diff --git "a/coreml/text2text-generation/decoder_float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel" "b/coreml/text2text-generation/decoder_float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel" new file mode 100644--- /dev/null +++ "b/coreml/text2text-generation/decoder_float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel" @@ -0,0 +1,12811 @@ +� +\ +decoder_input_ids:Indices of decoder input sequence tokens in the vocabulary * +��� + +decoder_attention_maskXMask to avoid performing attention on padding token indices (1 = not masked, 0 = masked) * +��� +t +encoder_last_hidden_stateHSequence of hidden states at the output of the last layer of the encoder * +���� + +encoder_attention_maskXMask to avoid performing attention on padding token indices (1 = not masked, 0 = masked) * +���R8 +logits&Classification scores (before softmax)*���� +1Helsinki-NLP/opus-mt-en-ru (text2text-generation)�3 +#com.github.apple.coremltools.source torch==2.0.1�- +$com.github.apple.coremltools.version7.0b1�; +co.huggingface.exporters.nameHelsinki-NLP/opus-mt-en-ru�5 +co.huggingface.exporters.tasktext2text-generation�6 +%co.huggingface.exporters.architecture MarianMTModel�- +"co.huggingface.exporters.frameworkpytorch�- +"co.huggingface.exporters.precisionfloat32�# +transformers_version 4.22.0.dev0����� +main�� +( +decoder_input_ids + + +� +- +decoder_attention_mask + + +� +7 +encoder_last_hidden_state +  + +� +� +- +encoder_attention_mask + + +�CoreML5�� +CoreML5��logits� +const2 +model_model_shared_weight +  +�� +�*/ +name' + +" +model_model_shared_weight*B +val; +  +�� +�*" +@model_path/weights/weight.bin@� +constC +2model_model_decoder_layers_0_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_0_self_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin��=� +constL +4model_model_decoder_layers_0_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_0_self_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin��=� +constC +2model_model_decoder_layers_0_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_0_self_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���=� +constL +4model_model_decoder_layers_0_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_0_self_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���=� +constC +2model_model_decoder_layers_0_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_0_self_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���>� +constL +4model_model_decoder_layers_0_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_0_self_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���>� +constE +4model_model_decoder_layers_0_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_0_self_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���>� +constN +6model_model_decoder_layers_0_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_0_self_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���>� +constG +6model_model_decoder_layers_0_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_0_self_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���?� +constI +8model_model_decoder_layers_0_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_0_self_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���?� +constF +5model_model_decoder_layers_0_encoder_attn_q_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_0_encoder_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin�Ƈ?� +constO +7model_model_decoder_layers_0_encoder_attn_q_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_0_encoder_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin�և?� +constF +5model_model_decoder_layers_0_encoder_attn_k_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_0_encoder_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���?� +constO +7model_model_decoder_layers_0_encoder_attn_k_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_0_encoder_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���?� +constF +5model_model_decoder_layers_0_encoder_attn_v_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_0_encoder_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin��@� +constO +7model_model_decoder_layers_0_encoder_attn_v_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_0_encoder_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���@� +constH +7model_model_decoder_layers_0_encoder_attn_out_proj_bias +   +�*M +nameE += +;"9 +7model_model_decoder_layers_0_encoder_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���@� +constQ +9model_model_decoder_layers_0_encoder_attn_out_proj_weight +  +� +�*O +nameG +? +="; +9model_model_decoder_layers_0_encoder_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���@� +constJ +9model_model_decoder_layers_0_encoder_attn_layer_norm_bias +   +�*O +nameG +? +="; +9model_model_decoder_layers_0_encoder_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���A� +constL +;model_model_decoder_layers_0_encoder_attn_layer_norm_weight +   +�*Q +nameI +A +?"= +;model_model_decoder_layers_0_encoder_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���A� +const6 +%model_model_decoder_layers_0_fc1_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_0_fc1_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���A� +const? +'model_model_decoder_layers_0_fc1_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_0_fc1_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin��A� +const6 +%model_model_decoder_layers_0_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_0_fc2_bias*= +val6 +   +�*% +@model_path/weights/weight.bin��C� +const? +'model_model_decoder_layers_0_fc2_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_0_fc2_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���C� +constC +2model_model_decoder_layers_0_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_0_final_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���E� +constE +4model_model_decoder_layers_0_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_0_final_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���E� +constC +2model_model_decoder_layers_1_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_1_self_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���E� +constL +4model_model_decoder_layers_1_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_1_self_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���E� +constC +2model_model_decoder_layers_1_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_1_self_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���E� +constL +4model_model_decoder_layers_1_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_1_self_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���E� +constC +2model_model_decoder_layers_1_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_1_self_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���F� +constL +4model_model_decoder_layers_1_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_1_self_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin�ЉF� +constE +4model_model_decoder_layers_1_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_1_self_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���F� +constN +6model_model_decoder_layers_1_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_1_self_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���F� +constG +6model_model_decoder_layers_1_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_1_self_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin��G� +constI +8model_model_decoder_layers_1_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_1_self_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin��G� +constF +5model_model_decoder_layers_1_encoder_attn_q_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_1_encoder_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���G� +constO +7model_model_decoder_layers_1_encoder_attn_q_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_1_encoder_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���G� +constF +5model_model_decoder_layers_1_encoder_attn_k_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_1_encoder_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���G� +constO +7model_model_decoder_layers_1_encoder_attn_k_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_1_encoder_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���G� +constF +5model_model_decoder_layers_1_encoder_attn_v_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_1_encoder_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���H� +constO +7model_model_decoder_layers_1_encoder_attn_v_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_1_encoder_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���H� +constH +7model_model_decoder_layers_1_encoder_attn_out_proj_bias +   +�*M +nameE += +;"9 +7model_model_decoder_layers_1_encoder_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���H� +constQ +9model_model_decoder_layers_1_encoder_attn_out_proj_weight +  +� +�*O +nameG +? +="; +9model_model_decoder_layers_1_encoder_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���H� +constJ +9model_model_decoder_layers_1_encoder_attn_layer_norm_bias +   +�*O +nameG +? +="; +9model_model_decoder_layers_1_encoder_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin�NJI� +constL +;model_model_decoder_layers_1_encoder_attn_layer_norm_weight +   +�*Q +nameI +A +?"= +;model_model_decoder_layers_1_encoder_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin�׊I� +const6 +%model_model_decoder_layers_1_fc1_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_1_fc1_bias*= +val6 +   +�*% +@model_path/weights/weight.bin��I� +const? +'model_model_decoder_layers_1_fc1_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_1_fc1_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���I� +const6 +%model_model_decoder_layers_1_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_1_fc2_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���K� +const? +'model_model_decoder_layers_1_fc2_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_1_fc2_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���K� +constC +2model_model_decoder_layers_1_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_1_final_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���M� +constE +4model_model_decoder_layers_1_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_1_final_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin�ʋM� +constC +2model_model_decoder_layers_2_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_2_self_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin�ۋM� +constL +4model_model_decoder_layers_2_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_2_self_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin��M� +constC +2model_model_decoder_layers_2_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_2_self_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���M� +constL +4model_model_decoder_layers_2_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_2_self_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���M� +constC +2model_model_decoder_layers_2_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_2_self_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���N� +constL +4model_model_decoder_layers_2_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_2_self_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���N� +constE +4model_model_decoder_layers_2_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_2_self_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���N� +constN +6model_model_decoder_layers_2_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_2_self_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���N� +constG +6model_model_decoder_layers_2_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_2_self_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���O� +constI +8model_model_decoder_layers_2_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_2_self_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���O� +constF +5model_model_decoder_layers_2_encoder_attn_q_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_2_encoder_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���O� +constO +7model_model_decoder_layers_2_encoder_attn_q_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_2_encoder_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin�ЌO� +constF +5model_model_decoder_layers_2_encoder_attn_k_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_2_encoder_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���O� +constO +7model_model_decoder_layers_2_encoder_attn_k_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_2_encoder_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���O� +constF +5model_model_decoder_layers_2_encoder_attn_v_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_2_encoder_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin��P� +constO +7model_model_decoder_layers_2_encoder_attn_v_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_2_encoder_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin��P� +constH +7model_model_decoder_layers_2_encoder_attn_out_proj_bias +   +�*M +nameE += +;"9 +7model_model_decoder_layers_2_encoder_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���P� +constQ +9model_model_decoder_layers_2_encoder_attn_out_proj_weight +  +� +�*O +nameG +? +="; +9model_model_decoder_layers_2_encoder_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���P� +constJ +9model_model_decoder_layers_2_encoder_attn_layer_norm_bias +   +�*O +nameG +? +="; +9model_model_decoder_layers_2_encoder_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���Q� +constL +;model_model_decoder_layers_2_encoder_attn_layer_norm_weight +   +�*Q +nameI +A +?"= +;model_model_decoder_layers_2_encoder_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���Q� +const6 +%model_model_decoder_layers_2_fc1_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_2_fc1_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���Q� +const? +'model_model_decoder_layers_2_fc1_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_2_fc1_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin��Q� +const6 +%model_model_decoder_layers_2_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_2_fc2_bias*= +val6 +   +�*% +@model_path/weights/weight.bin��S� +const? +'model_model_decoder_layers_2_fc2_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_2_fc2_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���S� +constC +2model_model_decoder_layers_2_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_2_final_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���U� +constE +4model_model_decoder_layers_2_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_2_final_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���U� +constC +2model_model_decoder_layers_3_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_3_self_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���U� +constL +4model_model_decoder_layers_3_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_3_self_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���U� +constC +2model_model_decoder_layers_3_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_3_self_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���U� +constL +4model_model_decoder_layers_3_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_3_self_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���U� +constC +2model_model_decoder_layers_3_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_3_self_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���V� +constL +4model_model_decoder_layers_3_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_3_self_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin�ʎV� +constE +4model_model_decoder_layers_3_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_3_self_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���V� +constN +6model_model_decoder_layers_3_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_3_self_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���V� +constG +6model_model_decoder_layers_3_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_3_self_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin�܎W� +constI +8model_model_decoder_layers_3_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_3_self_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin��W� +constF +5model_model_decoder_layers_3_encoder_attn_q_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_3_encoder_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���W� +constO +7model_model_decoder_layers_3_encoder_attn_q_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_3_encoder_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���W� +constF +5model_model_decoder_layers_3_encoder_attn_k_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_3_encoder_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���W� +constO +7model_model_decoder_layers_3_encoder_attn_k_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_3_encoder_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���W� +constF +5model_model_decoder_layers_3_encoder_attn_v_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_3_encoder_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���X� +constO +7model_model_decoder_layers_3_encoder_attn_v_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_3_encoder_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���X� +constH +7model_model_decoder_layers_3_encoder_attn_out_proj_bias +   +�*M +nameE += +;"9 +7model_model_decoder_layers_3_encoder_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���X� +constQ +9model_model_decoder_layers_3_encoder_attn_out_proj_weight +  +� +�*O +nameG +? +="; +9model_model_decoder_layers_3_encoder_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���X� +constJ +9model_model_decoder_layers_3_encoder_attn_layer_norm_bias +   +�*O +nameG +? +="; +9model_model_decoder_layers_3_encoder_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���Y� +constL +;model_model_decoder_layers_3_encoder_attn_layer_norm_weight +   +�*Q +nameI +A +?"= +;model_model_decoder_layers_3_encoder_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin�яY� +const6 +%model_model_decoder_layers_3_fc1_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_3_fc1_bias*= +val6 +   +�*% +@model_path/weights/weight.bin��Y� +const? +'model_model_decoder_layers_3_fc1_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_3_fc1_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���Y� +const6 +%model_model_decoder_layers_3_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_3_fc2_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���[� +const? +'model_model_decoder_layers_3_fc2_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_3_fc2_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���[� +constC +2model_model_decoder_layers_3_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_3_final_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���]� +constE +4model_model_decoder_layers_3_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_3_final_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin�Đ]� +constC +2model_model_decoder_layers_4_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_4_self_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin�Ր]� +constL +4model_model_decoder_layers_4_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_4_self_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin��]� +constC +2model_model_decoder_layers_4_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_4_self_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���]� +constL +4model_model_decoder_layers_4_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_4_self_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���]� +constC +2model_model_decoder_layers_4_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_4_self_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���^� +constL +4model_model_decoder_layers_4_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_4_self_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���^� +constE +4model_model_decoder_layers_4_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_4_self_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���^� +constN +6model_model_decoder_layers_4_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_4_self_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���^� +constG +6model_model_decoder_layers_4_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_4_self_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���_� +constI +8model_model_decoder_layers_4_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_4_self_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���_� +constF +5model_model_decoder_layers_4_encoder_attn_q_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_4_encoder_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���_� +constO +7model_model_decoder_layers_4_encoder_attn_q_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_4_encoder_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin�ʑ_� +constF +5model_model_decoder_layers_4_encoder_attn_k_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_4_encoder_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���_� +constO +7model_model_decoder_layers_4_encoder_attn_k_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_4_encoder_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���_� +constF +5model_model_decoder_layers_4_encoder_attn_v_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_4_encoder_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin�ܑ`� +constO +7model_model_decoder_layers_4_encoder_attn_v_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_4_encoder_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin��`� +constH +7model_model_decoder_layers_4_encoder_attn_out_proj_bias +   +�*M +nameE += +;"9 +7model_model_decoder_layers_4_encoder_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���`� +constQ +9model_model_decoder_layers_4_encoder_attn_out_proj_weight +  +� +�*O +nameG +? +="; +9model_model_decoder_layers_4_encoder_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���`� +constJ +9model_model_decoder_layers_4_encoder_attn_layer_norm_bias +   +�*O +nameG +? +="; +9model_model_decoder_layers_4_encoder_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���a� +constL +;model_model_decoder_layers_4_encoder_attn_layer_norm_weight +   +�*Q +nameI +A +?"= +;model_model_decoder_layers_4_encoder_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���a� +const6 +%model_model_decoder_layers_4_fc1_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_4_fc1_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���a� +const? +'model_model_decoder_layers_4_fc1_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_4_fc1_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin�ߒa� +const6 +%model_model_decoder_layers_4_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_4_fc2_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���c� +const? +'model_model_decoder_layers_4_fc2_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_4_fc2_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin��c� +constC +2model_model_decoder_layers_4_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_4_final_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin��e� +constE +4model_model_decoder_layers_4_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_4_final_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���e� +constC +2model_model_decoder_layers_5_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_5_self_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���e� +constL +4model_model_decoder_layers_5_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_5_self_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���e� +constC +2model_model_decoder_layers_5_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_5_self_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���e� +constL +4model_model_decoder_layers_5_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_5_self_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���e� +constC +2model_model_decoder_layers_5_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_5_self_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���f� +constL +4model_model_decoder_layers_5_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_5_self_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin�ēf� +constE +4model_model_decoder_layers_5_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_5_self_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���f� +constN +6model_model_decoder_layers_5_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_5_self_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���f� +constG +6model_model_decoder_layers_5_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_5_self_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin�֓g� +constI +8model_model_decoder_layers_5_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_5_self_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin��g� +constF +5model_model_decoder_layers_5_encoder_attn_q_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_5_encoder_attn_q_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���g� +constO +7model_model_decoder_layers_5_encoder_attn_q_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_5_encoder_attn_q_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���g� +constF +5model_model_decoder_layers_5_encoder_attn_k_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_5_encoder_attn_k_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���g� +constO +7model_model_decoder_layers_5_encoder_attn_k_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_5_encoder_attn_k_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���g� +constF +5model_model_decoder_layers_5_encoder_attn_v_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_5_encoder_attn_v_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���h� +constO +7model_model_decoder_layers_5_encoder_attn_v_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_5_encoder_attn_v_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���h� +constH +7model_model_decoder_layers_5_encoder_attn_out_proj_bias +   +�*M +nameE += +;"9 +7model_model_decoder_layers_5_encoder_attn_out_proj_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���h� +constQ +9model_model_decoder_layers_5_encoder_attn_out_proj_weight +  +� +�*O +nameG +? +="; +9model_model_decoder_layers_5_encoder_attn_out_proj_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���h� +constJ +9model_model_decoder_layers_5_encoder_attn_layer_norm_bias +   +�*O +nameG +? +="; +9model_model_decoder_layers_5_encoder_attn_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���i� +constL +;model_model_decoder_layers_5_encoder_attn_layer_norm_weight +   +�*Q +nameI +A +?"= +;model_model_decoder_layers_5_encoder_attn_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin�˔i� +const6 +%model_model_decoder_layers_5_fc1_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_5_fc1_bias*= +val6 +   +�*% +@model_path/weights/weight.bin�ܔi� +const? +'model_model_decoder_layers_5_fc1_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_5_fc1_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���i� +const6 +%model_model_decoder_layers_5_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_5_fc2_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���k� +const? +'model_model_decoder_layers_5_fc2_weight +  +� +�*= +name5 +- ++") +'model_model_decoder_layers_5_fc2_weight*D +val= +  +� +�*% +@model_path/weights/weight.bin���k� +constC +2model_model_decoder_layers_5_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_5_final_layer_norm_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���m� +constE +4model_model_decoder_layers_5_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_5_final_layer_norm_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���mO +const +var_11 + * +name + + " +op_11* +val +  + + + +��'7O +const +var_15 + * +name + + " +op_15* +val +  + + + +�?O +const +var_22 + * +name + + " +op_22* +val +  + + + +���U +const +var_25 +* +name + + " +op_25* +val + + + +���������Z +const + var_52_axis_0 +*" +name + +" + op_52_axis_0* +val + + +� +gather" +x + +model_model_shared_weight +indices + +decoder_input_ids +axis + + var_52_axis_0$ +var_52 +  + +� +�* +name + + " +op_52O +const +var_53 + * +name + + " +op_53* +val +  + + + +��Ay +mul +x + + +var_52 +y + + +var_53+ + inputs_embeds +  + +� +�*# +name + +" + inputs_embeds� +const* +var_68 +  + + +� +�* +name + + " +op_68*P +valI +  + + +� +�*% +@model_path/weights/weight.bin�ϕmj +const + var_79_axes_0 + + +*" +name + +" + op_79_axes_0* +val + + + + +� + expand_dims +x + +decoder_attention_mask +axes + + var_79_axes_0# +var_79 + + + +�* +name + + " +op_79j +const + var_80_axes_0 + + +*" +name + +" + op_80_axes_0* +val + + + + +� + expand_dims +x + + +var_79 +axes + + var_80_axes_0) +var_80 + + + + +�* +name + + " +op_80n +const + var_83_reps_0 + + +*" +name + +" + op_83_reps_0*" +val + + + +  +�{ +tile +x + + +var_80 +reps + + var_83_reps_0* +var_83 + + + +� +�* +name + + " +op_83_ +const +var_84_dtype_0 +*# +name + +" + op_84_dtype_0* +val + + +" +fp32� +cast +x + + +var_83 +dtype + +var_84_dtype_0, +cast_128 +  + + +� +�* +name + + " + +cast_128� +sub +x + + +var_15 +y + + +cast_1283 +inverted_mask_1 +  + + +� +�*% +name + +" +inverted_mask_1_ +const +var_86_dtype_0 +*# +name + +" + op_86_dtype_0* +val + + +" +bool� +cast +x + +inverted_mask_1 +dtype + +var_86_dtype_0, +cast_127 + + + +� +�* +name + + " + +cast_127� +select +cond + + +cast_127 +a + + +var_22 +b + +inverted_mask_1* +var_87 +  + + +� +�* +name + + " +op_87� +add +x + + +var_87 +y + + +var_684 +attention_mask_1 +  + + +� +�*& +name + +" +attention_mask_1j +const + var_93_axes_0 + + +*" +name + +" + op_93_axes_0* +val + + + + +� + expand_dims +x + +encoder_attention_mask +axes + + var_93_axes_0# +var_93 + + + +�* +name + + " +op_93j +const + var_94_axes_0 + + +*" +name + +" + op_94_axes_0* +val + + + + +� + expand_dims +x + + +var_93 +axes + + var_94_axes_0) +var_94 + + + + +�* +name + + " +op_94n +const + var_97_reps_0 + + +*" +name + +" + op_97_reps_0*" +val + + + +  +�{ +tile +x + + +var_94 +reps + + var_97_reps_0* +var_97 + + + +� +�* +name + + " +op_97_ +const +var_98_dtype_0 +*# +name + +" + op_98_dtype_0* +val + + +" +fp32� +cast +x + + +var_97 +dtype + +var_98_dtype_0, +cast_126 +  + + +� +�* +name + + " + +cast_126� +sub +x + + +var_15 +y + + +cast_1261 + inverted_mask +  + + +� +�*# +name + +" + inverted_maska +const +var_100_dtype_0 +*$ +name + +" +op_100_dtype_0* +val + + +" +bool� +cast +x + + inverted_mask +dtype + +var_100_dtype_0, +cast_125 + + + +� +�* +name + + " + +cast_125� +select +cond + + +cast_125 +a + + +var_22 +b + + inverted_mask2 +attention_mask +  + + +� +�*$ +name + +" +attention_mask� +const! + positions +  +� +�* +name + + " + positions*D +val= +  +� +�*% +@model_path/weights/weight.bin�ϙmw +add +x + + inputs_embeds +y + + positions% +input_5 +  + +� +�* +name + + " +input_5� +linear +x + +input_5B +weight8 +6 +4model_model_decoder_layers_0_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_0_self_attn_q_proj_bias% +var_126 +  + +� +�* +name + + +" +op_126Q +const +var_127 + * +name + + +" +op_127* +val +  + + + +>q +mul +x + +var_126 +y + +var_127& +tensor_5 +  + +� +�* +name + + " + +tensor_5� +linear +x + +input_5B +weight8 +6 +4model_model_decoder_layers_0_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_0_self_attn_k_proj_bias& +tensor_1 +  + +� +�* +name + + " + +tensor_1j +const +var_132 + + +* +name + + +" +op_132** +val# + + + + + ���������@| +reshape +x + + +tensor_1 +shape + +var_132* +var_133 +  + +� + +@* +name + + +" +op_133o +const +var_134_perm_0 + + +*# +name + +" + op_134_perm_0*! +val + + + + + +� +linear +x + +input_5B +weight8 +6 +4model_model_decoder_layers_0_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_0_self_attn_v_proj_bias& +tensor_3 +  + +� +�* +name + + " + +tensor_3j +const +var_139 + + +* +name + + +" +op_139** +val# + + + + + ���������@| +reshape +x + + +tensor_3 +shape + +var_139* +var_140 +  + +� + +@* +name + + +" +op_140o +const +var_141_perm_0 + + +*# +name + +" + op_141_perm_0*! +val + + + + + +b +const +var_148 + + +* +name + + +" +op_148*" +val + + + +  +�@| +reshape +x + + +tensor_5 +shape + +var_148* +var_149 +  + +� + +@* +name + + +" +op_149o +const +var_150_perm_0 + + +*# +name + +" + op_150_perm_0*! +val + + + + + +i +const +var_152 + + +* +name + + +" +op_152*) +val" + + + + + ���������@� + transpose +x + +var_149 +perm + +var_150_perm_0/ + transpose_57 +  + + +� +@*" +name + +" + transpose_57� +reshape +x + + transpose_57 +shape + +var_152+ +query_states_1 +  + +� +@*$ +name + +" +query_states_1i +const +var_154 + + +* +name + + +" +op_154*) +val" + + + + + ���������@� + transpose +x + +var_133 +perm + +var_134_perm_0/ + transpose_59 +  + + +� +@*" +name + +" + transpose_59� +reshape +x + + transpose_59 +shape + +var_154) + key_states_3 +  + +� +@*" +name + +" + key_states_3i +const +var_156 + + +* +name + + +" +op_156*) +val" + + + + + ���������@� + transpose +x + +var_140 +perm + +var_141_perm_0/ + transpose_58 +  + + +� +@*" +name + +" + transpose_58� +reshape +x + + transpose_58 +shape + +var_156+ +value_states_3 +  + +� +@*$ +name + +" +value_states_3n +const +var_159_perm_0 + + +*# +name + +" + op_159_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_159_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_161 + + +* +name + + +" +op_161*# +val + + + + + +��� +reshape +x + +attn_weights_1 +shape + +var_161+ +var_162 +  + + +� +�* +name + + +" +op_162� +add +x + +var_162 +y + +attention_mask_12 +attn_weights_3 +  + + +� +�*$ +name + +" +attn_weights_3b +const +var_167 + + +* +name + + +" +op_167*" +val + + + +  +��~ +reshape +x + +attn_weights_3 +shape + +var_167% +input_7 +  + +� +�* +name + + " +input_7u +softmax +x + +input_7 +axis + + +var_25% +input_9 +  + +� +�* +name + + " +input_9w +const# +attn_output_1_transpose_x_0 +*1 +name) +! +" +attn_output_1_transpose_x_0* +val + + +w +const# +attn_output_1_transpose_y_0 +*1 +name) +! +" +attn_output_1_transpose_y_0* +val + + +� +matmul +x + +input_9 +y + +value_states_3. + transpose_x + +attn_output_1_transpose_x_0. + transpose_y + +attn_output_1_transpose_y_0* + attn_output_1 +  + +� +@*# +name + +" + attn_output_1b +const +var_172 + + +* +name + + +" +op_172*" +val + + + +  +�@� +reshape +x + + attn_output_1 +shape + +var_1720 + attn_output_3 +  + + +� +@*# +name + +" + attn_output_3| +const$ +attn_output_5_perm_0 + + +** +name" + +" +attn_output_5_perm_0*! +val + + + + + +b +const +var_175 + + +* +name + + +" +op_175*" +val + + + +  +��� + transpose +x + + attn_output_3 +perm + +attn_output_5_perm_0/ + transpose_55 +  + +� + +@*" +name + +" + transpose_55~ +reshape +x + + transpose_55 +shape + +var_175& +input_11 +  + +� +�* +name + + " + +input_11� +linear +x + + +input_11D +weight: +8 +6model_model_decoder_layers_0_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_0_self_attn_out_proj_bias& +input_13 +  + +� +�* +name + + " + +input_13r +add +x + +input_5 +y + + +input_13& +input_15 +  + +� +�* +name + + " + +input_15� +const& +hidden_states_5_axes_0 + + +*, +name$ + +" +hidden_states_5_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_15" +axes + +hidden_states_5_axes_0E +gamma< +: +8model_model_decoder_layers_0_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_0_self_attn_layer_norm_bias +epsilon + + +var_11- +hidden_states_5 +  + +� +�*% +name + +" +hidden_states_5� +linear +x + +hidden_states_5E +weight; +9 +7model_model_decoder_layers_0_encoder_attn_q_proj_weightA +bias9 +7 +5model_model_decoder_layers_0_encoder_attn_q_proj_bias% +var_195 +  + +� +�* +name + + +" +op_195Q +const +var_196 + * +name + + +" +op_196* +val +  + + + +>s +mul +x + +var_195 +y + +var_196' + tensor_11 +  + +� +�* +name + + " + tensor_11� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_0_encoder_attn_k_proj_weightA +bias9 +7 +5model_model_decoder_layers_0_encoder_attn_k_proj_bias& +tensor_7 +  + +� +�* +name + + " + +tensor_7j +const +var_201 + + +* +name + + +" +op_201** +val# + + + + + ���������@| +reshape +x + + +tensor_7 +shape + +var_201* +var_202 +  + +� + +@* +name + + +" +op_202o +const +var_203_perm_0 + + +*# +name + +" + op_203_perm_0*! +val + + + + + +� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_0_encoder_attn_v_proj_weightA +bias9 +7 +5model_model_decoder_layers_0_encoder_attn_v_proj_bias& +tensor_9 +  + +� +�* +name + + " + +tensor_9j +const +var_208 + + +* +name + + +" +op_208** +val# + + + + + ���������@| +reshape +x + + +tensor_9 +shape + +var_208* +var_209 +  + +� + +@* +name + + +" +op_209o +const +var_210_perm_0 + + +*# +name + +" + op_210_perm_0*! +val + + + + + +b +const +var_217 + + +* +name + + +" +op_217*" +val + + + +  +�@} +reshape +x + + tensor_11 +shape + +var_217* +var_218 +  + +� + +@* +name + + +" +op_218o +const +var_219_perm_0 + + +*# +name + +" + op_219_perm_0*! +val + + + + + +i +const +var_221 + + +* +name + + +" +op_221*) +val" + + + + + ���������@� + transpose +x + +var_218 +perm + +var_219_perm_0/ + transpose_52 +  + + +� +@*" +name + +" + transpose_52� +reshape +x + + transpose_52 +shape + +var_221+ +query_states_3 +  + +� +@*$ +name + +" +query_states_3i +const +var_223 + + +* +name + + +" +op_223*) +val" + + + + + ���������@� + transpose +x + +var_202 +perm + +var_203_perm_0/ + transpose_54 +  + + +� +@*" +name + +" + transpose_54� +reshape +x + + transpose_54 +shape + +var_223) + key_states_7 +  + +� +@*" +name + +" + key_states_7i +const +var_225 + + +* +name + + +" +op_225*) +val" + + + + + ���������@� + transpose +x + +var_209 +perm + +var_210_perm_0/ + transpose_53 +  + + +� +@*" +name + +" + transpose_53� +reshape +x + + transpose_53 +shape + +var_225+ +value_states_7 +  + +� +@*$ +name + +" +value_states_7n +const +var_228_perm_0 + + +*# +name + +" + op_228_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_228_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_230 + + +* +name + + +" +op_230*# +val + + + + + +��� +reshape +x + +attn_weights_7 +shape + +var_230+ +var_231 +  + + +� +�* +name + + +" +op_231� +add +x + +var_231 +y + +attention_mask2 +attn_weights_9 +  + + +� +�*$ +name + +" +attn_weights_9b +const +var_236 + + +* +name + + +" +op_236*" +val + + + +  +��� +reshape +x + +attn_weights_9 +shape + +var_236& +input_19 +  + +� +�* +name + + " + +input_19x +softmax +x + + +input_19 +axis + + +var_25& +input_21 +  + +� +�* +name + + " + +input_21w +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_21 +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_241 + + +* +name + + +" +op_241*" +val + + + +  +�@� +reshape +x + + attn_output_7 +shape + +var_2410 + attn_output_9 +  + + +� +@*# +name + +" + attn_output_9~ +const% +attn_output_11_perm_0 + + +*+ +name# + +" +attn_output_11_perm_0*! +val + + + + + +b +const +var_244 + + +* +name + + +" +op_244*" +val + + + +  +��� + transpose +x + + attn_output_9! +perm + +attn_output_11_perm_0/ + transpose_50 +  + +� + +@*" +name + +" + transpose_50~ +reshape +x + + transpose_50 +shape + +var_244& +input_23 +  + +� +�* +name + + " + +input_23� +linear +x + + +input_23G +weight= +; +9model_model_decoder_layers_0_encoder_attn_out_proj_weightC +bias; +9 +7model_model_decoder_layers_0_encoder_attn_out_proj_bias& +input_25 +  + +� +�* +name + + " + +input_25z +add +x + +hidden_states_5 +y + + +input_25& +input_27 +  + +� +�* +name + + " + +input_27x +const +input_29_axes_0 + + +*% +name + +" +input_29_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_27 +axes + +input_29_axes_0H +gamma? += +;model_model_decoder_layers_0_encoder_attn_layer_norm_weightE +beta= +; +9model_model_decoder_layers_0_encoder_attn_layer_norm_bias +epsilon + + +var_11& +input_29 +  + +� +�* +name + + " + +input_29� +linear +x + + +input_295 +weight+ +) +'model_model_decoder_layers_0_fc1_weight1 +bias) +' +%model_model_decoder_layers_0_fc1_bias& +input_31 +  + +� +�* +name + + " + +input_31a +silu +x + + +input_31& +input_33 +  + +� +�* +name + + " + +input_33� +linear +x + + +input_335 +weight+ +) +'model_model_decoder_layers_0_fc2_weight1 +bias) +' +%model_model_decoder_layers_0_fc2_bias& +input_37 +  + +� +�* +name + + " + +input_37s +add +x + + +input_29 +y + + +input_37& +input_39 +  + +� +�* +name + + " + +input_39� +const' +hidden_states_11_axes_0 + + +*- +name% + +" +hidden_states_11_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_39# +axes + +hidden_states_11_axes_0A +gamma8 +6 +4model_model_decoder_layers_0_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_0_final_layer_norm_bias +epsilon + + +var_11. +hidden_states_11 +  + +� +�*& +name + +" +hidden_states_11� +linear +x + +hidden_states_11B +weight8 +6 +4model_model_decoder_layers_1_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_1_self_attn_q_proj_bias% +var_285 +  + +� +�* +name + + +" +op_285Q +const +var_286 + * +name + + +" +op_286* +val +  + + + +>s +mul +x + +var_285 +y + +var_286' + tensor_17 +  + +� +�* +name + + " + tensor_17� +linear +x + +hidden_states_11B +weight8 +6 +4model_model_decoder_layers_1_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_1_self_attn_k_proj_bias' + tensor_13 +  + +� +�* +name + + " + tensor_13j +const +var_291 + + +* +name + + +" +op_291** +val# + + + + + ���������@} +reshape +x + + tensor_13 +shape + +var_291* +var_292 +  + +� + +@* +name + + +" +op_292o +const +var_293_perm_0 + + +*# +name + +" + op_293_perm_0*! +val + + + + + +� +linear +x + +hidden_states_11B +weight8 +6 +4model_model_decoder_layers_1_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_1_self_attn_v_proj_bias' + tensor_15 +  + +� +�* +name + + " + tensor_15j +const +var_298 + + +* +name + + +" +op_298** +val# + + + + + ���������@} +reshape +x + + tensor_15 +shape + +var_298* +var_299 +  + +� + +@* +name + + +" +op_299o +const +var_300_perm_0 + + +*# +name + +" + op_300_perm_0*! +val + + + + + +b +const +var_307 + + +* +name + + +" +op_307*" +val + + + +  +�@} +reshape +x + + tensor_17 +shape + +var_307* +var_308 +  + +� + +@* +name + + +" +op_308o +const +var_309_perm_0 + + +*# +name + +" + op_309_perm_0*! +val + + + + + +i +const +var_311 + + +* +name + + +" +op_311*) +val" + + + + + ���������@� + transpose +x + +var_308 +perm + +var_309_perm_0/ + transpose_47 +  + + +� +@*" +name + +" + transpose_47� +reshape +x + + transpose_47 +shape + +var_311+ +query_states_5 +  + +� +@*$ +name + +" +query_states_5i +const +var_313 + + +* +name + + +" +op_313*) +val" + + + + + ���������@� + transpose +x + +var_292 +perm + +var_293_perm_0/ + transpose_49 +  + + +� +@*" +name + +" + transpose_49� +reshape +x + + transpose_49 +shape + +var_313* + key_states_11 +  + +� +@*# +name + +" + key_states_11i +const +var_315 + + +* +name + + +" +op_315*) +val" + + + + + ���������@� + transpose +x + +var_299 +perm + +var_300_perm_0/ + transpose_48 +  + + +� +@*" +name + +" + transpose_48� +reshape +x + + transpose_48 +shape + +var_315, +value_states_11 +  + +� +@*% +name + +" +value_states_11n +const +var_318_perm_0 + + +*# +name + +" + op_318_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_318_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_320 + + +* +name + + +" +op_320*# +val + + + + + +��� +reshape +x + +attn_weights_13 +shape + +var_320+ +var_321 +  + + +� +�* +name + + +" +op_321� +add +x + +var_321 +y + +attention_mask_13 +attn_weights_15 +  + + +� +�*% +name + +" +attn_weights_15b +const +var_326 + + +* +name + + +" +op_326*" +val + + + +  +��� +reshape +x + +attn_weights_15 +shape + +var_326& +input_41 +  + +� +�* +name + + " + +input_41x +softmax +x + + +input_41 +axis + + +var_25& +input_43 +  + +� +�* +name + + " + +input_43y +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_43 +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_331 + + +* +name + + +" +op_331*" +val + + + +  +�@� +reshape +x + +attn_output_13 +shape + +var_3311 +attn_output_15 +  + + +� +@*$ +name + +" +attn_output_15~ +const% +attn_output_17_perm_0 + + +*+ +name# + +" +attn_output_17_perm_0*! +val + + + + + +b +const +var_334 + + +* +name + + +" +op_334*" +val + + + +  +��� + transpose +x + +attn_output_15! +perm + +attn_output_17_perm_0/ + transpose_45 +  + +� + +@*" +name + +" + transpose_45~ +reshape +x + + transpose_45 +shape + +var_334& +input_45 +  + +� +�* +name + + " + +input_45� +linear +x + + +input_45D +weight: +8 +6model_model_decoder_layers_1_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_1_self_attn_out_proj_bias& +input_47 +  + +� +�* +name + + " + +input_47{ +add +x + +hidden_states_11 +y + + +input_47& +input_49 +  + +� +�* +name + + " + +input_49� +const' +hidden_states_15_axes_0 + + +*- +name% + +" +hidden_states_15_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_49# +axes + +hidden_states_15_axes_0E +gamma< +: +8model_model_decoder_layers_1_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_1_self_attn_layer_norm_bias +epsilon + + +var_11. +hidden_states_15 +  + +� +�*& +name + +" +hidden_states_15� +linear +x + +hidden_states_15E +weight; +9 +7model_model_decoder_layers_1_encoder_attn_q_proj_weightA +bias9 +7 +5model_model_decoder_layers_1_encoder_attn_q_proj_bias% +var_354 +  + +� +�* +name + + +" +op_354Q +const +var_355 + * +name + + +" +op_355* +val +  + + + +>s +mul +x + +var_354 +y + +var_355' + tensor_23 +  + +� +�* +name + + " + tensor_23� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_1_encoder_attn_k_proj_weightA +bias9 +7 +5model_model_decoder_layers_1_encoder_attn_k_proj_bias' + tensor_19 +  + +� +�* +name + + " + tensor_19j +const +var_360 + + +* +name + + +" +op_360** +val# + + + + + ���������@} +reshape +x + + tensor_19 +shape + +var_360* +var_361 +  + +� + +@* +name + + +" +op_361o +const +var_362_perm_0 + + +*# +name + +" + op_362_perm_0*! +val + + + + + +� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_1_encoder_attn_v_proj_weightA +bias9 +7 +5model_model_decoder_layers_1_encoder_attn_v_proj_bias' + tensor_21 +  + +� +�* +name + + " + tensor_21j +const +var_367 + + +* +name + + +" +op_367** +val# + + + + + ���������@} +reshape +x + + tensor_21 +shape + +var_367* +var_368 +  + +� + +@* +name + + +" +op_368o +const +var_369_perm_0 + + +*# +name + +" + op_369_perm_0*! +val + + + + + +b +const +var_376 + + +* +name + + +" +op_376*" +val + + + +  +�@} +reshape +x + + tensor_23 +shape + +var_376* +var_377 +  + +� + +@* +name + + +" +op_377o +const +var_378_perm_0 + + +*# +name + +" + op_378_perm_0*! +val + + + + + +i +const +var_380 + + +* +name + + +" +op_380*) +val" + + + + + ���������@� + transpose +x + +var_377 +perm + +var_378_perm_0/ + transpose_42 +  + + +� +@*" +name + +" + transpose_42� +reshape +x + + transpose_42 +shape + +var_380+ +query_states_7 +  + +� +@*$ +name + +" +query_states_7i +const +var_382 + + +* +name + + +" +op_382*) +val" + + + + + ���������@� + transpose +x + +var_361 +perm + +var_362_perm_0/ + transpose_44 +  + + +� +@*" +name + +" + transpose_44� +reshape +x + + transpose_44 +shape + +var_382* + key_states_15 +  + +� +@*# +name + +" + key_states_15i +const +var_384 + + +* +name + + +" +op_384*) +val" + + + + + ���������@� + transpose +x + +var_368 +perm + +var_369_perm_0/ + transpose_43 +  + + +� +@*" +name + +" + transpose_43� +reshape +x + + transpose_43 +shape + +var_384, +value_states_15 +  + +� +@*% +name + +" +value_states_15n +const +var_387_perm_0 + + +*# +name + +" + op_387_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_387_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_389 + + +* +name + + +" +op_389*# +val + + + + + +��� +reshape +x + +attn_weights_19 +shape + +var_389+ +var_390 +  + + +� +�* +name + + +" +op_390� +add +x + +var_390 +y + +attention_mask3 +attn_weights_21 +  + + +� +�*% +name + +" +attn_weights_21b +const +var_395 + + +* +name + + +" +op_395*" +val + + + +  +��� +reshape +x + +attn_weights_21 +shape + +var_395& +input_51 +  + +� +�* +name + + " + +input_51x +softmax +x + + +input_51 +axis + + +var_25& +input_53 +  + +� +�* +name + + " + +input_53y +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_53 +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_400 + + +* +name + + +" +op_400*" +val + + + +  +�@� +reshape +x + +attn_output_19 +shape + +var_4001 +attn_output_21 +  + + +� +@*$ +name + +" +attn_output_21~ +const% +attn_output_23_perm_0 + + +*+ +name# + +" +attn_output_23_perm_0*! +val + + + + + +b +const +var_403 + + +* +name + + +" +op_403*" +val + + + +  +��� + transpose +x + +attn_output_21! +perm + +attn_output_23_perm_0/ + transpose_40 +  + +� + +@*" +name + +" + transpose_40~ +reshape +x + + transpose_40 +shape + +var_403& +input_55 +  + +� +�* +name + + " + +input_55� +linear +x + + +input_55G +weight= +; +9model_model_decoder_layers_1_encoder_attn_out_proj_weightC +bias; +9 +7model_model_decoder_layers_1_encoder_attn_out_proj_bias& +input_57 +  + +� +�* +name + + " + +input_57{ +add +x + +hidden_states_15 +y + + +input_57& +input_59 +  + +� +�* +name + + " + +input_59x +const +input_61_axes_0 + + +*% +name + +" +input_61_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_59 +axes + +input_61_axes_0H +gamma? += +;model_model_decoder_layers_1_encoder_attn_layer_norm_weightE +beta= +; +9model_model_decoder_layers_1_encoder_attn_layer_norm_bias +epsilon + + +var_11& +input_61 +  + +� +�* +name + + " + +input_61� +linear +x + + +input_615 +weight+ +) +'model_model_decoder_layers_1_fc1_weight1 +bias) +' +%model_model_decoder_layers_1_fc1_bias& +input_63 +  + +� +�* +name + + " + +input_63a +silu +x + + +input_63& +input_65 +  + +� +�* +name + + " + +input_65� +linear +x + + +input_655 +weight+ +) +'model_model_decoder_layers_1_fc2_weight1 +bias) +' +%model_model_decoder_layers_1_fc2_bias& +input_69 +  + +� +�* +name + + " + +input_69s +add +x + + +input_61 +y + + +input_69& +input_71 +  + +� +�* +name + + " + +input_71� +const' +hidden_states_21_axes_0 + + +*- +name% + +" +hidden_states_21_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_71# +axes + +hidden_states_21_axes_0A +gamma8 +6 +4model_model_decoder_layers_1_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_1_final_layer_norm_bias +epsilon + + +var_11. +hidden_states_21 +  + +� +�*& +name + +" +hidden_states_21� +linear +x + +hidden_states_21B +weight8 +6 +4model_model_decoder_layers_2_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_2_self_attn_q_proj_bias% +var_444 +  + +� +�* +name + + +" +op_444Q +const +var_445 + * +name + + +" +op_445* +val +  + + + +>s +mul +x + +var_444 +y + +var_445' + tensor_29 +  + +� +�* +name + + " + tensor_29� +linear +x + +hidden_states_21B +weight8 +6 +4model_model_decoder_layers_2_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_2_self_attn_k_proj_bias' + tensor_25 +  + +� +�* +name + + " + tensor_25j +const +var_450 + + +* +name + + +" +op_450** +val# + + + + + ���������@} +reshape +x + + tensor_25 +shape + +var_450* +var_451 +  + +� + +@* +name + + +" +op_451o +const +var_452_perm_0 + + +*# +name + +" + op_452_perm_0*! +val + + + + + +� +linear +x + +hidden_states_21B +weight8 +6 +4model_model_decoder_layers_2_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_2_self_attn_v_proj_bias' + tensor_27 +  + +� +�* +name + + " + tensor_27j +const +var_457 + + +* +name + + +" +op_457** +val# + + + + + ���������@} +reshape +x + + tensor_27 +shape + +var_457* +var_458 +  + +� + +@* +name + + +" +op_458o +const +var_459_perm_0 + + +*# +name + +" + op_459_perm_0*! +val + + + + + +b +const +var_466 + + +* +name + + +" +op_466*" +val + + + +  +�@} +reshape +x + + tensor_29 +shape + +var_466* +var_467 +  + +� + +@* +name + + +" +op_467o +const +var_468_perm_0 + + +*# +name + +" + op_468_perm_0*! +val + + + + + +i +const +var_470 + + +* +name + + +" +op_470*) +val" + + + + + ���������@� + transpose +x + +var_467 +perm + +var_468_perm_0/ + transpose_37 +  + + +� +@*" +name + +" + transpose_37� +reshape +x + + transpose_37 +shape + +var_470+ +query_states_9 +  + +� +@*$ +name + +" +query_states_9i +const +var_472 + + +* +name + + +" +op_472*) +val" + + + + + ���������@� + transpose +x + +var_451 +perm + +var_452_perm_0/ + transpose_39 +  + + +� +@*" +name + +" + transpose_39� +reshape +x + + transpose_39 +shape + +var_472* + key_states_19 +  + +� +@*# +name + +" + key_states_19i +const +var_474 + + +* +name + + +" +op_474*) +val" + + + + + ���������@� + transpose +x + +var_458 +perm + +var_459_perm_0/ + transpose_38 +  + + +� +@*" +name + +" + transpose_38� +reshape +x + + transpose_38 +shape + +var_474, +value_states_19 +  + +� +@*% +name + +" +value_states_19n +const +var_477_perm_0 + + +*# +name + +" + op_477_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_477_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_479 + + +* +name + + +" +op_479*# +val + + + + + +��� +reshape +x + +attn_weights_25 +shape + +var_479+ +var_480 +  + + +� +�* +name + + +" +op_480� +add +x + +var_480 +y + +attention_mask_13 +attn_weights_27 +  + + +� +�*% +name + +" +attn_weights_27b +const +var_485 + + +* +name + + +" +op_485*" +val + + + +  +��� +reshape +x + +attn_weights_27 +shape + +var_485& +input_73 +  + +� +�* +name + + " + +input_73x +softmax +x + + +input_73 +axis + + +var_25& +input_75 +  + +� +�* +name + + " + +input_75y +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_75 +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_490 + + +* +name + + +" +op_490*" +val + + + +  +�@� +reshape +x + +attn_output_25 +shape + +var_4901 +attn_output_27 +  + + +� +@*$ +name + +" +attn_output_27~ +const% +attn_output_29_perm_0 + + +*+ +name# + +" +attn_output_29_perm_0*! +val + + + + + +b +const +var_493 + + +* +name + + +" +op_493*" +val + + + +  +��� + transpose +x + +attn_output_27! +perm + +attn_output_29_perm_0/ + transpose_35 +  + +� + +@*" +name + +" + transpose_35~ +reshape +x + + transpose_35 +shape + +var_493& +input_77 +  + +� +�* +name + + " + +input_77� +linear +x + + +input_77D +weight: +8 +6model_model_decoder_layers_2_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_2_self_attn_out_proj_bias& +input_79 +  + +� +�* +name + + " + +input_79{ +add +x + +hidden_states_21 +y + + +input_79& +input_81 +  + +� +�* +name + + " + +input_81� +const' +hidden_states_25_axes_0 + + +*- +name% + +" +hidden_states_25_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_81# +axes + +hidden_states_25_axes_0E +gamma< +: +8model_model_decoder_layers_2_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_2_self_attn_layer_norm_bias +epsilon + + +var_11. +hidden_states_25 +  + +� +�*& +name + +" +hidden_states_25� +linear +x + +hidden_states_25E +weight; +9 +7model_model_decoder_layers_2_encoder_attn_q_proj_weightA +bias9 +7 +5model_model_decoder_layers_2_encoder_attn_q_proj_bias% +var_513 +  + +� +�* +name + + +" +op_513Q +const +var_514 + * +name + + +" +op_514* +val +  + + + +>s +mul +x + +var_513 +y + +var_514' + tensor_35 +  + +� +�* +name + + " + tensor_35� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_2_encoder_attn_k_proj_weightA +bias9 +7 +5model_model_decoder_layers_2_encoder_attn_k_proj_bias' + tensor_31 +  + +� +�* +name + + " + tensor_31j +const +var_519 + + +* +name + + +" +op_519** +val# + + + + + ���������@} +reshape +x + + tensor_31 +shape + +var_519* +var_520 +  + +� + +@* +name + + +" +op_520o +const +var_521_perm_0 + + +*# +name + +" + op_521_perm_0*! +val + + + + + +� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_2_encoder_attn_v_proj_weightA +bias9 +7 +5model_model_decoder_layers_2_encoder_attn_v_proj_bias' + tensor_33 +  + +� +�* +name + + " + tensor_33j +const +var_526 + + +* +name + + +" +op_526** +val# + + + + + ���������@} +reshape +x + + tensor_33 +shape + +var_526* +var_527 +  + +� + +@* +name + + +" +op_527o +const +var_528_perm_0 + + +*# +name + +" + op_528_perm_0*! +val + + + + + +b +const +var_535 + + +* +name + + +" +op_535*" +val + + + +  +�@} +reshape +x + + tensor_35 +shape + +var_535* +var_536 +  + +� + +@* +name + + +" +op_536o +const +var_537_perm_0 + + +*# +name + +" + op_537_perm_0*! +val + + + + + +i +const +var_539 + + +* +name + + +" +op_539*) +val" + + + + + ���������@� + transpose +x + +var_536 +perm + +var_537_perm_0/ + transpose_32 +  + + +� +@*" +name + +" + transpose_32� +reshape +x + + transpose_32 +shape + +var_539, +query_states_11 +  + +� +@*% +name + +" +query_states_11i +const +var_541 + + +* +name + + +" +op_541*) +val" + + + + + ���������@� + transpose +x + +var_520 +perm + +var_521_perm_0/ + transpose_34 +  + + +� +@*" +name + +" + transpose_34� +reshape +x + + transpose_34 +shape + +var_541* + key_states_23 +  + +� +@*# +name + +" + key_states_23i +const +var_543 + + +* +name + + +" +op_543*) +val" + + + + + ���������@� + transpose +x + +var_527 +perm + +var_528_perm_0/ + transpose_33 +  + + +� +@*" +name + +" + transpose_33� +reshape +x + + transpose_33 +shape + +var_543, +value_states_23 +  + +� +@*% +name + +" +value_states_23n +const +var_546_perm_0 + + +*# +name + +" + op_546_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_546_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_548 + + +* +name + + +" +op_548*# +val + + + + + +��� +reshape +x + +attn_weights_31 +shape + +var_548+ +var_549 +  + + +� +�* +name + + +" +op_549� +add +x + +var_549 +y + +attention_mask3 +attn_weights_33 +  + + +� +�*% +name + +" +attn_weights_33b +const +var_554 + + +* +name + + +" +op_554*" +val + + + +  +��� +reshape +x + +attn_weights_33 +shape + +var_554& +input_83 +  + +� +�* +name + + " + +input_83x +softmax +x + + +input_83 +axis + + +var_25& +input_85 +  + +� +�* +name + + " + +input_85y +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_85 +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_559 + + +* +name + + +" +op_559*" +val + + + +  +�@� +reshape +x + +attn_output_31 +shape + +var_5591 +attn_output_33 +  + + +� +@*$ +name + +" +attn_output_33~ +const% +attn_output_35_perm_0 + + +*+ +name# + +" +attn_output_35_perm_0*! +val + + + + + +b +const +var_562 + + +* +name + + +" +op_562*" +val + + + +  +��� + transpose +x + +attn_output_33! +perm + +attn_output_35_perm_0/ + transpose_30 +  + +� + +@*" +name + +" + transpose_30~ +reshape +x + + transpose_30 +shape + +var_562& +input_87 +  + +� +�* +name + + " + +input_87� +linear +x + + +input_87G +weight= +; +9model_model_decoder_layers_2_encoder_attn_out_proj_weightC +bias; +9 +7model_model_decoder_layers_2_encoder_attn_out_proj_bias& +input_89 +  + +� +�* +name + + " + +input_89{ +add +x + +hidden_states_25 +y + + +input_89& +input_91 +  + +� +�* +name + + " + +input_91x +const +input_93_axes_0 + + +*% +name + +" +input_93_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_91 +axes + +input_93_axes_0H +gamma? += +;model_model_decoder_layers_2_encoder_attn_layer_norm_weightE +beta= +; +9model_model_decoder_layers_2_encoder_attn_layer_norm_bias +epsilon + + +var_11& +input_93 +  + +� +�* +name + + " + +input_93� +linear +x + + +input_935 +weight+ +) +'model_model_decoder_layers_2_fc1_weight1 +bias) +' +%model_model_decoder_layers_2_fc1_bias& +input_95 +  + +� +�* +name + + " + +input_95a +silu +x + + +input_95& +input_97 +  + +� +�* +name + + " + +input_97� +linear +x + + +input_975 +weight+ +) +'model_model_decoder_layers_2_fc2_weight1 +bias) +' +%model_model_decoder_layers_2_fc2_bias' + input_101 +  + +� +�* +name + + " + input_101v +add +x + + +input_93 +y + + input_101' + input_103 +  + +� +�* +name + + " + input_103� +const' +hidden_states_31_axes_0 + + +*- +name% + +" +hidden_states_31_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_103# +axes + +hidden_states_31_axes_0A +gamma8 +6 +4model_model_decoder_layers_2_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_2_final_layer_norm_bias +epsilon + + +var_11. +hidden_states_31 +  + +� +�*& +name + +" +hidden_states_31� +linear +x + +hidden_states_31B +weight8 +6 +4model_model_decoder_layers_3_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_3_self_attn_q_proj_bias% +var_603 +  + +� +�* +name + + +" +op_603Q +const +var_604 + * +name + + +" +op_604* +val +  + + + +>s +mul +x + +var_603 +y + +var_604' + tensor_41 +  + +� +�* +name + + " + tensor_41� +linear +x + +hidden_states_31B +weight8 +6 +4model_model_decoder_layers_3_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_3_self_attn_k_proj_bias' + tensor_37 +  + +� +�* +name + + " + tensor_37j +const +var_609 + + +* +name + + +" +op_609** +val# + + + + + ���������@} +reshape +x + + tensor_37 +shape + +var_609* +var_610 +  + +� + +@* +name + + +" +op_610o +const +var_611_perm_0 + + +*# +name + +" + op_611_perm_0*! +val + + + + + +� +linear +x + +hidden_states_31B +weight8 +6 +4model_model_decoder_layers_3_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_3_self_attn_v_proj_bias' + tensor_39 +  + +� +�* +name + + " + tensor_39j +const +var_616 + + +* +name + + +" +op_616** +val# + + + + + ���������@} +reshape +x + + tensor_39 +shape + +var_616* +var_617 +  + +� + +@* +name + + +" +op_617o +const +var_618_perm_0 + + +*# +name + +" + op_618_perm_0*! +val + + + + + +b +const +var_625 + + +* +name + + +" +op_625*" +val + + + +  +�@} +reshape +x + + tensor_41 +shape + +var_625* +var_626 +  + +� + +@* +name + + +" +op_626o +const +var_627_perm_0 + + +*# +name + +" + op_627_perm_0*! +val + + + + + +i +const +var_629 + + +* +name + + +" +op_629*) +val" + + + + + ���������@� + transpose +x + +var_626 +perm + +var_627_perm_0/ + transpose_27 +  + + +� +@*" +name + +" + transpose_27� +reshape +x + + transpose_27 +shape + +var_629, +query_states_13 +  + +� +@*% +name + +" +query_states_13i +const +var_631 + + +* +name + + +" +op_631*) +val" + + + + + ���������@� + transpose +x + +var_610 +perm + +var_611_perm_0/ + transpose_29 +  + + +� +@*" +name + +" + transpose_29� +reshape +x + + transpose_29 +shape + +var_631* + key_states_27 +  + +� +@*# +name + +" + key_states_27i +const +var_633 + + +* +name + + +" +op_633*) +val" + + + + + ���������@� + transpose +x + +var_617 +perm + +var_618_perm_0/ + transpose_28 +  + + +� +@*" +name + +" + transpose_28� +reshape +x + + transpose_28 +shape + +var_633, +value_states_27 +  + +� +@*% +name + +" +value_states_27n +const +var_636_perm_0 + + +*# +name + +" + op_636_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_636_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_638 + + +* +name + + +" +op_638*# +val + + + + + +��� +reshape +x + +attn_weights_37 +shape + +var_638+ +var_639 +  + + +� +�* +name + + +" +op_639� +add +x + +var_639 +y + +attention_mask_13 +attn_weights_39 +  + + +� +�*% +name + +" +attn_weights_39b +const +var_644 + + +* +name + + +" +op_644*" +val + + + +  +��� +reshape +x + +attn_weights_39 +shape + +var_644' + input_105 +  + +� +�* +name + + " + input_105{ +softmax +x + + input_105 +axis + + +var_25' + input_107 +  + +� +�* +name + + " + input_107y +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_107 +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_649 + + +* +name + + +" +op_649*" +val + + + +  +�@� +reshape +x + +attn_output_37 +shape + +var_6491 +attn_output_39 +  + + +� +@*$ +name + +" +attn_output_39~ +const% +attn_output_41_perm_0 + + +*+ +name# + +" +attn_output_41_perm_0*! +val + + + + + +b +const +var_652 + + +* +name + + +" +op_652*" +val + + + +  +��� + transpose +x + +attn_output_39! +perm + +attn_output_41_perm_0/ + transpose_25 +  + +� + +@*" +name + +" + transpose_25� +reshape +x + + transpose_25 +shape + +var_652' + input_109 +  + +� +�* +name + + " + input_109� +linear +x + + input_109D +weight: +8 +6model_model_decoder_layers_3_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_3_self_attn_out_proj_bias' + input_111 +  + +� +�* +name + + " + input_111~ +add +x + +hidden_states_31 +y + + input_111' + input_113 +  + +� +�* +name + + " + input_113� +const' +hidden_states_35_axes_0 + + +*- +name% + +" +hidden_states_35_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_113# +axes + +hidden_states_35_axes_0E +gamma< +: +8model_model_decoder_layers_3_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_3_self_attn_layer_norm_bias +epsilon + + +var_11. +hidden_states_35 +  + +� +�*& +name + +" +hidden_states_35� +linear +x + +hidden_states_35E +weight; +9 +7model_model_decoder_layers_3_encoder_attn_q_proj_weightA +bias9 +7 +5model_model_decoder_layers_3_encoder_attn_q_proj_bias% +var_672 +  + +� +�* +name + + +" +op_672Q +const +var_673 + * +name + + +" +op_673* +val +  + + + +>s +mul +x + +var_672 +y + +var_673' + tensor_47 +  + +� +�* +name + + " + tensor_47� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_3_encoder_attn_k_proj_weightA +bias9 +7 +5model_model_decoder_layers_3_encoder_attn_k_proj_bias' + tensor_43 +  + +� +�* +name + + " + tensor_43j +const +var_678 + + +* +name + + +" +op_678** +val# + + + + + ���������@} +reshape +x + + tensor_43 +shape + +var_678* +var_679 +  + +� + +@* +name + + +" +op_679o +const +var_680_perm_0 + + +*# +name + +" + op_680_perm_0*! +val + + + + + +� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_3_encoder_attn_v_proj_weightA +bias9 +7 +5model_model_decoder_layers_3_encoder_attn_v_proj_bias' + tensor_45 +  + +� +�* +name + + " + tensor_45j +const +var_685 + + +* +name + + +" +op_685** +val# + + + + + ���������@} +reshape +x + + tensor_45 +shape + +var_685* +var_686 +  + +� + +@* +name + + +" +op_686o +const +var_687_perm_0 + + +*# +name + +" + op_687_perm_0*! +val + + + + + +b +const +var_694 + + +* +name + + +" +op_694*" +val + + + +  +�@} +reshape +x + + tensor_47 +shape + +var_694* +var_695 +  + +� + +@* +name + + +" +op_695o +const +var_696_perm_0 + + +*# +name + +" + op_696_perm_0*! +val + + + + + +i +const +var_698 + + +* +name + + +" +op_698*) +val" + + + + + ���������@� + transpose +x + +var_695 +perm + +var_696_perm_0/ + transpose_22 +  + + +� +@*" +name + +" + transpose_22� +reshape +x + + transpose_22 +shape + +var_698, +query_states_15 +  + +� +@*% +name + +" +query_states_15i +const +var_700 + + +* +name + + +" +op_700*) +val" + + + + + ���������@� + transpose +x + +var_679 +perm + +var_680_perm_0/ + transpose_24 +  + + +� +@*" +name + +" + transpose_24� +reshape +x + + transpose_24 +shape + +var_700* + key_states_31 +  + +� +@*# +name + +" + key_states_31i +const +var_702 + + +* +name + + +" +op_702*) +val" + + + + + ���������@� + transpose +x + +var_686 +perm + +var_687_perm_0/ + transpose_23 +  + + +� +@*" +name + +" + transpose_23� +reshape +x + + transpose_23 +shape + +var_702, +value_states_31 +  + +� +@*% +name + +" +value_states_31n +const +var_705_perm_0 + + +*# +name + +" + op_705_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_705_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_707 + + +* +name + + +" +op_707*# +val + + + + + +��� +reshape +x + +attn_weights_43 +shape + +var_707+ +var_708 +  + + +� +�* +name + + +" +op_708� +add +x + +var_708 +y + +attention_mask3 +attn_weights_45 +  + + +� +�*% +name + +" +attn_weights_45b +const +var_713 + + +* +name + + +" +op_713*" +val + + + +  +��� +reshape +x + +attn_weights_45 +shape + +var_713' + input_115 +  + +� +�* +name + + " + input_115{ +softmax +x + + input_115 +axis + + +var_25' + input_117 +  + +� +�* +name + + " + input_117y +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_117 +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_718 + + +* +name + + +" +op_718*" +val + + + +  +�@� +reshape +x + +attn_output_43 +shape + +var_7181 +attn_output_45 +  + + +� +@*$ +name + +" +attn_output_45~ +const% +attn_output_47_perm_0 + + +*+ +name# + +" +attn_output_47_perm_0*! +val + + + + + +b +const +var_721 + + +* +name + + +" +op_721*" +val + + + +  +��� + transpose +x + +attn_output_45! +perm + +attn_output_47_perm_0/ + transpose_20 +  + +� + +@*" +name + +" + transpose_20� +reshape +x + + transpose_20 +shape + +var_721' + input_119 +  + +� +�* +name + + " + input_119� +linear +x + + input_119G +weight= +; +9model_model_decoder_layers_3_encoder_attn_out_proj_weightC +bias; +9 +7model_model_decoder_layers_3_encoder_attn_out_proj_bias' + input_121 +  + +� +�* +name + + " + input_121~ +add +x + +hidden_states_35 +y + + input_121' + input_123 +  + +� +�* +name + + " + input_123z +const +input_125_axes_0 + + +*& +name + +" +input_125_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_123 +axes + +input_125_axes_0H +gamma? += +;model_model_decoder_layers_3_encoder_attn_layer_norm_weightE +beta= +; +9model_model_decoder_layers_3_encoder_attn_layer_norm_bias +epsilon + + +var_11' + input_125 +  + +� +�* +name + + " + input_125� +linear +x + + input_1255 +weight+ +) +'model_model_decoder_layers_3_fc1_weight1 +bias) +' +%model_model_decoder_layers_3_fc1_bias' + input_127 +  + +� +�* +name + + " + input_127d +silu +x + + input_127' + input_129 +  + +� +�* +name + + " + input_129� +linear +x + + input_1295 +weight+ +) +'model_model_decoder_layers_3_fc2_weight1 +bias) +' +%model_model_decoder_layers_3_fc2_bias' + input_133 +  + +� +�* +name + + " + input_133w +add +x + + input_125 +y + + input_133' + input_135 +  + +� +�* +name + + " + input_135� +const' +hidden_states_41_axes_0 + + +*- +name% + +" +hidden_states_41_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_135# +axes + +hidden_states_41_axes_0A +gamma8 +6 +4model_model_decoder_layers_3_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_3_final_layer_norm_bias +epsilon + + +var_11. +hidden_states_41 +  + +� +�*& +name + +" +hidden_states_41� +linear +x + +hidden_states_41B +weight8 +6 +4model_model_decoder_layers_4_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_4_self_attn_q_proj_bias% +var_762 +  + +� +�* +name + + +" +op_762Q +const +var_763 + * +name + + +" +op_763* +val +  + + + +>s +mul +x + +var_762 +y + +var_763' + tensor_53 +  + +� +�* +name + + " + tensor_53� +linear +x + +hidden_states_41B +weight8 +6 +4model_model_decoder_layers_4_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_4_self_attn_k_proj_bias' + tensor_49 +  + +� +�* +name + + " + tensor_49j +const +var_768 + + +* +name + + +" +op_768** +val# + + + + + ���������@} +reshape +x + + tensor_49 +shape + +var_768* +var_769 +  + +� + +@* +name + + +" +op_769o +const +var_770_perm_0 + + +*# +name + +" + op_770_perm_0*! +val + + + + + +� +linear +x + +hidden_states_41B +weight8 +6 +4model_model_decoder_layers_4_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_4_self_attn_v_proj_bias' + tensor_51 +  + +� +�* +name + + " + tensor_51j +const +var_775 + + +* +name + + +" +op_775** +val# + + + + + ���������@} +reshape +x + + tensor_51 +shape + +var_775* +var_776 +  + +� + +@* +name + + +" +op_776o +const +var_777_perm_0 + + +*# +name + +" + op_777_perm_0*! +val + + + + + +b +const +var_784 + + +* +name + + +" +op_784*" +val + + + +  +�@} +reshape +x + + tensor_53 +shape + +var_784* +var_785 +  + +� + +@* +name + + +" +op_785o +const +var_786_perm_0 + + +*# +name + +" + op_786_perm_0*! +val + + + + + +i +const +var_788 + + +* +name + + +" +op_788*) +val" + + + + + ���������@� + transpose +x + +var_785 +perm + +var_786_perm_0/ + transpose_17 +  + + +� +@*" +name + +" + transpose_17� +reshape +x + + transpose_17 +shape + +var_788, +query_states_17 +  + +� +@*% +name + +" +query_states_17i +const +var_790 + + +* +name + + +" +op_790*) +val" + + + + + ���������@� + transpose +x + +var_769 +perm + +var_770_perm_0/ + transpose_19 +  + + +� +@*" +name + +" + transpose_19� +reshape +x + + transpose_19 +shape + +var_790* + key_states_35 +  + +� +@*# +name + +" + key_states_35i +const +var_792 + + +* +name + + +" +op_792*) +val" + + + + + ���������@� + transpose +x + +var_776 +perm + +var_777_perm_0/ + transpose_18 +  + + +� +@*" +name + +" + transpose_18� +reshape +x + + transpose_18 +shape + +var_792, +value_states_35 +  + +� +@*% +name + +" +value_states_35n +const +var_795_perm_0 + + +*# +name + +" + op_795_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_795_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_797 + + +* +name + + +" +op_797*# +val + + + + + +��� +reshape +x + +attn_weights_49 +shape + +var_797+ +var_798 +  + + +� +�* +name + + +" +op_798� +add +x + +var_798 +y + +attention_mask_13 +attn_weights_51 +  + + +� +�*% +name + +" +attn_weights_51b +const +var_803 + + +* +name + + +" +op_803*" +val + + + +  +��� +reshape +x + +attn_weights_51 +shape + +var_803' + input_137 +  + +� +�* +name + + " + input_137{ +softmax +x + + input_137 +axis + + +var_25' + input_139 +  + +� +�* +name + + " + input_139y +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_139 +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_808 + + +* +name + + +" +op_808*" +val + + + +  +�@� +reshape +x + +attn_output_49 +shape + +var_8081 +attn_output_51 +  + + +� +@*$ +name + +" +attn_output_51~ +const% +attn_output_53_perm_0 + + +*+ +name# + +" +attn_output_53_perm_0*! +val + + + + + +b +const +var_811 + + +* +name + + +" +op_811*" +val + + + +  +��� + transpose +x + +attn_output_51! +perm + +attn_output_53_perm_0/ + transpose_15 +  + +� + +@*" +name + +" + transpose_15� +reshape +x + + transpose_15 +shape + +var_811' + input_141 +  + +� +�* +name + + " + input_141� +linear +x + + input_141D +weight: +8 +6model_model_decoder_layers_4_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_4_self_attn_out_proj_bias' + input_143 +  + +� +�* +name + + " + input_143~ +add +x + +hidden_states_41 +y + + input_143' + input_145 +  + +� +�* +name + + " + input_145� +const' +hidden_states_45_axes_0 + + +*- +name% + +" +hidden_states_45_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_145# +axes + +hidden_states_45_axes_0E +gamma< +: +8model_model_decoder_layers_4_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_4_self_attn_layer_norm_bias +epsilon + + +var_11. +hidden_states_45 +  + +� +�*& +name + +" +hidden_states_45� +linear +x + +hidden_states_45E +weight; +9 +7model_model_decoder_layers_4_encoder_attn_q_proj_weightA +bias9 +7 +5model_model_decoder_layers_4_encoder_attn_q_proj_bias% +var_831 +  + +� +�* +name + + +" +op_831Q +const +var_832 + * +name + + +" +op_832* +val +  + + + +>s +mul +x + +var_831 +y + +var_832' + tensor_59 +  + +� +�* +name + + " + tensor_59� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_4_encoder_attn_k_proj_weightA +bias9 +7 +5model_model_decoder_layers_4_encoder_attn_k_proj_bias' + tensor_55 +  + +� +�* +name + + " + tensor_55j +const +var_837 + + +* +name + + +" +op_837** +val# + + + + + ���������@} +reshape +x + + tensor_55 +shape + +var_837* +var_838 +  + +� + +@* +name + + +" +op_838o +const +var_839_perm_0 + + +*# +name + +" + op_839_perm_0*! +val + + + + + +� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_4_encoder_attn_v_proj_weightA +bias9 +7 +5model_model_decoder_layers_4_encoder_attn_v_proj_bias' + tensor_57 +  + +� +�* +name + + " + tensor_57j +const +var_844 + + +* +name + + +" +op_844** +val# + + + + + ���������@} +reshape +x + + tensor_57 +shape + +var_844* +var_845 +  + +� + +@* +name + + +" +op_845o +const +var_846_perm_0 + + +*# +name + +" + op_846_perm_0*! +val + + + + + +b +const +var_853 + + +* +name + + +" +op_853*" +val + + + +  +�@} +reshape +x + + tensor_59 +shape + +var_853* +var_854 +  + +� + +@* +name + + +" +op_854o +const +var_855_perm_0 + + +*# +name + +" + op_855_perm_0*! +val + + + + + +i +const +var_857 + + +* +name + + +" +op_857*) +val" + + + + + ���������@� + transpose +x + +var_854 +perm + +var_855_perm_0/ + transpose_12 +  + + +� +@*" +name + +" + transpose_12� +reshape +x + + transpose_12 +shape + +var_857, +query_states_19 +  + +� +@*% +name + +" +query_states_19i +const +var_859 + + +* +name + + +" +op_859*) +val" + + + + + ���������@� + transpose +x + +var_838 +perm + +var_839_perm_0/ + transpose_14 +  + + +� +@*" +name + +" + transpose_14� +reshape +x + + transpose_14 +shape + +var_859* + key_states_39 +  + +� +@*# +name + +" + key_states_39i +const +var_861 + + +* +name + + +" +op_861*) +val" + + + + + ���������@� + transpose +x + +var_845 +perm + +var_846_perm_0/ + transpose_13 +  + + +� +@*" +name + +" + transpose_13� +reshape +x + + transpose_13 +shape + +var_861, +value_states_39 +  + +� +@*% +name + +" +value_states_39n +const +var_864_perm_0 + + +*# +name + +" + op_864_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_864_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_866 + + +* +name + + +" +op_866*# +val + + + + + +��� +reshape +x + +attn_weights_55 +shape + +var_866+ +var_867 +  + + +� +�* +name + + +" +op_867� +add +x + +var_867 +y + +attention_mask3 +attn_weights_57 +  + + +� +�*% +name + +" +attn_weights_57b +const +var_872 + + +* +name + + +" +op_872*" +val + + + +  +��� +reshape +x + +attn_weights_57 +shape + +var_872' + input_147 +  + +� +�* +name + + " + input_147{ +softmax +x + + input_147 +axis + + +var_25' + input_149 +  + +� +�* +name + + " + input_149y +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_149 +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_877 + + +* +name + + +" +op_877*" +val + + + +  +�@� +reshape +x + +attn_output_55 +shape + +var_8771 +attn_output_57 +  + + +� +@*$ +name + +" +attn_output_57~ +const% +attn_output_59_perm_0 + + +*+ +name# + +" +attn_output_59_perm_0*! +val + + + + + +b +const +var_880 + + +* +name + + +" +op_880*" +val + + + +  +��� + transpose +x + +attn_output_57! +perm + +attn_output_59_perm_0/ + transpose_10 +  + +� + +@*" +name + +" + transpose_10� +reshape +x + + transpose_10 +shape + +var_880' + input_151 +  + +� +�* +name + + " + input_151� +linear +x + + input_151G +weight= +; +9model_model_decoder_layers_4_encoder_attn_out_proj_weightC +bias; +9 +7model_model_decoder_layers_4_encoder_attn_out_proj_bias' + input_153 +  + +� +�* +name + + " + input_153~ +add +x + +hidden_states_45 +y + + input_153' + input_155 +  + +� +�* +name + + " + input_155z +const +input_157_axes_0 + + +*& +name + +" +input_157_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_155 +axes + +input_157_axes_0H +gamma? += +;model_model_decoder_layers_4_encoder_attn_layer_norm_weightE +beta= +; +9model_model_decoder_layers_4_encoder_attn_layer_norm_bias +epsilon + + +var_11' + input_157 +  + +� +�* +name + + " + input_157� +linear +x + + input_1575 +weight+ +) +'model_model_decoder_layers_4_fc1_weight1 +bias) +' +%model_model_decoder_layers_4_fc1_bias' + input_159 +  + +� +�* +name + + " + input_159d +silu +x + + input_159' + input_161 +  + +� +�* +name + + " + input_161� +linear +x + + input_1615 +weight+ +) +'model_model_decoder_layers_4_fc2_weight1 +bias) +' +%model_model_decoder_layers_4_fc2_bias' + input_165 +  + +� +�* +name + + " + input_165w +add +x + + input_157 +y + + input_165' + input_167 +  + +� +�* +name + + " + input_167� +const' +hidden_states_51_axes_0 + + +*- +name% + +" +hidden_states_51_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_167# +axes + +hidden_states_51_axes_0A +gamma8 +6 +4model_model_decoder_layers_4_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_4_final_layer_norm_bias +epsilon + + +var_11. +hidden_states_51 +  + +� +�*& +name + +" +hidden_states_51� +linear +x + +hidden_states_51B +weight8 +6 +4model_model_decoder_layers_5_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_5_self_attn_q_proj_bias% +var_921 +  + +� +�* +name + + +" +op_921Q +const +var_922 + * +name + + +" +op_922* +val +  + + + +>s +mul +x + +var_921 +y + +var_922' + tensor_65 +  + +� +�* +name + + " + tensor_65� +linear +x + +hidden_states_51B +weight8 +6 +4model_model_decoder_layers_5_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_5_self_attn_k_proj_bias' + tensor_61 +  + +� +�* +name + + " + tensor_61j +const +var_927 + + +* +name + + +" +op_927** +val# + + + + + ���������@} +reshape +x + + tensor_61 +shape + +var_927* +var_928 +  + +� + +@* +name + + +" +op_928o +const +var_929_perm_0 + + +*# +name + +" + op_929_perm_0*! +val + + + + + +� +linear +x + +hidden_states_51B +weight8 +6 +4model_model_decoder_layers_5_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_5_self_attn_v_proj_bias' + tensor_63 +  + +� +�* +name + + " + tensor_63j +const +var_934 + + +* +name + + +" +op_934** +val# + + + + + ���������@} +reshape +x + + tensor_63 +shape + +var_934* +var_935 +  + +� + +@* +name + + +" +op_935o +const +var_936_perm_0 + + +*# +name + +" + op_936_perm_0*! +val + + + + + +b +const +var_943 + + +* +name + + +" +op_943*" +val + + + +  +�@} +reshape +x + + tensor_65 +shape + +var_943* +var_944 +  + +� + +@* +name + + +" +op_944o +const +var_945_perm_0 + + +*# +name + +" + op_945_perm_0*! +val + + + + + +i +const +var_947 + + +* +name + + +" +op_947*) +val" + + + + + ���������@� + transpose +x + +var_944 +perm + +var_945_perm_0. + transpose_7 +  + + +� +@*! +name + +" + transpose_7� +reshape +x + + transpose_7 +shape + +var_947, +query_states_21 +  + +� +@*% +name + +" +query_states_21i +const +var_949 + + +* +name + + +" +op_949*) +val" + + + + + ���������@� + transpose +x + +var_928 +perm + +var_929_perm_0. + transpose_9 +  + + +� +@*! +name + +" + transpose_9� +reshape +x + + transpose_9 +shape + +var_949* + key_states_43 +  + +� +@*# +name + +" + key_states_43i +const +var_951 + + +* +name + + +" +op_951*) +val" + + + + + ���������@� + transpose +x + +var_935 +perm + +var_936_perm_0. + transpose_8 +  + + +� +@*! +name + +" + transpose_8� +reshape +x + + transpose_8 +shape + +var_951, +value_states_43 +  + +� +@*% +name + +" +value_states_43n +const +var_954_perm_0 + + +*# +name + +" + op_954_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_954_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_61c +const +var_956 + + +* +name + + +" +op_956*# +val + + + + + +��� +reshape +x + +attn_weights_61 +shape + +var_956+ +var_957 +  + + +� +�* +name + + +" +op_957� +add +x + +var_957 +y + +attention_mask_13 +attn_weights_63 +  + + +� +�*% +name + +" +attn_weights_63b +const +var_962 + + +* +name + + +" +op_962*" +val + + + +  +��� +reshape +x + +attn_weights_63 +shape + +var_962' + input_169 +  + +� +�* +name + + " + input_169{ +softmax +x + + input_169 +axis + + +var_25' + input_171 +  + +� +�* +name + + " + input_171y +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_171 +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_61b +const +var_967 + + +* +name + + +" +op_967*" +val + + + +  +�@� +reshape +x + +attn_output_61 +shape + +var_9671 +attn_output_63 +  + + +� +@*$ +name + +" +attn_output_63~ +const% +attn_output_65_perm_0 + + +*+ +name# + +" +attn_output_65_perm_0*! +val + + + + + +b +const +var_970 + + +* +name + + +" +op_970*" +val + + + +  +��� + transpose +x + +attn_output_63! +perm + +attn_output_65_perm_0. + transpose_5 +  + +� + +@*! +name + +" + transpose_5 +reshape +x + + transpose_5 +shape + +var_970' + input_173 +  + +� +�* +name + + " + input_173� +linear +x + + input_173D +weight: +8 +6model_model_decoder_layers_5_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_5_self_attn_out_proj_bias' + input_175 +  + +� +�* +name + + " + input_175~ +add +x + +hidden_states_51 +y + + input_175' + input_177 +  + +� +�* +name + + " + input_177� +const' +hidden_states_55_axes_0 + + +*- +name% + +" +hidden_states_55_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_177# +axes + +hidden_states_55_axes_0E +gamma< +: +8model_model_decoder_layers_5_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_5_self_attn_layer_norm_bias +epsilon + + +var_11. +hidden_states_55 +  + +� +�*& +name + +" +hidden_states_55� +linear +x + +hidden_states_55E +weight; +9 +7model_model_decoder_layers_5_encoder_attn_q_proj_weightA +bias9 +7 +5model_model_decoder_layers_5_encoder_attn_q_proj_bias% +var_990 +  + +� +�* +name + + +" +op_990Q +const +var_991 + * +name + + +" +op_991* +val +  + + + +>� +mul +x + +var_990 +y + +var_991/ +tensor_workaround +  + +� +�*' +name + +" +tensor_workaround� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_5_encoder_attn_k_proj_weightA +bias9 +7 +5model_model_decoder_layers_5_encoder_attn_k_proj_bias' + tensor_67 +  + +� +�* +name + + " + tensor_67j +const +var_996 + + +* +name + + +" +op_996** +val# + + + + + ���������@} +reshape +x + + tensor_67 +shape + +var_996* +var_997 +  + +� + +@* +name + + +" +op_997o +const +var_998_perm_0 + + +*# +name + +" + op_998_perm_0*! +val + + + + + +� +linear" +x + +encoder_last_hidden_stateE +weight; +9 +7model_model_decoder_layers_5_encoder_attn_v_proj_weightA +bias9 +7 +5model_model_decoder_layers_5_encoder_attn_v_proj_bias' + tensor_69 +  + +� +�* +name + + " + tensor_69l +const +var_1003 + + +* +name + + " +op_1003** +val# + + + + + ���������@� +reshape +x + + tensor_69 +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_workaround +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_2 +  + + +� +@*! +name + +" + transpose_2� +reshape +x + + transpose_2 +shape + + +var_1016) + query_states +  + +� +@*" +name + +" + query_statesk +const +var_1018 + + +* +name + + " +op_1018*) +val" + + + + + ���������@� + transpose +x + +var_997 +perm + +var_998_perm_0. + transpose_4 +  + + +� +@*! +name + +" + transpose_4� +reshape +x + + transpose_4 +shape + + +var_1018' + +key_states +  + +� +@* +name + +" + +key_statesk +const +var_1020 + + +* +name + + " +op_1020*) +val" + + + + + ���������@� + transpose +x + + +var_1004 +perm + +var_1005_perm_0. + transpose_3 +  + + +� +@*! +name + +" + transpose_3� +reshape +x + + transpose_3 +shape + + +var_1020) + value_states +  + +� +@*" +name + +" + value_statesp +const +var_1023_perm_0 + + +*$ +name + +" +op_1023_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_1023_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_1025 + + +* +name + + " +op_1025*# +val + + + + + +��� +reshape +x + +attn_weights_67 +shape + + +var_1025, +var_1026 +  + + +� +�* +name + + " +op_1026� +add +x + + +var_1026 +y + +attention_mask3 +attn_weights_69 +  + + +� +�*% +name + +" +attn_weights_69d +const +var_1031 + + +* +name + + " +op_1031*" +val + + + +  +��� +reshape +x + +attn_weights_69 +shape + + +var_1031' + input_179 +  + +� +�* +name + + " + input_179{ +softmax +x + + input_179 +axis + + +var_25' + input_181 +  + +� +�* +name + + " + input_181y +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_181 +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_1036 + + +* +name + + " +op_1036*" +val + + + +  +�@� +reshape +x + +attn_output_67 +shape + + +var_10361 +attn_output_69 +  + + +� +@*$ +name + +" +attn_output_69x +const" +attn_output_perm_0 + + +*( +name  + +" +attn_output_perm_0*! +val + + + + + +d +const +var_1039 + + +* +name + + " +op_1039*" +val + + + +  +��� + transpose +x + +attn_output_69 +perm + +attn_output_perm_0. + transpose_0 +  + +� + +@*! +name + +" + transpose_0� +reshape +x + + transpose_0 +shape + + +var_1039' + input_183 +  + +� +�* +name + + " + input_183� +linear +x + + input_183G +weight= +; +9model_model_decoder_layers_5_encoder_attn_out_proj_weightC +bias; +9 +7model_model_decoder_layers_5_encoder_attn_out_proj_bias' + input_185 +  + +� +�* +name + + " + input_185~ +add +x + +hidden_states_55 +y + + input_185' + input_187 +  + +� +�* +name + + " + input_187z +const +input_189_axes_0 + + +*& +name + +" +input_189_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_187 +axes + +input_189_axes_0H +gamma? += +;model_model_decoder_layers_5_encoder_attn_layer_norm_weightE +beta= +; +9model_model_decoder_layers_5_encoder_attn_layer_norm_bias +epsilon + + +var_11' + input_189 +  + +� +�* +name + + " + input_189� +linear +x + + input_1895 +weight+ +) +'model_model_decoder_layers_5_fc1_weight1 +bias) +' +%model_model_decoder_layers_5_fc1_bias' + input_191 +  + +� +�* +name + + " + input_191d +silu +x + + input_191' + input_193 +  + +� +�* +name + + " + input_193� +linear +x + + input_1935 +weight+ +) +'model_model_decoder_layers_5_fc2_weight1 +bias) +' +%model_model_decoder_layers_5_fc2_bias' + input_197 +  + +� +�* +name + + " + input_197w +add +x + + input_189 +y + + input_197' + input_199 +  + +� +�* +name + + " + input_199r +const + input_axes_0 + + +*" +name + +" + input_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_199 +axes + + input_axes_0A +gamma8 +6 +4model_model_decoder_layers_5_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_5_final_layer_norm_bias +epsilon + + +var_11# +input +  + +� +�* +name + + " +input� +const! +var_1066_bias_0 +   +��*% +name + +" +var_1066_bias_0*> +val7 +   +��*% +@model_path/weights/weight.bin�Щm� +linear +x + +input' +weight + +model_model_shared_weight +bias + +var_1066_bias_0% +logits +  + +� +��* +name + + " + +var_1066"� + buildInfo�" + + +~"| +8 +! + +" +coremltools-version + + " +7.0b1 +@ +) +! +" +coremltools-component-torch + + " +2.0.1 \ No newline at end of file