diff --git "a/coreml/text2text-generation/encoder_float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel" "b/coreml/text2text-generation/encoder_float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel" new file mode 100644--- /dev/null +++ "b/coreml/text2text-generation/encoder_float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel" @@ -0,0 +1,13828 @@ +� +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) * +���Rj +last_hidden_stateFSequence of hidden-states at the output of the last layer of the model * +������ +4sshleifer/distilbart-cnn-12-6 (text2text-generation)�3 +#com.github.apple.coremltools.source torch==2.0.1�+ +$com.github.apple.coremltools.version7.0�> +co.huggingface.exporters.namesshleifer/distilbart-cnn-12-6�5 +co.huggingface.exporters.tasktext2text-generation�E +%co.huggingface.exporters.architectureBartForConditionalGeneration�- +"co.huggingface.exporters.frameworkpytorch�- +"co.huggingface.exporters.precisionfloat32�" +transformers_version +4.7.0.dev0����� +main�� + + input_ids + + +� +% +attention_mask + + +�CoreML5�� +CoreML5��last_hidden_state� +const2 +model_model_shared_weight +  +؈ +�*/ +name' + +" +model_model_shared_weight*B +val; +  +؈ +�*" +@model_path/weights/weight.bin@� +const= +,model_model_encoder_layernorm_embedding_bias +   +�*B +name: +2 +0". +,model_model_encoder_layernorm_embedding_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���b� +const? +.model_model_encoder_layernorm_embedding_weight +   +�*D +name< +4 +2"0 +.model_model_encoder_layernorm_embedding_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���b� +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�–b� +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��b� +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��d� +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���d� +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���f� +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���f� +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���h� +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�ŗh� +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�Ɨj� +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��j� +const6 +%model_model_encoder_layers_0_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_0_fc1_bias*= +val6 +   +� *% +@model_path/weights/weight.bin���j� +const? +'model_model_encoder_layers_0_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_0_fc1_weight*D +val= +  +�  +�*% +@model_path/weights/weight.bin���j� +const6 +%model_model_encoder_layers_0_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_0_fc2_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���r� +const? +'model_model_encoder_layers_0_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_0_fc2_weight*D +val= +  +� +� *% +@model_path/weights/weight.bin���r� +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���z� +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�əz� +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��z� +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���z� +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���|� +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���|� +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���~� +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�̚~� +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*> +val7 +   +�*& +@model_path/weights/weight.bin�͚�� +constN +6model_model_encoder_layers_1_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_encoder_layers_1_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�횀� +constG +6model_model_encoder_layers_1_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_encoder_layers_1_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�� +constI +8model_model_encoder_layers_1_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_encoder_layers_1_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_1_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_1_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_1_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_1_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_1_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_1_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_1_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_1_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin�М�� +constC +2model_model_encoder_layers_1_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_1_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ќ�� +constE +4model_model_encoder_layers_1_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_1_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin��� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin�ӝ�� +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*> +val7 +   +�*& +@model_path/weights/weight.bin�ԝ�� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_encoder_layers_2_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_encoder_layers_2_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_encoder_layers_2_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_encoder_layers_2_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_encoder_layers_2_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_encoder_layers_2_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_2_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_2_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin�מ�� +const? +'model_model_encoder_layers_2_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_2_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin�ן�� +const6 +%model_model_encoder_layers_2_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_2_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�؟�� +const? +'model_model_encoder_layers_2_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_2_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_2_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_2_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_encoder_layers_2_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_2_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin�ڠ�� +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*> +val7 +   +�*& +@model_path/weights/weight.bin�۠�� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_encoder_layers_3_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_encoder_layers_3_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_encoder_layers_3_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_encoder_layers_3_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_encoder_layers_3_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_encoder_layers_3_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�ޡ�� +const6 +%model_model_encoder_layers_3_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_3_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_3_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_3_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_3_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_3_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_3_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_3_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_3_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_3_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_encoder_layers_3_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_3_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin���� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin�Ĥ�� +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*> +val7 +   +�*& +@model_path/weights/weight.bin�Ť�� +constN +6model_model_encoder_layers_4_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_encoder_layers_4_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin���� +constG +6model_model_encoder_layers_4_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_encoder_layers_4_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin���� +constI +8model_model_encoder_layers_4_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_encoder_layers_4_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_4_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_4_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_4_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_4_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_4_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_4_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_4_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_4_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin�Ȧ�� +constC +2model_model_encoder_layers_4_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_4_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ɦ�� +constE +4model_model_encoder_layers_4_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_4_final_layer_norm_weight*> +val7 +   +�*& +@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*> +val7 +   +�*& +@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*E +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*> +val7 +   +�*& +@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*E +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*> +val7 +   +�*& +@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*E +val> +  +� +�*& +@model_path/weights/weight.bin���� +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*> +val7 +   +�*& +@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*E +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_encoder_layers_5_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_encoder_layers_5_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_5_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_5_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin�Ϩ�� +const? +'model_model_encoder_layers_5_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_5_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin�ϩ�� +const6 +%model_model_encoder_layers_5_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_5_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�Щ�� +const? +'model_model_encoder_layers_5_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_5_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin���� +constC +2model_model_encoder_layers_5_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_5_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin���� +constE +4model_model_encoder_layers_5_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_5_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_6_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_6_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_encoder_layers_6_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_6_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�Ҫ�� +constC +2model_model_encoder_layers_6_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_6_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�Ӫ�� +constL +4model_model_encoder_layers_6_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_6_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin���� +constC +2model_model_encoder_layers_6_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_6_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_encoder_layers_6_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_6_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constE +4model_model_encoder_layers_6_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_6_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_encoder_layers_6_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_encoder_layers_6_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_encoder_layers_6_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_encoder_layers_6_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_encoder_layers_6_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_encoder_layers_6_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�֫�� +const6 +%model_model_encoder_layers_6_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_6_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_6_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_6_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_6_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_6_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_6_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_6_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_6_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_6_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_encoder_layers_6_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_6_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_7_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_7_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ڭ�� +constL +4model_model_encoder_layers_7_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_7_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_7_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_7_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_encoder_layers_7_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_7_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_7_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_7_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_encoder_layers_7_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_7_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constE +4model_model_encoder_layers_7_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_7_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_encoder_layers_7_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_encoder_layers_7_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�ݮ�� +constG +6model_model_encoder_layers_7_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_encoder_layers_7_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ޮ�� +constI +8model_model_encoder_layers_7_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_encoder_layers_7_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_7_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_7_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_7_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_7_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_7_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_7_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_encoder_layers_7_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_7_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_7_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_7_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_encoder_layers_7_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_7_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�ᰢ� +constC +2model_model_encoder_layers_8_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_8_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_encoder_layers_8_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_8_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_8_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_8_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_encoder_layers_8_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_8_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�ñ�� +constC +2model_model_encoder_layers_8_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_8_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ı�� +constL +4model_model_encoder_layers_8_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_8_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�䱦� +constE +4model_model_encoder_layers_8_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_8_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�屨� +constN +6model_model_encoder_layers_8_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_encoder_layers_8_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_encoder_layers_8_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_encoder_layers_8_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_encoder_layers_8_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_encoder_layers_8_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_encoder_layers_8_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_8_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin�Dz�� +const? +'model_model_encoder_layers_8_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_8_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin�dz�� +const6 +%model_model_encoder_layers_8_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_8_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ȳ�� +const? +'model_model_encoder_layers_8_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_8_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin�賲� +constC +2model_model_encoder_layers_8_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_8_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�鳺� +constE +4model_model_encoder_layers_8_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_8_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_9_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_9_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_encoder_layers_9_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_9_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�ʴ�� +constC +2model_model_encoder_layers_9_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_9_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�˴�� +constL +4model_model_encoder_layers_9_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_9_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�봼� +constC +2model_model_encoder_layers_9_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_9_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�촾� +constL +4model_model_encoder_layers_9_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_encoder_layers_9_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constE +4model_model_encoder_layers_9_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_9_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_encoder_layers_9_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_encoder_layers_9_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_encoder_layers_9_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_encoder_layers_9_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_encoder_layers_9_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_encoder_layers_9_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�ε�� +const6 +%model_model_encoder_layers_9_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_encoder_layers_9_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin���� +const? +'model_model_encoder_layers_9_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_encoder_layers_9_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin���� +const6 +%model_model_encoder_layers_9_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_encoder_layers_9_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin���� +const? +'model_model_encoder_layers_9_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_encoder_layers_9_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_encoder_layers_9_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_encoder_layers_9_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_encoder_layers_9_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_encoder_layers_9_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constD +3model_model_encoder_layers_10_self_attn_q_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_encoder_layers_10_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ҷ�� +constM +5model_model_encoder_layers_10_self_attn_q_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_encoder_layers_10_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin���� +constD +3model_model_encoder_layers_10_self_attn_k_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_encoder_layers_10_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin���� +constM +5model_model_encoder_layers_10_self_attn_k_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_encoder_layers_10_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constD +3model_model_encoder_layers_10_self_attn_v_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_encoder_layers_10_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constM +5model_model_encoder_layers_10_self_attn_v_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_encoder_layers_10_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constF +5model_model_encoder_layers_10_self_attn_out_proj_bias +   +�*K +nameC +; +9"7 +5model_model_encoder_layers_10_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constO +7model_model_encoder_layers_10_self_attn_out_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_encoder_layers_10_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�ո�� +constH +7model_model_encoder_layers_10_self_attn_layer_norm_bias +   +�*M +nameE += +;"9 +7model_model_encoder_layers_10_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ָ�� +constJ +9model_model_encoder_layers_10_self_attn_layer_norm_weight +   +�*O +nameG +? +="; +9model_model_encoder_layers_10_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const7 +&model_model_encoder_layers_10_fc1_bias +   +� *< +name4 +, +*"( +&model_model_encoder_layers_10_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const@ +(model_model_encoder_layers_10_fc1_weight +  +�  +�*> +name6 +. +,"* +(model_model_encoder_layers_10_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const7 +&model_model_encoder_layers_10_fc2_bias +   +�*< +name4 +, +*"( +&model_model_encoder_layers_10_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const@ +(model_model_encoder_layers_10_fc2_weight +  +� +� *> +name6 +. +,"* +(model_model_encoder_layers_10_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constD +3model_model_encoder_layers_10_final_layer_norm_bias +   +�*I +nameA +9 +7"5 +3model_model_encoder_layers_10_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constF +5model_model_encoder_layers_10_final_layer_norm_weight +   +�*K +nameC +; +9"7 +5model_model_encoder_layers_10_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�ٺ�� +constD +3model_model_encoder_layers_11_self_attn_q_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_encoder_layers_11_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constM +5model_model_encoder_layers_11_self_attn_q_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_encoder_layers_11_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constD +3model_model_encoder_layers_11_self_attn_k_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_encoder_layers_11_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constM +5model_model_encoder_layers_11_self_attn_k_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_encoder_layers_11_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constD +3model_model_encoder_layers_11_self_attn_v_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_encoder_layers_11_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constM +5model_model_encoder_layers_11_self_attn_v_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_encoder_layers_11_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�ܻ�� +constF +5model_model_encoder_layers_11_self_attn_out_proj_bias +   +�*K +nameC +; +9"7 +5model_model_encoder_layers_11_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ݻ�� +constO +7model_model_encoder_layers_11_self_attn_out_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_encoder_layers_11_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constH +7model_model_encoder_layers_11_self_attn_layer_norm_bias +   +�*M +nameE += +;"9 +7model_model_encoder_layers_11_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constJ +9model_model_encoder_layers_11_self_attn_layer_norm_weight +   +�*O +nameG +? +="; +9model_model_encoder_layers_11_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const7 +&model_model_encoder_layers_11_fc1_bias +   +� *< +name4 +, +*"( +&model_model_encoder_layers_11_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const@ +(model_model_encoder_layers_11_fc1_weight +  +�  +�*> +name6 +. +,"* +(model_model_encoder_layers_11_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const7 +&model_model_encoder_layers_11_fc2_bias +   +�*< +name4 +, +*"( +&model_model_encoder_layers_11_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const@ +(model_model_encoder_layers_11_fc2_weight +  +� +� *> +name6 +. +,"* +(model_model_encoder_layers_11_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin���� +constD +3model_model_encoder_layers_11_final_layer_norm_bias +   +�*I +nameA +9 +7"5 +3model_model_encoder_layers_11_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ὂ� +constF +5model_model_encoder_layers_11_final_layer_norm_weight +   +�*K +nameC +; +9"7 +5model_model_encoder_layers_11_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����O +const +var_14 + * +name + + " +op_14* +val +  + + + +���O +const +var_16 + * +name + + " +op_16* +val +  + + + +�?O +const +var_22 + * +name + + " +op_22* +val +  + + + +��'7U +const +var_30 +* +name + + " +op_30* +val + + + +���������Z +const + var_61_axis_0 +*" +name + +" + op_61_axis_0* +val + + +� +gather" +x + +model_model_shared_weight +indices + + input_ids +axis + + var_61_axis_0$ +var_61 +  + +� +�* +name + + " +op_61� +const' + embed_pos +  + +� +�* +name + + " + embed_pos*K +valD +  + +� +�*& +@model_path/weights/weight.bin����p +add +x + + +var_61 +y + + embed_pos% +input_5 +  + +� +�* +name + + " +input_5v +const +input_7_axes_0 + + +*$ +name + +" +input_7_axes_0*' +val  + + + + + +���������� + +layer_norm +x + +input_5 +axes + +input_7_axes_0; +gamma2 +0 +.model_model_encoder_layernorm_embedding_weight8 +beta0 +. +,model_model_encoder_layernorm_embedding_bias +epsilon + + +var_22% +input_7 +  + +� +�* +name + + " +input_7j +const + var_87_axes_0 + + +*" +name + +" + op_87_axes_0* +val + + + + +� + expand_dims +x + +attention_mask +axes + + var_87_axes_0# +var_87 + + + +�* +name + + " +op_87j +const + var_88_axes_0 + + +*" +name + +" + op_88_axes_0* +val + + + + +� + expand_dims +x + + +var_87 +axes + + var_88_axes_0) +var_88 + + + + +�* +name + + " +op_88n +const + var_91_reps_0 + + +*" +name + +" + op_91_reps_0*" +val + + + +  +�{ +tile +x + + +var_88 +reps + + var_91_reps_0* +var_91 + + + +� +�* +name + + " +op_91_ +const +var_92_dtype_0 +*# +name + +" + op_92_dtype_0* +val + + +" +fp32� +cast +x + + +var_91 +dtype + +var_92_dtype_0, +cast_124 +  + + +� +�* +name + + " + +cast_124� +sub +x + + +var_16 +y + + +cast_1241 + inverted_mask +  + + +� +�*# +name + +" + inverted_mask_ +const +var_94_dtype_0 +*# +name + +" + op_94_dtype_0* +val + + +" +bool� +cast +x + + inverted_mask +dtype + +var_94_dtype_0, +cast_123 + + + +� +�* +name + + " + +cast_123� +select +cond + + +cast_123 +a + + +var_14 +b + + inverted_mask4 +attention_mask_1 +  + + +� +�*$ +name + +" +attention_mask� +linear +x + +input_7B +weight8 +6 +4model_model_encoder_layers_0_self_attn_q_proj_weight> +bias6 +4 +2model_model_encoder_layers_0_self_attn_q_proj_bias& +linear_0 +  + +� +�* +name + + " + +linear_0Q +const +var_111 + * +name + + +" +op_111* +val +  + + + +>r +mul +x + + +linear_0 +y + +var_111& +tensor_5 +  + +� +�* +name + + " + +tensor_5� +linear +x + +input_7B +weight8 +6 +4model_model_encoder_layers_0_self_attn_k_proj_weight> +bias6 +4 +2model_model_encoder_layers_0_self_attn_k_proj_bias& +linear_1 +  + +� +�* +name + + " + +linear_1j +const +var_116 + + +* +name + + +" +op_116** +val# + + + + + ���������@| +reshape +x + + +linear_1 +shape + +var_116* +var_117 +  + +� + +@* +name + + +" +op_117o +const +var_118_perm_0 + + +*# +name + +" + op_118_perm_0*! +val + + + + + +� +linear +x + +input_7B +weight8 +6 +4model_model_encoder_layers_0_self_attn_v_proj_weight> +bias6 +4 +2model_model_encoder_layers_0_self_attn_v_proj_bias& +linear_2 +  + +� +�* +name + + " + +linear_2j +const +var_123 + + +* +name + + +" +op_123** +val# + + + + + ���������@| +reshape +x + + +linear_2 +shape + +var_123* +var_124 +  + +� + +@* +name + + +" +op_124o +const +var_125_perm_0 + + +*# +name + +" + op_125_perm_0*! +val + + + + + +b +const +var_132 + + +* +name + + +" +op_132*" +val + + + +  +�@| +reshape +x + + +tensor_5 +shape + +var_132* +var_133 +  + +� + +@* +name + + +" +op_133o +const +var_134_perm_0 + + +*# +name + +" + op_134_perm_0*! +val + + + + + +i +const +var_136 + + +* +name + + +" +op_136*) +val" + + + + + ���������@� + transpose +x + +var_133 +perm + +var_134_perm_0/ + transpose_57 +  + + +� +@*" +name + +" + transpose_57� +reshape +x + + transpose_57 +shape + +var_136+ +query_states_1 +  + +� +@*$ +name + +" +query_states_1i +const +var_138 + + +* +name + + +" +op_138*) +val" + + + + + ���������@� + transpose +x + +var_117 +perm + +var_118_perm_0/ + transpose_59 +  + + +� +@*" +name + +" + transpose_59� +reshape +x + + transpose_59 +shape + +var_138) + key_states_3 +  + +� +@*" +name + +" + key_states_3i +const +var_140 + + +* +name + + +" +op_140*) +val" + + + + + ���������@� + transpose +x + +var_124 +perm + +var_125_perm_0/ + transpose_58 +  + + +� +@*" +name + +" + transpose_58� +reshape +x + + transpose_58 +shape + +var_140+ +value_states_3 +  + +� +@*$ +name + +" +value_states_3n +const +var_143_perm_0 + + +*# +name + +" + op_143_perm_0* +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_143_perm_0) + transpose_56 +  + +@ +�*" +name + +" + transpose_56� +matmul +x + +query_states_1 +y + + transpose_56/ + 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_145 + + +* +name + + +" +op_145*# +val + + + + + +��� +reshape +x + +attn_weights_1 +shape + +var_145+ +var_146 +  + + +� +�* +name + + +" +op_146� +add +x + +var_146 +y + +attention_mask_12 +attn_weights_3 +  + + +� +�*$ +name + +" +attn_weights_3b +const +var_151 + + +* +name + + +" +op_151*" +val + + + +  +��~ +reshape +x + +attn_weights_3 +shape + +var_151% +input_9 +  + +� +�* +name + + " +input_9w +softmax +x + +input_9 +axis + + +var_30& +input_11 +  + +� +�* +name + + " + +input_11w +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_11 +y + +value_states_3. + transpose_x + +attn_output_1_transpose_x_0. + transpose_y + +attn_output_1_transpose_y_0* + attn_output_1 +  + +� +@*# +name + +" + attn_output_1b +const +var_156 + + +* +name + + +" +op_156*" +val + + + +  +�@� +reshape +x + + attn_output_1 +shape + +var_1560 + attn_output_3 +  + + +� +@*# +name + +" + attn_output_3| +const$ +attn_output_5_perm_0 + + +** +name" + +" +attn_output_5_perm_0*! +val + + + + + +b +const +var_159 + + +* +name + + +" +op_159*" +val + + + +  +��� + transpose +x + + attn_output_3 +perm + +attn_output_5_perm_0/ + transpose_55 +  + +� + +@*" +name + +" + transpose_55~ +reshape +x + + transpose_55 +shape + +var_159& +input_13 +  + +� +�* +name + + " + +input_13� +linear +x + + +input_13D +weight: +8 +6model_model_encoder_layers_0_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_0_self_attn_out_proj_bias& +linear_3 +  + +� +�* +name + + " + +linear_3r +add +x + +input_7 +y + + +linear_3& +input_17 +  + +� +�* +name + + " + +input_17x +const +input_19_axes_0 + + +*% +name + +" +input_19_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_17 +axes + +input_19_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_22& +input_19 +  + +� +�* +name + + " + +input_19� +linear +x + + +input_195 +weight+ +) +'model_model_encoder_layers_0_fc1_weight1 +bias) +' +%model_model_encoder_layers_0_fc1_bias& +linear_4 +  + +� +� * +name + + " + +linear_4c +const +input_23_mode_0 +*% +name + +" +input_23_mode_0* +val + + " +EXACT~ +gelu +x + + +linear_4 +mode + +input_23_mode_0& +input_23 +  + +� +� * +name + + " + +input_23� +linear +x + + +input_235 +weight+ +) +'model_model_encoder_layers_0_fc2_weight1 +bias) +' +%model_model_encoder_layers_0_fc2_bias& +linear_5 +  + +� +�* +name + + " + +linear_5s +add +x + + +input_19 +y + + +linear_5& +input_29 +  + +� +�* +name + + " + +input_29� +const& +hidden_states_7_axes_0 + + +*, +name$ + +" +hidden_states_7_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_29" +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_22- +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& +linear_6 +  + +� +�* +name + + " + +linear_6Q +const +var_199 + * +name + + +" +op_199* +val +  + + + +>t +mul +x + + +linear_6 +y + +var_199' + 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& +linear_7 +  + +� +�* +name + + " + +linear_7j +const +var_204 + + +* +name + + +" +op_204** +val# + + + + + ���������@| +reshape +x + + +linear_7 +shape + +var_204* +var_205 +  + +� + +@* +name + + +" +op_205o +const +var_206_perm_0 + + +*# +name + +" + op_206_perm_0*! +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& +linear_8 +  + +� +�* +name + + " + +linear_8j +const +var_211 + + +* +name + + +" +op_211** +val# + + + + + ���������@| +reshape +x + + +linear_8 +shape + +var_211* +var_212 +  + +� + +@* +name + + +" +op_212o +const +var_213_perm_0 + + +*# +name + +" + op_213_perm_0*! +val + + + + + +b +const +var_220 + + +* +name + + +" +op_220*" +val + + + +  +�@} +reshape +x + + tensor_11 +shape + +var_220* +var_221 +  + +� + +@* +name + + +" +op_221o +const +var_222_perm_0 + + +*# +name + +" + op_222_perm_0*! +val + + + + + +i +const +var_224 + + +* +name + + +" +op_224*) +val" + + + + + ���������@� + transpose +x + +var_221 +perm + +var_222_perm_0/ + transpose_52 +  + + +� +@*" +name + +" + transpose_52� +reshape +x + + transpose_52 +shape + +var_224+ +query_states_3 +  + +� +@*$ +name + +" +query_states_3i +const +var_226 + + +* +name + + +" +op_226*) +val" + + + + + ���������@� + transpose +x + +var_205 +perm + +var_206_perm_0/ + transpose_54 +  + + +� +@*" +name + +" + transpose_54� +reshape +x + + transpose_54 +shape + +var_226) + key_states_7 +  + +� +@*" +name + +" + key_states_7i +const +var_228 + + +* +name + + +" +op_228*) +val" + + + + + ���������@� + transpose +x + +var_212 +perm + +var_213_perm_0/ + transpose_53 +  + + +� +@*" +name + +" + transpose_53� +reshape +x + + transpose_53 +shape + +var_228+ +value_states_7 +  + +� +@*$ +name + +" +value_states_7n +const +var_231_perm_0 + + +*# +name + +" + op_231_perm_0* +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_231_perm_0) + transpose_51 +  + +@ +�*" +name + +" + transpose_51� +matmul +x + +query_states_3 +y + + transpose_51/ + 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_233 + + +* +name + + +" +op_233*# +val + + + + + +��� +reshape +x + +attn_weights_7 +shape + +var_233+ +var_234 +  + + +� +�* +name + + +" +op_234� +add +x + +var_234 +y + +attention_mask_12 +attn_weights_9 +  + + +� +�*$ +name + +" +attn_weights_9b +const +var_239 + + +* +name + + +" +op_239*" +val + + + +  +��� +reshape +x + +attn_weights_9 +shape + +var_239& +input_31 +  + +� +�* +name + + " + +input_31x +softmax +x + + +input_31 +axis + + +var_30& +input_33 +  + +� +�* +name + + " + +input_33w +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_33 +y + +value_states_7. + transpose_x + +attn_output_7_transpose_x_0. + transpose_y + +attn_output_7_transpose_y_0* + attn_output_7 +  + +� +@*# +name + +" + attn_output_7b +const +var_244 + + +* +name + + +" +op_244*" +val + + + +  +�@� +reshape +x + + attn_output_7 +shape + +var_2440 + attn_output_9 +  + + +� +@*# +name + +" + attn_output_9~ +const% +attn_output_11_perm_0 + + +*+ +name# + +" +attn_output_11_perm_0*! +val + + + + + +b +const +var_247 + + +* +name + + +" +op_247*" +val + + + +  +��� + transpose +x + + attn_output_9! +perm + +attn_output_11_perm_0/ + transpose_50 +  + +� + +@*" +name + +" + transpose_50~ +reshape +x + + transpose_50 +shape + +var_247& +input_35 +  + +� +�* +name + + " + +input_35� +linear +x + + +input_35D +weight: +8 +6model_model_encoder_layers_1_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_1_self_attn_out_proj_bias& +linear_9 +  + +� +�* +name + + " + +linear_9z +add +x + +hidden_states_7 +y + + +linear_9& +input_39 +  + +� +�* +name + + " + +input_39x +const +input_41_axes_0 + + +*% +name + +" +input_41_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_39 +axes + +input_41_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_22& +input_41 +  + +� +�* +name + + " + +input_41� +linear +x + + +input_415 +weight+ +) +'model_model_encoder_layers_1_fc1_weight1 +bias) +' +%model_model_encoder_layers_1_fc1_bias' + linear_10 +  + +� +� * +name + + " + linear_10c +const +input_45_mode_0 +*% +name + +" +input_45_mode_0* +val + + " +EXACT +gelu +x + + linear_10 +mode + +input_45_mode_0& +input_45 +  + +� +� * +name + + " + +input_45� +linear +x + + +input_455 +weight+ +) +'model_model_encoder_layers_1_fc2_weight1 +bias) +' +%model_model_encoder_layers_1_fc2_bias' + linear_11 +  + +� +�* +name + + " + linear_11t +add +x + + +input_41 +y + + linear_11& +input_51 +  + +� +�* +name + + " + +input_51� +const' +hidden_states_13_axes_0 + + +*- +name% + +" +hidden_states_13_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_51# +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_22. +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' + linear_12 +  + +� +�* +name + + " + linear_12Q +const +var_287 + * +name + + +" +op_287* +val +  + + + +>u +mul +x + + linear_12 +y + +var_287' + 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' + linear_13 +  + +� +�* +name + + " + linear_13j +const +var_292 + + +* +name + + +" +op_292** +val# + + + + + ���������@} +reshape +x + + linear_13 +shape + +var_292* +var_293 +  + +� + +@* +name + + +" +op_293o +const +var_294_perm_0 + + +*# +name + +" + op_294_perm_0*! +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' + linear_14 +  + +� +�* +name + + " + linear_14j +const +var_299 + + +* +name + + +" +op_299** +val# + + + + + ���������@} +reshape +x + + linear_14 +shape + +var_299* +var_300 +  + +� + +@* +name + + +" +op_300o +const +var_301_perm_0 + + +*# +name + +" + op_301_perm_0*! +val + + + + + +b +const +var_308 + + +* +name + + +" +op_308*" +val + + + +  +�@} +reshape +x + + tensor_17 +shape + +var_308* +var_309 +  + +� + +@* +name + + +" +op_309o +const +var_310_perm_0 + + +*# +name + +" + op_310_perm_0*! +val + + + + + +i +const +var_312 + + +* +name + + +" +op_312*) +val" + + + + + ���������@� + transpose +x + +var_309 +perm + +var_310_perm_0/ + transpose_47 +  + + +� +@*" +name + +" + transpose_47� +reshape +x + + transpose_47 +shape + +var_312+ +query_states_5 +  + +� +@*$ +name + +" +query_states_5i +const +var_314 + + +* +name + + +" +op_314*) +val" + + + + + ���������@� + transpose +x + +var_293 +perm + +var_294_perm_0/ + transpose_49 +  + + +� +@*" +name + +" + transpose_49� +reshape +x + + transpose_49 +shape + +var_314* + key_states_11 +  + +� +@*# +name + +" + key_states_11i +const +var_316 + + +* +name + + +" +op_316*) +val" + + + + + ���������@� + transpose +x + +var_300 +perm + +var_301_perm_0/ + transpose_48 +  + + +� +@*" +name + +" + transpose_48� +reshape +x + + transpose_48 +shape + +var_316, +value_states_11 +  + +� +@*% +name + +" +value_states_11n +const +var_319_perm_0 + + +*# +name + +" + op_319_perm_0* +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 + + key_states_11 +perm + +var_319_perm_0) + transpose_46 +  + +@ +�*" +name + +" + transpose_46� +matmul +x + +query_states_5 +y + + transpose_460 + 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_321 + + +* +name + + +" +op_321*# +val + + + + + +��� +reshape +x + +attn_weights_13 +shape + +var_321+ +var_322 +  + + +� +�* +name + + +" +op_322� +add +x + +var_322 +y + +attention_mask_13 +attn_weights_15 +  + + +� +�*% +name + +" +attn_weights_15b +const +var_327 + + +* +name + + +" +op_327*" +val + + + +  +��� +reshape +x + +attn_weights_15 +shape + +var_327& +input_53 +  + +� +�* +name + + " + +input_53x +softmax +x + + +input_53 +axis + + +var_30& +input_55 +  + +� +�* +name + + " + +input_55y +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_55 +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_332 + + +* +name + + +" +op_332*" +val + + + +  +�@� +reshape +x + +attn_output_13 +shape + +var_3321 +attn_output_15 +  + + +� +@*$ +name + +" +attn_output_15~ +const% +attn_output_17_perm_0 + + +*+ +name# + +" +attn_output_17_perm_0*! +val + + + + + +b +const +var_335 + + +* +name + + +" +op_335*" +val + + + +  +��� + transpose +x + +attn_output_15! +perm + +attn_output_17_perm_0/ + transpose_45 +  + +� + +@*" +name + +" + transpose_45~ +reshape +x + + transpose_45 +shape + +var_335& +input_57 +  + +� +�* +name + + " + +input_57� +linear +x + + +input_57D +weight: +8 +6model_model_encoder_layers_2_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_2_self_attn_out_proj_bias' + linear_15 +  + +� +�* +name + + " + linear_15| +add +x + +hidden_states_13 +y + + linear_15& +input_61 +  + +� +�* +name + + " + +input_61x +const +input_63_axes_0 + + +*% +name + +" +input_63_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_61 +axes + +input_63_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_22& +input_63 +  + +� +�* +name + + " + +input_63� +linear +x + + +input_635 +weight+ +) +'model_model_encoder_layers_2_fc1_weight1 +bias) +' +%model_model_encoder_layers_2_fc1_bias' + linear_16 +  + +� +� * +name + + " + linear_16c +const +input_67_mode_0 +*% +name + +" +input_67_mode_0* +val + + " +EXACT +gelu +x + + linear_16 +mode + +input_67_mode_0& +input_67 +  + +� +� * +name + + " + +input_67� +linear +x + + +input_675 +weight+ +) +'model_model_encoder_layers_2_fc2_weight1 +bias) +' +%model_model_encoder_layers_2_fc2_bias' + linear_17 +  + +� +�* +name + + " + linear_17t +add +x + + +input_63 +y + + linear_17& +input_73 +  + +� +�* +name + + " + +input_73� +const' +hidden_states_19_axes_0 + + +*- +name% + +" +hidden_states_19_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_73# +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_22. +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' + linear_18 +  + +� +�* +name + + " + linear_18Q +const +var_375 + * +name + + +" +op_375* +val +  + + + +>u +mul +x + + linear_18 +y + +var_375' + 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' + linear_19 +  + +� +�* +name + + " + linear_19j +const +var_380 + + +* +name + + +" +op_380** +val# + + + + + ���������@} +reshape +x + + linear_19 +shape + +var_380* +var_381 +  + +� + +@* +name + + +" +op_381o +const +var_382_perm_0 + + +*# +name + +" + op_382_perm_0*! +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' + linear_20 +  + +� +�* +name + + " + linear_20j +const +var_387 + + +* +name + + +" +op_387** +val# + + + + + ���������@} +reshape +x + + linear_20 +shape + +var_387* +var_388 +  + +� + +@* +name + + +" +op_388o +const +var_389_perm_0 + + +*# +name + +" + op_389_perm_0*! +val + + + + + +b +const +var_396 + + +* +name + + +" +op_396*" +val + + + +  +�@} +reshape +x + + tensor_23 +shape + +var_396* +var_397 +  + +� + +@* +name + + +" +op_397o +const +var_398_perm_0 + + +*# +name + +" + op_398_perm_0*! +val + + + + + +i +const +var_400 + + +* +name + + +" +op_400*) +val" + + + + + ���������@� + transpose +x + +var_397 +perm + +var_398_perm_0/ + transpose_42 +  + + +� +@*" +name + +" + transpose_42� +reshape +x + + transpose_42 +shape + +var_400+ +query_states_7 +  + +� +@*$ +name + +" +query_states_7i +const +var_402 + + +* +name + + +" +op_402*) +val" + + + + + ���������@� + transpose +x + +var_381 +perm + +var_382_perm_0/ + transpose_44 +  + + +� +@*" +name + +" + transpose_44� +reshape +x + + transpose_44 +shape + +var_402* + key_states_15 +  + +� +@*# +name + +" + key_states_15i +const +var_404 + + +* +name + + +" +op_404*) +val" + + + + + ���������@� + transpose +x + +var_388 +perm + +var_389_perm_0/ + transpose_43 +  + + +� +@*" +name + +" + transpose_43� +reshape +x + + transpose_43 +shape + +var_404, +value_states_15 +  + +� +@*% +name + +" +value_states_15n +const +var_407_perm_0 + + +*# +name + +" + op_407_perm_0* +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 + + key_states_15 +perm + +var_407_perm_0) + transpose_41 +  + +@ +�*" +name + +" + transpose_41� +matmul +x + +query_states_7 +y + + transpose_410 + 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_409 + + +* +name + + +" +op_409*# +val + + + + + +��� +reshape +x + +attn_weights_19 +shape + +var_409+ +var_410 +  + + +� +�* +name + + +" +op_410� +add +x + +var_410 +y + +attention_mask_13 +attn_weights_21 +  + + +� +�*% +name + +" +attn_weights_21b +const +var_415 + + +* +name + + +" +op_415*" +val + + + +  +��� +reshape +x + +attn_weights_21 +shape + +var_415& +input_75 +  + +� +�* +name + + " + +input_75x +softmax +x + + +input_75 +axis + + +var_30& +input_77 +  + +� +�* +name + + " + +input_77y +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_77 +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_420 + + +* +name + + +" +op_420*" +val + + + +  +�@� +reshape +x + +attn_output_19 +shape + +var_4201 +attn_output_21 +  + + +� +@*$ +name + +" +attn_output_21~ +const% +attn_output_23_perm_0 + + +*+ +name# + +" +attn_output_23_perm_0*! +val + + + + + +b +const +var_423 + + +* +name + + +" +op_423*" +val + + + +  +��� + transpose +x + +attn_output_21! +perm + +attn_output_23_perm_0/ + transpose_40 +  + +� + +@*" +name + +" + transpose_40~ +reshape +x + + transpose_40 +shape + +var_423& +input_79 +  + +� +�* +name + + " + +input_79� +linear +x + + +input_79D +weight: +8 +6model_model_encoder_layers_3_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_3_self_attn_out_proj_bias' + linear_21 +  + +� +�* +name + + " + linear_21| +add +x + +hidden_states_19 +y + + linear_21& +input_83 +  + +� +�* +name + + " + +input_83x +const +input_85_axes_0 + + +*% +name + +" +input_85_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_83 +axes + +input_85_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_22& +input_85 +  + +� +�* +name + + " + +input_85� +linear +x + + +input_855 +weight+ +) +'model_model_encoder_layers_3_fc1_weight1 +bias) +' +%model_model_encoder_layers_3_fc1_bias' + linear_22 +  + +� +� * +name + + " + linear_22c +const +input_89_mode_0 +*% +name + +" +input_89_mode_0* +val + + " +EXACT +gelu +x + + linear_22 +mode + +input_89_mode_0& +input_89 +  + +� +� * +name + + " + +input_89� +linear +x + + +input_895 +weight+ +) +'model_model_encoder_layers_3_fc2_weight1 +bias) +' +%model_model_encoder_layers_3_fc2_bias' + linear_23 +  + +� +�* +name + + " + linear_23t +add +x + + +input_85 +y + + linear_23& +input_95 +  + +� +�* +name + + " + +input_95� +const' +hidden_states_25_axes_0 + + +*- +name% + +" +hidden_states_25_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_95# +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_22. +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' + linear_24 +  + +� +�* +name + + " + linear_24Q +const +var_463 + * +name + + +" +op_463* +val +  + + + +>u +mul +x + + linear_24 +y + +var_463' + 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' + linear_25 +  + +� +�* +name + + " + linear_25j +const +var_468 + + +* +name + + +" +op_468** +val# + + + + + ���������@} +reshape +x + + linear_25 +shape + +var_468* +var_469 +  + +� + +@* +name + + +" +op_469o +const +var_470_perm_0 + + +*# +name + +" + op_470_perm_0*! +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' + linear_26 +  + +� +�* +name + + " + linear_26j +const +var_475 + + +* +name + + +" +op_475** +val# + + + + + ���������@} +reshape +x + + linear_26 +shape + +var_475* +var_476 +  + +� + +@* +name + + +" +op_476o +const +var_477_perm_0 + + +*# +name + +" + op_477_perm_0*! +val + + + + + +b +const +var_484 + + +* +name + + +" +op_484*" +val + + + +  +�@} +reshape +x + + tensor_29 +shape + +var_484* +var_485 +  + +� + +@* +name + + +" +op_485o +const +var_486_perm_0 + + +*# +name + +" + op_486_perm_0*! +val + + + + + +i +const +var_488 + + +* +name + + +" +op_488*) +val" + + + + + ���������@� + transpose +x + +var_485 +perm + +var_486_perm_0/ + transpose_37 +  + + +� +@*" +name + +" + transpose_37� +reshape +x + + transpose_37 +shape + +var_488+ +query_states_9 +  + +� +@*$ +name + +" +query_states_9i +const +var_490 + + +* +name + + +" +op_490*) +val" + + + + + ���������@� + transpose +x + +var_469 +perm + +var_470_perm_0/ + transpose_39 +  + + +� +@*" +name + +" + transpose_39� +reshape +x + + transpose_39 +shape + +var_490* + key_states_19 +  + +� +@*# +name + +" + key_states_19i +const +var_492 + + +* +name + + +" +op_492*) +val" + + + + + ���������@� + transpose +x + +var_476 +perm + +var_477_perm_0/ + transpose_38 +  + + +� +@*" +name + +" + transpose_38� +reshape +x + + transpose_38 +shape + +var_492, +value_states_19 +  + +� +@*% +name + +" +value_states_19n +const +var_495_perm_0 + + +*# +name + +" + op_495_perm_0* +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 + + key_states_19 +perm + +var_495_perm_0) + transpose_36 +  + +@ +�*" +name + +" + transpose_36� +matmul +x + +query_states_9 +y + + transpose_360 + 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_497 + + +* +name + + +" +op_497*# +val + + + + + +��� +reshape +x + +attn_weights_25 +shape + +var_497+ +var_498 +  + + +� +�* +name + + +" +op_498� +add +x + +var_498 +y + +attention_mask_13 +attn_weights_27 +  + + +� +�*% +name + +" +attn_weights_27b +const +var_503 + + +* +name + + +" +op_503*" +val + + + +  +��� +reshape +x + +attn_weights_27 +shape + +var_503& +input_97 +  + +� +�* +name + + " + +input_97x +softmax +x + + +input_97 +axis + + +var_30& +input_99 +  + +� +�* +name + + " + +input_99y +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_99 +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_508 + + +* +name + + +" +op_508*" +val + + + +  +�@� +reshape +x + +attn_output_25 +shape + +var_5081 +attn_output_27 +  + + +� +@*$ +name + +" +attn_output_27~ +const% +attn_output_29_perm_0 + + +*+ +name# + +" +attn_output_29_perm_0*! +val + + + + + +b +const +var_511 + + +* +name + + +" +op_511*" +val + + + +  +��� + transpose +x + +attn_output_27! +perm + +attn_output_29_perm_0/ + transpose_35 +  + +� + +@*" +name + +" + transpose_35� +reshape +x + + transpose_35 +shape + +var_511' + input_101 +  + +� +�* +name + + " + input_101� +linear +x + + input_101D +weight: +8 +6model_model_encoder_layers_4_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_4_self_attn_out_proj_bias' + linear_27 +  + +� +�* +name + + " + linear_27~ +add +x + +hidden_states_25 +y + + linear_27' + input_105 +  + +� +�* +name + + " + input_105z +const +input_107_axes_0 + + +*& +name + +" +input_107_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_105 +axes + +input_107_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_22' + input_107 +  + +� +�* +name + + " + input_107� +linear +x + + input_1075 +weight+ +) +'model_model_encoder_layers_4_fc1_weight1 +bias) +' +%model_model_encoder_layers_4_fc1_bias' + linear_28 +  + +� +� * +name + + " + linear_28e +const +input_111_mode_0 +*& +name + +" +input_111_mode_0* +val + + " +EXACT� +gelu +x + + linear_28 +mode + +input_111_mode_0' + input_111 +  + +� +� * +name + + " + input_111� +linear +x + + input_1115 +weight+ +) +'model_model_encoder_layers_4_fc2_weight1 +bias) +' +%model_model_encoder_layers_4_fc2_bias' + linear_29 +  + +� +�* +name + + " + linear_29w +add +x + + input_107 +y + + linear_29' + input_117 +  + +� +�* +name + + " + input_117� +const' +hidden_states_31_axes_0 + + +*- +name% + +" +hidden_states_31_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_117# +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_22. +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' + linear_30 +  + +� +�* +name + + " + linear_30Q +const +var_551 + * +name + + +" +op_551* +val +  + + + +>u +mul +x + + linear_30 +y + +var_551' + tensor_35 +  + +� +�* +name + + " + tensor_35� +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' + linear_31 +  + +� +�* +name + + " + linear_31j +const +var_556 + + +* +name + + +" +op_556** +val# + + + + + ���������@} +reshape +x + + linear_31 +shape + +var_556* +var_557 +  + +� + +@* +name + + +" +op_557o +const +var_558_perm_0 + + +*# +name + +" + op_558_perm_0*! +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' + linear_32 +  + +� +�* +name + + " + linear_32j +const +var_563 + + +* +name + + +" +op_563** +val# + + + + + ���������@} +reshape +x + + linear_32 +shape + +var_563* +var_564 +  + +� + +@* +name + + +" +op_564o +const +var_565_perm_0 + + +*# +name + +" + op_565_perm_0*! +val + + + + + +b +const +var_572 + + +* +name + + +" +op_572*" +val + + + +  +�@} +reshape +x + + tensor_35 +shape + +var_572* +var_573 +  + +� + +@* +name + + +" +op_573o +const +var_574_perm_0 + + +*# +name + +" + op_574_perm_0*! +val + + + + + +i +const +var_576 + + +* +name + + +" +op_576*) +val" + + + + + ���������@� + transpose +x + +var_573 +perm + +var_574_perm_0/ + transpose_32 +  + + +� +@*" +name + +" + transpose_32� +reshape +x + + transpose_32 +shape + +var_576, +query_states_11 +  + +� +@*% +name + +" +query_states_11i +const +var_578 + + +* +name + + +" +op_578*) +val" + + + + + ���������@� + transpose +x + +var_557 +perm + +var_558_perm_0/ + transpose_34 +  + + +� +@*" +name + +" + transpose_34� +reshape +x + + transpose_34 +shape + +var_578* + key_states_23 +  + +� +@*# +name + +" + key_states_23i +const +var_580 + + +* +name + + +" +op_580*) +val" + + + + + ���������@� + transpose +x + +var_564 +perm + +var_565_perm_0/ + transpose_33 +  + + +� +@*" +name + +" + transpose_33� +reshape +x + + transpose_33 +shape + +var_580, +value_states_23 +  + +� +@*% +name + +" +value_states_23n +const +var_583_perm_0 + + +*# +name + +" + op_583_perm_0* +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_23 +perm + +var_583_perm_0) + transpose_31 +  + +@ +�*" +name + +" + transpose_31� +matmul +x + +query_states_11 +y + + transpose_310 + 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_585 + + +* +name + + +" +op_585*# +val + + + + + +��� +reshape +x + +attn_weights_31 +shape + +var_585+ +var_586 +  + + +� +�* +name + + +" +op_586� +add +x + +var_586 +y + +attention_mask_13 +attn_weights_33 +  + + +� +�*% +name + +" +attn_weights_33b +const +var_591 + + +* +name + + +" +op_591*" +val + + + +  +��� +reshape +x + +attn_weights_33 +shape + +var_591' + input_119 +  + +� +�* +name + + " + input_119{ +softmax +x + + input_119 +axis + + +var_30' + input_121 +  + +� +�* +name + + " + input_121y +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_121 +y + +value_states_23/ + 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_596 + + +* +name + + +" +op_596*" +val + + + +  +�@� +reshape +x + +attn_output_31 +shape + +var_5961 +attn_output_33 +  + + +� +@*$ +name + +" +attn_output_33~ +const% +attn_output_35_perm_0 + + +*+ +name# + +" +attn_output_35_perm_0*! +val + + + + + +b +const +var_599 + + +* +name + + +" +op_599*" +val + + + +  +��� + transpose +x + +attn_output_33! +perm + +attn_output_35_perm_0/ + transpose_30 +  + +� + +@*" +name + +" + transpose_30� +reshape +x + + transpose_30 +shape + +var_599' + input_123 +  + +� +�* +name + + " + input_123� +linear +x + + input_123D +weight: +8 +6model_model_encoder_layers_5_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_5_self_attn_out_proj_bias' + linear_33 +  + +� +�* +name + + " + linear_33~ +add +x + +hidden_states_31 +y + + linear_33' + input_127 +  + +� +�* +name + + " + input_127z +const +input_129_axes_0 + + +*& +name + +" +input_129_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_127 +axes + +input_129_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_22' + input_129 +  + +� +�* +name + + " + input_129� +linear +x + + input_1295 +weight+ +) +'model_model_encoder_layers_5_fc1_weight1 +bias) +' +%model_model_encoder_layers_5_fc1_bias' + linear_34 +  + +� +� * +name + + " + linear_34e +const +input_133_mode_0 +*& +name + +" +input_133_mode_0* +val + + " +EXACT� +gelu +x + + linear_34 +mode + +input_133_mode_0' + input_133 +  + +� +� * +name + + " + input_133� +linear +x + + input_1335 +weight+ +) +'model_model_encoder_layers_5_fc2_weight1 +bias) +' +%model_model_encoder_layers_5_fc2_bias' + linear_35 +  + +� +�* +name + + " + linear_35w +add +x + + input_129 +y + + linear_35' + input_139 +  + +� +�* +name + + " + input_139� +const' +hidden_states_37_axes_0 + + +*- +name% + +" +hidden_states_37_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_139# +axes + +hidden_states_37_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_22. +hidden_states_37 +  + +� +�*& +name + +" +hidden_states_37� +linear +x + +hidden_states_37B +weight8 +6 +4model_model_encoder_layers_6_self_attn_q_proj_weight> +bias6 +4 +2model_model_encoder_layers_6_self_attn_q_proj_bias' + linear_36 +  + +� +�* +name + + " + linear_36Q +const +var_639 + * +name + + +" +op_639* +val +  + + + +>u +mul +x + + linear_36 +y + +var_639' + tensor_41 +  + +� +�* +name + + " + tensor_41� +linear +x + +hidden_states_37B +weight8 +6 +4model_model_encoder_layers_6_self_attn_k_proj_weight> +bias6 +4 +2model_model_encoder_layers_6_self_attn_k_proj_bias' + linear_37 +  + +� +�* +name + + " + linear_37j +const +var_644 + + +* +name + + +" +op_644** +val# + + + + + ���������@} +reshape +x + + linear_37 +shape + +var_644* +var_645 +  + +� + +@* +name + + +" +op_645o +const +var_646_perm_0 + + +*# +name + +" + op_646_perm_0*! +val + + + + + +� +linear +x + +hidden_states_37B +weight8 +6 +4model_model_encoder_layers_6_self_attn_v_proj_weight> +bias6 +4 +2model_model_encoder_layers_6_self_attn_v_proj_bias' + linear_38 +  + +� +�* +name + + " + linear_38j +const +var_651 + + +* +name + + +" +op_651** +val# + + + + + ���������@} +reshape +x + + linear_38 +shape + +var_651* +var_652 +  + +� + +@* +name + + +" +op_652o +const +var_653_perm_0 + + +*# +name + +" + op_653_perm_0*! +val + + + + + +b +const +var_660 + + +* +name + + +" +op_660*" +val + + + +  +�@} +reshape +x + + tensor_41 +shape + +var_660* +var_661 +  + +� + +@* +name + + +" +op_661o +const +var_662_perm_0 + + +*# +name + +" + op_662_perm_0*! +val + + + + + +i +const +var_664 + + +* +name + + +" +op_664*) +val" + + + + + ���������@� + transpose +x + +var_661 +perm + +var_662_perm_0/ + transpose_27 +  + + +� +@*" +name + +" + transpose_27� +reshape +x + + transpose_27 +shape + +var_664, +query_states_13 +  + +� +@*% +name + +" +query_states_13i +const +var_666 + + +* +name + + +" +op_666*) +val" + + + + + ���������@� + transpose +x + +var_645 +perm + +var_646_perm_0/ + transpose_29 +  + + +� +@*" +name + +" + transpose_29� +reshape +x + + transpose_29 +shape + +var_666* + key_states_27 +  + +� +@*# +name + +" + key_states_27i +const +var_668 + + +* +name + + +" +op_668*) +val" + + + + + ���������@� + transpose +x + +var_652 +perm + +var_653_perm_0/ + transpose_28 +  + + +� +@*" +name + +" + transpose_28� +reshape +x + + transpose_28 +shape + +var_668, +value_states_27 +  + +� +@*% +name + +" +value_states_27n +const +var_671_perm_0 + + +*# +name + +" + op_671_perm_0* +val + + + + +{ +const% +attn_weights_37_transpose_x_0 +*3 +name+ +# +!" +attn_weights_37_transpose_x_0* +val + + +{ +const% +attn_weights_37_transpose_y_0 +*3 +name+ +# +!" +attn_weights_37_transpose_y_0* +val + + +� + transpose +x + + key_states_27 +perm + +var_671_perm_0) + transpose_26 +  + +@ +�*" +name + +" + transpose_26� +matmul +x + +query_states_13 +y + + transpose_260 + transpose_x! + +attn_weights_37_transpose_x_00 + transpose_y! + +attn_weights_37_transpose_y_0- +attn_weights_37 +  + +� +�*% +name + +" +attn_weights_37c +const +var_673 + + +* +name + + +" +op_673*# +val + + + + + +��� +reshape +x + +attn_weights_37 +shape + +var_673+ +var_674 +  + + +� +�* +name + + +" +op_674� +add +x + +var_674 +y + +attention_mask_13 +attn_weights_39 +  + + +� +�*% +name + +" +attn_weights_39b +const +var_679 + + +* +name + + +" +op_679*" +val + + + +  +��� +reshape +x + +attn_weights_39 +shape + +var_679' + input_141 +  + +� +�* +name + + " + input_141{ +softmax +x + + input_141 +axis + + +var_30' + input_143 +  + +� +�* +name + + " + input_143y +const$ +attn_output_37_transpose_x_0 +*2 +name* +" + " +attn_output_37_transpose_x_0* +val + + +y +const$ +attn_output_37_transpose_y_0 +*2 +name* +" + " +attn_output_37_transpose_y_0* +val + + +� +matmul +x + + input_143 +y + +value_states_27/ + transpose_x + +attn_output_37_transpose_x_0/ + transpose_y + +attn_output_37_transpose_y_0+ +attn_output_37 +  + +� +@*$ +name + +" +attn_output_37b +const +var_684 + + +* +name + + +" +op_684*" +val + + + +  +�@� +reshape +x + +attn_output_37 +shape + +var_6841 +attn_output_39 +  + + +� +@*$ +name + +" +attn_output_39~ +const% +attn_output_41_perm_0 + + +*+ +name# + +" +attn_output_41_perm_0*! +val + + + + + +b +const +var_687 + + +* +name + + +" +op_687*" +val + + + +  +��� + transpose +x + +attn_output_39! +perm + +attn_output_41_perm_0/ + transpose_25 +  + +� + +@*" +name + +" + transpose_25� +reshape +x + + transpose_25 +shape + +var_687' + input_145 +  + +� +�* +name + + " + input_145� +linear +x + + input_145D +weight: +8 +6model_model_encoder_layers_6_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_6_self_attn_out_proj_bias' + linear_39 +  + +� +�* +name + + " + linear_39~ +add +x + +hidden_states_37 +y + + linear_39' + input_149 +  + +� +�* +name + + " + input_149z +const +input_151_axes_0 + + +*& +name + +" +input_151_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_149 +axes + +input_151_axes_0E +gamma< +: +8model_model_encoder_layers_6_self_attn_layer_norm_weightB +beta: +8 +6model_model_encoder_layers_6_self_attn_layer_norm_bias +epsilon + + +var_22' + input_151 +  + +� +�* +name + + " + input_151� +linear +x + + input_1515 +weight+ +) +'model_model_encoder_layers_6_fc1_weight1 +bias) +' +%model_model_encoder_layers_6_fc1_bias' + linear_40 +  + +� +� * +name + + " + linear_40e +const +input_155_mode_0 +*& +name + +" +input_155_mode_0* +val + + " +EXACT� +gelu +x + + linear_40 +mode + +input_155_mode_0' + input_155 +  + +� +� * +name + + " + input_155� +linear +x + + input_1555 +weight+ +) +'model_model_encoder_layers_6_fc2_weight1 +bias) +' +%model_model_encoder_layers_6_fc2_bias' + linear_41 +  + +� +�* +name + + " + linear_41w +add +x + + input_151 +y + + linear_41' + input_161 +  + +� +�* +name + + " + input_161� +const' +hidden_states_43_axes_0 + + +*- +name% + +" +hidden_states_43_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_161# +axes + +hidden_states_43_axes_0A +gamma8 +6 +4model_model_encoder_layers_6_final_layer_norm_weight> +beta6 +4 +2model_model_encoder_layers_6_final_layer_norm_bias +epsilon + + +var_22. +hidden_states_43 +  + +� +�*& +name + +" +hidden_states_43� +linear +x + +hidden_states_43B +weight8 +6 +4model_model_encoder_layers_7_self_attn_q_proj_weight> +bias6 +4 +2model_model_encoder_layers_7_self_attn_q_proj_bias' + linear_42 +  + +� +�* +name + + " + linear_42Q +const +var_727 + * +name + + +" +op_727* +val +  + + + +>u +mul +x + + linear_42 +y + +var_727' + tensor_47 +  + +� +�* +name + + " + tensor_47� +linear +x + +hidden_states_43B +weight8 +6 +4model_model_encoder_layers_7_self_attn_k_proj_weight> +bias6 +4 +2model_model_encoder_layers_7_self_attn_k_proj_bias' + linear_43 +  + +� +�* +name + + " + linear_43j +const +var_732 + + +* +name + + +" +op_732** +val# + + + + + ���������@} +reshape +x + + linear_43 +shape + +var_732* +var_733 +  + +� + +@* +name + + +" +op_733o +const +var_734_perm_0 + + +*# +name + +" + op_734_perm_0*! +val + + + + + +� +linear +x + +hidden_states_43B +weight8 +6 +4model_model_encoder_layers_7_self_attn_v_proj_weight> +bias6 +4 +2model_model_encoder_layers_7_self_attn_v_proj_bias' + linear_44 +  + +� +�* +name + + " + linear_44j +const +var_739 + + +* +name + + +" +op_739** +val# + + + + + ���������@} +reshape +x + + linear_44 +shape + +var_739* +var_740 +  + +� + +@* +name + + +" +op_740o +const +var_741_perm_0 + + +*# +name + +" + op_741_perm_0*! +val + + + + + +b +const +var_748 + + +* +name + + +" +op_748*" +val + + + +  +�@} +reshape +x + + tensor_47 +shape + +var_748* +var_749 +  + +� + +@* +name + + +" +op_749o +const +var_750_perm_0 + + +*# +name + +" + op_750_perm_0*! +val + + + + + +i +const +var_752 + + +* +name + + +" +op_752*) +val" + + + + + ���������@� + transpose +x + +var_749 +perm + +var_750_perm_0/ + transpose_22 +  + + +� +@*" +name + +" + transpose_22� +reshape +x + + transpose_22 +shape + +var_752, +query_states_15 +  + +� +@*% +name + +" +query_states_15i +const +var_754 + + +* +name + + +" +op_754*) +val" + + + + + ���������@� + transpose +x + +var_733 +perm + +var_734_perm_0/ + transpose_24 +  + + +� +@*" +name + +" + transpose_24� +reshape +x + + transpose_24 +shape + +var_754* + key_states_31 +  + +� +@*# +name + +" + key_states_31i +const +var_756 + + +* +name + + +" +op_756*) +val" + + + + + ���������@� + transpose +x + +var_740 +perm + +var_741_perm_0/ + transpose_23 +  + + +� +@*" +name + +" + transpose_23� +reshape +x + + transpose_23 +shape + +var_756, +value_states_31 +  + +� +@*% +name + +" +value_states_31n +const +var_759_perm_0 + + +*# +name + +" + op_759_perm_0* +val + + + + +{ +const% +attn_weights_43_transpose_x_0 +*3 +name+ +# +!" +attn_weights_43_transpose_x_0* +val + + +{ +const% +attn_weights_43_transpose_y_0 +*3 +name+ +# +!" +attn_weights_43_transpose_y_0* +val + + +� + transpose +x + + key_states_31 +perm + +var_759_perm_0) + transpose_21 +  + +@ +�*" +name + +" + transpose_21� +matmul +x + +query_states_15 +y + + transpose_210 + transpose_x! + +attn_weights_43_transpose_x_00 + transpose_y! + +attn_weights_43_transpose_y_0- +attn_weights_43 +  + +� +�*% +name + +" +attn_weights_43c +const +var_761 + + +* +name + + +" +op_761*# +val + + + + + +��� +reshape +x + +attn_weights_43 +shape + +var_761+ +var_762 +  + + +� +�* +name + + +" +op_762� +add +x + +var_762 +y + +attention_mask_13 +attn_weights_45 +  + + +� +�*% +name + +" +attn_weights_45b +const +var_767 + + +* +name + + +" +op_767*" +val + + + +  +��� +reshape +x + +attn_weights_45 +shape + +var_767' + input_163 +  + +� +�* +name + + " + input_163{ +softmax +x + + input_163 +axis + + +var_30' + input_165 +  + +� +�* +name + + " + input_165y +const$ +attn_output_43_transpose_x_0 +*2 +name* +" + " +attn_output_43_transpose_x_0* +val + + +y +const$ +attn_output_43_transpose_y_0 +*2 +name* +" + " +attn_output_43_transpose_y_0* +val + + +� +matmul +x + + input_165 +y + +value_states_31/ + transpose_x + +attn_output_43_transpose_x_0/ + transpose_y + +attn_output_43_transpose_y_0+ +attn_output_43 +  + +� +@*$ +name + +" +attn_output_43b +const +var_772 + + +* +name + + +" +op_772*" +val + + + +  +�@� +reshape +x + +attn_output_43 +shape + +var_7721 +attn_output_45 +  + + +� +@*$ +name + +" +attn_output_45~ +const% +attn_output_47_perm_0 + + +*+ +name# + +" +attn_output_47_perm_0*! +val + + + + + +b +const +var_775 + + +* +name + + +" +op_775*" +val + + + +  +��� + transpose +x + +attn_output_45! +perm + +attn_output_47_perm_0/ + transpose_20 +  + +� + +@*" +name + +" + transpose_20� +reshape +x + + transpose_20 +shape + +var_775' + input_167 +  + +� +�* +name + + " + input_167� +linear +x + + input_167D +weight: +8 +6model_model_encoder_layers_7_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_7_self_attn_out_proj_bias' + linear_45 +  + +� +�* +name + + " + linear_45~ +add +x + +hidden_states_43 +y + + linear_45' + input_171 +  + +� +�* +name + + " + input_171z +const +input_173_axes_0 + + +*& +name + +" +input_173_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_171 +axes + +input_173_axes_0E +gamma< +: +8model_model_encoder_layers_7_self_attn_layer_norm_weightB +beta: +8 +6model_model_encoder_layers_7_self_attn_layer_norm_bias +epsilon + + +var_22' + input_173 +  + +� +�* +name + + " + input_173� +linear +x + + input_1735 +weight+ +) +'model_model_encoder_layers_7_fc1_weight1 +bias) +' +%model_model_encoder_layers_7_fc1_bias' + linear_46 +  + +� +� * +name + + " + linear_46e +const +input_177_mode_0 +*& +name + +" +input_177_mode_0* +val + + " +EXACT� +gelu +x + + linear_46 +mode + +input_177_mode_0' + input_177 +  + +� +� * +name + + " + input_177� +linear +x + + input_1775 +weight+ +) +'model_model_encoder_layers_7_fc2_weight1 +bias) +' +%model_model_encoder_layers_7_fc2_bias' + linear_47 +  + +� +�* +name + + " + linear_47w +add +x + + input_173 +y + + linear_47' + input_183 +  + +� +�* +name + + " + input_183� +const' +hidden_states_49_axes_0 + + +*- +name% + +" +hidden_states_49_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_183# +axes + +hidden_states_49_axes_0A +gamma8 +6 +4model_model_encoder_layers_7_final_layer_norm_weight> +beta6 +4 +2model_model_encoder_layers_7_final_layer_norm_bias +epsilon + + +var_22. +hidden_states_49 +  + +� +�*& +name + +" +hidden_states_49� +linear +x + +hidden_states_49B +weight8 +6 +4model_model_encoder_layers_8_self_attn_q_proj_weight> +bias6 +4 +2model_model_encoder_layers_8_self_attn_q_proj_bias' + linear_48 +  + +� +�* +name + + " + linear_48Q +const +var_815 + * +name + + +" +op_815* +val +  + + + +>u +mul +x + + linear_48 +y + +var_815' + tensor_53 +  + +� +�* +name + + " + tensor_53� +linear +x + +hidden_states_49B +weight8 +6 +4model_model_encoder_layers_8_self_attn_k_proj_weight> +bias6 +4 +2model_model_encoder_layers_8_self_attn_k_proj_bias' + linear_49 +  + +� +�* +name + + " + linear_49j +const +var_820 + + +* +name + + +" +op_820** +val# + + + + + ���������@} +reshape +x + + linear_49 +shape + +var_820* +var_821 +  + +� + +@* +name + + +" +op_821o +const +var_822_perm_0 + + +*# +name + +" + op_822_perm_0*! +val + + + + + +� +linear +x + +hidden_states_49B +weight8 +6 +4model_model_encoder_layers_8_self_attn_v_proj_weight> +bias6 +4 +2model_model_encoder_layers_8_self_attn_v_proj_bias' + linear_50 +  + +� +�* +name + + " + linear_50j +const +var_827 + + +* +name + + +" +op_827** +val# + + + + + ���������@} +reshape +x + + linear_50 +shape + +var_827* +var_828 +  + +� + +@* +name + + +" +op_828o +const +var_829_perm_0 + + +*# +name + +" + op_829_perm_0*! +val + + + + + +b +const +var_836 + + +* +name + + +" +op_836*" +val + + + +  +�@} +reshape +x + + tensor_53 +shape + +var_836* +var_837 +  + +� + +@* +name + + +" +op_837o +const +var_838_perm_0 + + +*# +name + +" + op_838_perm_0*! +val + + + + + +i +const +var_840 + + +* +name + + +" +op_840*) +val" + + + + + ���������@� + transpose +x + +var_837 +perm + +var_838_perm_0/ + transpose_17 +  + + +� +@*" +name + +" + transpose_17� +reshape +x + + transpose_17 +shape + +var_840, +query_states_17 +  + +� +@*% +name + +" +query_states_17i +const +var_842 + + +* +name + + +" +op_842*) +val" + + + + + ���������@� + transpose +x + +var_821 +perm + +var_822_perm_0/ + transpose_19 +  + + +� +@*" +name + +" + transpose_19� +reshape +x + + transpose_19 +shape + +var_842* + key_states_35 +  + +� +@*# +name + +" + key_states_35i +const +var_844 + + +* +name + + +" +op_844*) +val" + + + + + ���������@� + transpose +x + +var_828 +perm + +var_829_perm_0/ + transpose_18 +  + + +� +@*" +name + +" + transpose_18� +reshape +x + + transpose_18 +shape + +var_844, +value_states_35 +  + +� +@*% +name + +" +value_states_35n +const +var_847_perm_0 + + +*# +name + +" + op_847_perm_0* +val + + + + +{ +const% +attn_weights_49_transpose_x_0 +*3 +name+ +# +!" +attn_weights_49_transpose_x_0* +val + + +{ +const% +attn_weights_49_transpose_y_0 +*3 +name+ +# +!" +attn_weights_49_transpose_y_0* +val + + +� + transpose +x + + key_states_35 +perm + +var_847_perm_0) + transpose_16 +  + +@ +�*" +name + +" + transpose_16� +matmul +x + +query_states_17 +y + + transpose_160 + transpose_x! + +attn_weights_49_transpose_x_00 + transpose_y! + +attn_weights_49_transpose_y_0- +attn_weights_49 +  + +� +�*% +name + +" +attn_weights_49c +const +var_849 + + +* +name + + +" +op_849*# +val + + + + + +��� +reshape +x + +attn_weights_49 +shape + +var_849+ +var_850 +  + + +� +�* +name + + +" +op_850� +add +x + +var_850 +y + +attention_mask_13 +attn_weights_51 +  + + +� +�*% +name + +" +attn_weights_51b +const +var_855 + + +* +name + + +" +op_855*" +val + + + +  +��� +reshape +x + +attn_weights_51 +shape + +var_855' + input_185 +  + +� +�* +name + + " + input_185{ +softmax +x + + input_185 +axis + + +var_30' + input_187 +  + +� +�* +name + + " + input_187y +const$ +attn_output_49_transpose_x_0 +*2 +name* +" + " +attn_output_49_transpose_x_0* +val + + +y +const$ +attn_output_49_transpose_y_0 +*2 +name* +" + " +attn_output_49_transpose_y_0* +val + + +� +matmul +x + + input_187 +y + +value_states_35/ + transpose_x + +attn_output_49_transpose_x_0/ + transpose_y + +attn_output_49_transpose_y_0+ +attn_output_49 +  + +� +@*$ +name + +" +attn_output_49b +const +var_860 + + +* +name + + +" +op_860*" +val + + + +  +�@� +reshape +x + +attn_output_49 +shape + +var_8601 +attn_output_51 +  + + +� +@*$ +name + +" +attn_output_51~ +const% +attn_output_53_perm_0 + + +*+ +name# + +" +attn_output_53_perm_0*! +val + + + + + +b +const +var_863 + + +* +name + + +" +op_863*" +val + + + +  +��� + transpose +x + +attn_output_51! +perm + +attn_output_53_perm_0/ + transpose_15 +  + +� + +@*" +name + +" + transpose_15� +reshape +x + + transpose_15 +shape + +var_863' + input_189 +  + +� +�* +name + + " + input_189� +linear +x + + input_189D +weight: +8 +6model_model_encoder_layers_8_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_8_self_attn_out_proj_bias' + linear_51 +  + +� +�* +name + + " + linear_51~ +add +x + +hidden_states_49 +y + + linear_51' + input_193 +  + +� +�* +name + + " + input_193z +const +input_195_axes_0 + + +*& +name + +" +input_195_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_193 +axes + +input_195_axes_0E +gamma< +: +8model_model_encoder_layers_8_self_attn_layer_norm_weightB +beta: +8 +6model_model_encoder_layers_8_self_attn_layer_norm_bias +epsilon + + +var_22' + input_195 +  + +� +�* +name + + " + input_195� +linear +x + + input_1955 +weight+ +) +'model_model_encoder_layers_8_fc1_weight1 +bias) +' +%model_model_encoder_layers_8_fc1_bias' + linear_52 +  + +� +� * +name + + " + linear_52e +const +input_199_mode_0 +*& +name + +" +input_199_mode_0* +val + + " +EXACT� +gelu +x + + linear_52 +mode + +input_199_mode_0' + input_199 +  + +� +� * +name + + " + input_199� +linear +x + + input_1995 +weight+ +) +'model_model_encoder_layers_8_fc2_weight1 +bias) +' +%model_model_encoder_layers_8_fc2_bias' + linear_53 +  + +� +�* +name + + " + linear_53w +add +x + + input_195 +y + + linear_53' + input_205 +  + +� +�* +name + + " + input_205� +const' +hidden_states_55_axes_0 + + +*- +name% + +" +hidden_states_55_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_205# +axes + +hidden_states_55_axes_0A +gamma8 +6 +4model_model_encoder_layers_8_final_layer_norm_weight> +beta6 +4 +2model_model_encoder_layers_8_final_layer_norm_bias +epsilon + + +var_22. +hidden_states_55 +  + +� +�*& +name + +" +hidden_states_55� +linear +x + +hidden_states_55B +weight8 +6 +4model_model_encoder_layers_9_self_attn_q_proj_weight> +bias6 +4 +2model_model_encoder_layers_9_self_attn_q_proj_bias' + linear_54 +  + +� +�* +name + + " + linear_54Q +const +var_903 + * +name + + +" +op_903* +val +  + + + +>u +mul +x + + linear_54 +y + +var_903' + tensor_59 +  + +� +�* +name + + " + tensor_59� +linear +x + +hidden_states_55B +weight8 +6 +4model_model_encoder_layers_9_self_attn_k_proj_weight> +bias6 +4 +2model_model_encoder_layers_9_self_attn_k_proj_bias' + linear_55 +  + +� +�* +name + + " + linear_55j +const +var_908 + + +* +name + + +" +op_908** +val# + + + + + ���������@} +reshape +x + + linear_55 +shape + +var_908* +var_909 +  + +� + +@* +name + + +" +op_909o +const +var_910_perm_0 + + +*# +name + +" + op_910_perm_0*! +val + + + + + +� +linear +x + +hidden_states_55B +weight8 +6 +4model_model_encoder_layers_9_self_attn_v_proj_weight> +bias6 +4 +2model_model_encoder_layers_9_self_attn_v_proj_bias' + linear_56 +  + +� +�* +name + + " + linear_56j +const +var_915 + + +* +name + + +" +op_915** +val# + + + + + ���������@} +reshape +x + + linear_56 +shape + +var_915* +var_916 +  + +� + +@* +name + + +" +op_916o +const +var_917_perm_0 + + +*# +name + +" + op_917_perm_0*! +val + + + + + +b +const +var_924 + + +* +name + + +" +op_924*" +val + + + +  +�@} +reshape +x + + tensor_59 +shape + +var_924* +var_925 +  + +� + +@* +name + + +" +op_925o +const +var_926_perm_0 + + +*# +name + +" + op_926_perm_0*! +val + + + + + +i +const +var_928 + + +* +name + + +" +op_928*) +val" + + + + + ���������@� + transpose +x + +var_925 +perm + +var_926_perm_0/ + transpose_12 +  + + +� +@*" +name + +" + transpose_12� +reshape +x + + transpose_12 +shape + +var_928, +query_states_19 +  + +� +@*% +name + +" +query_states_19i +const +var_930 + + +* +name + + +" +op_930*) +val" + + + + + ���������@� + transpose +x + +var_909 +perm + +var_910_perm_0/ + transpose_14 +  + + +� +@*" +name + +" + transpose_14� +reshape +x + + transpose_14 +shape + +var_930* + key_states_39 +  + +� +@*# +name + +" + key_states_39i +const +var_932 + + +* +name + + +" +op_932*) +val" + + + + + ���������@� + transpose +x + +var_916 +perm + +var_917_perm_0/ + transpose_13 +  + + +� +@*" +name + +" + transpose_13� +reshape +x + + transpose_13 +shape + +var_932, +value_states_39 +  + +� +@*% +name + +" +value_states_39n +const +var_935_perm_0 + + +*# +name + +" + op_935_perm_0* +val + + + + +{ +const% +attn_weights_55_transpose_x_0 +*3 +name+ +# +!" +attn_weights_55_transpose_x_0* +val + + +{ +const% +attn_weights_55_transpose_y_0 +*3 +name+ +# +!" +attn_weights_55_transpose_y_0* +val + + +� + transpose +x + + key_states_39 +perm + +var_935_perm_0) + transpose_11 +  + +@ +�*" +name + +" + transpose_11� +matmul +x + +query_states_19 +y + + transpose_110 + transpose_x! + +attn_weights_55_transpose_x_00 + transpose_y! + +attn_weights_55_transpose_y_0- +attn_weights_55 +  + +� +�*% +name + +" +attn_weights_55c +const +var_937 + + +* +name + + +" +op_937*# +val + + + + + +��� +reshape +x + +attn_weights_55 +shape + +var_937+ +var_938 +  + + +� +�* +name + + +" +op_938� +add +x + +var_938 +y + +attention_mask_13 +attn_weights_57 +  + + +� +�*% +name + +" +attn_weights_57b +const +var_943 + + +* +name + + +" +op_943*" +val + + + +  +��� +reshape +x + +attn_weights_57 +shape + +var_943' + input_207 +  + +� +�* +name + + " + input_207{ +softmax +x + + input_207 +axis + + +var_30' + input_209 +  + +� +�* +name + + " + input_209y +const$ +attn_output_55_transpose_x_0 +*2 +name* +" + " +attn_output_55_transpose_x_0* +val + + +y +const$ +attn_output_55_transpose_y_0 +*2 +name* +" + " +attn_output_55_transpose_y_0* +val + + +� +matmul +x + + input_209 +y + +value_states_39/ + transpose_x + +attn_output_55_transpose_x_0/ + transpose_y + +attn_output_55_transpose_y_0+ +attn_output_55 +  + +� +@*$ +name + +" +attn_output_55b +const +var_948 + + +* +name + + +" +op_948*" +val + + + +  +�@� +reshape +x + +attn_output_55 +shape + +var_9481 +attn_output_57 +  + + +� +@*$ +name + +" +attn_output_57~ +const% +attn_output_59_perm_0 + + +*+ +name# + +" +attn_output_59_perm_0*! +val + + + + + +b +const +var_951 + + +* +name + + +" +op_951*" +val + + + +  +��� + transpose +x + +attn_output_57! +perm + +attn_output_59_perm_0/ + transpose_10 +  + +� + +@*" +name + +" + transpose_10� +reshape +x + + transpose_10 +shape + +var_951' + input_211 +  + +� +�* +name + + " + input_211� +linear +x + + input_211D +weight: +8 +6model_model_encoder_layers_9_self_attn_out_proj_weight@ +bias8 +6 +4model_model_encoder_layers_9_self_attn_out_proj_bias' + linear_57 +  + +� +�* +name + + " + linear_57~ +add +x + +hidden_states_55 +y + + linear_57' + input_215 +  + +� +�* +name + + " + input_215z +const +input_217_axes_0 + + +*& +name + +" +input_217_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_215 +axes + +input_217_axes_0E +gamma< +: +8model_model_encoder_layers_9_self_attn_layer_norm_weightB +beta: +8 +6model_model_encoder_layers_9_self_attn_layer_norm_bias +epsilon + + +var_22' + input_217 +  + +� +�* +name + + " + input_217� +linear +x + + input_2175 +weight+ +) +'model_model_encoder_layers_9_fc1_weight1 +bias) +' +%model_model_encoder_layers_9_fc1_bias' + linear_58 +  + +� +� * +name + + " + linear_58e +const +input_221_mode_0 +*& +name + +" +input_221_mode_0* +val + + " +EXACT� +gelu +x + + linear_58 +mode + +input_221_mode_0' + input_221 +  + +� +� * +name + + " + input_221� +linear +x + + input_2215 +weight+ +) +'model_model_encoder_layers_9_fc2_weight1 +bias) +' +%model_model_encoder_layers_9_fc2_bias' + linear_59 +  + +� +�* +name + + " + linear_59w +add +x + + input_217 +y + + linear_59' + input_227 +  + +� +�* +name + + " + input_227� +const' +hidden_states_61_axes_0 + + +*- +name% + +" +hidden_states_61_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_227# +axes + +hidden_states_61_axes_0A +gamma8 +6 +4model_model_encoder_layers_9_final_layer_norm_weight> +beta6 +4 +2model_model_encoder_layers_9_final_layer_norm_bias +epsilon + + +var_22. +hidden_states_61 +  + +� +�*& +name + +" +hidden_states_61� +linear +x + +hidden_states_61C +weight9 +7 +5model_model_encoder_layers_10_self_attn_q_proj_weight? +bias7 +5 +3model_model_encoder_layers_10_self_attn_q_proj_bias' + linear_60 +  + +� +�* +name + + " + linear_60Q +const +var_991 + * +name + + +" +op_991* +val +  + + + +>u +mul +x + + linear_60 +y + +var_991' + tensor_65 +  + +� +�* +name + + " + tensor_65� +linear +x + +hidden_states_61C +weight9 +7 +5model_model_encoder_layers_10_self_attn_k_proj_weight? +bias7 +5 +3model_model_encoder_layers_10_self_attn_k_proj_bias' + linear_61 +  + +� +�* +name + + " + linear_61j +const +var_996 + + +* +name + + +" +op_996** +val# + + + + + ���������@} +reshape +x + + linear_61 +shape + +var_996* +var_997 +  + +� + +@* +name + + +" +op_997o +const +var_998_perm_0 + + +*# +name + +" + op_998_perm_0*! +val + + + + + +� +linear +x + +hidden_states_61C +weight9 +7 +5model_model_encoder_layers_10_self_attn_v_proj_weight? +bias7 +5 +3model_model_encoder_layers_10_self_attn_v_proj_bias' + linear_62 +  + +� +�* +name + + " + linear_62l +const +var_1003 + + +* +name + + " +op_1003** +val# + + + + + ���������@� +reshape +x + + linear_62 +shape + + +var_1003+ +var_1004 +  + +� + +@* +name + + " +op_1004q +const +var_1005_perm_0 + + +*$ +name + +" +op_1005_perm_0*! +val + + + + + +d +const +var_1012 + + +* +name + + " +op_1012*" +val + + + +  +�@� +reshape +x + + tensor_65 +shape + + +var_1012+ +var_1013 +  + +� + +@* +name + + " +op_1013q +const +var_1014_perm_0 + + +*$ +name + +" +op_1014_perm_0*! +val + + + + + +k +const +var_1016 + + +* +name + + " +op_1016*) +val" + + + + + ���������@� + transpose +x + + +var_1013 +perm + +var_1014_perm_0. + transpose_7 +  + + +� +@*! +name + +" + transpose_7� +reshape +x + + transpose_7 +shape + + +var_1016, +query_states_21 +  + +� +@*% +name + +" +query_states_21k +const +var_1018 + + +* +name + + " +op_1018*) +val" + + + + + ���������@� + transpose +x + +var_997 +perm + +var_998_perm_0. + transpose_9 +  + + +� +@*! +name + +" + transpose_9� +reshape +x + + transpose_9 +shape + + +var_1018* + key_states_43 +  + +� +@*# +name + +" + key_states_43k +const +var_1020 + + +* +name + + " +op_1020*) +val" + + + + + ���������@� + transpose +x + + +var_1004 +perm + +var_1005_perm_0. + transpose_8 +  + + +� +@*! +name + +" + transpose_8� +reshape +x + + transpose_8 +shape + + +var_1020, +value_states_43 +  + +� +@*% +name + +" +value_states_43p +const +var_1023_perm_0 + + +*$ +name + +" +op_1023_perm_0* +val + + + + +{ +const% +attn_weights_61_transpose_x_0 +*3 +name+ +# +!" +attn_weights_61_transpose_x_0* +val + + +{ +const% +attn_weights_61_transpose_y_0 +*3 +name+ +# +!" +attn_weights_61_transpose_y_0* +val + + +� + transpose +x + + key_states_43 +perm + +var_1023_perm_0( + transpose_6 +  + +@ +�*! +name + +" + transpose_6� +matmul +x + +query_states_21 +y + + transpose_60 + transpose_x! + +attn_weights_61_transpose_x_00 + transpose_y! + +attn_weights_61_transpose_y_0- +attn_weights_61 +  + +� +�*% +name + +" +attn_weights_61e +const +var_1025 + + +* +name + + " +op_1025*# +val + + + + + +��� +reshape +x + +attn_weights_61 +shape + + +var_1025, +var_1026 +  + + +� +�* +name + + " +op_1026� +add +x + + +var_1026 +y + +attention_mask_13 +attn_weights_63 +  + + +� +�*% +name + +" +attn_weights_63d +const +var_1031 + + +* +name + + " +op_1031*" +val + + + +  +��� +reshape +x + +attn_weights_63 +shape + + +var_1031' + input_229 +  + +� +�* +name + + " + input_229{ +softmax +x + + input_229 +axis + + +var_30' + input_231 +  + +� +�* +name + + " + input_231y +const$ +attn_output_61_transpose_x_0 +*2 +name* +" + " +attn_output_61_transpose_x_0* +val + + +y +const$ +attn_output_61_transpose_y_0 +*2 +name* +" + " +attn_output_61_transpose_y_0* +val + + +� +matmul +x + + input_231 +y + +value_states_43/ + transpose_x + +attn_output_61_transpose_x_0/ + transpose_y + +attn_output_61_transpose_y_0+ +attn_output_61 +  + +� +@*$ +name + +" +attn_output_61d +const +var_1036 + + +* +name + + " +op_1036*" +val + + + +  +�@� +reshape +x + +attn_output_61 +shape + + +var_10361 +attn_output_63 +  + + +� +@*$ +name + +" +attn_output_63~ +const% +attn_output_65_perm_0 + + +*+ +name# + +" +attn_output_65_perm_0*! +val + + + + + +d +const +var_1039 + + +* +name + + " +op_1039*" +val + + + +  +��� + transpose +x + +attn_output_63! +perm + +attn_output_65_perm_0. + transpose_5 +  + +� + +@*! +name + +" + transpose_5� +reshape +x + + transpose_5 +shape + + +var_1039' + input_233 +  + +� +�* +name + + " + input_233� +linear +x + + input_233E +weight; +9 +7model_model_encoder_layers_10_self_attn_out_proj_weightA +bias9 +7 +5model_model_encoder_layers_10_self_attn_out_proj_bias' + linear_63 +  + +� +�* +name + + " + linear_63~ +add +x + +hidden_states_61 +y + + linear_63' + input_237 +  + +� +�* +name + + " + input_237z +const +input_239_axes_0 + + +*& +name + +" +input_239_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_237 +axes + +input_239_axes_0F +gamma= +; +9model_model_encoder_layers_10_self_attn_layer_norm_weightC +beta; +9 +7model_model_encoder_layers_10_self_attn_layer_norm_bias +epsilon + + +var_22' + input_239 +  + +� +�* +name + + " + input_239� +linear +x + + input_2396 +weight, +* +(model_model_encoder_layers_10_fc1_weight2 +bias* +( +&model_model_encoder_layers_10_fc1_bias' + linear_64 +  + +� +� * +name + + " + linear_64e +const +input_243_mode_0 +*& +name + +" +input_243_mode_0* +val + + " +EXACT� +gelu +x + + linear_64 +mode + +input_243_mode_0' + input_243 +  + +� +� * +name + + " + input_243� +linear +x + + input_2436 +weight, +* +(model_model_encoder_layers_10_fc2_weight2 +bias* +( +&model_model_encoder_layers_10_fc2_bias' + linear_65 +  + +� +�* +name + + " + linear_65w +add +x + + input_239 +y + + linear_65' + input_249 +  + +� +�* +name + + " + input_249� +const' +hidden_states_67_axes_0 + + +*- +name% + +" +hidden_states_67_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_249# +axes + +hidden_states_67_axes_0B +gamma9 +7 +5model_model_encoder_layers_10_final_layer_norm_weight? +beta7 +5 +3model_model_encoder_layers_10_final_layer_norm_bias +epsilon + + +var_22. +hidden_states_67 +  + +� +�*& +name + +" +hidden_states_67� +linear +x + +hidden_states_67C +weight9 +7 +5model_model_encoder_layers_11_self_attn_q_proj_weight? +bias7 +5 +3model_model_encoder_layers_11_self_attn_q_proj_bias' + linear_66 +  + +� +�* +name + + " + linear_66S +const +var_1079 + * +name + + " +op_1079* +val +  + + + +>� +mul +x + + linear_66 +y + + +var_1079/ +tensor_workaround +  + +� +�*' +name + +" +tensor_workaround� +linear +x + +hidden_states_67C +weight9 +7 +5model_model_encoder_layers_11_self_attn_k_proj_weight? +bias7 +5 +3model_model_encoder_layers_11_self_attn_k_proj_bias' + linear_67 +  + +� +�* +name + + " + linear_67l +const +var_1084 + + +* +name + + " +op_1084** +val# + + + + + ���������@� +reshape +x + + linear_67 +shape + + +var_1084+ +var_1085 +  + +� + +@* +name + + " +op_1085q +const +var_1086_perm_0 + + +*$ +name + +" +op_1086_perm_0*! +val + + + + + +� +linear +x + +hidden_states_67C +weight9 +7 +5model_model_encoder_layers_11_self_attn_v_proj_weight? +bias7 +5 +3model_model_encoder_layers_11_self_attn_v_proj_bias' + linear_68 +  + +� +�* +name + + " + linear_68l +const +var_1091 + + +* +name + + " +op_1091** +val# + + + + + ���������@� +reshape +x + + linear_68 +shape + + +var_1091+ +var_1092 +  + +� + +@* +name + + " +op_1092q +const +var_1093_perm_0 + + +*$ +name + +" +op_1093_perm_0*! +val + + + + + +d +const +var_1100 + + +* +name + + " +op_1100*" +val + + + +  +�@� +reshape +x + +tensor_workaround +shape + + +var_1100+ +var_1101 +  + +� + +@* +name + + " +op_1101q +const +var_1102_perm_0 + + +*$ +name + +" +op_1102_perm_0*! +val + + + + + +k +const +var_1104 + + +* +name + + " +op_1104*) +val" + + + + + ���������@� + transpose +x + + +var_1101 +perm + +var_1102_perm_0. + transpose_2 +  + + +� +@*! +name + +" + transpose_2� +reshape +x + + transpose_2 +shape + + +var_1104) + query_states +  + +� +@*" +name + +" + query_statesk +const +var_1106 + + +* +name + + " +op_1106*) +val" + + + + + ���������@� + transpose +x + + +var_1085 +perm + +var_1086_perm_0. + transpose_4 +  + + +� +@*! +name + +" + transpose_4� +reshape +x + + transpose_4 +shape + + +var_1106' + +key_states +  + +� +@* +name + +" + +key_statesk +const +var_1108 + + +* +name + + " +op_1108*) +val" + + + + + ���������@� + transpose +x + + +var_1092 +perm + +var_1093_perm_0. + transpose_3 +  + + +� +@*! +name + +" + transpose_3� +reshape +x + + transpose_3 +shape + + +var_1108) + value_states +  + +� +@*" +name + +" + value_statesp +const +var_1111_perm_0 + + +*$ +name + +" +op_1111_perm_0* +val + + + + +{ +const% +attn_weights_67_transpose_x_0 +*3 +name+ +# +!" +attn_weights_67_transpose_x_0* +val + + +{ +const% +attn_weights_67_transpose_y_0 +*3 +name+ +# +!" +attn_weights_67_transpose_y_0* +val + + +� + transpose +x + + +key_states +perm + +var_1111_perm_0( + transpose_1 +  + +@ +�*! +name + +" + transpose_1� +matmul +x + + query_states +y + + transpose_10 + transpose_x! + +attn_weights_67_transpose_x_00 + transpose_y! + +attn_weights_67_transpose_y_0- +attn_weights_67 +  + +� +�*% +name + +" +attn_weights_67e +const +var_1113 + + +* +name + + " +op_1113*# +val + + + + + +��� +reshape +x + +attn_weights_67 +shape + + +var_1113, +var_1114 +  + + +� +�* +name + + " +op_1114� +add +x + + +var_1114 +y + +attention_mask_13 +attn_weights_69 +  + + +� +�*% +name + +" +attn_weights_69d +const +var_1119 + + +* +name + + " +op_1119*" +val + + + +  +��� +reshape +x + +attn_weights_69 +shape + + +var_1119' + input_251 +  + +� +�* +name + + " + input_251{ +softmax +x + + input_251 +axis + + +var_30' + input_253 +  + +� +�* +name + + " + input_253y +const$ +attn_output_67_transpose_x_0 +*2 +name* +" + " +attn_output_67_transpose_x_0* +val + + +y +const$ +attn_output_67_transpose_y_0 +*2 +name* +" + " +attn_output_67_transpose_y_0* +val + + +� +matmul +x + + input_253 +y + + value_states/ + transpose_x + +attn_output_67_transpose_x_0/ + transpose_y + +attn_output_67_transpose_y_0+ +attn_output_67 +  + +� +@*$ +name + +" +attn_output_67d +const +var_1124 + + +* +name + + " +op_1124*" +val + + + +  +�@� +reshape +x + +attn_output_67 +shape + + +var_11241 +attn_output_69 +  + + +� +@*$ +name + +" +attn_output_69x +const" +attn_output_perm_0 + + +*( +name  + +" +attn_output_perm_0*! +val + + + + + +d +const +var_1127 + + +* +name + + " +op_1127*" +val + + + +  +��� + transpose +x + +attn_output_69 +perm + +attn_output_perm_0. + transpose_0 +  + +� + +@*! +name + +" + transpose_0� +reshape +x + + transpose_0 +shape + + +var_1127' + input_255 +  + +� +�* +name + + " + input_255� +linear +x + + input_255E +weight; +9 +7model_model_encoder_layers_11_self_attn_out_proj_weightA +bias9 +7 +5model_model_encoder_layers_11_self_attn_out_proj_bias' + linear_69 +  + +� +�* +name + + " + linear_69~ +add +x + +hidden_states_67 +y + + linear_69' + input_259 +  + +� +�* +name + + " + input_259z +const +input_261_axes_0 + + +*& +name + +" +input_261_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_259 +axes + +input_261_axes_0F +gamma= +; +9model_model_encoder_layers_11_self_attn_layer_norm_weightC +beta; +9 +7model_model_encoder_layers_11_self_attn_layer_norm_bias +epsilon + + +var_22' + input_261 +  + +� +�* +name + + " + input_261� +linear +x + + input_2616 +weight, +* +(model_model_encoder_layers_11_fc1_weight2 +bias* +( +&model_model_encoder_layers_11_fc1_bias' + linear_70 +  + +� +� * +name + + " + linear_70e +const +input_265_mode_0 +*& +name + +" +input_265_mode_0* +val + + " +EXACT� +gelu +x + + linear_70 +mode + +input_265_mode_0' + input_265 +  + +� +� * +name + + " + input_265� +linear +x + + input_2656 +weight, +* +(model_model_encoder_layers_11_fc2_weight2 +bias* +( +&model_model_encoder_layers_11_fc2_bias' + linear_71 +  + +� +�* +name + + " + linear_71o +add +x + + input_261 +y + + linear_71# +input +  + +� +�* +name + + " +inputw +const +var_1151_axes_0 + + +*$ +name + +" +op_1151_axes_0*' +val  + + + + + +���������� + +layer_norm +x + +input +axes + +var_1151_axes_0B +gamma9 +7 +5model_model_encoder_layers_11_final_layer_norm_weight? +beta7 +5 +3model_model_encoder_layers_11_final_layer_norm_bias +epsilon + + +var_22/ +last_hidden_state +  + +� +�* +name + + " +op_1151"� + buildInfo�" + + +|"z +6 +! + +" +coremltools-version + +" +7.0 +@ +) +! +" +coremltools-component-torch + + " +2.0.1 \ No newline at end of file