Add Core ML conversion

Core ML conversion, task=text2text-generation, precision=float32

coreml/text2text-generation/decoder_float32_model.mlpackage/Data/com.apple.CoreML/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61504dd91d8eba9d514702d13855a8c05d2244221bd32b37258667cab0f84d57
+size 229521688

coreml/text2text-generation/decoder_float32_model.mlpackage/Manifest.json

@@ -0,0 +1,18 @@
+{
+    "fileFormatVersion": "1.0.0",
+    "itemInfoEntries": {
+        "AA7474CA-FE6E-4D50-9A5F-B22F12CB0F52": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Specification",
+            "name": "model.mlmodel",
+            "path": "com.apple.CoreML/model.mlmodel"
+        },
+        "D7A5A471-5A5D-418F-AC58-FFC869603A38": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Weights",
+            "name": "weights",
+            "path": "com.apple.CoreML/weights"
+        }
+    },
+    "rootModelIdentifier": "AA7474CA-FE6E-4D50-9A5F-B22F12CB0F52"
+}

coreml/text2text-generation/encoder_float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel

@@ -0,0 +1,7070 @@
+�
+L
+	input_ids2Indices of input sequence tokens in the vocabulary*	
+
�
��
+w
+attention_maskXMask to avoid performing attention on padding token indices (1 = not masked, 0 = masked)*	
+
�
��Rc
+last_hidden_stateFSequence of hidden-states at the output of the last layer of the model*����
+1Helsinki-NLP/opus-mt-en-ru (text2text-generation)�3
+#com.github.apple.coremltools.sourcetorch==2.0.1�-
+$com.github.apple.coremltools.version7.0b1�;
+co.huggingface.exporters.nameHelsinki-NLP/opus-mt-en-ru�5
+co.huggingface.exporters.tasktext2text-generation�6
+%co.huggingface.exporters.architecture
+"co.huggingface.exporters.frameworkpytorch�-
+"co.huggingface.exporters.precisionfloat32�#
+transformers_version4.22.0.dev0�ʤ
��
+main��
+ 
+	input_ids
+
+

+�

+%
+attention_mask
+
+

+�
CoreML5â
+CoreML5��last_hidden_state�

+const2
+model_model_shared_weight
+
+��
+�*/
+name'
+
+"
+model_model_shared_weight*B
+val;
+
+��
+�*"
+@model_path/weights/weight.bin@�

+constC
+2model_model_encoder_layers_0_self_attn_q_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_0_self_attn_q_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin��=�

+constL
+4model_model_encoder_layers_0_self_attn_q_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_0_self_attn_q_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin��=�

+constC
+2model_model_encoder_layers_0_self_attn_k_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_0_self_attn_k_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���=�

+constL
+4model_model_encoder_layers_0_self_attn_k_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_0_self_attn_k_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���=�

+constC
+2model_model_encoder_layers_0_self_attn_v_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_0_self_attn_v_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���>�

+constL
+4model_model_encoder_layers_0_self_attn_v_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_0_self_attn_v_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���>�

+constE
+4model_model_encoder_layers_0_self_attn_out_proj_bias
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_0_self_attn_out_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���>�

+constN
+6model_model_encoder_layers_0_self_attn_out_proj_weight
+
+�
+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_0_self_attn_out_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���>�

+constG
+6model_model_encoder_layers_0_self_attn_layer_norm_bias
+

+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_0_self_attn_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���?�

+constI
+8model_model_encoder_layers_0_self_attn_layer_norm_weight
+

+�*N
+nameF
+>
+<":
+8model_model_encoder_layers_0_self_attn_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin���?�

+const6
+%model_model_encoder_layers_0_fc1_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_0_fc1_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin�Ƈ?�

+const?
+'model_model_encoder_layers_0_fc1_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_0_fc1_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���?�

+const6
+%model_model_encoder_layers_0_fc2_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_0_fc2_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���A�

+const?
+'model_model_encoder_layers_0_fc2_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_0_fc2_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���A�

+constC
+2model_model_encoder_layers_0_final_layer_norm_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_0_final_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���C�

+constE
+4model_model_encoder_layers_0_final_layer_norm_weight
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_0_final_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin���C�

+constC
+2model_model_encoder_layers_1_self_attn_q_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_1_self_attn_q_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���C�

+constL
+4model_model_encoder_layers_1_self_attn_q_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_1_self_attn_q_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin�ɈC�

+constC
+2model_model_encoder_layers_1_self_attn_k_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_1_self_attn_k_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���C�

+constL
+4model_model_encoder_layers_1_self_attn_k_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_1_self_attn_k_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���C�

+constC
+2model_model_encoder_layers_1_self_attn_v_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_1_self_attn_v_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin�ۈD�

+constL
+4model_model_encoder_layers_1_self_attn_v_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_1_self_attn_v_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin��D�

+constE
+4model_model_encoder_layers_1_self_attn_out_proj_bias
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_1_self_attn_out_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���D�

+constN
+6model_model_encoder_layers_1_self_attn_out_proj_weight
+
+�
+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_1_self_attn_out_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���D�

+constG
+6model_model_encoder_layers_1_self_attn_layer_norm_bias
+

+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_1_self_attn_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���E�

+constI
+8model_model_encoder_layers_1_self_attn_layer_norm_weight
+

+�*N
+nameF
+>
+<":
+8model_model_encoder_layers_1_self_attn_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin���E�

+const6
+%model_model_encoder_layers_1_fc1_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_1_fc1_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���E�

+const?
+'model_model_encoder_layers_1_fc1_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_1_fc1_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin�މE�

+const6
+%model_model_encoder_layers_1_fc2_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_1_fc2_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin�߉G�

+const?
+'model_model_encoder_layers_1_fc2_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_1_fc2_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin��G�

+constC
+2model_model_encoder_layers_1_final_layer_norm_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_1_final_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���I�

+constE
+4model_model_encoder_layers_1_final_layer_norm_weight
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_1_final_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin���I�

+constC
+2model_model_encoder_layers_2_self_attn_q_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_2_self_attn_q_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���I�

+constL
+4model_model_encoder_layers_2_self_attn_q_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_2_self_attn_q_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���I�

+constC
+2model_model_encoder_layers_2_self_attn_k_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_2_self_attn_k_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���I�

+constL
+4model_model_encoder_layers_2_self_attn_k_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_2_self_attn_k_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���I�

+constC
+2model_model_encoder_layers_2_self_attn_v_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_2_self_attn_v_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���J�

+constL
+4model_model_encoder_layers_2_self_attn_v_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_2_self_attn_v_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin�ÊJ�

+constE
+4model_model_encoder_layers_2_self_attn_out_proj_bias
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_2_self_attn_out_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���J�

+constN
+6model_model_encoder_layers_2_self_attn_out_proj_weight
+
+�
+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_2_self_attn_out_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���J�

+constG
+6model_model_encoder_layers_2_self_attn_layer_norm_bias
+

+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_2_self_attn_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin�ՊK�

+constI
+8model_model_encoder_layers_2_self_attn_layer_norm_weight
+

+�*N
+nameF
+>
+<":
+8model_model_encoder_layers_2_self_attn_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin��K�

+const6
+%model_model_encoder_layers_2_fc1_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_2_fc1_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���K�

+const?
+'model_model_encoder_layers_2_fc1_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_2_fc1_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���K�

+const6
+%model_model_encoder_layers_2_fc2_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_2_fc2_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���M�

+const?
+'model_model_encoder_layers_2_fc2_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_2_fc2_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin�NjM�

+constC
+2model_model_encoder_layers_2_final_layer_norm_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_2_final_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin�ȋO�

+constE
+4model_model_encoder_layers_2_final_layer_norm_weight
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_2_final_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin�؋O�

+constC
+2model_model_encoder_layers_3_self_attn_q_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_3_self_attn_q_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin��O�

+constL
+4model_model_encoder_layers_3_self_attn_q_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_3_self_attn_q_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���O�

+constC
+2model_model_encoder_layers_3_self_attn_k_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_3_self_attn_k_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���O�

+constL
+4model_model_encoder_layers_3_self_attn_k_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_3_self_attn_k_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���O�

+constC
+2model_model_encoder_layers_3_self_attn_v_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_3_self_attn_v_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���P�

+constL
+4model_model_encoder_layers_3_self_attn_v_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_3_self_attn_v_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���P�

+constE
+4model_model_encoder_layers_3_self_attn_out_proj_bias
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_3_self_attn_out_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���P�

+constN
+6model_model_encoder_layers_3_self_attn_out_proj_weight
+
+�
+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_3_self_attn_out_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���P�

+constG
+6model_model_encoder_layers_3_self_attn_layer_norm_bias
+

+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_3_self_attn_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���Q�

+constI
+8model_model_encoder_layers_3_self_attn_layer_norm_weight
+

+�*N
+nameF
+>
+<":
+8model_model_encoder_layers_3_self_attn_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin���Q�

+const6
+%model_model_encoder_layers_3_fc1_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_3_fc1_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin�ΌQ�

+const?
+'model_model_encoder_layers_3_fc1_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_3_fc1_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���Q�

+const6
+%model_model_encoder_layers_3_fc2_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_3_fc2_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���S�

+const?
+'model_model_encoder_layers_3_fc2_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_3_fc2_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���S�

+constC
+2model_model_encoder_layers_3_final_layer_norm_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_3_final_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���U�

+constE
+4model_model_encoder_layers_3_final_layer_norm_weight
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_3_final_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin���U�

+constC
+2model_model_encoder_layers_4_self_attn_q_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_4_self_attn_q_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���U�

+constL
+4model_model_encoder_layers_4_self_attn_q_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_4_self_attn_q_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin�эU�

+constC
+2model_model_encoder_layers_4_self_attn_k_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_4_self_attn_k_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���U�

+constL
+4model_model_encoder_layers_4_self_attn_k_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_4_self_attn_k_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���U�

+constC
+2model_model_encoder_layers_4_self_attn_v_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_4_self_attn_v_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin��V�

+constL
+4model_model_encoder_layers_4_self_attn_v_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_4_self_attn_v_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin��V�

+constE
+4model_model_encoder_layers_4_self_attn_out_proj_bias
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_4_self_attn_out_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���V�

+constN
+6model_model_encoder_layers_4_self_attn_out_proj_weight
+
+�
+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_4_self_attn_out_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���V�

+constG
+6model_model_encoder_layers_4_self_attn_layer_norm_bias
+

+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_4_self_attn_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���W�

+constI
+8model_model_encoder_layers_4_self_attn_layer_norm_weight
+

+�*N
+nameF
+>
+<":
+8model_model_encoder_layers_4_self_attn_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin���W�

+const6
+%model_model_encoder_layers_4_fc1_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_4_fc1_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���W�

+const?
+'model_model_encoder_layers_4_fc1_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_4_fc1_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin��W�

+const6
+%model_model_encoder_layers_4_fc2_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_4_fc2_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin��Y�

+const?
+'model_model_encoder_layers_4_fc2_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_4_fc2_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���Y�

+constC
+2model_model_encoder_layers_4_final_layer_norm_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_4_final_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���[�

+constE
+4model_model_encoder_layers_4_final_layer_norm_weight
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_4_final_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin���[�

+constC
+2model_model_encoder_layers_5_self_attn_q_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_5_self_attn_q_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���[�

+constL
+4model_model_encoder_layers_5_self_attn_q_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_5_self_attn_q_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���[�

+constC
+2model_model_encoder_layers_5_self_attn_k_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_5_self_attn_k_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���[�

+constL
+4model_model_encoder_layers_5_self_attn_k_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_5_self_attn_k_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���[�

+constC
+2model_model_encoder_layers_5_self_attn_v_proj_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_5_self_attn_v_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���\�

+constL
+4model_model_encoder_layers_5_self_attn_v_proj_weight
+
+�
+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_5_self_attn_v_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin�ˏ\�

+constE
+4model_model_encoder_layers_5_self_attn_out_proj_bias
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_5_self_attn_out_proj_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���\�

+constN
+6model_model_encoder_layers_5_self_attn_out_proj_weight
+
+�
+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_5_self_attn_out_proj_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���\�

+constG
+6model_model_encoder_layers_5_self_attn_layer_norm_bias
+

+�*L
+nameD
+<
+:"8
+6model_model_encoder_layers_5_self_attn_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin�ݏ]�

+constI
+8model_model_encoder_layers_5_self_attn_layer_norm_weight
+

+�*N
+nameF
+>
+<":
+8model_model_encoder_layers_5_self_attn_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin��]�

+const6
+%model_model_encoder_layers_5_fc1_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_5_fc1_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���]�

+const?
+'model_model_encoder_layers_5_fc1_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_5_fc1_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin���]�

+const6
+%model_model_encoder_layers_5_fc2_bias
+

+�*;
+name3
++
+)"'
+%model_model_encoder_layers_5_fc2_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin���_�

+const?
+'model_model_encoder_layers_5_fc2_weight
+
+�
+�*=
+name5
+-
++")
+'model_model_encoder_layers_5_fc2_weight*D
+val=
+
+�
+�*%
+@model_path/weights/weight.bin�ϐ_�

+constC
+2model_model_encoder_layers_5_final_layer_norm_bias
+

+�*H
+name@
+8
+6"4
+2model_model_encoder_layers_5_final_layer_norm_bias*=
+val6
+

+�*%
+@model_path/weights/weight.bin�Аa�

+constE
+4model_model_encoder_layers_5_final_layer_norm_weight
+

+�*J
+nameB
+:
+8"6
+4model_model_encoder_layers_5_final_layer_norm_weight*=
+val6
+

+�*%
+@model_path/weights/weight.bin���aO
+const
+var_15
+*
+name
+
+	"
+op_15*
+val
+
+
+
+
+��'7O
+const
+var_16
+*
+name
+
+	"
+op_16*
+val
+
+
+
+
+���O
+const
+var_18
+*
+name
+
+	"
+op_18*
+val
+
+
+
+
+�?U
+const
+var_30
+*
+name
+
+	"
+op_30*
+val
+
+
+
+���������
Z
+const
+
+*"
+name
+
+"
+op_49_axis_0*
+val
+
+
+
�

+gather"
+
x
+
+model_model_shared_weight
+indices
+
+	input_ids
+axis
+
+
+var_49
+
+

+�

+�*
+name
+
+	"
+op_49O
+const
+var_50
+*
+name
+
+	"
+op_50*
+val
+
+
+
+
+��Ay
+mul
+
x
+
+
+var_49
+
y
+
+
+var_50+
+
+
+

+�

+�*#
+name
+
+"
+
+const!
+	embed_pos
+
+�

+�*
+name
+
+
+	embed_pos*D
+val=
+
+�

+�*%
+@model_path/weights/weight.bin��aw
+add
+
x
+
+
+
y
+
+	embed_pos%
+input_5
+
+

+�

+�*
+name
+
+"	
+input_5j
+const
+
+
+

+
*"
+name
+
+"
+op_63_axes_0*
+val
+
+

+

+
+

�

+expand_dims
+
x
+
+attention_mask
+axes
+
+
+var_63
+
+

+

+�
*
+name
+
+	"
+op_63j
+const
+
+
+

+
*"
+name
+
+"
+op_64_axes_0*
+val
+
+

+

+
+
�

+expand_dims
+
x
+
+
+var_63
+axes
+
+
+var_64
+
+

+

+

+�
*
+name
+
+	"
+op_64n
+const
+
+
+

+*"
+name
+
+"
+op_67_reps_0*"
+val
+
+

+
+	
+

�

{
+tile
+
x
+
+
+var_64
+reps
+
+
+var_67 
+
+

+

+�

+�
*
+name
+
+	"
+op_67_
+const
+var_68_dtype_0
+*#
+name
+
+"
+
+val
+
+
+"
+fp32�

+cast
+
x
+
+
+var_67
+dtype
+
+var_68_dtype_0+
+cast_64 
+
+

+

+�

+�
*
+name
+
+"	
+cast_64�

+sub
+
x
+
+
+var_18
+
y
+	
+cast_641
+
+
+

+

+�

+�
*#
+name
+
+"
+
+const
+var_70_dtype_0
+*#
+name
+
+"
+
+val
+
+
+"
+bool�

+cast
+
x
+
+
+dtype
+
+var_70_dtype_0+
+cast_63 
+

+

+

+�

+�
*
+name
+
+"	
+cast_63�

+select
+cond
+	
+cast_63
+
a
+
+
+var_16
+
b
+
+
+attention_mask_1 
+
+

+

+�

+�
*$
+name
+
+"
+attention_mask�

+linear
+
x
+	
+input_5B
+weight8
+6
+4model_model_encoder_layers_0_self_attn_q_proj_weight>
+bias6
+4
+2model_model_encoder_layers_0_self_attn_q_proj_bias$
+var_86
+
+

+�

+�*
+name
+
+	"
+op_86O
+const
+var_87
+*
+name
+
+	"
+op_87*
+val
+
+
+
+
+>o
+mul
+
x
+
+
+var_86
+
y
+
+
+var_87&
+tensor_5
+
+

+�

+�*
+name
+
+"
+
+tensor_5�

+linear
+
x
+	
+input_5B
+weight8
+6
+4model_model_encoder_layers_0_self_attn_k_proj_weight>
+bias6
+4
+2model_model_encoder_layers_0_self_attn_k_proj_bias&
+tensor_1
+
+

+�

+�*
+name
+
+"
+
+tensor_1h
+const
+var_92
+
+

+*
+name
+
+	"
+op_92**
+val#
+
+

+
+
+
+reshape
+
x
+
+
+tensor_1
+shape
+
+
+var_92)
+var_93
+
+

+�

+
+@*
+name
+
+	"
+op_93m
+const
+
+
+

+*"
+name
+
+"
+op_94_perm_0*!
+val
+
+

+
+
+
+
�

+linear
+
x
+	
+input_5B
+weight8
+6
+4model_model_encoder_layers_0_self_attn_v_proj_weight>
+bias6
+4
+2model_model_encoder_layers_0_self_attn_v_proj_bias&
+tensor_3
+
+

+�

+�*
+name
+
+"
+
+tensor_3h
+const
+var_99
+
+

+*
+name
+
+	"
+op_99**
+val#
+
+

+
+
+
+reshape
+
x
+
+
+tensor_3
+shape
+
+
+var_99*
+var_100
+
+

+�

+
+@*
+name
+
+
+"
+op_100o
+const
+var_101_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
b
+const
+var_108
+
+

+*
+name
+
+
+"
+op_108*"
+val
+
+

+
+	
+
�
@|
+reshape
+
x
+
+
+tensor_5
+shape
+	
+var_108*
+var_109
+
+

+�

+
+@*
+name
+
+
+"
+op_109o
+const
+var_110_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
i
+const
+var_112
+
+

+*
+name
+
+
+"
+op_112*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_109
+perm
+
+var_110_perm_0/
+transpose_27
+
+

+
+�

+@*"
+name
+
+"
+transpose_27�

+reshape
+
x
+
+transpose_27
+shape
+	
+var_112+
+query_states_1
+
+
+�

+@*$
+name
+
+"
+query_states_1i
+const
+var_114
+
+

+*
+name
+
+
+"
+op_114*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+
+
+var_93
+perm
+
+
+transpose_29
+
+

+
+�

+@*"
+name
+
+"
+transpose_29�

+reshape
+
x
+
+transpose_29
+shape
+	
+var_114)
+key_states_3
+
+
+�

+@*"
+name
+
+"
+key_states_3i
+const
+var_116
+
+

+*
+name
+
+
+"
+op_116*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_100
+perm
+
+var_101_perm_0/
+transpose_28
+
+

+
+�

+@*"
+name
+
+"
+transpose_28�

+reshape
+
x
+
+transpose_28
+shape
+	
+var_116+
+value_states_3
+
+
+�

+@*$
+name
+
+"
+value_states_3n
+const
+var_119_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+	
+
+
y
+const$
+attn_weights_1_transpose_x_0
+
*2
+name*
+"
+ "
+attn_weights_1_transpose_x_0*
+val
+

+
+
y
+const$
+attn_weights_1_transpose_y_0
+
*2
+name*
+"
+ "
+attn_weights_1_transpose_y_0*
+val
+

+
+
�

+	transpose
+
x
+
+key_states_3
+perm
+
+var_119_perm_0)
+transpose_26
+
+
+@
+�
*"
+name
+
+"
+transpose_26�

+matmul
+
x
+
+query_states_1
+
y
+
+transpose_26/
+transpose_x 
+
+attn_weights_1_transpose_x_0/
+transpose_y 
+
+attn_weights_1_transpose_y_0,
+attn_weights_1
+
+
+�

+�
*$
+name
+
+"
+attn_weights_1c
+const
+var_121
+
+

+*
+name
+
+
+"
+op_121*#
+val
+
+

+
+
+
+
�
�
�

+reshape
+
x
+
+attn_weights_1
+shape
+	
+var_121+
+var_122 
+
+

+
+�

+�
*
+name
+
+
+"
+op_122�

+add
+
x
+	
+var_122
+
y
+
+attention_mask_12
+attn_weights_3 
+
+

+
+�

+�
*$
+name
+
+"
+attn_weights_3b
+const
+var_127
+
+

+*
+name
+
+
+"
+op_127*"
+val
+
+

+
+	
+�
�
~
+reshape
+
x
+
+attn_weights_3
+shape
+	
+var_127%
+input_7
+
+
+�

+�
*
+name
+
+"	
+input_7u
+softmax
+
x
+	
+input_7
+axis
+
+
+var_30%
+input_9
+
+
+�

+�
*
+name
+
+"	
+input_9w
+const#
+attn_output_1_transpose_x_0
+
*1
+name)
+!
+"
+attn_output_1_transpose_x_0*
+val
+

+
+
w
+const#
+attn_output_1_transpose_y_0
+
*1
+name)
+!
+"
+attn_output_1_transpose_y_0*
+val
+

+
+
�

+matmul
+
x
+	
+input_9
+
y
+
+value_states_3.
+transpose_x
+
+attn_output_1_transpose_x_0.
+transpose_y
+
+attn_output_1_transpose_y_0*
+
+
+
+�

+@*#
+name
+
+"
+
+const
+var_132
+
+

+*
+name
+
+
+"
+op_132*"
+val
+
+

+
+	
+
�
@�

+reshape
+
x
+
+
+shape
+	
+var_1320
+
+
+

+
+�

+@*#
+name
+
+"
+
+const$
+attn_output_5_perm_0
+
+

+**
+name"
+
+"
+attn_output_5_perm_0*!
+val
+
+

+
+
+
+
b
+const
+var_135
+
+

+*
+name
+
+
+"
+op_135*"
+val
+
+

+
+	
+
�
��

+	transpose
+
x
+
+
+perm
+
+attn_output_5_perm_0/
+transpose_25
+
+

+�

+
+@*"
+name
+
+"
+transpose_25~
+reshape
+
x
+
+transpose_25
+shape
+	
+var_135&
+input_11
+
+

+�

+�*
+name
+
+"
+
+input_11�

+linear
+
x
+
+
+input_11D
+weight:
+8
+6model_model_encoder_layers_0_self_attn_out_proj_weight@
+bias8
+6
+4model_model_encoder_layers_0_self_attn_out_proj_bias&
+input_13
+
+

+�

+�*
+name
+
+"
+
+input_13r
+add
+
x
+	
+input_5
+
y
+
+
+input_13&
+input_15
+
+

+�

+�*
+name
+
+"
+
+input_15x
+const
+input_17_axes_0
+
+

+
*%
+name
+
+"
+input_17_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+
+input_15
+axes
+
+input_17_axes_0E
+gamma<
+:
+8model_model_encoder_layers_0_self_attn_layer_norm_weightB
+beta:
+8
+6model_model_encoder_layers_0_self_attn_layer_norm_bias
+epsilon
+
+
+var_15&
+input_17
+
+

+�

+�*
+name
+
+"
+
+input_17�

+linear
+
x
+
+
+input_175
+weight+
+)
+'model_model_encoder_layers_0_fc1_weight1
+bias)
+'
+%model_model_encoder_layers_0_fc1_bias&
+input_19
+
+

+�

+�*
+name
+
+"
+
+input_19a
+silu
+
x
+
+
+input_19&
+input_21
+
+

+�

+�*
+name
+
+"
+
+input_21�

+linear
+
x
+
+
+input_215
+weight+
+)
+'model_model_encoder_layers_0_fc2_weight1
+bias)
+'
+%model_model_encoder_layers_0_fc2_bias&
+input_25
+
+

+�

+�*
+name
+
+"
+
+input_25s
+add
+
x
+
+
+input_17
+
y
+
+
+input_25&
+input_27
+
+

+�

+�*
+name
+
+"
+
+input_27�

+const&
+hidden_states_7_axes_0
+
+

+
*,
+name$
+
+"
+hidden_states_7_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+
+input_27"
+axes
+
+hidden_states_7_axes_0A
+gamma8
+6
+4model_model_encoder_layers_0_final_layer_norm_weight>
+beta6
+4
+2model_model_encoder_layers_0_final_layer_norm_bias
+epsilon
+
+
+var_15-
+hidden_states_7
+
+

+�

+�*%
+name
+
+"
+hidden_states_7�

+linear
+
x
+
+hidden_states_7B
+weight8
+6
+4model_model_encoder_layers_1_self_attn_q_proj_weight>
+bias6
+4
+2model_model_encoder_layers_1_self_attn_q_proj_bias%
+var_174
+
+

+�

+�*
+name
+
+
+"
+op_174Q
+const
+var_175
+*
+name
+
+
+"
+op_175*
+val
+
+
+
+
+>s
+mul
+
x
+	
+var_174
+
y
+	
+var_175'
+	tensor_11
+
+

+�

+�*
+name
+
+
+	tensor_11�

+linear
+
x
+
+hidden_states_7B
+weight8
+6
+4model_model_encoder_layers_1_self_attn_k_proj_weight>
+bias6
+4
+2model_model_encoder_layers_1_self_attn_k_proj_bias&
+tensor_7
+
+

+�

+�*
+name
+
+"
+
+tensor_7j
+const
+var_180
+
+

+*
+name
+
+
+"
+op_180**
+val#
+
+

+
+
+
+reshape
+
x
+
+
+tensor_7
+shape
+	
+var_180*
+var_181
+
+

+�

+
+@*
+name
+
+
+"
+op_181o
+const
+var_182_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
�

+linear
+
x
+
+hidden_states_7B
+weight8
+6
+4model_model_encoder_layers_1_self_attn_v_proj_weight>
+bias6
+4
+2model_model_encoder_layers_1_self_attn_v_proj_bias&
+tensor_9
+
+

+�

+�*
+name
+
+"
+
+tensor_9j
+const
+var_187
+
+

+*
+name
+
+
+"
+op_187**
+val#
+
+

+
+
+
+reshape
+
x
+
+
+tensor_9
+shape
+	
+var_187*
+var_188
+
+

+�

+
+@*
+name
+
+
+"
+op_188o
+const
+var_189_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
b
+const
+var_196
+
+

+*
+name
+
+
+"
+op_196*"
+val
+
+

+
+	
+
�
@}
+reshape
+
x
+
+	tensor_11
+shape
+	
+var_196*
+var_197
+
+

+�

+
+@*
+name
+
+
+"
+op_197o
+const
+var_198_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
i
+const
+var_200
+
+

+*
+name
+
+
+"
+op_200*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_197
+perm
+
+var_198_perm_0/
+transpose_22
+
+

+
+�

+@*"
+name
+
+"
+transpose_22�

+reshape
+
x
+
+transpose_22
+shape
+	
+var_200+
+query_states_3
+
+
+�

+@*$
+name
+
+"
+query_states_3i
+const
+var_202
+
+

+*
+name
+
+
+"
+op_202*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_181
+perm
+
+var_182_perm_0/
+transpose_24
+
+

+
+�

+@*"
+name
+
+"
+transpose_24�

+reshape
+
x
+
+transpose_24
+shape
+	
+var_202)
+key_states_7
+
+
+�

+@*"
+name
+
+"
+key_states_7i
+const
+var_204
+
+

+*
+name
+
+
+"
+op_204*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_188
+perm
+
+var_189_perm_0/
+transpose_23
+
+

+
+�

+@*"
+name
+
+"
+transpose_23�

+reshape
+
x
+
+transpose_23
+shape
+	
+var_204+
+value_states_7
+
+
+�

+@*$
+name
+
+"
+value_states_7n
+const
+var_207_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+	
+
+
y
+const$
+attn_weights_7_transpose_x_0
+
*2
+name*
+"
+ "
+attn_weights_7_transpose_x_0*
+val
+

+
+
y
+const$
+attn_weights_7_transpose_y_0
+
*2
+name*
+"
+ "
+attn_weights_7_transpose_y_0*
+val
+

+
+
�

+	transpose
+
x
+
+key_states_7
+perm
+
+var_207_perm_0)
+transpose_21
+
+
+@
+�
*"
+name
+
+"
+transpose_21�

+matmul
+
x
+
+query_states_3
+
y
+
+transpose_21/
+transpose_x 
+
+attn_weights_7_transpose_x_0/
+transpose_y 
+
+attn_weights_7_transpose_y_0,
+attn_weights_7
+
+
+�

+�
*$
+name
+
+"
+attn_weights_7c
+const
+var_209
+
+

+*
+name
+
+
+"
+op_209*#
+val
+
+

+
+
+
+
�
�
�

+reshape
+
x
+
+attn_weights_7
+shape
+	
+var_209+
+var_210 
+
+

+
+�

+�
*
+name
+
+
+"
+op_210�

+add
+
x
+	
+var_210
+
y
+
+attention_mask_12
+attn_weights_9 
+
+

+
+�

+�
*$
+name
+
+"
+attn_weights_9b
+const
+var_215
+
+

+*
+name
+
+
+"
+op_215*"
+val
+
+

+
+	
+�
�
�

+reshape
+
x
+
+attn_weights_9
+shape
+	
+var_215&
+input_29
+
+
+�

+�
*
+name
+
+"
+
+input_29x
+softmax
+
x
+
+
+input_29
+axis
+
+
+var_30&
+input_31
+
+
+�

+�
*
+name
+
+"
+
+input_31w
+const#
+attn_output_7_transpose_x_0
+
*1
+name)
+!
+"
+attn_output_7_transpose_x_0*
+val
+

+
+
w
+const#
+attn_output_7_transpose_y_0
+
*1
+name)
+!
+"
+attn_output_7_transpose_y_0*
+val
+

+
+
�

+matmul
+
x
+
+
+input_31
+
y
+
+value_states_7.
+transpose_x
+
+attn_output_7_transpose_x_0.
+transpose_y
+
+attn_output_7_transpose_y_0*
+
+
+
+�

+@*#
+name
+
+"
+
+const
+var_220
+
+

+*
+name
+
+
+"
+op_220*"
+val
+
+

+
+	
+
�
@�

+reshape
+
x
+
+
+shape
+	
+var_2200
+
+
+

+
+�

+@*#
+name
+
+"
+
+const%
+attn_output_11_perm_0
+
+

+*+
+name#
+
+"
+attn_output_11_perm_0*!
+val
+
+

+
+
+
+
b
+const
+var_223
+
+

+*
+name
+
+
+"
+op_223*"
+val
+
+

+
+	
+
�
��

+	transpose
+
x
+
+
+perm
+
+attn_output_11_perm_0/
+transpose_20
+
+

+�

+
+@*"
+name
+
+"
+transpose_20~
+reshape
+
x
+
+transpose_20
+shape
+	
+var_223&
+input_33
+
+

+�

+�*
+name
+
+"
+
+input_33�

+linear
+
x
+
+
+input_33D
+weight:
+8
+6model_model_encoder_layers_1_self_attn_out_proj_weight@
+bias8
+6
+4model_model_encoder_layers_1_self_attn_out_proj_bias&
+input_35
+
+

+�

+�*
+name
+
+"
+
+input_35z
+add
+
x
+
+hidden_states_7
+
y
+
+
+input_35&
+input_37
+
+

+�

+�*
+name
+
+"
+
+input_37x
+const
+input_39_axes_0
+
+

+
*%
+name
+
+"
+input_39_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+
+input_37
+axes
+
+input_39_axes_0E
+gamma<
+:
+8model_model_encoder_layers_1_self_attn_layer_norm_weightB
+beta:
+8
+6model_model_encoder_layers_1_self_attn_layer_norm_bias
+epsilon
+
+
+var_15&
+input_39
+
+

+�

+�*
+name
+
+"
+
+input_39�

+linear
+
x
+
+
+input_395
+weight+
+)
+'model_model_encoder_layers_1_fc1_weight1
+bias)
+'
+%model_model_encoder_layers_1_fc1_bias&
+input_41
+
+

+�

+�*
+name
+
+"
+
+input_41a
+silu
+
x
+
+
+input_41&
+input_43
+
+

+�

+�*
+name
+
+"
+
+input_43�

+linear
+
x
+
+
+input_435
+weight+
+)
+'model_model_encoder_layers_1_fc2_weight1
+bias)
+'
+%model_model_encoder_layers_1_fc2_bias&
+input_47
+
+

+�

+�*
+name
+
+"
+
+input_47s
+add
+
x
+
+
+input_39
+
y
+
+
+input_47&
+input_49
+
+

+�

+�*
+name
+
+"
+
+input_49�

+const'
+hidden_states_13_axes_0
+
+

+
*-
+name%
+
+"
+hidden_states_13_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+
+input_49#
+axes
+
+hidden_states_13_axes_0A
+gamma8
+6
+4model_model_encoder_layers_1_final_layer_norm_weight>
+beta6
+4
+2model_model_encoder_layers_1_final_layer_norm_bias
+epsilon
+
+
+var_15.
+hidden_states_13
+
+

+�

+�*&
+name
+
+"
+hidden_states_13�

+linear
+
x
+
+hidden_states_13B
+weight8
+6
+4model_model_encoder_layers_2_self_attn_q_proj_weight>
+bias6
+4
+2model_model_encoder_layers_2_self_attn_q_proj_bias%
+var_262
+
+

+�

+�*
+name
+
+
+"
+op_262Q
+const
+var_263
+*
+name
+
+
+"
+op_263*
+val
+
+
+
+
+>s
+mul
+
x
+	
+var_262
+
y
+	
+var_263'
+	tensor_17
+
+

+�

+�*
+name
+
+
+	tensor_17�

+linear
+
x
+
+hidden_states_13B
+weight8
+6
+4model_model_encoder_layers_2_self_attn_k_proj_weight>
+bias6
+4
+2model_model_encoder_layers_2_self_attn_k_proj_bias'
+	tensor_13
+
+

+�

+�*
+name
+
+
+	tensor_13j
+const
+var_268
+
+

+*
+name
+
+
+"
+op_268**
+val#
+
+

+
+
+
+reshape
+
x
+
+	tensor_13
+shape
+	
+var_268*
+var_269
+
+

+�

+
+@*
+name
+
+
+"
+op_269o
+const
+var_270_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
�

+linear
+
x
+
+hidden_states_13B
+weight8
+6
+4model_model_encoder_layers_2_self_attn_v_proj_weight>
+bias6
+4
+2model_model_encoder_layers_2_self_attn_v_proj_bias'
+	tensor_15
+
+

+�

+�*
+name
+
+
+	tensor_15j
+const
+var_275
+
+

+*
+name
+
+
+"
+op_275**
+val#
+
+

+
+
+
+reshape
+
x
+
+	tensor_15
+shape
+	
+var_275*
+var_276
+
+

+�

+
+@*
+name
+
+
+"
+op_276o
+const
+var_277_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
b
+const
+var_284
+
+

+*
+name
+
+
+"
+op_284*"
+val
+
+

+
+	
+
�
@}
+reshape
+
x
+
+	tensor_17
+shape
+	
+var_284*
+var_285
+
+

+�

+
+@*
+name
+
+
+"
+op_285o
+const
+var_286_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
i
+const
+var_288
+
+

+*
+name
+
+
+"
+op_288*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_285
+perm
+
+var_286_perm_0/
+transpose_17
+
+

+
+�

+@*"
+name
+
+"
+transpose_17�

+reshape
+
x
+
+transpose_17
+shape
+	
+var_288+
+query_states_5
+
+
+�

+@*$
+name
+
+"
+query_states_5i
+const
+var_290
+
+

+*
+name
+
+
+"
+op_290*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_269
+perm
+
+var_270_perm_0/
+transpose_19
+
+

+
+�

+@*"
+name
+
+"
+transpose_19�

+reshape
+
x
+
+transpose_19
+shape
+	
+var_290*
+
+
+
+�

+@*#
+name
+
+"
+
+const
+var_292
+
+

+*
+name
+
+
+"
+op_292*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_276
+perm
+
+var_277_perm_0/
+transpose_18
+
+

+
+�

+@*"
+name
+
+"
+transpose_18�

+reshape
+
x
+
+transpose_18
+shape
+	
+var_292,
+value_states_11
+
+
+�

+@*%
+name
+
+"
+value_states_11n
+const
+var_295_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+	
+
+
{
+const%
+attn_weights_13_transpose_x_0
+
*3
+name+
+#
+!"
+attn_weights_13_transpose_x_0*
+val
+

+
+
{
+const%
+attn_weights_13_transpose_y_0
+
*3
+name+
+#
+!"
+attn_weights_13_transpose_y_0*
+val
+

+
+
�

+	transpose
+
x
+
+
+perm
+
+var_295_perm_0)
+transpose_16
+
+
+@
+�
*"
+name
+
+"
+transpose_16�

+matmul
+
x
+
+query_states_5
+
y
+
+transpose_160
+transpose_x!
+
+attn_weights_13_transpose_x_00
+transpose_y!
+
+attn_weights_13_transpose_y_0-
+attn_weights_13
+
+
+�

+�
*%
+name
+
+"
+attn_weights_13c
+const
+var_297
+
+

+*
+name
+
+
+"
+op_297*#
+val
+
+

+
+
+
+
�
�
�

+reshape
+
x
+
+attn_weights_13
+shape
+	
+var_297+
+var_298 
+
+

+
+�

+�
*
+name
+
+
+"
+op_298�

+add
+
x
+	
+var_298
+
y
+
+attention_mask_13
+attn_weights_15 
+
+

+
+�

+�
*%
+name
+
+"
+attn_weights_15b
+const
+var_303
+
+

+*
+name
+
+
+"
+op_303*"
+val
+
+

+
+	
+�
�
�

+reshape
+
x
+
+attn_weights_15
+shape
+	
+var_303&
+input_51
+
+
+�

+�
*
+name
+
+"
+
+input_51x
+softmax
+
x
+
+
+input_51
+axis
+
+
+var_30&
+input_53
+
+
+�

+�
*
+name
+
+"
+
+input_53y
+const$
+attn_output_13_transpose_x_0
+
*2
+name*
+"
+ "
+attn_output_13_transpose_x_0*
+val
+

+
+
y
+const$
+attn_output_13_transpose_y_0
+
*2
+name*
+"
+ "
+attn_output_13_transpose_y_0*
+val
+

+
+
�

+matmul
+
x
+
+
+input_53
+
y
+
+value_states_11/
+transpose_x 
+
+attn_output_13_transpose_x_0/
+transpose_y 
+
+attn_output_13_transpose_y_0+
+attn_output_13
+
+
+�

+@*$
+name
+
+"
+attn_output_13b
+const
+var_308
+
+

+*
+name
+
+
+"
+op_308*"
+val
+
+

+
+	
+
�
@�

+reshape
+
x
+
+attn_output_13
+shape
+	
+var_3081
+attn_output_15
+
+

+
+�

+@*$
+name
+
+"
+attn_output_15~
+const%
+attn_output_17_perm_0
+
+

+*+
+name#
+
+"
+attn_output_17_perm_0*!
+val
+
+

+
+
+
+
b
+const
+var_311
+
+

+*
+name
+
+
+"
+op_311*"
+val
+
+

+
+	
+
�
��

+	transpose
+
x
+
+attn_output_15!
+perm
+
+attn_output_17_perm_0/
+transpose_15
+
+

+�

+
+@*"
+name
+
+"
+transpose_15~
+reshape
+
x
+
+transpose_15
+shape
+	
+var_311&
+input_55
+
+

+�

+�*
+name
+
+"
+
+input_55�

+linear
+
x
+
+
+input_55D
+weight:
+8
+6model_model_encoder_layers_2_self_attn_out_proj_weight@
+bias8
+6
+4model_model_encoder_layers_2_self_attn_out_proj_bias&
+input_57
+
+

+�

+�*
+name
+
+"
+
+input_57{
+add
+
x
+
+hidden_states_13
+
y
+
+
+input_57&
+input_59
+
+

+�

+�*
+name
+
+"
+
+input_59x
+const
+input_61_axes_0
+
+

+
*%
+name
+
+"
+input_61_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+
+input_59
+axes
+
+input_61_axes_0E
+gamma<
+:
+8model_model_encoder_layers_2_self_attn_layer_norm_weightB
+beta:
+8
+6model_model_encoder_layers_2_self_attn_layer_norm_bias
+epsilon
+
+
+var_15&
+input_61
+
+

+�

+�*
+name
+
+"
+
+input_61�

+linear
+
x
+
+
+input_615
+weight+
+)
+'model_model_encoder_layers_2_fc1_weight1
+bias)
+'
+%model_model_encoder_layers_2_fc1_bias&
+input_63
+
+

+�

+�*
+name
+
+"
+
+input_63a
+silu
+
x
+
+
+input_63&
+input_65
+
+

+�

+�*
+name
+
+"
+
+input_65�

+linear
+
x
+
+
+input_655
+weight+
+)
+'model_model_encoder_layers_2_fc2_weight1
+bias)
+'
+%model_model_encoder_layers_2_fc2_bias&
+input_69
+
+

+�

+�*
+name
+
+"
+
+input_69s
+add
+
x
+
+
+input_61
+
y
+
+
+input_69&
+input_71
+
+

+�

+�*
+name
+
+"
+
+input_71�

+const'
+hidden_states_19_axes_0
+
+

+
*-
+name%
+
+"
+hidden_states_19_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+
+input_71#
+axes
+
+hidden_states_19_axes_0A
+gamma8
+6
+4model_model_encoder_layers_2_final_layer_norm_weight>
+beta6
+4
+2model_model_encoder_layers_2_final_layer_norm_bias
+epsilon
+
+
+var_15.
+hidden_states_19
+
+

+�

+�*&
+name
+
+"
+hidden_states_19�

+linear
+
x
+
+hidden_states_19B
+weight8
+6
+4model_model_encoder_layers_3_self_attn_q_proj_weight>
+bias6
+4
+2model_model_encoder_layers_3_self_attn_q_proj_bias%
+var_350
+
+

+�

+�*
+name
+
+
+"
+op_350Q
+const
+var_351
+*
+name
+
+
+"
+op_351*
+val
+
+
+
+
+>s
+mul
+
x
+	
+var_350
+
y
+	
+var_351'
+	tensor_23
+
+

+�

+�*
+name
+
+
+	tensor_23�

+linear
+
x
+
+hidden_states_19B
+weight8
+6
+4model_model_encoder_layers_3_self_attn_k_proj_weight>
+bias6
+4
+2model_model_encoder_layers_3_self_attn_k_proj_bias'
+	tensor_19
+
+

+�

+�*
+name
+
+
+	tensor_19j
+const
+var_356
+
+

+*
+name
+
+
+"
+op_356**
+val#
+
+

+
+
+
+reshape
+
x
+
+	tensor_19
+shape
+	
+var_356*
+var_357
+
+

+�

+
+@*
+name
+
+
+"
+op_357o
+const
+var_358_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
�

+linear
+
x
+
+hidden_states_19B
+weight8
+6
+4model_model_encoder_layers_3_self_attn_v_proj_weight>
+bias6
+4
+2model_model_encoder_layers_3_self_attn_v_proj_bias'
+	tensor_21
+
+

+�

+�*
+name
+
+
+	tensor_21j
+const
+var_363
+
+

+*
+name
+
+
+"
+op_363**
+val#
+
+

+
+
+
+reshape
+
x
+
+	tensor_21
+shape
+	
+var_363*
+var_364
+
+

+�

+
+@*
+name
+
+
+"
+op_364o
+const
+var_365_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
b
+const
+var_372
+
+

+*
+name
+
+
+"
+op_372*"
+val
+
+

+
+	
+
�
@}
+reshape
+
x
+
+	tensor_23
+shape
+	
+var_372*
+var_373
+
+

+�

+
+@*
+name
+
+
+"
+op_373o
+const
+var_374_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
i
+const
+var_376
+
+

+*
+name
+
+
+"
+op_376*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_373
+perm
+
+var_374_perm_0/
+transpose_12
+
+

+
+�

+@*"
+name
+
+"
+transpose_12�

+reshape
+
x
+
+transpose_12
+shape
+	
+var_376+
+query_states_7
+
+
+�

+@*$
+name
+
+"
+query_states_7i
+const
+var_378
+
+

+*
+name
+
+
+"
+op_378*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_357
+perm
+
+var_358_perm_0/
+transpose_14
+
+

+
+�

+@*"
+name
+
+"
+transpose_14�

+reshape
+
x
+
+transpose_14
+shape
+	
+var_378*
+
+
+
+�

+@*#
+name
+
+"
+
+const
+var_380
+
+

+*
+name
+
+
+"
+op_380*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_364
+perm
+
+var_365_perm_0/
+transpose_13
+
+

+
+�

+@*"
+name
+
+"
+transpose_13�

+reshape
+
x
+
+transpose_13
+shape
+	
+var_380,
+value_states_15
+
+
+�

+@*%
+name
+
+"
+value_states_15n
+const
+var_383_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+	
+
+
{
+const%
+attn_weights_19_transpose_x_0
+
*3
+name+
+#
+!"
+attn_weights_19_transpose_x_0*
+val
+

+
+
{
+const%
+attn_weights_19_transpose_y_0
+
*3
+name+
+#
+!"
+attn_weights_19_transpose_y_0*
+val
+

+
+
�

+	transpose
+
x
+
+
+perm
+
+var_383_perm_0)
+transpose_11
+
+
+@
+�
*"
+name
+
+"
+transpose_11�

+matmul
+
x
+
+query_states_7
+
y
+
+transpose_110
+transpose_x!
+
+attn_weights_19_transpose_x_00
+transpose_y!
+
+attn_weights_19_transpose_y_0-
+attn_weights_19
+
+
+�

+�
*%
+name
+
+"
+attn_weights_19c
+const
+var_385
+
+

+*
+name
+
+
+"
+op_385*#
+val
+
+

+
+
+
+
�
�
�

+reshape
+
x
+
+attn_weights_19
+shape
+	
+var_385+
+var_386 
+
+

+
+�

+�
*
+name
+
+
+"
+op_386�

+add
+
x
+	
+var_386
+
y
+
+attention_mask_13
+attn_weights_21 
+
+

+
+�

+�
*%
+name
+
+"
+attn_weights_21b
+const
+var_391
+
+

+*
+name
+
+
+"
+op_391*"
+val
+
+

+
+	
+�
�
�

+reshape
+
x
+
+attn_weights_21
+shape
+	
+var_391&
+input_73
+
+
+�

+�
*
+name
+
+"
+
+input_73x
+softmax
+
x
+
+
+input_73
+axis
+
+
+var_30&
+input_75
+
+
+�

+�
*
+name
+
+"
+
+input_75y
+const$
+attn_output_19_transpose_x_0
+
*2
+name*
+"
+ "
+attn_output_19_transpose_x_0*
+val
+

+
+
y
+const$
+attn_output_19_transpose_y_0
+
*2
+name*
+"
+ "
+attn_output_19_transpose_y_0*
+val
+

+
+
�

+matmul
+
x
+
+
+input_75
+
y
+
+value_states_15/
+transpose_x 
+
+attn_output_19_transpose_x_0/
+transpose_y 
+
+attn_output_19_transpose_y_0+
+attn_output_19
+
+
+�

+@*$
+name
+
+"
+attn_output_19b
+const
+var_396
+
+

+*
+name
+
+
+"
+op_396*"
+val
+
+

+
+	
+
�
@�

+reshape
+
x
+
+attn_output_19
+shape
+	
+var_3961
+attn_output_21
+
+

+
+�

+@*$
+name
+
+"
+attn_output_21~
+const%
+attn_output_23_perm_0
+
+

+*+
+name#
+
+"
+attn_output_23_perm_0*!
+val
+
+

+
+
+
+
b
+const
+var_399
+
+

+*
+name
+
+
+"
+op_399*"
+val
+
+

+
+	
+
�
��

+	transpose
+
x
+
+attn_output_21!
+perm
+
+attn_output_23_perm_0/
+transpose_10
+
+

+�

+
+@*"
+name
+
+"
+transpose_10~
+reshape
+
x
+
+transpose_10
+shape
+	
+var_399&
+input_77
+
+

+�

+�*
+name
+
+"
+
+input_77�

+linear
+
x
+
+
+input_77D
+weight:
+8
+6model_model_encoder_layers_3_self_attn_out_proj_weight@
+bias8
+6
+4model_model_encoder_layers_3_self_attn_out_proj_bias&
+input_79
+
+

+�

+�*
+name
+
+"
+
+input_79{
+add
+
x
+
+hidden_states_19
+
y
+
+
+input_79&
+input_81
+
+

+�

+�*
+name
+
+"
+
+input_81x
+const
+input_83_axes_0
+
+

+
*%
+name
+
+"
+input_83_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+
+input_81
+axes
+
+input_83_axes_0E
+gamma<
+:
+8model_model_encoder_layers_3_self_attn_layer_norm_weightB
+beta:
+8
+6model_model_encoder_layers_3_self_attn_layer_norm_bias
+epsilon
+
+
+var_15&
+input_83
+
+

+�

+�*
+name
+
+"
+
+input_83�

+linear
+
x
+
+
+input_835
+weight+
+)
+'model_model_encoder_layers_3_fc1_weight1
+bias)
+'
+%model_model_encoder_layers_3_fc1_bias&
+input_85
+
+

+�

+�*
+name
+
+"
+
+input_85a
+silu
+
x
+
+
+input_85&
+input_87
+
+

+�

+�*
+name
+
+"
+
+input_87�

+linear
+
x
+
+
+input_875
+weight+
+)
+'model_model_encoder_layers_3_fc2_weight1
+bias)
+'
+%model_model_encoder_layers_3_fc2_bias&
+input_91
+
+

+�

+�*
+name
+
+"
+
+input_91s
+add
+
x
+
+
+input_83
+
y
+
+
+input_91&
+input_93
+
+

+�

+�*
+name
+
+"
+
+input_93�

+const'
+hidden_states_25_axes_0
+
+

+
*-
+name%
+
+"
+hidden_states_25_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+
+input_93#
+axes
+
+hidden_states_25_axes_0A
+gamma8
+6
+4model_model_encoder_layers_3_final_layer_norm_weight>
+beta6
+4
+2model_model_encoder_layers_3_final_layer_norm_bias
+epsilon
+
+
+var_15.
+hidden_states_25
+
+

+�

+�*&
+name
+
+"
+hidden_states_25�

+linear
+
x
+
+hidden_states_25B
+weight8
+6
+4model_model_encoder_layers_4_self_attn_q_proj_weight>
+bias6
+4
+2model_model_encoder_layers_4_self_attn_q_proj_bias%
+var_438
+
+

+�

+�*
+name
+
+
+"
+op_438Q
+const
+var_439
+*
+name
+
+
+"
+op_439*
+val
+
+
+
+
+>s
+mul
+
x
+	
+var_438
+
y
+	
+var_439'
+	tensor_29
+
+

+�

+�*
+name
+
+
+	tensor_29�

+linear
+
x
+
+hidden_states_25B
+weight8
+6
+4model_model_encoder_layers_4_self_attn_k_proj_weight>
+bias6
+4
+2model_model_encoder_layers_4_self_attn_k_proj_bias'
+	tensor_25
+
+

+�

+�*
+name
+
+
+	tensor_25j
+const
+var_444
+
+

+*
+name
+
+
+"
+op_444**
+val#
+
+

+
+
+
+reshape
+
x
+
+	tensor_25
+shape
+	
+var_444*
+var_445
+
+

+�

+
+@*
+name
+
+
+"
+op_445o
+const
+var_446_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
�

+linear
+
x
+
+hidden_states_25B
+weight8
+6
+4model_model_encoder_layers_4_self_attn_v_proj_weight>
+bias6
+4
+2model_model_encoder_layers_4_self_attn_v_proj_bias'
+	tensor_27
+
+

+�

+�*
+name
+
+
+	tensor_27j
+const
+var_451
+
+

+*
+name
+
+
+"
+op_451**
+val#
+
+

+
+
+
+reshape
+
x
+
+	tensor_27
+shape
+	
+var_451*
+var_452
+
+

+�

+
+@*
+name
+
+
+"
+op_452o
+const
+var_453_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
b
+const
+var_460
+
+

+*
+name
+
+
+"
+op_460*"
+val
+
+

+
+	
+
�
@}
+reshape
+
x
+
+	tensor_29
+shape
+	
+var_460*
+var_461
+
+

+�

+
+@*
+name
+
+
+"
+op_461o
+const
+var_462_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
i
+const
+var_464
+
+

+*
+name
+
+
+"
+op_464*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_461
+perm
+
+var_462_perm_0.
+transpose_7
+
+

+
+�

+@*!
+name
+
+"
+transpose_7�

+reshape
+
x
+
+transpose_7
+shape
+	
+var_464+
+query_states_9
+
+
+�

+@*$
+name
+
+"
+query_states_9i
+const
+var_466
+
+

+*
+name
+
+
+"
+op_466*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_445
+perm
+
+var_446_perm_0.
+transpose_9
+
+

+
+�

+@*!
+name
+
+"
+transpose_9�

+reshape
+
x
+
+transpose_9
+shape
+	
+var_466*
+
+
+
+�

+@*#
+name
+
+"
+
+const
+var_468
+
+

+*
+name
+
+
+"
+op_468*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_452
+perm
+
+var_453_perm_0.
+transpose_8
+
+

+
+�

+@*!
+name
+
+"
+transpose_8�

+reshape
+
x
+
+transpose_8
+shape
+	
+var_468,
+value_states_19
+
+
+�

+@*%
+name
+
+"
+value_states_19n
+const
+var_471_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+	
+
+
{
+const%
+attn_weights_25_transpose_x_0
+
*3
+name+
+#
+!"
+attn_weights_25_transpose_x_0*
+val
+

+
+
{
+const%
+attn_weights_25_transpose_y_0
+
*3
+name+
+#
+!"
+attn_weights_25_transpose_y_0*
+val
+

+
+
�

+	transpose
+
x
+
+
+perm
+
+var_471_perm_0(
+transpose_6
+
+
+@
+�
*!
+name
+
+"
+transpose_6�

+matmul
+
x
+
+query_states_9
+
y
+
+transpose_60
+transpose_x!
+
+attn_weights_25_transpose_x_00
+transpose_y!
+
+attn_weights_25_transpose_y_0-
+attn_weights_25
+
+
+�

+�
*%
+name
+
+"
+attn_weights_25c
+const
+var_473
+
+

+*
+name
+
+
+"
+op_473*#
+val
+
+

+
+
+
+
�
�
�

+reshape
+
x
+
+attn_weights_25
+shape
+	
+var_473+
+var_474 
+
+

+
+�

+�
*
+name
+
+
+"
+op_474�

+add
+
x
+	
+var_474
+
y
+
+attention_mask_13
+attn_weights_27 
+
+

+
+�

+�
*%
+name
+
+"
+attn_weights_27b
+const
+var_479
+
+

+*
+name
+
+
+"
+op_479*"
+val
+
+

+
+	
+�
�
�

+reshape
+
x
+
+attn_weights_27
+shape
+	
+var_479&
+input_95
+
+
+�

+�
*
+name
+
+"
+
+input_95x
+softmax
+
x
+
+
+input_95
+axis
+
+
+var_30&
+input_97
+
+
+�

+�
*
+name
+
+"
+
+input_97y
+const$
+attn_output_25_transpose_x_0
+
*2
+name*
+"
+ "
+attn_output_25_transpose_x_0*
+val
+

+
+
y
+const$
+attn_output_25_transpose_y_0
+
*2
+name*
+"
+ "
+attn_output_25_transpose_y_0*
+val
+

+
+
�

+matmul
+
x
+
+
+input_97
+
y
+
+value_states_19/
+transpose_x 
+
+attn_output_25_transpose_x_0/
+transpose_y 
+
+attn_output_25_transpose_y_0+
+attn_output_25
+
+
+�

+@*$
+name
+
+"
+attn_output_25b
+const
+var_484
+
+

+*
+name
+
+
+"
+op_484*"
+val
+
+

+
+	
+
�
@�

+reshape
+
x
+
+attn_output_25
+shape
+	
+var_4841
+attn_output_27
+
+

+
+�

+@*$
+name
+
+"
+attn_output_27~
+const%
+attn_output_29_perm_0
+
+

+*+
+name#
+
+"
+attn_output_29_perm_0*!
+val
+
+

+
+
+
+
b
+const
+var_487
+
+

+*
+name
+
+
+"
+op_487*"
+val
+
+

+
+	
+
�
��

+	transpose
+
x
+
+attn_output_27!
+perm
+
+attn_output_29_perm_0.
+transpose_5
+
+

+�

+
+@*!
+name
+
+"
+transpose_5}
+reshape
+
x
+
+transpose_5
+shape
+	
+var_487&
+input_99
+
+

+�

+�*
+name
+
+"
+
+input_99�

+linear
+
x
+
+
+input_99D
+weight:
+8
+6model_model_encoder_layers_4_self_attn_out_proj_weight@
+bias8
+6
+4model_model_encoder_layers_4_self_attn_out_proj_bias'
+	input_101
+
+

+�

+�*
+name
+
+
+	input_101~
+add
+
x
+
+hidden_states_25
+
y
+
+	input_101'
+	input_103
+
+

+�

+�*
+name
+
+
+	input_103z
+const 
+input_105_axes_0
+
+

+
*&
+name
+
+"
+input_105_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+	input_103
+axes
+
+input_105_axes_0E
+gamma<
+:
+8model_model_encoder_layers_4_self_attn_layer_norm_weightB
+beta:
+8
+6model_model_encoder_layers_4_self_attn_layer_norm_bias
+epsilon
+
+
+var_15'
+	input_105
+
+

+�

+�*
+name
+
+
+	input_105�

+linear
+
x
+
+	input_1055
+weight+
+)
+'model_model_encoder_layers_4_fc1_weight1
+bias)
+'
+%model_model_encoder_layers_4_fc1_bias'
+	input_107
+
+

+�

+�*
+name
+
+
+	input_107d
+silu
+
x
+
+	input_107'
+	input_109
+
+

+�

+�*
+name
+
+
+	input_109�

+linear
+
x
+
+	input_1095
+weight+
+)
+'model_model_encoder_layers_4_fc2_weight1
+bias)
+'
+%model_model_encoder_layers_4_fc2_bias'
+	input_113
+
+

+�

+�*
+name
+
+
+	input_113w
+add
+
x
+
+	input_105
+
y
+
+	input_113'
+	input_115
+
+

+�

+�*
+name
+
+
+	input_115�

+const'
+hidden_states_31_axes_0
+
+

+
*-
+name%
+
+"
+hidden_states_31_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+	input_115#
+axes
+
+hidden_states_31_axes_0A
+gamma8
+6
+4model_model_encoder_layers_4_final_layer_norm_weight>
+beta6
+4
+2model_model_encoder_layers_4_final_layer_norm_bias
+epsilon
+
+
+var_15.
+hidden_states_31
+
+

+�

+�*&
+name
+
+"
+hidden_states_31�

+linear
+
x
+
+hidden_states_31B
+weight8
+6
+4model_model_encoder_layers_5_self_attn_q_proj_weight>
+bias6
+4
+2model_model_encoder_layers_5_self_attn_q_proj_bias%
+var_526
+
+

+�

+�*
+name
+
+
+"
+op_526Q
+const
+var_527
+*
+name
+
+
+"
+op_527*
+val
+
+
+
+
+>�

+mul
+
x
+	
+var_526
+
y
+	
+var_527/
+tensor_workaround
+
+

+�

+�*'
+name
+
+"
+tensor_workaround�

+linear
+
x
+
+hidden_states_31B
+weight8
+6
+4model_model_encoder_layers_5_self_attn_k_proj_weight>
+bias6
+4
+2model_model_encoder_layers_5_self_attn_k_proj_bias'
+	tensor_31
+
+

+�

+�*
+name
+
+
+	tensor_31j
+const
+var_532
+
+

+*
+name
+
+
+"
+op_532**
+val#
+
+

+
+
+
+reshape
+
x
+
+	tensor_31
+shape
+	
+var_532*
+var_533
+
+

+�

+
+@*
+name
+
+
+"
+op_533o
+const
+var_534_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
�

+linear
+
x
+
+hidden_states_31B
+weight8
+6
+4model_model_encoder_layers_5_self_attn_v_proj_weight>
+bias6
+4
+2model_model_encoder_layers_5_self_attn_v_proj_bias'
+	tensor_33
+
+

+�

+�*
+name
+
+
+	tensor_33j
+const
+var_539
+
+

+*
+name
+
+
+"
+op_539**
+val#
+
+

+
+
+
+reshape
+
x
+
+	tensor_33
+shape
+	
+var_539*
+var_540
+
+

+�

+
+@*
+name
+
+
+"
+op_540o
+const
+var_541_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
b
+const
+var_548
+
+

+*
+name
+
+
+"
+op_548*"
+val
+
+

+
+	
+
�
@�

+reshape
+
x
+
+tensor_workaround
+shape
+	
+var_548*
+var_549
+
+

+�

+
+@*
+name
+
+
+"
+op_549o
+const
+var_550_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+
+
+
+
i
+const
+var_552
+
+

+*
+name
+
+
+"
+op_552*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_549
+perm
+
+var_550_perm_0.
+transpose_2
+
+

+
+�

+@*!
+name
+
+"
+transpose_2�

+reshape
+
x
+
+transpose_2
+shape
+	
+var_552)
+query_states
+
+
+�

+@*"
+name
+
+"
+query_statesi
+const
+var_554
+
+

+*
+name
+
+
+"
+op_554*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_533
+perm
+
+var_534_perm_0.
+transpose_4
+
+

+
+�

+@*!
+name
+
+"
+transpose_4�

+reshape
+
x
+
+transpose_4
+shape
+	
+var_554'
+
+key_states
+
+
+�

+@* 
+name
+
+"
+
+key_statesi
+const
+var_556
+
+

+*
+name
+
+
+"
+op_556*)
+val"
+
+

+
+
+���������
@�

+	transpose
+
x
+	
+var_540
+perm
+
+var_541_perm_0.
+transpose_3
+
+

+
+�

+@*!
+name
+
+"
+transpose_3�

+reshape
+
x
+
+transpose_3
+shape
+	
+var_556)
+value_states
+
+
+�

+@*"
+name
+
+"
+value_statesn
+const
+var_559_perm_0
+
+

+*#
+name
+
+"
+
+val
+
+

+	
+
+
{
+const%
+attn_weights_31_transpose_x_0
+
*3
+name+
+#
+!"
+attn_weights_31_transpose_x_0*
+val
+

+
+
{
+const%
+attn_weights_31_transpose_y_0
+
*3
+name+
+#
+!"
+attn_weights_31_transpose_y_0*
+val
+

+
+
�

+	transpose
+
x
+
+
+key_states
+perm
+
+var_559_perm_0(
+transpose_1
+
+
+@
+�
*!
+name
+
+"
+transpose_1�

+matmul
+
x
+
+query_states
+
y
+
+transpose_10
+transpose_x!
+
+attn_weights_31_transpose_x_00
+transpose_y!
+
+attn_weights_31_transpose_y_0-
+attn_weights_31
+
+
+�

+�
*%
+name
+
+"
+attn_weights_31c
+const
+var_561
+
+

+*
+name
+
+
+"
+op_561*#
+val
+
+

+
+
+
+
�
�
�

+reshape
+
x
+
+attn_weights_31
+shape
+	
+var_561+
+var_562 
+
+

+
+�

+�
*
+name
+
+
+"
+op_562�

+add
+
x
+	
+var_562
+
y
+
+attention_mask_13
+attn_weights_33 
+
+

+
+�

+�
*%
+name
+
+"
+attn_weights_33b
+const
+var_567
+
+

+*
+name
+
+
+"
+op_567*"
+val
+
+

+
+	
+�
�
�

+reshape
+
x
+
+attn_weights_33
+shape
+	
+var_567'
+	input_117
+
+
+�

+�
*
+name
+
+
+	input_117{
+softmax
+
x
+
+	input_117
+axis
+
+
+var_30'
+	input_119
+
+
+�

+�
*
+name
+
+
+	input_119y
+const$
+attn_output_31_transpose_x_0
+
*2
+name*
+"
+ "
+attn_output_31_transpose_x_0*
+val
+

+
+
y
+const$
+attn_output_31_transpose_y_0
+
*2
+name*
+"
+ "
+attn_output_31_transpose_y_0*
+val
+

+
+
�

+matmul
+
x
+
+	input_119
+
y
+
+value_states/
+transpose_x 
+
+attn_output_31_transpose_x_0/
+transpose_y 
+
+attn_output_31_transpose_y_0+
+attn_output_31
+
+
+�

+@*$
+name
+
+"
+attn_output_31b
+const
+var_572
+
+

+*
+name
+
+
+"
+op_572*"
+val
+
+

+
+	
+
�
@�

+reshape
+
x
+
+attn_output_31
+shape
+	
+var_5721
+attn_output_33
+
+

+
+�

+@*$
+name
+
+"
+attn_output_33x
+const"
+attn_output_perm_0
+
+

+*(
+name 
+
+"
+attn_output_perm_0*!
+val
+
+

+
+
+
+
b
+const
+var_575
+
+

+*
+name
+
+
+"
+op_575*"
+val
+
+

+
+	
+
�
��

+	transpose
+
x
+
+attn_output_33
+perm
+
+attn_output_perm_0.
+transpose_0
+
+

+�

+
+@*!
+name
+
+"
+transpose_0
+reshape
+
x
+
+transpose_0
+shape
+	
+var_575'
+	input_121
+
+

+�

+�*
+name
+
+
+	input_121�

+linear
+
x
+
+	input_121D
+weight:
+8
+6model_model_encoder_layers_5_self_attn_out_proj_weight@
+bias8
+6
+4model_model_encoder_layers_5_self_attn_out_proj_bias'
+	input_123
+
+

+�

+�*
+name
+
+
+	input_123~
+add
+
x
+
+hidden_states_31
+
y
+
+	input_123'
+	input_125
+
+

+�

+�*
+name
+
+
+	input_125z
+const 
+input_127_axes_0
+
+

+
*&
+name
+
+"
+input_127_axes_0*'
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x
+
+	input_125
+axes
+
+input_127_axes_0E
+gamma<
+:
+8model_model_encoder_layers_5_self_attn_layer_norm_weightB
+beta:
+8
+6model_model_encoder_layers_5_self_attn_layer_norm_bias
+epsilon
+
+
+var_15'
+	input_127
+
+

+�

+�*
+name
+
+
+	input_127�

+linear
+
x
+
+	input_1275
+weight+
+)
+'model_model_encoder_layers_5_fc1_weight1
+bias)
+'
+%model_model_encoder_layers_5_fc1_bias'
+	input_129
+
+

+�

+�*
+name
+
+
+	input_129d
+silu
+
x
+
+	input_129'
+	input_131
+
+

+�

+�*
+name
+
+
+	input_131�

+linear
+
x
+
+	input_1315
+weight+
+)
+'model_model_encoder_layers_5_fc2_weight1
+bias)
+'
+%model_model_encoder_layers_5_fc2_bias'
+	input_135
+
+

+�

+�*
+name
+
+
+	input_135o
+add
+
x
+
+	input_127
+
y
+
+	input_135#
+input
+
+

+�

+�*
+name
+
+	"
+inputu
+const
+var_599_axes_0
+
+

+
*#
+name
+
+"
+
+val 
+
+

+

+
+
+���������
�
+
+layer_norm
+
x	
+
+input
+axes
+
+var_599_axes_0A
+gamma8
+6
+4model_model_encoder_layers_5_final_layer_norm_weight>
+beta6
+4
+2model_model_encoder_layers_5_final_layer_norm_bias
+epsilon
+
+
+var_15/
+last_hidden_state
+
+

+�

+�*
+name
+
+
+"
+op_599"�

+	buildInfo�
"
+
+
+~"|
+8
+!
+
+"
+coremltools-version
+
+	"
+7.0b1
+@
+)
+!
+"
+coremltools-component-torch
+
+	"
+2.0.1

coreml/text2text-generation/encoder_float32_model.mlpackage/Data/com.apple.CoreML/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d0d77b746a93d054c184d53f3038da38042ff80fa6a4d7ac6ea9f64802a408d
+size 203962560

coreml/text2text-generation/encoder_float32_model.mlpackage/Manifest.json

@@ -0,0 +1,18 @@
+{
+    "fileFormatVersion": "1.0.0",
+    "itemInfoEntries": {
+        "189D1ABB-AB08-418D-936C-750F69C20B79": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Weights",
+            "name": "weights",
+            "path": "com.apple.CoreML/weights"
+        },
+        "95D67307-FEB0-44E9-8093-3281381FEB6D": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Specification",
+            "name": "model.mlmodel",
+            "path": "com.apple.CoreML/model.mlmodel"
+        }
+    },
+    "rootModelIdentifier": "95D67307-FEB0-44E9-8093-3281381FEB6D"
+}