="2" to-layer="312" to-port="0" />
		<edge from-layer="310" from-port="0" to-layer="311" to-port="0" />
		<edge from-layer="311" from-port="1" to-layer="312" to-port="1" />
		<edge from-layer="312" from-port="2" to-layer="315" to-port="0" />
		<edge from-layer="313" from-port="0" 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="316" to-port="0" />
		<edge from-layer="316" from-port="2" to-layer="318" to-port="0" />
		<edge from-layer="317" from-port="0" to-layer="318" to-port="1" />
		<edge from-layer="318" from-port="2" to-layer="321" 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="321" to-port="1" />
		<edge from-layer="321" from-port="2" to-layer="324" to-port="0" />
		<edge from-layer="322" from-port="0" to-layer="323" to-port="0" />
		<edge from-layer="323" from-port="1" to-layer="324" to-port="1" />
		<edge from-layer="324" from-port="2" to-layer="327" to-port="0" />
		<edge from-layer="324" from-port="2" to-layer="354" to-port="0" />
		<edge from-layer="324" from-port="2" to-layer="373" to-port="1" />
		<edge from-layer="324" from-port="2" to-layer="337" to-port="0" />
		<edge from-layer="325" from-port="0" to-layer="326" to-port="0" />
		<edge from-layer="326" from-port="1" to-layer="327" to-port="1" />
		<edge from-layer="327" from-port="2" to-layer="330" to-port="0" />
		<edge from-layer="328" from-port="0" to-layer="329" to-port="0" />
		<edge from-layer="329" from-port="1" to-layer="330" to-port="1" />
		<edge from-layer="330" from-port="2" to-layer="332" to-port="0" />
		<edge from-layer="331" from-port="0" to-layer="332" to-port="1" />
		<edge from-layer="332" from-port="2" to-layer="334" to-port="0" />
		<edge from-layer="333" from-port="0" to-layer="334" to-port="1" />
		<edge from-layer="334" from-port="2" to-layer="345" to-port="0" />
		<edge from-layer="335" from-port="0" to-layer="336" to-port="0" />
		<edge from-layer="336" from-port="1" to-layer="337" to-port="1" />
		<edge from-layer="337" from-port="2" to-layer="340" to-port="0" />
		<edge from-layer="338" from-port="0" to-layer="339" to-port="0" />
		<edge from-layer="339" from-port="1" to-layer="340" to-port="1" />
		<edge from-layer="340" from-port="2" to-layer="342" to-port="0" />
		<edge from-layer="341" from-port="0" to-layer="342" to-port="1" />
		<edge from-layer="342" from-port="2" to-layer="344" to-port="0" />
		<edge from-layer="343" from-port="0" to-layer="344" to-port="1" />
		<edge from-layer="344" from-port="2" to-layer="345" to-port="1" />
		<edge from-layer="345" from-port="2" to-layer="348" to-port="0" />
		<edge from-layer="346" from-port="0" to-layer="347" to-port="0" />
		<edge from-layer="347" from-port="1" to-layer="348" to-port="1" />
		<edge from-layer="348" from-port="2" to-layer="349" to-port="0" />
		<edge from-layer="349" from-port="2" to-layer="350" to-port="0" />
		<edge from-layer="350" from-port="2" to-layer="351" to-port="0" />
		<edge from-layer="351" from-port="1" to-layer="362" to-port="0" />
		<edge from-layer="352" from-port="0" to-layer="353" to-port="0" />
		<edge from-layer="353" from-port="1" to-layer="354" to-port="1" />
		<edge from-layer="354" from-port="2" to-layer="357" to-port="0" />
		<edge from-layer="355" from-port="0" to-layer="356" to-port="0" />
		<edge from-layer="356" from-port="1" to-layer="357" to-port="1" />
		<edge from-layer="357" from-port="2" to-layer="359" to-port="0" />
		<edge from-layer="358" from-port="0" to-layer="359" to-port="1" />
		<edge from-layer="359" from-port="2" to-layer="361" to-port="0" />
		<edge from-layer="360" from-port="0" to-layer="361" to-port="1" />
		<edge from-layer="361" from-port="2" to-layer="362" to-port="1" />
		<edge from-layer="362" from-port="2" to-layer="364" to-port="0" />
		<edge from-layer="363" from-port="0" to-layer="364" to-port="1" />
		<edge from-layer="364" from-port="2" to-layer="366" to-port="0" />
		<edge from-layer="365" from-port="0" to-layer="366" to-port="1" />
		<edge from-layer="366" from-port="2" to-layer="369" to-port="0" />
		<edge from-layer="367" from-port="0" to-layer="368" to-port="0" />
		<edge from-layer="368" from-port="1" to-layer="369" to-port="1" />
		<edge from-layer="369" from-port="2" to-layer="372" to-port="0" />
		<edge from-layer="370" from-port="0" to-layer="371" to-port="0" />
		<edge from-layer="371" from-port="1" to-layer="372" to-port="1" />
		<edge from-layer="372" from-port="2" to-layer="373" to-port="0" />
		<edge from-layer="373" from-port="2" to-layer="375" to-port="0" />
		<edge from-layer="374" from-port="0" to-layer="375" to-port="1" />
		<edge from-layer="375" from-port="2" to-layer="378" to-port="0" />
		<edge from-layer="376" from-port="0" to-layer="377" to-port="0" />
		<edge from-layer="377" from-port="1" to-layer="378" to-port="1" />
		<edge from-layer="378" from-port="2" to-layer="381" to-port="0" />
		<edge from-layer="379" from-port="0" to-layer="380" to-port="0" />
		<edge from-layer="380" from-port="1" to-layer="381" to-port="1" />
		<edge from-layer="381" from-port="2" to-layer="384" to-port="0" />
		<edge from-layer="381" from-port="2" to-layer="403" to-port="1" />
		<edge from-layer="382" from-port="0" to-layer="383" to-port="0" />
		<edge from-layer="383" from-port="1" to-layer="384" to-port="1" />
		<edge from-layer="384" from-port="2" to-layer="389" to-port="0" />
		<edge from-layer="384" from-port="2" to-layer="395" to-port="0" />
		<edge from-layer="385" from-port="0" to-layer="389" to-port="1" />
		<edge from-layer="386" from-port="0" to-layer="389" to-port="2" />
		<edge from-layer="387" from-port="0" to-layer="389" to-port="3" />
		<edge from-layer="388" from-port="0" to-layer="389" to-port="4" />
		<edge from-layer="389" from-port="5" to-layer="390" to-port="0" />
		<edge from-layer="390" from-port="1" to-layer="396" to-port="0" />
		<edge from-layer="391" from-port="0" to-layer="395" to-port="1" />
		<edge from-layer="392" from-port="0" to-layer="395" to-port="2" />
		<edge from-layer="393" from-port="0" to-layer="395" to-port="3" />
		<edge from-layer="394" from-port="0" to-layer="395" to-port="4" />
		<edge from-layer="395" from-port="5" to-layer="396" to-port="1" />
		<edge from-layer="396" from-port="2" to-layer="399" to-port="0" />
		<edge from-layer="397" from-port="0" to-layer="398" to-port="0" />
		<edge from-layer="398" from-port="1" to-layer="399" to-port="1" />
		<edge from-layer="399" from-port="2" to-layer="402" to-port="0" />
		<edge from-layer="400" from-port="0" to-layer="401" to-port="0" />
		<edge from-layer="401" from-port="1" to-layer="402" to-port="1" />
		<edge from-layer="402" from-port="2" to-layer="403" to-port="0" />
		<edge from-layer="403" from-port="2" to-layer="405" to-port="0" />
		<edge from-layer="404" from-port="0" to-layer="405" to-port="1" />
		<edge from-layer="405" from-port="2" to-layer="408" to-port="0" />
		<edge from-layer="406" from-port="0" to-layer="407" to-port="0" />
		<edge from-layer="407" from-port="1" to-layer="408" to-port="1" />
		<edge from-layer="408" from-port="2" to-layer="411" to-port="0" />
		<edge from-layer="409" from-port="0" to-layer="410" to-port="0" />
		<edge from-layer="410" from-port="1" to-layer="411" to-port="1" />
		<edge from-layer="411" from-port="2" to-layer="412" to-port="0" />
		<edge from-layer="412" from-port="3" 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="418" to-port="0" />
		<edge from-layer="416" from-port="0" to-layer="417" to-port="0" />
		<edge from-layer="417" from-port="1" to-layer="418" to-port="1" />
		<edge from-layer="418" from-port="2" to-layer="419" to-port="0" />
		<edge from-layer="419" from-port="1" to-layer="422" to-port="0" />
		<edge from-layer="420" from-port="0" to-layer="421" to-port="0" />
		<edge from-layer="421" from-port="1" to-layer="422" to-port="1" />
		<edge from-layer="422" from-port="2" to-layer="425" to-port="0" />
		<edge from-layer="423" from-port="0" to-layer="424" to-port="0" />
		<edge from-layer="424" from-port="1" to-layer="425" to-port="1" />
		<edge from-layer="425" from-port="2" to-layer="426" 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>
	</rt_info>
</net>