diff --git "a/coreml/text-generation/float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel" "b/coreml/text-generation/float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel" new file mode 100644--- /dev/null +++ "b/coreml/text-generation/float32_model.mlpackage/Data/com.apple.CoreML/model.mlmodel" @@ -0,0 +1,13896 @@ +� +L + input_ids2Indices of input sequence tokens in the vocabulary * +��� +w +attention_maskXMask to avoid performing attention on padding token indices (1 = not masked, 0 = masked) * +���R@ +logits&Classification scores (before softmax)* +�و���� +*facebook/bart-large-mnli (text-generation)�3 +#com.github.apple.coremltools.source torch==2.0.1�+ +$com.github.apple.coremltools.version7.0�9 +co.huggingface.exporters.namefacebook/bart-large-mnli�0 +co.huggingface.exporters.tasktext-generation�F +%co.huggingface.exporters.architectureBartForSequenceClassification�- +"co.huggingface.exporters.frameworkpytorch�- +"co.huggingface.exporters.precisionfloat32�" +transformers_version +4.7.0.dev0����� +main�� + + input_ids + + +� +% +attention_mask + + +�CoreML5�� +CoreML5��logits� +const@ +'model_model_decoder_embed_tokens_weight +  +و +�*= +name5 +- ++") +'model_model_decoder_embed_tokens_weight*B +val; +  +و +�*" +@model_path/weights/weight.bin@� +const= +,model_model_decoder_layernorm_embedding_bias +   +�*B +name: +2 +0". +,model_model_decoder_layernorm_embedding_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���b� +const? +.model_model_decoder_layernorm_embedding_weight +   +�*D +name< +4 +2"0 +.model_model_decoder_layernorm_embedding_weight*= +val6 +   +�*% +@model_path/weights/weight.bin���b� +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��b� +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���b� +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���d� +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���d� +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���f� +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�ėf� +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�ŗh� +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��h� +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��j� +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���j� +const6 +%model_model_decoder_layers_0_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_0_fc1_bias*= +val6 +   +� *% +@model_path/weights/weight.bin���j� +const? +'model_model_decoder_layers_0_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_0_fc1_weight*D +val= +  +�  +�*% +@model_path/weights/weight.bin���j� +const6 +%model_model_decoder_layers_0_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_0_fc2_bias*= +val6 +   +�*% +@model_path/weights/weight.bin���r� +const? +'model_model_decoder_layers_0_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_0_fc2_weight*D +val= +  +� +� *% +@model_path/weights/weight.bin�șr� +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�əz� +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��z� +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���z� +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���z� +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���|� +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�˚|� +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�̚~� +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��~� +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*> +val7 +   +�*& +@model_path/weights/weight.bin�횀� +constN +6model_model_decoder_layers_1_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_1_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_decoder_layers_1_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_1_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_decoder_layers_1_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_1_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_1_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_1_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin�ϛ�� +const? +'model_model_decoder_layers_1_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_1_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin�Ϝ�� +const6 +%model_model_decoder_layers_1_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_1_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�М�� +const? +'model_model_decoder_layers_1_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_1_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin��� +constC +2model_model_decoder_layers_1_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_1_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin��� +constE +4model_model_decoder_layers_1_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_1_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin�ҝ�� +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*> +val7 +   +�*& +@model_path/weights/weight.bin�ӝ�� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin��� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_decoder_layers_2_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_2_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_decoder_layers_2_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_2_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_decoder_layers_2_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_2_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�֞�� +const6 +%model_model_decoder_layers_2_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_2_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_2_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_2_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_2_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_2_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_2_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_2_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_2_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_2_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_decoder_layers_2_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_2_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin�ڠ�� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_decoder_layers_3_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_3_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�ݡ�� +constG +6model_model_decoder_layers_3_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_3_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ޡ�� +constI +8model_model_decoder_layers_3_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_3_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_3_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_3_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_3_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_3_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_3_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_3_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_3_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_3_fc2_weight*E +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*> +val7 +   +�*& +@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*> +val7 +   +�*& +@model_path/weights/weight.bin���� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +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*> +val7 +   +�*& +@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*E +val> +  +� +�*& +@model_path/weights/weight.bin�ä�� +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*> +val7 +   +�*& +@model_path/weights/weight.bin�Ĥ�� +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*E +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*> +val7 +   +�*& +@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*E +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*> +val7 +   +�*& +@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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_4_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_4_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin�ǥ�� +const? +'model_model_decoder_layers_4_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_4_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin�Ǧ�� +const6 +%model_model_decoder_layers_4_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_4_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�Ȧ�� +const? +'model_model_decoder_layers_4_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_4_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin���� +constC +2model_model_decoder_layers_4_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_4_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin���� +constE +4model_model_decoder_layers_4_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_4_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin�ʧ�� +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*> +val7 +   +�*& +@model_path/weights/weight.bin�˧�� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin���� +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*> +val7 +   +�*& +@model_path/weights/weight.bin���� +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*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +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*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_decoder_layers_5_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_5_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_decoder_layers_5_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_5_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_decoder_layers_5_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_5_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�Ψ�� +const6 +%model_model_decoder_layers_5_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_5_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin���� +const? +'model_model_decoder_layers_5_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_5_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin���� +const6 +%model_model_decoder_layers_5_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_5_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin���� +const? +'model_model_decoder_layers_5_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_5_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_5_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_5_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_decoder_layers_5_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_5_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_6_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_6_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�Ҫ�� +constL +4model_model_decoder_layers_6_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_6_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin���� +constC +2model_model_decoder_layers_6_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_6_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin���� +constL +4model_model_decoder_layers_6_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_6_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_6_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_6_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_decoder_layers_6_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_6_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constE +4model_model_decoder_layers_6_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_6_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_decoder_layers_6_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_6_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�ի�� +constG +6model_model_decoder_layers_6_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_6_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�֫�� +constI +8model_model_decoder_layers_6_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_6_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_6_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_6_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_6_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_6_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_6_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_6_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_6_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_6_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_6_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_6_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_decoder_layers_6_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_6_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�٭�� +constC +2model_model_decoder_layers_7_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_7_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_decoder_layers_7_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_7_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_7_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_7_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_decoder_layers_7_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_7_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_7_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_7_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_decoder_layers_7_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_7_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�ܮ�� +constE +4model_model_decoder_layers_7_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_7_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ݮ�� +constN +6model_model_decoder_layers_7_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_7_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_decoder_layers_7_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_7_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_decoder_layers_7_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_7_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_7_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_7_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_7_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_7_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_7_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_7_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_7_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_7_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin�చ� +constC +2model_model_decoder_layers_7_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_7_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ᰢ� +constE +4model_model_decoder_layers_7_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_7_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_8_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_8_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_decoder_layers_8_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_8_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�±�� +constC +2model_model_decoder_layers_8_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_8_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ñ�� +constL +4model_model_decoder_layers_8_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_8_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�㱤� +constC +2model_model_decoder_layers_8_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_8_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�䱦� +constL +4model_model_decoder_layers_8_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_8_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constE +4model_model_decoder_layers_8_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_8_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_decoder_layers_8_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_8_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constG +6model_model_decoder_layers_8_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_8_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constI +8model_model_decoder_layers_8_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_8_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�Ʋ�� +const6 +%model_model_decoder_layers_8_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_8_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin�粪� +const? +'model_model_decoder_layers_8_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_8_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin�糪� +const6 +%model_model_decoder_layers_8_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_8_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�賲� +const? +'model_model_decoder_layers_8_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_8_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_8_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_8_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_decoder_layers_8_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_8_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_9_self_attn_q_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_9_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ʴ�� +constL +4model_model_decoder_layers_9_self_attn_q_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_9_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�괺� +constC +2model_model_decoder_layers_9_self_attn_k_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_9_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�봼� +constL +4model_model_decoder_layers_9_self_attn_k_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_9_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_9_self_attn_v_proj_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_9_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constL +4model_model_decoder_layers_9_self_attn_v_proj_weight +  +� +�*J +nameB +: +8"6 +4model_model_decoder_layers_9_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constE +4model_model_decoder_layers_9_self_attn_out_proj_bias +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_9_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constN +6model_model_decoder_layers_9_self_attn_out_proj_weight +  +� +�*L +nameD +< +:"8 +6model_model_decoder_layers_9_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�͵�� +constG +6model_model_decoder_layers_9_self_attn_layer_norm_bias +   +�*L +nameD +< +:"8 +6model_model_decoder_layers_9_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ε�� +constI +8model_model_decoder_layers_9_self_attn_layer_norm_weight +   +�*N +nameF +> +<": +8model_model_decoder_layers_9_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin���� +const6 +%model_model_decoder_layers_9_fc1_bias +   +� *; +name3 ++ +)"' +%model_model_decoder_layers_9_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_9_fc1_weight +  +�  +�*= +name5 +- ++") +'model_model_decoder_layers_9_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const6 +%model_model_decoder_layers_9_fc2_bias +   +�*; +name3 ++ +)"' +%model_model_decoder_layers_9_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const? +'model_model_decoder_layers_9_fc2_weight +  +� +� *= +name5 +- ++") +'model_model_decoder_layers_9_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constC +2model_model_decoder_layers_9_final_layer_norm_bias +   +�*H +name@ +8 +6"4 +2model_model_decoder_layers_9_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constE +4model_model_decoder_layers_9_final_layer_norm_weight +   +�*J +nameB +: +8"6 +4model_model_decoder_layers_9_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin�ѷ�� +constD +3model_model_decoder_layers_10_self_attn_q_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_decoder_layers_10_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin���� +constM +5model_model_decoder_layers_10_self_attn_q_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_decoder_layers_10_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constD +3model_model_decoder_layers_10_self_attn_k_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_decoder_layers_10_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constM +5model_model_decoder_layers_10_self_attn_k_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_decoder_layers_10_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constD +3model_model_decoder_layers_10_self_attn_v_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_decoder_layers_10_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constM +5model_model_decoder_layers_10_self_attn_v_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_decoder_layers_10_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�Ը�� +constF +5model_model_decoder_layers_10_self_attn_out_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_10_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ո�� +constO +7model_model_decoder_layers_10_self_attn_out_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_10_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constH +7model_model_decoder_layers_10_self_attn_layer_norm_bias +   +�*M +nameE += +;"9 +7model_model_decoder_layers_10_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constJ +9model_model_decoder_layers_10_self_attn_layer_norm_weight +   +�*O +nameG +? +="; +9model_model_decoder_layers_10_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const7 +&model_model_decoder_layers_10_fc1_bias +   +� *< +name4 +, +*"( +&model_model_decoder_layers_10_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin����� +const@ +(model_model_decoder_layers_10_fc1_weight +  +�  +�*> +name6 +. +,"* +(model_model_decoder_layers_10_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin����� +const7 +&model_model_decoder_layers_10_fc2_bias +   +�*< +name4 +, +*"( +&model_model_decoder_layers_10_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const@ +(model_model_decoder_layers_10_fc2_weight +  +� +� *> +name6 +. +,"* +(model_model_decoder_layers_10_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin�غ�� +constD +3model_model_decoder_layers_10_final_layer_norm_bias +   +�*I +nameA +9 +7"5 +3model_model_decoder_layers_10_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ٺ�� +constF +5model_model_decoder_layers_10_final_layer_norm_weight +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_10_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constD +3model_model_decoder_layers_11_self_attn_q_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_decoder_layers_11_self_attn_q_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constM +5model_model_decoder_layers_11_self_attn_q_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_decoder_layers_11_self_attn_q_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constD +3model_model_decoder_layers_11_self_attn_k_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_decoder_layers_11_self_attn_k_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constM +5model_model_decoder_layers_11_self_attn_k_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_decoder_layers_11_self_attn_k_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin�ۻ�� +constD +3model_model_decoder_layers_11_self_attn_v_proj_bias +   +�*I +nameA +9 +7"5 +3model_model_decoder_layers_11_self_attn_v_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin�ܻ�� +constM +5model_model_decoder_layers_11_self_attn_v_proj_weight +  +� +�*K +nameC +; +9"7 +5model_model_decoder_layers_11_self_attn_v_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constF +5model_model_decoder_layers_11_self_attn_out_proj_bias +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_11_self_attn_out_proj_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constO +7model_model_decoder_layers_11_self_attn_out_proj_weight +  +� +�*M +nameE += +;"9 +7model_model_decoder_layers_11_self_attn_out_proj_weight*E +val> +  +� +�*& +@model_path/weights/weight.bin����� +constH +7model_model_decoder_layers_11_self_attn_layer_norm_bias +   +�*M +nameE += +;"9 +7model_model_decoder_layers_11_self_attn_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constJ +9model_model_decoder_layers_11_self_attn_layer_norm_weight +   +�*O +nameG +? +="; +9model_model_decoder_layers_11_self_attn_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����� +const7 +&model_model_decoder_layers_11_fc1_bias +   +� *< +name4 +, +*"( +&model_model_decoder_layers_11_fc1_bias*> +val7 +   +� *& +@model_path/weights/weight.bin�߼�� +const@ +(model_model_decoder_layers_11_fc1_weight +  +�  +�*> +name6 +. +,"* +(model_model_decoder_layers_11_fc1_weight*E +val> +  +�  +�*& +@model_path/weights/weight.bin�߽�� +const7 +&model_model_decoder_layers_11_fc2_bias +   +�*< +name4 +, +*"( +&model_model_decoder_layers_11_fc2_bias*> +val7 +   +�*& +@model_path/weights/weight.bin���� +const@ +(model_model_decoder_layers_11_fc2_weight +  +� +� *> +name6 +. +,"* +(model_model_decoder_layers_11_fc2_weight*E +val> +  +� +� *& +@model_path/weights/weight.bin����� +constD +3model_model_decoder_layers_11_final_layer_norm_bias +   +�*I +nameA +9 +7"5 +3model_model_decoder_layers_11_final_layer_norm_bias*> +val7 +   +�*& +@model_path/weights/weight.bin����� +constF +5model_model_decoder_layers_11_final_layer_norm_weight +   +�*K +nameC +; +9"7 +5model_model_decoder_layers_11_final_layer_norm_weight*> +val7 +   +�*& +@model_path/weights/weight.bin����M +const +var_7 + * +name + + +" +op_7* +val +  + + + +���U +const +var_10 +* +name + + " +op_10* +val + + + +���������O +const +var_15 + * +name + + " +op_15* +val +  + + + +�?O +const +var_19 + * +name + + " +op_19* +val +  + + + +��'7Z +const + var_60_axis_0 +*" +name + +" + op_60_axis_0* +val + + +� +gather0 +x+ +) +'model_model_decoder_embed_tokens_weight +indices + + input_ids +axis + + var_60_axis_0$ +var_60 +  + +� +�* +name + + " +op_60� +const* +var_76 +  + + +� +�* +name + + " +op_76*Q +valJ +  + + +� +�*& +@model_path/weights/weight.bin�¾�j +const + var_87_axes_0 + + +*" +name + +" + op_87_axes_0* +val + + + + +� + expand_dims +x + +attention_mask +axes + + var_87_axes_0# +var_87 + + + +�* +name + + " +op_87j +const + var_88_axes_0 + + +*" +name + +" + op_88_axes_0* +val + + + + +� + expand_dims +x + + +var_87 +axes + + var_88_axes_0) +var_88 + + + + +�* +name + + " +op_88n +const + var_91_reps_0 + + +*" +name + +" + op_91_reps_0*" +val + + + +  +�{ +tile +x + + +var_88 +reps + + var_91_reps_0* +var_91 + + + +� +�* +name + + " +op_91_ +const +var_92_dtype_0 +*# +name + +" + op_92_dtype_0* +val + + +" +fp32� +cast +x + + +var_91 +dtype + +var_92_dtype_0, +cast_125 +  + + +� +�* +name + + " + +cast_125� +sub +x + + +var_15 +y + + +cast_1251 + inverted_mask +  + + +� +�*# +name + +" + inverted_mask_ +const +var_94_dtype_0 +*# +name + +" + op_94_dtype_0* +val + + +" +bool� +cast +x + + inverted_mask +dtype + +var_94_dtype_0, +cast_124 + + + +� +�* +name + + " + +cast_124� +select +cond + + +cast_124 +a + +var_7 +b + + inverted_mask* +var_95 +  + + +� +�* +name + + " +op_95� +add +x + + +var_95 +y + + +var_764 +attention_mask_1 +  + + +� +�*$ +name + +" +attention_mask� +const' + positions +  + +� +�* +name + + " + positions*K +valD +  + +� +�*& +@model_path/weights/weight.bin��‚p +add +x + + +var_60 +y + + positions% +input_3 +  + +� +�* +name + + " +input_3v +const +input_5_axes_0 + + +*$ +name + +" +input_5_axes_0*' +val  + + + + + +���������� + +layer_norm +x + +input_3 +axes + +input_5_axes_0; +gamma2 +0 +.model_model_decoder_layernorm_embedding_weight8 +beta0 +. +,model_model_decoder_layernorm_embedding_bias +epsilon + + +var_19% +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& +linear_0 +  + +� +�* +name + + " + +linear_0Q +const +var_133 + * +name + + +" +op_133* +val +  + + + +>r +mul +x + + +linear_0 +y + +var_133& +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& +linear_1 +  + +� +�* +name + + " + +linear_1j +const +var_138 + + +* +name + + +" +op_138** +val# + + + + + ���������@| +reshape +x + + +linear_1 +shape + +var_138* +var_139 +  + +� + +@* +name + + +" +op_139o +const +var_140_perm_0 + + +*# +name + +" + op_140_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& +linear_2 +  + +� +�* +name + + " + +linear_2j +const +var_145 + + +* +name + + +" +op_145** +val# + + + + + ���������@| +reshape +x + + +linear_2 +shape + +var_145* +var_146 +  + +� + +@* +name + + +" +op_146o +const +var_147_perm_0 + + +*# +name + +" + op_147_perm_0*! +val + + + + + +b +const +var_154 + + +* +name + + +" +op_154*" +val + + + +  +�@| +reshape +x + + +tensor_5 +shape + +var_154* +var_155 +  + +� + +@* +name + + +" +op_155o +const +var_156_perm_0 + + +*# +name + +" + op_156_perm_0*! +val + + + + + +i +const +var_158 + + +* +name + + +" +op_158*) +val" + + + + + ���������@� + transpose +x + +var_155 +perm + +var_156_perm_0/ + transpose_57 +  + + +� +@*" +name + +" + transpose_57� +reshape +x + + transpose_57 +shape + +var_158+ +query_states_1 +  + +� +@*$ +name + +" +query_states_1i +const +var_160 + + +* +name + + +" +op_160*) +val" + + + + + ���������@� + transpose +x + +var_139 +perm + +var_140_perm_0/ + transpose_59 +  + + +� +@*" +name + +" + transpose_59� +reshape +x + + transpose_59 +shape + +var_160) + key_states_3 +  + +� +@*" +name + +" + key_states_3i +const +var_162 + + +* +name + + +" +op_162*) +val" + + + + + ���������@� + transpose +x + +var_146 +perm + +var_147_perm_0/ + transpose_58 +  + + +� +@*" +name + +" + transpose_58� +reshape +x + + transpose_58 +shape + +var_162+ +value_states_3 +  + +� +@*$ +name + +" +value_states_3n +const +var_165_perm_0 + + +*# +name + +" + op_165_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_165_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_167 + + +* +name + + +" +op_167*# +val + + + + + +��� +reshape +x + +attn_weights_1 +shape + +var_167+ +var_168 +  + + +� +�* +name + + +" +op_168� +add +x + +var_168 +y + +attention_mask_12 +attn_weights_3 +  + + +� +�*$ +name + +" +attn_weights_3b +const +var_173 + + +* +name + + +" +op_173*" +val + + + +  +��~ +reshape +x + +attn_weights_3 +shape + +var_173% +input_7 +  + +� +�* +name + + " +input_7u +softmax +x + +input_7 +axis + + +var_10% +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_178 + + +* +name + + +" +op_178*" +val + + + +  +�@� +reshape +x + + attn_output_1 +shape + +var_1780 + attn_output_3 +  + + +� +@*# +name + +" + attn_output_3| +const$ +attn_output_5_perm_0 + + +** +name" + +" +attn_output_5_perm_0*! +val + + + + + +b +const +var_181 + + +* +name + + +" +op_181*" +val + + + +  +��� + transpose +x + + attn_output_3 +perm + +attn_output_5_perm_0/ + transpose_55 +  + +� + +@*" +name + +" + transpose_55~ +reshape +x + + transpose_55 +shape + +var_181& +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& +linear_3 +  + +� +�* +name + + " + +linear_3r +add +x + +input_5 +y + + +linear_3& +input_15 +  + +� +�* +name + + " + +input_15x +const +input_17_axes_0 + + +*% +name + +" +input_17_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_15 +axes + +input_17_axes_0E +gamma< +: +8model_model_decoder_layers_0_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_0_self_attn_layer_norm_bias +epsilon + + +var_19& +input_17 +  + +� +�* +name + + " + +input_17� +linear +x + + +input_175 +weight+ +) +'model_model_decoder_layers_0_fc1_weight1 +bias) +' +%model_model_decoder_layers_0_fc1_bias& +linear_4 +  + +� +� * +name + + " + +linear_4c +const +input_21_mode_0 +*% +name + +" +input_21_mode_0* +val + + " +EXACT~ +gelu +x + + +linear_4 +mode + +input_21_mode_0& +input_21 +  + +� +� * +name + + " + +input_21� +linear +x + + +input_215 +weight+ +) +'model_model_decoder_layers_0_fc2_weight1 +bias) +' +%model_model_decoder_layers_0_fc2_bias& +linear_5 +  + +� +�* +name + + " + +linear_5s +add +x + + +input_17 +y + + +linear_5& +input_27 +  + +� +�* +name + + " + +input_27� +const& +hidden_states_7_axes_0 + + +*, +name$ + +" +hidden_states_7_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_27" +axes + +hidden_states_7_axes_0A +gamma8 +6 +4model_model_decoder_layers_0_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_0_final_layer_norm_bias +epsilon + + +var_19- +hidden_states_7 +  + +� +�*% +name + +" +hidden_states_7� +linear +x + +hidden_states_7B +weight8 +6 +4model_model_decoder_layers_1_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_1_self_attn_q_proj_bias& +linear_6 +  + +� +�* +name + + " + +linear_6Q +const +var_221 + * +name + + +" +op_221* +val +  + + + +>t +mul +x + + +linear_6 +y + +var_221' + tensor_11 +  + +� +�* +name + + " + tensor_11� +linear +x + +hidden_states_7B +weight8 +6 +4model_model_decoder_layers_1_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_1_self_attn_k_proj_bias& +linear_7 +  + +� +�* +name + + " + +linear_7j +const +var_226 + + +* +name + + +" +op_226** +val# + + + + + ���������@| +reshape +x + + +linear_7 +shape + +var_226* +var_227 +  + +� + +@* +name + + +" +op_227o +const +var_228_perm_0 + + +*# +name + +" + op_228_perm_0*! +val + + + + + +� +linear +x + +hidden_states_7B +weight8 +6 +4model_model_decoder_layers_1_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_1_self_attn_v_proj_bias& +linear_8 +  + +� +�* +name + + " + +linear_8j +const +var_233 + + +* +name + + +" +op_233** +val# + + + + + ���������@| +reshape +x + + +linear_8 +shape + +var_233* +var_234 +  + +� + +@* +name + + +" +op_234o +const +var_235_perm_0 + + +*# +name + +" + op_235_perm_0*! +val + + + + + +b +const +var_242 + + +* +name + + +" +op_242*" +val + + + +  +�@} +reshape +x + + tensor_11 +shape + +var_242* +var_243 +  + +� + +@* +name + + +" +op_243o +const +var_244_perm_0 + + +*# +name + +" + op_244_perm_0*! +val + + + + + +i +const +var_246 + + +* +name + + +" +op_246*) +val" + + + + + ���������@� + transpose +x + +var_243 +perm + +var_244_perm_0/ + transpose_52 +  + + +� +@*" +name + +" + transpose_52� +reshape +x + + transpose_52 +shape + +var_246+ +query_states_3 +  + +� +@*$ +name + +" +query_states_3i +const +var_248 + + +* +name + + +" +op_248*) +val" + + + + + ���������@� + transpose +x + +var_227 +perm + +var_228_perm_0/ + transpose_54 +  + + +� +@*" +name + +" + transpose_54� +reshape +x + + transpose_54 +shape + +var_248) + key_states_7 +  + +� +@*" +name + +" + key_states_7i +const +var_250 + + +* +name + + +" +op_250*) +val" + + + + + ���������@� + transpose +x + +var_234 +perm + +var_235_perm_0/ + transpose_53 +  + + +� +@*" +name + +" + transpose_53� +reshape +x + + transpose_53 +shape + +var_250+ +value_states_7 +  + +� +@*$ +name + +" +value_states_7n +const +var_253_perm_0 + + +*# +name + +" + op_253_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_253_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_255 + + +* +name + + +" +op_255*# +val + + + + + +��� +reshape +x + +attn_weights_7 +shape + +var_255+ +var_256 +  + + +� +�* +name + + +" +op_256� +add +x + +var_256 +y + +attention_mask_12 +attn_weights_9 +  + + +� +�*$ +name + +" +attn_weights_9b +const +var_261 + + +* +name + + +" +op_261*" +val + + + +  +��� +reshape +x + +attn_weights_9 +shape + +var_261& +input_29 +  + +� +�* +name + + " + +input_29x +softmax +x + + +input_29 +axis + + +var_10& +input_31 +  + +� +�* +name + + " + +input_31w +const# +attn_output_7_transpose_x_0 +*1 +name) +! +" +attn_output_7_transpose_x_0* +val + + +w +const# +attn_output_7_transpose_y_0 +*1 +name) +! +" +attn_output_7_transpose_y_0* +val + + +� +matmul +x + + +input_31 +y + +value_states_7. + transpose_x + +attn_output_7_transpose_x_0. + transpose_y + +attn_output_7_transpose_y_0* + attn_output_7 +  + +� +@*# +name + +" + attn_output_7b +const +var_266 + + +* +name + + +" +op_266*" +val + + + +  +�@� +reshape +x + + attn_output_7 +shape + +var_2660 + attn_output_9 +  + + +� +@*# +name + +" + attn_output_9~ +const% +attn_output_11_perm_0 + + +*+ +name# + +" +attn_output_11_perm_0*! +val + + + + + +b +const +var_269 + + +* +name + + +" +op_269*" +val + + + +  +��� + transpose +x + + attn_output_9! +perm + +attn_output_11_perm_0/ + transpose_50 +  + +� + +@*" +name + +" + transpose_50~ +reshape +x + + transpose_50 +shape + +var_269& +input_33 +  + +� +�* +name + + " + +input_33� +linear +x + + +input_33D +weight: +8 +6model_model_decoder_layers_1_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_1_self_attn_out_proj_bias& +linear_9 +  + +� +�* +name + + " + +linear_9z +add +x + +hidden_states_7 +y + + +linear_9& +input_37 +  + +� +�* +name + + " + +input_37x +const +input_39_axes_0 + + +*% +name + +" +input_39_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_37 +axes + +input_39_axes_0E +gamma< +: +8model_model_decoder_layers_1_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_1_self_attn_layer_norm_bias +epsilon + + +var_19& +input_39 +  + +� +�* +name + + " + +input_39� +linear +x + + +input_395 +weight+ +) +'model_model_decoder_layers_1_fc1_weight1 +bias) +' +%model_model_decoder_layers_1_fc1_bias' + linear_10 +  + +� +� * +name + + " + linear_10c +const +input_43_mode_0 +*% +name + +" +input_43_mode_0* +val + + " +EXACT +gelu +x + + linear_10 +mode + +input_43_mode_0& +input_43 +  + +� +� * +name + + " + +input_43� +linear +x + + +input_435 +weight+ +) +'model_model_decoder_layers_1_fc2_weight1 +bias) +' +%model_model_decoder_layers_1_fc2_bias' + linear_11 +  + +� +�* +name + + " + linear_11t +add +x + + +input_39 +y + + linear_11& +input_49 +  + +� +�* +name + + " + +input_49� +const' +hidden_states_13_axes_0 + + +*- +name% + +" +hidden_states_13_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_49# +axes + +hidden_states_13_axes_0A +gamma8 +6 +4model_model_decoder_layers_1_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_1_final_layer_norm_bias +epsilon + + +var_19. +hidden_states_13 +  + +� +�*& +name + +" +hidden_states_13� +linear +x + +hidden_states_13B +weight8 +6 +4model_model_decoder_layers_2_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_2_self_attn_q_proj_bias' + linear_12 +  + +� +�* +name + + " + linear_12Q +const +var_309 + * +name + + +" +op_309* +val +  + + + +>u +mul +x + + linear_12 +y + +var_309' + tensor_17 +  + +� +�* +name + + " + tensor_17� +linear +x + +hidden_states_13B +weight8 +6 +4model_model_decoder_layers_2_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_2_self_attn_k_proj_bias' + linear_13 +  + +� +�* +name + + " + linear_13j +const +var_314 + + +* +name + + +" +op_314** +val# + + + + + ���������@} +reshape +x + + linear_13 +shape + +var_314* +var_315 +  + +� + +@* +name + + +" +op_315o +const +var_316_perm_0 + + +*# +name + +" + op_316_perm_0*! +val + + + + + +� +linear +x + +hidden_states_13B +weight8 +6 +4model_model_decoder_layers_2_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_2_self_attn_v_proj_bias' + linear_14 +  + +� +�* +name + + " + linear_14j +const +var_321 + + +* +name + + +" +op_321** +val# + + + + + ���������@} +reshape +x + + linear_14 +shape + +var_321* +var_322 +  + +� + +@* +name + + +" +op_322o +const +var_323_perm_0 + + +*# +name + +" + op_323_perm_0*! +val + + + + + +b +const +var_330 + + +* +name + + +" +op_330*" +val + + + +  +�@} +reshape +x + + tensor_17 +shape + +var_330* +var_331 +  + +� + +@* +name + + +" +op_331o +const +var_332_perm_0 + + +*# +name + +" + op_332_perm_0*! +val + + + + + +i +const +var_334 + + +* +name + + +" +op_334*) +val" + + + + + ���������@� + transpose +x + +var_331 +perm + +var_332_perm_0/ + transpose_47 +  + + +� +@*" +name + +" + transpose_47� +reshape +x + + transpose_47 +shape + +var_334+ +query_states_5 +  + +� +@*$ +name + +" +query_states_5i +const +var_336 + + +* +name + + +" +op_336*) +val" + + + + + ���������@� + transpose +x + +var_315 +perm + +var_316_perm_0/ + transpose_49 +  + + +� +@*" +name + +" + transpose_49� +reshape +x + + transpose_49 +shape + +var_336* + key_states_11 +  + +� +@*# +name + +" + key_states_11i +const +var_338 + + +* +name + + +" +op_338*) +val" + + + + + ���������@� + transpose +x + +var_322 +perm + +var_323_perm_0/ + transpose_48 +  + + +� +@*" +name + +" + transpose_48� +reshape +x + + transpose_48 +shape + +var_338, +value_states_11 +  + +� +@*% +name + +" +value_states_11n +const +var_341_perm_0 + + +*# +name + +" + op_341_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_341_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_343 + + +* +name + + +" +op_343*# +val + + + + + +��� +reshape +x + +attn_weights_13 +shape + +var_343+ +var_344 +  + + +� +�* +name + + +" +op_344� +add +x + +var_344 +y + +attention_mask_13 +attn_weights_15 +  + + +� +�*% +name + +" +attn_weights_15b +const +var_349 + + +* +name + + +" +op_349*" +val + + + +  +��� +reshape +x + +attn_weights_15 +shape + +var_349& +input_51 +  + +� +�* +name + + " + +input_51x +softmax +x + + +input_51 +axis + + +var_10& +input_53 +  + +� +�* +name + + " + +input_53y +const$ +attn_output_13_transpose_x_0 +*2 +name* +" + " +attn_output_13_transpose_x_0* +val + + +y +const$ +attn_output_13_transpose_y_0 +*2 +name* +" + " +attn_output_13_transpose_y_0* +val + + +� +matmul +x + + +input_53 +y + +value_states_11/ + transpose_x + +attn_output_13_transpose_x_0/ + transpose_y + +attn_output_13_transpose_y_0+ +attn_output_13 +  + +� +@*$ +name + +" +attn_output_13b +const +var_354 + + +* +name + + +" +op_354*" +val + + + +  +�@� +reshape +x + +attn_output_13 +shape + +var_3541 +attn_output_15 +  + + +� +@*$ +name + +" +attn_output_15~ +const% +attn_output_17_perm_0 + + +*+ +name# + +" +attn_output_17_perm_0*! +val + + + + + +b +const +var_357 + + +* +name + + +" +op_357*" +val + + + +  +��� + transpose +x + +attn_output_15! +perm + +attn_output_17_perm_0/ + transpose_45 +  + +� + +@*" +name + +" + transpose_45~ +reshape +x + + transpose_45 +shape + +var_357& +input_55 +  + +� +�* +name + + " + +input_55� +linear +x + + +input_55D +weight: +8 +6model_model_decoder_layers_2_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_2_self_attn_out_proj_bias' + linear_15 +  + +� +�* +name + + " + linear_15| +add +x + +hidden_states_13 +y + + linear_15& +input_59 +  + +� +�* +name + + " + +input_59x +const +input_61_axes_0 + + +*% +name + +" +input_61_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_59 +axes + +input_61_axes_0E +gamma< +: +8model_model_decoder_layers_2_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_2_self_attn_layer_norm_bias +epsilon + + +var_19& +input_61 +  + +� +�* +name + + " + +input_61� +linear +x + + +input_615 +weight+ +) +'model_model_decoder_layers_2_fc1_weight1 +bias) +' +%model_model_decoder_layers_2_fc1_bias' + linear_16 +  + +� +� * +name + + " + linear_16c +const +input_65_mode_0 +*% +name + +" +input_65_mode_0* +val + + " +EXACT +gelu +x + + linear_16 +mode + +input_65_mode_0& +input_65 +  + +� +� * +name + + " + +input_65� +linear +x + + +input_655 +weight+ +) +'model_model_decoder_layers_2_fc2_weight1 +bias) +' +%model_model_decoder_layers_2_fc2_bias' + linear_17 +  + +� +�* +name + + " + linear_17t +add +x + + +input_61 +y + + linear_17& +input_71 +  + +� +�* +name + + " + +input_71� +const' +hidden_states_19_axes_0 + + +*- +name% + +" +hidden_states_19_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_71# +axes + +hidden_states_19_axes_0A +gamma8 +6 +4model_model_decoder_layers_2_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_2_final_layer_norm_bias +epsilon + + +var_19. +hidden_states_19 +  + +� +�*& +name + +" +hidden_states_19� +linear +x + +hidden_states_19B +weight8 +6 +4model_model_decoder_layers_3_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_3_self_attn_q_proj_bias' + linear_18 +  + +� +�* +name + + " + linear_18Q +const +var_397 + * +name + + +" +op_397* +val +  + + + +>u +mul +x + + linear_18 +y + +var_397' + tensor_23 +  + +� +�* +name + + " + tensor_23� +linear +x + +hidden_states_19B +weight8 +6 +4model_model_decoder_layers_3_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_3_self_attn_k_proj_bias' + linear_19 +  + +� +�* +name + + " + linear_19j +const +var_402 + + +* +name + + +" +op_402** +val# + + + + + ���������@} +reshape +x + + linear_19 +shape + +var_402* +var_403 +  + +� + +@* +name + + +" +op_403o +const +var_404_perm_0 + + +*# +name + +" + op_404_perm_0*! +val + + + + + +� +linear +x + +hidden_states_19B +weight8 +6 +4model_model_decoder_layers_3_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_3_self_attn_v_proj_bias' + linear_20 +  + +� +�* +name + + " + linear_20j +const +var_409 + + +* +name + + +" +op_409** +val# + + + + + ���������@} +reshape +x + + linear_20 +shape + +var_409* +var_410 +  + +� + +@* +name + + +" +op_410o +const +var_411_perm_0 + + +*# +name + +" + op_411_perm_0*! +val + + + + + +b +const +var_418 + + +* +name + + +" +op_418*" +val + + + +  +�@} +reshape +x + + tensor_23 +shape + +var_418* +var_419 +  + +� + +@* +name + + +" +op_419o +const +var_420_perm_0 + + +*# +name + +" + op_420_perm_0*! +val + + + + + +i +const +var_422 + + +* +name + + +" +op_422*) +val" + + + + + ���������@� + transpose +x + +var_419 +perm + +var_420_perm_0/ + transpose_42 +  + + +� +@*" +name + +" + transpose_42� +reshape +x + + transpose_42 +shape + +var_422+ +query_states_7 +  + +� +@*$ +name + +" +query_states_7i +const +var_424 + + +* +name + + +" +op_424*) +val" + + + + + ���������@� + transpose +x + +var_403 +perm + +var_404_perm_0/ + transpose_44 +  + + +� +@*" +name + +" + transpose_44� +reshape +x + + transpose_44 +shape + +var_424* + key_states_15 +  + +� +@*# +name + +" + key_states_15i +const +var_426 + + +* +name + + +" +op_426*) +val" + + + + + ���������@� + transpose +x + +var_410 +perm + +var_411_perm_0/ + transpose_43 +  + + +� +@*" +name + +" + transpose_43� +reshape +x + + transpose_43 +shape + +var_426, +value_states_15 +  + +� +@*% +name + +" +value_states_15n +const +var_429_perm_0 + + +*# +name + +" + op_429_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_429_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_431 + + +* +name + + +" +op_431*# +val + + + + + +��� +reshape +x + +attn_weights_19 +shape + +var_431+ +var_432 +  + + +� +�* +name + + +" +op_432� +add +x + +var_432 +y + +attention_mask_13 +attn_weights_21 +  + + +� +�*% +name + +" +attn_weights_21b +const +var_437 + + +* +name + + +" +op_437*" +val + + + +  +��� +reshape +x + +attn_weights_21 +shape + +var_437& +input_73 +  + +� +�* +name + + " + +input_73x +softmax +x + + +input_73 +axis + + +var_10& +input_75 +  + +� +�* +name + + " + +input_75y +const$ +attn_output_19_transpose_x_0 +*2 +name* +" + " +attn_output_19_transpose_x_0* +val + + +y +const$ +attn_output_19_transpose_y_0 +*2 +name* +" + " +attn_output_19_transpose_y_0* +val + + +� +matmul +x + + +input_75 +y + +value_states_15/ + transpose_x + +attn_output_19_transpose_x_0/ + transpose_y + +attn_output_19_transpose_y_0+ +attn_output_19 +  + +� +@*$ +name + +" +attn_output_19b +const +var_442 + + +* +name + + +" +op_442*" +val + + + +  +�@� +reshape +x + +attn_output_19 +shape + +var_4421 +attn_output_21 +  + + +� +@*$ +name + +" +attn_output_21~ +const% +attn_output_23_perm_0 + + +*+ +name# + +" +attn_output_23_perm_0*! +val + + + + + +b +const +var_445 + + +* +name + + +" +op_445*" +val + + + +  +��� + transpose +x + +attn_output_21! +perm + +attn_output_23_perm_0/ + transpose_40 +  + +� + +@*" +name + +" + transpose_40~ +reshape +x + + transpose_40 +shape + +var_445& +input_77 +  + +� +�* +name + + " + +input_77� +linear +x + + +input_77D +weight: +8 +6model_model_decoder_layers_3_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_3_self_attn_out_proj_bias' + linear_21 +  + +� +�* +name + + " + linear_21| +add +x + +hidden_states_19 +y + + linear_21& +input_81 +  + +� +�* +name + + " + +input_81x +const +input_83_axes_0 + + +*% +name + +" +input_83_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_81 +axes + +input_83_axes_0E +gamma< +: +8model_model_decoder_layers_3_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_3_self_attn_layer_norm_bias +epsilon + + +var_19& +input_83 +  + +� +�* +name + + " + +input_83� +linear +x + + +input_835 +weight+ +) +'model_model_decoder_layers_3_fc1_weight1 +bias) +' +%model_model_decoder_layers_3_fc1_bias' + linear_22 +  + +� +� * +name + + " + linear_22c +const +input_87_mode_0 +*% +name + +" +input_87_mode_0* +val + + " +EXACT +gelu +x + + linear_22 +mode + +input_87_mode_0& +input_87 +  + +� +� * +name + + " + +input_87� +linear +x + + +input_875 +weight+ +) +'model_model_decoder_layers_3_fc2_weight1 +bias) +' +%model_model_decoder_layers_3_fc2_bias' + linear_23 +  + +� +�* +name + + " + linear_23t +add +x + + +input_83 +y + + linear_23& +input_93 +  + +� +�* +name + + " + +input_93� +const' +hidden_states_25_axes_0 + + +*- +name% + +" +hidden_states_25_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + +input_93# +axes + +hidden_states_25_axes_0A +gamma8 +6 +4model_model_decoder_layers_3_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_3_final_layer_norm_bias +epsilon + + +var_19. +hidden_states_25 +  + +� +�*& +name + +" +hidden_states_25� +linear +x + +hidden_states_25B +weight8 +6 +4model_model_decoder_layers_4_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_4_self_attn_q_proj_bias' + linear_24 +  + +� +�* +name + + " + linear_24Q +const +var_485 + * +name + + +" +op_485* +val +  + + + +>u +mul +x + + linear_24 +y + +var_485' + tensor_29 +  + +� +�* +name + + " + tensor_29� +linear +x + +hidden_states_25B +weight8 +6 +4model_model_decoder_layers_4_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_4_self_attn_k_proj_bias' + linear_25 +  + +� +�* +name + + " + linear_25j +const +var_490 + + +* +name + + +" +op_490** +val# + + + + + ���������@} +reshape +x + + linear_25 +shape + +var_490* +var_491 +  + +� + +@* +name + + +" +op_491o +const +var_492_perm_0 + + +*# +name + +" + op_492_perm_0*! +val + + + + + +� +linear +x + +hidden_states_25B +weight8 +6 +4model_model_decoder_layers_4_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_4_self_attn_v_proj_bias' + linear_26 +  + +� +�* +name + + " + linear_26j +const +var_497 + + +* +name + + +" +op_497** +val# + + + + + ���������@} +reshape +x + + linear_26 +shape + +var_497* +var_498 +  + +� + +@* +name + + +" +op_498o +const +var_499_perm_0 + + +*# +name + +" + op_499_perm_0*! +val + + + + + +b +const +var_506 + + +* +name + + +" +op_506*" +val + + + +  +�@} +reshape +x + + tensor_29 +shape + +var_506* +var_507 +  + +� + +@* +name + + +" +op_507o +const +var_508_perm_0 + + +*# +name + +" + op_508_perm_0*! +val + + + + + +i +const +var_510 + + +* +name + + +" +op_510*) +val" + + + + + ���������@� + transpose +x + +var_507 +perm + +var_508_perm_0/ + transpose_37 +  + + +� +@*" +name + +" + transpose_37� +reshape +x + + transpose_37 +shape + +var_510+ +query_states_9 +  + +� +@*$ +name + +" +query_states_9i +const +var_512 + + +* +name + + +" +op_512*) +val" + + + + + ���������@� + transpose +x + +var_491 +perm + +var_492_perm_0/ + transpose_39 +  + + +� +@*" +name + +" + transpose_39� +reshape +x + + transpose_39 +shape + +var_512* + key_states_19 +  + +� +@*# +name + +" + key_states_19i +const +var_514 + + +* +name + + +" +op_514*) +val" + + + + + ���������@� + transpose +x + +var_498 +perm + +var_499_perm_0/ + transpose_38 +  + + +� +@*" +name + +" + transpose_38� +reshape +x + + transpose_38 +shape + +var_514, +value_states_19 +  + +� +@*% +name + +" +value_states_19n +const +var_517_perm_0 + + +*# +name + +" + op_517_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_517_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_519 + + +* +name + + +" +op_519*# +val + + + + + +��� +reshape +x + +attn_weights_25 +shape + +var_519+ +var_520 +  + + +� +�* +name + + +" +op_520� +add +x + +var_520 +y + +attention_mask_13 +attn_weights_27 +  + + +� +�*% +name + +" +attn_weights_27b +const +var_525 + + +* +name + + +" +op_525*" +val + + + +  +��� +reshape +x + +attn_weights_27 +shape + +var_525& +input_95 +  + +� +�* +name + + " + +input_95x +softmax +x + + +input_95 +axis + + +var_10& +input_97 +  + +� +�* +name + + " + +input_97y +const$ +attn_output_25_transpose_x_0 +*2 +name* +" + " +attn_output_25_transpose_x_0* +val + + +y +const$ +attn_output_25_transpose_y_0 +*2 +name* +" + " +attn_output_25_transpose_y_0* +val + + +� +matmul +x + + +input_97 +y + +value_states_19/ + transpose_x + +attn_output_25_transpose_x_0/ + transpose_y + +attn_output_25_transpose_y_0+ +attn_output_25 +  + +� +@*$ +name + +" +attn_output_25b +const +var_530 + + +* +name + + +" +op_530*" +val + + + +  +�@� +reshape +x + +attn_output_25 +shape + +var_5301 +attn_output_27 +  + + +� +@*$ +name + +" +attn_output_27~ +const% +attn_output_29_perm_0 + + +*+ +name# + +" +attn_output_29_perm_0*! +val + + + + + +b +const +var_533 + + +* +name + + +" +op_533*" +val + + + +  +��� + transpose +x + +attn_output_27! +perm + +attn_output_29_perm_0/ + transpose_35 +  + +� + +@*" +name + +" + transpose_35~ +reshape +x + + transpose_35 +shape + +var_533& +input_99 +  + +� +�* +name + + " + +input_99� +linear +x + + +input_99D +weight: +8 +6model_model_decoder_layers_4_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_4_self_attn_out_proj_bias' + linear_27 +  + +� +�* +name + + " + linear_27~ +add +x + +hidden_states_25 +y + + linear_27' + input_103 +  + +� +�* +name + + " + input_103z +const +input_105_axes_0 + + +*& +name + +" +input_105_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_103 +axes + +input_105_axes_0E +gamma< +: +8model_model_decoder_layers_4_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_4_self_attn_layer_norm_bias +epsilon + + +var_19' + input_105 +  + +� +�* +name + + " + input_105� +linear +x + + input_1055 +weight+ +) +'model_model_decoder_layers_4_fc1_weight1 +bias) +' +%model_model_decoder_layers_4_fc1_bias' + linear_28 +  + +� +� * +name + + " + linear_28e +const +input_109_mode_0 +*& +name + +" +input_109_mode_0* +val + + " +EXACT� +gelu +x + + linear_28 +mode + +input_109_mode_0' + input_109 +  + +� +� * +name + + " + input_109� +linear +x + + input_1095 +weight+ +) +'model_model_decoder_layers_4_fc2_weight1 +bias) +' +%model_model_decoder_layers_4_fc2_bias' + linear_29 +  + +� +�* +name + + " + linear_29w +add +x + + input_105 +y + + linear_29' + input_115 +  + +� +�* +name + + " + input_115� +const' +hidden_states_31_axes_0 + + +*- +name% + +" +hidden_states_31_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_115# +axes + +hidden_states_31_axes_0A +gamma8 +6 +4model_model_decoder_layers_4_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_4_final_layer_norm_bias +epsilon + + +var_19. +hidden_states_31 +  + +� +�*& +name + +" +hidden_states_31� +linear +x + +hidden_states_31B +weight8 +6 +4model_model_decoder_layers_5_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_5_self_attn_q_proj_bias' + linear_30 +  + +� +�* +name + + " + linear_30Q +const +var_573 + * +name + + +" +op_573* +val +  + + + +>u +mul +x + + linear_30 +y + +var_573' + tensor_35 +  + +� +�* +name + + " + tensor_35� +linear +x + +hidden_states_31B +weight8 +6 +4model_model_decoder_layers_5_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_5_self_attn_k_proj_bias' + linear_31 +  + +� +�* +name + + " + linear_31j +const +var_578 + + +* +name + + +" +op_578** +val# + + + + + ���������@} +reshape +x + + linear_31 +shape + +var_578* +var_579 +  + +� + +@* +name + + +" +op_579o +const +var_580_perm_0 + + +*# +name + +" + op_580_perm_0*! +val + + + + + +� +linear +x + +hidden_states_31B +weight8 +6 +4model_model_decoder_layers_5_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_5_self_attn_v_proj_bias' + linear_32 +  + +� +�* +name + + " + linear_32j +const +var_585 + + +* +name + + +" +op_585** +val# + + + + + ���������@} +reshape +x + + linear_32 +shape + +var_585* +var_586 +  + +� + +@* +name + + +" +op_586o +const +var_587_perm_0 + + +*# +name + +" + op_587_perm_0*! +val + + + + + +b +const +var_594 + + +* +name + + +" +op_594*" +val + + + +  +�@} +reshape +x + + tensor_35 +shape + +var_594* +var_595 +  + +� + +@* +name + + +" +op_595o +const +var_596_perm_0 + + +*# +name + +" + op_596_perm_0*! +val + + + + + +i +const +var_598 + + +* +name + + +" +op_598*) +val" + + + + + ���������@� + transpose +x + +var_595 +perm + +var_596_perm_0/ + transpose_32 +  + + +� +@*" +name + +" + transpose_32� +reshape +x + + transpose_32 +shape + +var_598, +query_states_11 +  + +� +@*% +name + +" +query_states_11i +const +var_600 + + +* +name + + +" +op_600*) +val" + + + + + ���������@� + transpose +x + +var_579 +perm + +var_580_perm_0/ + transpose_34 +  + + +� +@*" +name + +" + transpose_34� +reshape +x + + transpose_34 +shape + +var_600* + key_states_23 +  + +� +@*# +name + +" + key_states_23i +const +var_602 + + +* +name + + +" +op_602*) +val" + + + + + ���������@� + transpose +x + +var_586 +perm + +var_587_perm_0/ + transpose_33 +  + + +� +@*" +name + +" + transpose_33� +reshape +x + + transpose_33 +shape + +var_602, +value_states_23 +  + +� +@*% +name + +" +value_states_23n +const +var_605_perm_0 + + +*# +name + +" + op_605_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_605_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_607 + + +* +name + + +" +op_607*# +val + + + + + +��� +reshape +x + +attn_weights_31 +shape + +var_607+ +var_608 +  + + +� +�* +name + + +" +op_608� +add +x + +var_608 +y + +attention_mask_13 +attn_weights_33 +  + + +� +�*% +name + +" +attn_weights_33b +const +var_613 + + +* +name + + +" +op_613*" +val + + + +  +��� +reshape +x + +attn_weights_33 +shape + +var_613' + input_117 +  + +� +�* +name + + " + input_117{ +softmax +x + + input_117 +axis + + +var_10' + input_119 +  + +� +�* +name + + " + input_119y +const$ +attn_output_31_transpose_x_0 +*2 +name* +" + " +attn_output_31_transpose_x_0* +val + + +y +const$ +attn_output_31_transpose_y_0 +*2 +name* +" + " +attn_output_31_transpose_y_0* +val + + +� +matmul +x + + input_119 +y + +value_states_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_618 + + +* +name + + +" +op_618*" +val + + + +  +�@� +reshape +x + +attn_output_31 +shape + +var_6181 +attn_output_33 +  + + +� +@*$ +name + +" +attn_output_33~ +const% +attn_output_35_perm_0 + + +*+ +name# + +" +attn_output_35_perm_0*! +val + + + + + +b +const +var_621 + + +* +name + + +" +op_621*" +val + + + +  +��� + transpose +x + +attn_output_33! +perm + +attn_output_35_perm_0/ + transpose_30 +  + +� + +@*" +name + +" + transpose_30� +reshape +x + + transpose_30 +shape + +var_621' + input_121 +  + +� +�* +name + + " + input_121� +linear +x + + input_121D +weight: +8 +6model_model_decoder_layers_5_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_5_self_attn_out_proj_bias' + linear_33 +  + +� +�* +name + + " + linear_33~ +add +x + +hidden_states_31 +y + + linear_33' + input_125 +  + +� +�* +name + + " + input_125z +const +input_127_axes_0 + + +*& +name + +" +input_127_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_125 +axes + +input_127_axes_0E +gamma< +: +8model_model_decoder_layers_5_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_5_self_attn_layer_norm_bias +epsilon + + +var_19' + input_127 +  + +� +�* +name + + " + input_127� +linear +x + + input_1275 +weight+ +) +'model_model_decoder_layers_5_fc1_weight1 +bias) +' +%model_model_decoder_layers_5_fc1_bias' + linear_34 +  + +� +� * +name + + " + linear_34e +const +input_131_mode_0 +*& +name + +" +input_131_mode_0* +val + + " +EXACT� +gelu +x + + linear_34 +mode + +input_131_mode_0' + input_131 +  + +� +� * +name + + " + input_131� +linear +x + + input_1315 +weight+ +) +'model_model_decoder_layers_5_fc2_weight1 +bias) +' +%model_model_decoder_layers_5_fc2_bias' + linear_35 +  + +� +�* +name + + " + linear_35w +add +x + + input_127 +y + + linear_35' + input_137 +  + +� +�* +name + + " + input_137� +const' +hidden_states_37_axes_0 + + +*- +name% + +" +hidden_states_37_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_137# +axes + +hidden_states_37_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_19. +hidden_states_37 +  + +� +�*& +name + +" +hidden_states_37� +linear +x + +hidden_states_37B +weight8 +6 +4model_model_decoder_layers_6_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_6_self_attn_q_proj_bias' + linear_36 +  + +� +�* +name + + " + linear_36Q +const +var_661 + * +name + + +" +op_661* +val +  + + + +>u +mul +x + + linear_36 +y + +var_661' + tensor_41 +  + +� +�* +name + + " + tensor_41� +linear +x + +hidden_states_37B +weight8 +6 +4model_model_decoder_layers_6_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_6_self_attn_k_proj_bias' + linear_37 +  + +� +�* +name + + " + linear_37j +const +var_666 + + +* +name + + +" +op_666** +val# + + + + + ���������@} +reshape +x + + linear_37 +shape + +var_666* +var_667 +  + +� + +@* +name + + +" +op_667o +const +var_668_perm_0 + + +*# +name + +" + op_668_perm_0*! +val + + + + + +� +linear +x + +hidden_states_37B +weight8 +6 +4model_model_decoder_layers_6_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_6_self_attn_v_proj_bias' + linear_38 +  + +� +�* +name + + " + linear_38j +const +var_673 + + +* +name + + +" +op_673** +val# + + + + + ���������@} +reshape +x + + linear_38 +shape + +var_673* +var_674 +  + +� + +@* +name + + +" +op_674o +const +var_675_perm_0 + + +*# +name + +" + op_675_perm_0*! +val + + + + + +b +const +var_682 + + +* +name + + +" +op_682*" +val + + + +  +�@} +reshape +x + + tensor_41 +shape + +var_682* +var_683 +  + +� + +@* +name + + +" +op_683o +const +var_684_perm_0 + + +*# +name + +" + op_684_perm_0*! +val + + + + + +i +const +var_686 + + +* +name + + +" +op_686*) +val" + + + + + ���������@� + transpose +x + +var_683 +perm + +var_684_perm_0/ + transpose_27 +  + + +� +@*" +name + +" + transpose_27� +reshape +x + + transpose_27 +shape + +var_686, +query_states_13 +  + +� +@*% +name + +" +query_states_13i +const +var_688 + + +* +name + + +" +op_688*) +val" + + + + + ���������@� + transpose +x + +var_667 +perm + +var_668_perm_0/ + transpose_29 +  + + +� +@*" +name + +" + transpose_29� +reshape +x + + transpose_29 +shape + +var_688* + key_states_27 +  + +� +@*# +name + +" + key_states_27i +const +var_690 + + +* +name + + +" +op_690*) +val" + + + + + ���������@� + transpose +x + +var_674 +perm + +var_675_perm_0/ + transpose_28 +  + + +� +@*" +name + +" + transpose_28� +reshape +x + + transpose_28 +shape + +var_690, +value_states_27 +  + +� +@*% +name + +" +value_states_27n +const +var_693_perm_0 + + +*# +name + +" + op_693_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_693_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_695 + + +* +name + + +" +op_695*# +val + + + + + +��� +reshape +x + +attn_weights_37 +shape + +var_695+ +var_696 +  + + +� +�* +name + + +" +op_696� +add +x + +var_696 +y + +attention_mask_13 +attn_weights_39 +  + + +� +�*% +name + +" +attn_weights_39b +const +var_701 + + +* +name + + +" +op_701*" +val + + + +  +��� +reshape +x + +attn_weights_39 +shape + +var_701' + input_139 +  + +� +�* +name + + " + input_139{ +softmax +x + + input_139 +axis + + +var_10' + input_141 +  + +� +�* +name + + " + input_141y +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_141 +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_706 + + +* +name + + +" +op_706*" +val + + + +  +�@� +reshape +x + +attn_output_37 +shape + +var_7061 +attn_output_39 +  + + +� +@*$ +name + +" +attn_output_39~ +const% +attn_output_41_perm_0 + + +*+ +name# + +" +attn_output_41_perm_0*! +val + + + + + +b +const +var_709 + + +* +name + + +" +op_709*" +val + + + +  +��� + transpose +x + +attn_output_39! +perm + +attn_output_41_perm_0/ + transpose_25 +  + +� + +@*" +name + +" + transpose_25� +reshape +x + + transpose_25 +shape + +var_709' + input_143 +  + +� +�* +name + + " + input_143� +linear +x + + input_143D +weight: +8 +6model_model_decoder_layers_6_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_6_self_attn_out_proj_bias' + linear_39 +  + +� +�* +name + + " + linear_39~ +add +x + +hidden_states_37 +y + + linear_39' + input_147 +  + +� +�* +name + + " + input_147z +const +input_149_axes_0 + + +*& +name + +" +input_149_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_147 +axes + +input_149_axes_0E +gamma< +: +8model_model_decoder_layers_6_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_6_self_attn_layer_norm_bias +epsilon + + +var_19' + input_149 +  + +� +�* +name + + " + input_149� +linear +x + + input_1495 +weight+ +) +'model_model_decoder_layers_6_fc1_weight1 +bias) +' +%model_model_decoder_layers_6_fc1_bias' + linear_40 +  + +� +� * +name + + " + linear_40e +const +input_153_mode_0 +*& +name + +" +input_153_mode_0* +val + + " +EXACT� +gelu +x + + linear_40 +mode + +input_153_mode_0' + input_153 +  + +� +� * +name + + " + input_153� +linear +x + + input_1535 +weight+ +) +'model_model_decoder_layers_6_fc2_weight1 +bias) +' +%model_model_decoder_layers_6_fc2_bias' + linear_41 +  + +� +�* +name + + " + linear_41w +add +x + + input_149 +y + + linear_41' + input_159 +  + +� +�* +name + + " + input_159� +const' +hidden_states_43_axes_0 + + +*- +name% + +" +hidden_states_43_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_159# +axes + +hidden_states_43_axes_0A +gamma8 +6 +4model_model_decoder_layers_6_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_6_final_layer_norm_bias +epsilon + + +var_19. +hidden_states_43 +  + +� +�*& +name + +" +hidden_states_43� +linear +x + +hidden_states_43B +weight8 +6 +4model_model_decoder_layers_7_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_7_self_attn_q_proj_bias' + linear_42 +  + +� +�* +name + + " + linear_42Q +const +var_749 + * +name + + +" +op_749* +val +  + + + +>u +mul +x + + linear_42 +y + +var_749' + tensor_47 +  + +� +�* +name + + " + tensor_47� +linear +x + +hidden_states_43B +weight8 +6 +4model_model_decoder_layers_7_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_7_self_attn_k_proj_bias' + linear_43 +  + +� +�* +name + + " + linear_43j +const +var_754 + + +* +name + + +" +op_754** +val# + + + + + ���������@} +reshape +x + + linear_43 +shape + +var_754* +var_755 +  + +� + +@* +name + + +" +op_755o +const +var_756_perm_0 + + +*# +name + +" + op_756_perm_0*! +val + + + + + +� +linear +x + +hidden_states_43B +weight8 +6 +4model_model_decoder_layers_7_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_7_self_attn_v_proj_bias' + linear_44 +  + +� +�* +name + + " + linear_44j +const +var_761 + + +* +name + + +" +op_761** +val# + + + + + ���������@} +reshape +x + + linear_44 +shape + +var_761* +var_762 +  + +� + +@* +name + + +" +op_762o +const +var_763_perm_0 + + +*# +name + +" + op_763_perm_0*! +val + + + + + +b +const +var_770 + + +* +name + + +" +op_770*" +val + + + +  +�@} +reshape +x + + tensor_47 +shape + +var_770* +var_771 +  + +� + +@* +name + + +" +op_771o +const +var_772_perm_0 + + +*# +name + +" + op_772_perm_0*! +val + + + + + +i +const +var_774 + + +* +name + + +" +op_774*) +val" + + + + + ���������@� + transpose +x + +var_771 +perm + +var_772_perm_0/ + transpose_22 +  + + +� +@*" +name + +" + transpose_22� +reshape +x + + transpose_22 +shape + +var_774, +query_states_15 +  + +� +@*% +name + +" +query_states_15i +const +var_776 + + +* +name + + +" +op_776*) +val" + + + + + ���������@� + transpose +x + +var_755 +perm + +var_756_perm_0/ + transpose_24 +  + + +� +@*" +name + +" + transpose_24� +reshape +x + + transpose_24 +shape + +var_776* + key_states_31 +  + +� +@*# +name + +" + key_states_31i +const +var_778 + + +* +name + + +" +op_778*) +val" + + + + + ���������@� + transpose +x + +var_762 +perm + +var_763_perm_0/ + transpose_23 +  + + +� +@*" +name + +" + transpose_23� +reshape +x + + transpose_23 +shape + +var_778, +value_states_31 +  + +� +@*% +name + +" +value_states_31n +const +var_781_perm_0 + + +*# +name + +" + op_781_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_781_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_783 + + +* +name + + +" +op_783*# +val + + + + + +��� +reshape +x + +attn_weights_43 +shape + +var_783+ +var_784 +  + + +� +�* +name + + +" +op_784� +add +x + +var_784 +y + +attention_mask_13 +attn_weights_45 +  + + +� +�*% +name + +" +attn_weights_45b +const +var_789 + + +* +name + + +" +op_789*" +val + + + +  +��� +reshape +x + +attn_weights_45 +shape + +var_789' + input_161 +  + +� +�* +name + + " + input_161{ +softmax +x + + input_161 +axis + + +var_10' + input_163 +  + +� +�* +name + + " + input_163y +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_163 +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_794 + + +* +name + + +" +op_794*" +val + + + +  +�@� +reshape +x + +attn_output_43 +shape + +var_7941 +attn_output_45 +  + + +� +@*$ +name + +" +attn_output_45~ +const% +attn_output_47_perm_0 + + +*+ +name# + +" +attn_output_47_perm_0*! +val + + + + + +b +const +var_797 + + +* +name + + +" +op_797*" +val + + + +  +��� + transpose +x + +attn_output_45! +perm + +attn_output_47_perm_0/ + transpose_20 +  + +� + +@*" +name + +" + transpose_20� +reshape +x + + transpose_20 +shape + +var_797' + input_165 +  + +� +�* +name + + " + input_165� +linear +x + + input_165D +weight: +8 +6model_model_decoder_layers_7_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_7_self_attn_out_proj_bias' + linear_45 +  + +� +�* +name + + " + linear_45~ +add +x + +hidden_states_43 +y + + linear_45' + input_169 +  + +� +�* +name + + " + input_169z +const +input_171_axes_0 + + +*& +name + +" +input_171_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_169 +axes + +input_171_axes_0E +gamma< +: +8model_model_decoder_layers_7_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_7_self_attn_layer_norm_bias +epsilon + + +var_19' + input_171 +  + +� +�* +name + + " + input_171� +linear +x + + input_1715 +weight+ +) +'model_model_decoder_layers_7_fc1_weight1 +bias) +' +%model_model_decoder_layers_7_fc1_bias' + linear_46 +  + +� +� * +name + + " + linear_46e +const +input_175_mode_0 +*& +name + +" +input_175_mode_0* +val + + " +EXACT� +gelu +x + + linear_46 +mode + +input_175_mode_0' + input_175 +  + +� +� * +name + + " + input_175� +linear +x + + input_1755 +weight+ +) +'model_model_decoder_layers_7_fc2_weight1 +bias) +' +%model_model_decoder_layers_7_fc2_bias' + linear_47 +  + +� +�* +name + + " + linear_47w +add +x + + input_171 +y + + linear_47' + input_181 +  + +� +�* +name + + " + input_181� +const' +hidden_states_49_axes_0 + + +*- +name% + +" +hidden_states_49_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_181# +axes + +hidden_states_49_axes_0A +gamma8 +6 +4model_model_decoder_layers_7_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_7_final_layer_norm_bias +epsilon + + +var_19. +hidden_states_49 +  + +� +�*& +name + +" +hidden_states_49� +linear +x + +hidden_states_49B +weight8 +6 +4model_model_decoder_layers_8_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_8_self_attn_q_proj_bias' + linear_48 +  + +� +�* +name + + " + linear_48Q +const +var_837 + * +name + + +" +op_837* +val +  + + + +>u +mul +x + + linear_48 +y + +var_837' + tensor_53 +  + +� +�* +name + + " + tensor_53� +linear +x + +hidden_states_49B +weight8 +6 +4model_model_decoder_layers_8_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_8_self_attn_k_proj_bias' + linear_49 +  + +� +�* +name + + " + linear_49j +const +var_842 + + +* +name + + +" +op_842** +val# + + + + + ���������@} +reshape +x + + linear_49 +shape + +var_842* +var_843 +  + +� + +@* +name + + +" +op_843o +const +var_844_perm_0 + + +*# +name + +" + op_844_perm_0*! +val + + + + + +� +linear +x + +hidden_states_49B +weight8 +6 +4model_model_decoder_layers_8_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_8_self_attn_v_proj_bias' + linear_50 +  + +� +�* +name + + " + linear_50j +const +var_849 + + +* +name + + +" +op_849** +val# + + + + + ���������@} +reshape +x + + linear_50 +shape + +var_849* +var_850 +  + +� + +@* +name + + +" +op_850o +const +var_851_perm_0 + + +*# +name + +" + op_851_perm_0*! +val + + + + + +b +const +var_858 + + +* +name + + +" +op_858*" +val + + + +  +�@} +reshape +x + + tensor_53 +shape + +var_858* +var_859 +  + +� + +@* +name + + +" +op_859o +const +var_860_perm_0 + + +*# +name + +" + op_860_perm_0*! +val + + + + + +i +const +var_862 + + +* +name + + +" +op_862*) +val" + + + + + ���������@� + transpose +x + +var_859 +perm + +var_860_perm_0/ + transpose_17 +  + + +� +@*" +name + +" + transpose_17� +reshape +x + + transpose_17 +shape + +var_862, +query_states_17 +  + +� +@*% +name + +" +query_states_17i +const +var_864 + + +* +name + + +" +op_864*) +val" + + + + + ���������@� + transpose +x + +var_843 +perm + +var_844_perm_0/ + transpose_19 +  + + +� +@*" +name + +" + transpose_19� +reshape +x + + transpose_19 +shape + +var_864* + key_states_35 +  + +� +@*# +name + +" + key_states_35i +const +var_866 + + +* +name + + +" +op_866*) +val" + + + + + ���������@� + transpose +x + +var_850 +perm + +var_851_perm_0/ + transpose_18 +  + + +� +@*" +name + +" + transpose_18� +reshape +x + + transpose_18 +shape + +var_866, +value_states_35 +  + +� +@*% +name + +" +value_states_35n +const +var_869_perm_0 + + +*# +name + +" + op_869_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_869_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_871 + + +* +name + + +" +op_871*# +val + + + + + +��� +reshape +x + +attn_weights_49 +shape + +var_871+ +var_872 +  + + +� +�* +name + + +" +op_872� +add +x + +var_872 +y + +attention_mask_13 +attn_weights_51 +  + + +� +�*% +name + +" +attn_weights_51b +const +var_877 + + +* +name + + +" +op_877*" +val + + + +  +��� +reshape +x + +attn_weights_51 +shape + +var_877' + input_183 +  + +� +�* +name + + " + input_183{ +softmax +x + + input_183 +axis + + +var_10' + input_185 +  + +� +�* +name + + " + input_185y +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_185 +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_882 + + +* +name + + +" +op_882*" +val + + + +  +�@� +reshape +x + +attn_output_49 +shape + +var_8821 +attn_output_51 +  + + +� +@*$ +name + +" +attn_output_51~ +const% +attn_output_53_perm_0 + + +*+ +name# + +" +attn_output_53_perm_0*! +val + + + + + +b +const +var_885 + + +* +name + + +" +op_885*" +val + + + +  +��� + transpose +x + +attn_output_51! +perm + +attn_output_53_perm_0/ + transpose_15 +  + +� + +@*" +name + +" + transpose_15� +reshape +x + + transpose_15 +shape + +var_885' + input_187 +  + +� +�* +name + + " + input_187� +linear +x + + input_187D +weight: +8 +6model_model_decoder_layers_8_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_8_self_attn_out_proj_bias' + linear_51 +  + +� +�* +name + + " + linear_51~ +add +x + +hidden_states_49 +y + + linear_51' + input_191 +  + +� +�* +name + + " + input_191z +const +input_193_axes_0 + + +*& +name + +" +input_193_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_191 +axes + +input_193_axes_0E +gamma< +: +8model_model_decoder_layers_8_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_8_self_attn_layer_norm_bias +epsilon + + +var_19' + input_193 +  + +� +�* +name + + " + input_193� +linear +x + + input_1935 +weight+ +) +'model_model_decoder_layers_8_fc1_weight1 +bias) +' +%model_model_decoder_layers_8_fc1_bias' + linear_52 +  + +� +� * +name + + " + linear_52e +const +input_197_mode_0 +*& +name + +" +input_197_mode_0* +val + + " +EXACT� +gelu +x + + linear_52 +mode + +input_197_mode_0' + input_197 +  + +� +� * +name + + " + input_197� +linear +x + + input_1975 +weight+ +) +'model_model_decoder_layers_8_fc2_weight1 +bias) +' +%model_model_decoder_layers_8_fc2_bias' + linear_53 +  + +� +�* +name + + " + linear_53w +add +x + + input_193 +y + + linear_53' + input_203 +  + +� +�* +name + + " + input_203� +const' +hidden_states_55_axes_0 + + +*- +name% + +" +hidden_states_55_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_203# +axes + +hidden_states_55_axes_0A +gamma8 +6 +4model_model_decoder_layers_8_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_8_final_layer_norm_bias +epsilon + + +var_19. +hidden_states_55 +  + +� +�*& +name + +" +hidden_states_55� +linear +x + +hidden_states_55B +weight8 +6 +4model_model_decoder_layers_9_self_attn_q_proj_weight> +bias6 +4 +2model_model_decoder_layers_9_self_attn_q_proj_bias' + linear_54 +  + +� +�* +name + + " + linear_54Q +const +var_925 + * +name + + +" +op_925* +val +  + + + +>u +mul +x + + linear_54 +y + +var_925' + tensor_59 +  + +� +�* +name + + " + tensor_59� +linear +x + +hidden_states_55B +weight8 +6 +4model_model_decoder_layers_9_self_attn_k_proj_weight> +bias6 +4 +2model_model_decoder_layers_9_self_attn_k_proj_bias' + linear_55 +  + +� +�* +name + + " + linear_55j +const +var_930 + + +* +name + + +" +op_930** +val# + + + + + ���������@} +reshape +x + + linear_55 +shape + +var_930* +var_931 +  + +� + +@* +name + + +" +op_931o +const +var_932_perm_0 + + +*# +name + +" + op_932_perm_0*! +val + + + + + +� +linear +x + +hidden_states_55B +weight8 +6 +4model_model_decoder_layers_9_self_attn_v_proj_weight> +bias6 +4 +2model_model_decoder_layers_9_self_attn_v_proj_bias' + linear_56 +  + +� +�* +name + + " + linear_56j +const +var_937 + + +* +name + + +" +op_937** +val# + + + + + ���������@} +reshape +x + + linear_56 +shape + +var_937* +var_938 +  + +� + +@* +name + + +" +op_938o +const +var_939_perm_0 + + +*# +name + +" + op_939_perm_0*! +val + + + + + +b +const +var_946 + + +* +name + + +" +op_946*" +val + + + +  +�@} +reshape +x + + tensor_59 +shape + +var_946* +var_947 +  + +� + +@* +name + + +" +op_947o +const +var_948_perm_0 + + +*# +name + +" + op_948_perm_0*! +val + + + + + +i +const +var_950 + + +* +name + + +" +op_950*) +val" + + + + + ���������@� + transpose +x + +var_947 +perm + +var_948_perm_0/ + transpose_12 +  + + +� +@*" +name + +" + transpose_12� +reshape +x + + transpose_12 +shape + +var_950, +query_states_19 +  + +� +@*% +name + +" +query_states_19i +const +var_952 + + +* +name + + +" +op_952*) +val" + + + + + ���������@� + transpose +x + +var_931 +perm + +var_932_perm_0/ + transpose_14 +  + + +� +@*" +name + +" + transpose_14� +reshape +x + + transpose_14 +shape + +var_952* + key_states_39 +  + +� +@*# +name + +" + key_states_39i +const +var_954 + + +* +name + + +" +op_954*) +val" + + + + + ���������@� + transpose +x + +var_938 +perm + +var_939_perm_0/ + transpose_13 +  + + +� +@*" +name + +" + transpose_13� +reshape +x + + transpose_13 +shape + +var_954, +value_states_39 +  + +� +@*% +name + +" +value_states_39n +const +var_957_perm_0 + + +*# +name + +" + op_957_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_957_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_959 + + +* +name + + +" +op_959*# +val + + + + + +��� +reshape +x + +attn_weights_55 +shape + +var_959+ +var_960 +  + + +� +�* +name + + +" +op_960� +add +x + +var_960 +y + +attention_mask_13 +attn_weights_57 +  + + +� +�*% +name + +" +attn_weights_57b +const +var_965 + + +* +name + + +" +op_965*" +val + + + +  +��� +reshape +x + +attn_weights_57 +shape + +var_965' + input_205 +  + +� +�* +name + + " + input_205{ +softmax +x + + input_205 +axis + + +var_10' + input_207 +  + +� +�* +name + + " + input_207y +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_207 +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_970 + + +* +name + + +" +op_970*" +val + + + +  +�@� +reshape +x + +attn_output_55 +shape + +var_9701 +attn_output_57 +  + + +� +@*$ +name + +" +attn_output_57~ +const% +attn_output_59_perm_0 + + +*+ +name# + +" +attn_output_59_perm_0*! +val + + + + + +b +const +var_973 + + +* +name + + +" +op_973*" +val + + + +  +��� + transpose +x + +attn_output_57! +perm + +attn_output_59_perm_0/ + transpose_10 +  + +� + +@*" +name + +" + transpose_10� +reshape +x + + transpose_10 +shape + +var_973' + input_209 +  + +� +�* +name + + " + input_209� +linear +x + + input_209D +weight: +8 +6model_model_decoder_layers_9_self_attn_out_proj_weight@ +bias8 +6 +4model_model_decoder_layers_9_self_attn_out_proj_bias' + linear_57 +  + +� +�* +name + + " + linear_57~ +add +x + +hidden_states_55 +y + + linear_57' + input_213 +  + +� +�* +name + + " + input_213z +const +input_215_axes_0 + + +*& +name + +" +input_215_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_213 +axes + +input_215_axes_0E +gamma< +: +8model_model_decoder_layers_9_self_attn_layer_norm_weightB +beta: +8 +6model_model_decoder_layers_9_self_attn_layer_norm_bias +epsilon + + +var_19' + input_215 +  + +� +�* +name + + " + input_215� +linear +x + + input_2155 +weight+ +) +'model_model_decoder_layers_9_fc1_weight1 +bias) +' +%model_model_decoder_layers_9_fc1_bias' + linear_58 +  + +� +� * +name + + " + linear_58e +const +input_219_mode_0 +*& +name + +" +input_219_mode_0* +val + + " +EXACT� +gelu +x + + linear_58 +mode + +input_219_mode_0' + input_219 +  + +� +� * +name + + " + input_219� +linear +x + + input_2195 +weight+ +) +'model_model_decoder_layers_9_fc2_weight1 +bias) +' +%model_model_decoder_layers_9_fc2_bias' + linear_59 +  + +� +�* +name + + " + linear_59w +add +x + + input_215 +y + + linear_59' + input_225 +  + +� +�* +name + + " + input_225� +const' +hidden_states_61_axes_0 + + +*- +name% + +" +hidden_states_61_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_225# +axes + +hidden_states_61_axes_0A +gamma8 +6 +4model_model_decoder_layers_9_final_layer_norm_weight> +beta6 +4 +2model_model_decoder_layers_9_final_layer_norm_bias +epsilon + + +var_19. +hidden_states_61 +  + +� +�*& +name + +" +hidden_states_61� +linear +x + +hidden_states_61C +weight9 +7 +5model_model_decoder_layers_10_self_attn_q_proj_weight? +bias7 +5 +3model_model_decoder_layers_10_self_attn_q_proj_bias' + linear_60 +  + +� +�* +name + + " + linear_60S +const +var_1013 + * +name + + " +op_1013* +val +  + + + +>v +mul +x + + linear_60 +y + + +var_1013' + tensor_65 +  + +� +�* +name + + " + tensor_65� +linear +x + +hidden_states_61C +weight9 +7 +5model_model_decoder_layers_10_self_attn_k_proj_weight? +bias7 +5 +3model_model_decoder_layers_10_self_attn_k_proj_bias' + linear_61 +  + +� +�* +name + + " + linear_61l +const +var_1018 + + +* +name + + " +op_1018** +val# + + + + + ���������@� +reshape +x + + linear_61 +shape + + +var_1018+ +var_1019 +  + +� + +@* +name + + " +op_1019q +const +var_1020_perm_0 + + +*$ +name + +" +op_1020_perm_0*! +val + + + + + +� +linear +x + +hidden_states_61C +weight9 +7 +5model_model_decoder_layers_10_self_attn_v_proj_weight? +bias7 +5 +3model_model_decoder_layers_10_self_attn_v_proj_bias' + linear_62 +  + +� +�* +name + + " + linear_62l +const +var_1025 + + +* +name + + " +op_1025** +val# + + + + + ���������@� +reshape +x + + linear_62 +shape + + +var_1025+ +var_1026 +  + +� + +@* +name + + " +op_1026q +const +var_1027_perm_0 + + +*$ +name + +" +op_1027_perm_0*! +val + + + + + +d +const +var_1034 + + +* +name + + " +op_1034*" +val + + + +  +�@� +reshape +x + + tensor_65 +shape + + +var_1034+ +var_1035 +  + +� + +@* +name + + " +op_1035q +const +var_1036_perm_0 + + +*$ +name + +" +op_1036_perm_0*! +val + + + + + +k +const +var_1038 + + +* +name + + " +op_1038*) +val" + + + + + ���������@� + transpose +x + + +var_1035 +perm + +var_1036_perm_0. + transpose_7 +  + + +� +@*! +name + +" + transpose_7� +reshape +x + + transpose_7 +shape + + +var_1038, +query_states_21 +  + +� +@*% +name + +" +query_states_21k +const +var_1040 + + +* +name + + " +op_1040*) +val" + + + + + ���������@� + transpose +x + + +var_1019 +perm + +var_1020_perm_0. + transpose_9 +  + + +� +@*! +name + +" + transpose_9� +reshape +x + + transpose_9 +shape + + +var_1040* + key_states_43 +  + +� +@*# +name + +" + key_states_43k +const +var_1042 + + +* +name + + " +op_1042*) +val" + + + + + ���������@� + transpose +x + + +var_1026 +perm + +var_1027_perm_0. + transpose_8 +  + + +� +@*! +name + +" + transpose_8� +reshape +x + + transpose_8 +shape + + +var_1042, +value_states_43 +  + +� +@*% +name + +" +value_states_43p +const +var_1045_perm_0 + + +*$ +name + +" +op_1045_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_1045_perm_0( + transpose_6 +  + +@ +�*! +name + +" + transpose_6� +matmul +x + +query_states_21 +y + + transpose_60 + transpose_x! + +attn_weights_61_transpose_x_00 + transpose_y! + +attn_weights_61_transpose_y_0- +attn_weights_61 +  + +� +�*% +name + +" +attn_weights_61e +const +var_1047 + + +* +name + + " +op_1047*# +val + + + + + +��� +reshape +x + +attn_weights_61 +shape + + +var_1047, +var_1048 +  + + +� +�* +name + + " +op_1048� +add +x + + +var_1048 +y + +attention_mask_13 +attn_weights_63 +  + + +� +�*% +name + +" +attn_weights_63d +const +var_1053 + + +* +name + + " +op_1053*" +val + + + +  +��� +reshape +x + +attn_weights_63 +shape + + +var_1053' + input_227 +  + +� +�* +name + + " + input_227{ +softmax +x + + input_227 +axis + + +var_10' + input_229 +  + +� +�* +name + + " + input_229y +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_229 +y + +value_states_43/ + transpose_x + +attn_output_61_transpose_x_0/ + transpose_y + +attn_output_61_transpose_y_0+ +attn_output_61 +  + +� +@*$ +name + +" +attn_output_61d +const +var_1058 + + +* +name + + " +op_1058*" +val + + + +  +�@� +reshape +x + +attn_output_61 +shape + + +var_10581 +attn_output_63 +  + + +� +@*$ +name + +" +attn_output_63~ +const% +attn_output_65_perm_0 + + +*+ +name# + +" +attn_output_65_perm_0*! +val + + + + + +d +const +var_1061 + + +* +name + + " +op_1061*" +val + + + +  +��� + transpose +x + +attn_output_63! +perm + +attn_output_65_perm_0. + transpose_5 +  + +� + +@*! +name + +" + transpose_5� +reshape +x + + transpose_5 +shape + + +var_1061' + input_231 +  + +� +�* +name + + " + input_231� +linear +x + + input_231E +weight; +9 +7model_model_decoder_layers_10_self_attn_out_proj_weightA +bias9 +7 +5model_model_decoder_layers_10_self_attn_out_proj_bias' + linear_63 +  + +� +�* +name + + " + linear_63~ +add +x + +hidden_states_61 +y + + linear_63' + input_235 +  + +� +�* +name + + " + input_235z +const +input_237_axes_0 + + +*& +name + +" +input_237_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_235 +axes + +input_237_axes_0F +gamma= +; +9model_model_decoder_layers_10_self_attn_layer_norm_weightC +beta; +9 +7model_model_decoder_layers_10_self_attn_layer_norm_bias +epsilon + + +var_19' + input_237 +  + +� +�* +name + + " + input_237� +linear +x + + input_2376 +weight, +* +(model_model_decoder_layers_10_fc1_weight2 +bias* +( +&model_model_decoder_layers_10_fc1_bias' + linear_64 +  + +� +� * +name + + " + linear_64e +const +input_241_mode_0 +*& +name + +" +input_241_mode_0* +val + + " +EXACT� +gelu +x + + linear_64 +mode + +input_241_mode_0' + input_241 +  + +� +� * +name + + " + input_241� +linear +x + + input_2416 +weight, +* +(model_model_decoder_layers_10_fc2_weight2 +bias* +( +&model_model_decoder_layers_10_fc2_bias' + linear_65 +  + +� +�* +name + + " + linear_65w +add +x + + input_237 +y + + linear_65' + input_247 +  + +� +�* +name + + " + input_247� +const' +hidden_states_67_axes_0 + + +*- +name% + +" +hidden_states_67_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_247# +axes + +hidden_states_67_axes_0B +gamma9 +7 +5model_model_decoder_layers_10_final_layer_norm_weight? +beta7 +5 +3model_model_decoder_layers_10_final_layer_norm_bias +epsilon + + +var_19. +hidden_states_67 +  + +� +�*& +name + +" +hidden_states_67� +linear +x + +hidden_states_67C +weight9 +7 +5model_model_decoder_layers_11_self_attn_q_proj_weight? +bias7 +5 +3model_model_decoder_layers_11_self_attn_q_proj_bias' + linear_66 +  + +� +�* +name + + " + linear_66S +const +var_1101 + * +name + + " +op_1101* +val +  + + + +>� +mul +x + + linear_66 +y + + +var_1101/ +tensor_workaround +  + +� +�*' +name + +" +tensor_workaround� +linear +x + +hidden_states_67C +weight9 +7 +5model_model_decoder_layers_11_self_attn_k_proj_weight? +bias7 +5 +3model_model_decoder_layers_11_self_attn_k_proj_bias' + linear_67 +  + +� +�* +name + + " + linear_67l +const +var_1106 + + +* +name + + " +op_1106** +val# + + + + + ���������@� +reshape +x + + linear_67 +shape + + +var_1106+ +var_1107 +  + +� + +@* +name + + " +op_1107q +const +var_1108_perm_0 + + +*$ +name + +" +op_1108_perm_0*! +val + + + + + +� +linear +x + +hidden_states_67C +weight9 +7 +5model_model_decoder_layers_11_self_attn_v_proj_weight? +bias7 +5 +3model_model_decoder_layers_11_self_attn_v_proj_bias' + linear_68 +  + +� +�* +name + + " + linear_68l +const +var_1113 + + +* +name + + " +op_1113** +val# + + + + + ���������@� +reshape +x + + linear_68 +shape + + +var_1113+ +var_1114 +  + +� + +@* +name + + " +op_1114q +const +var_1115_perm_0 + + +*$ +name + +" +op_1115_perm_0*! +val + + + + + +d +const +var_1122 + + +* +name + + " +op_1122*" +val + + + +  +�@� +reshape +x + +tensor_workaround +shape + + +var_1122+ +var_1123 +  + +� + +@* +name + + " +op_1123q +const +var_1124_perm_0 + + +*$ +name + +" +op_1124_perm_0*! +val + + + + + +k +const +var_1126 + + +* +name + + " +op_1126*) +val" + + + + + ���������@� + transpose +x + + +var_1123 +perm + +var_1124_perm_0. + transpose_2 +  + + +� +@*! +name + +" + transpose_2� +reshape +x + + transpose_2 +shape + + +var_1126) + query_states +  + +� +@*" +name + +" + query_statesk +const +var_1128 + + +* +name + + " +op_1128*) +val" + + + + + ���������@� + transpose +x + + +var_1107 +perm + +var_1108_perm_0. + transpose_4 +  + + +� +@*! +name + +" + transpose_4� +reshape +x + + transpose_4 +shape + + +var_1128' + +key_states +  + +� +@* +name + +" + +key_statesk +const +var_1130 + + +* +name + + " +op_1130*) +val" + + + + + ���������@� + transpose +x + + +var_1114 +perm + +var_1115_perm_0. + transpose_3 +  + + +� +@*! +name + +" + transpose_3� +reshape +x + + transpose_3 +shape + + +var_1130) + value_states +  + +� +@*" +name + +" + value_statesp +const +var_1133_perm_0 + + +*$ +name + +" +op_1133_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_1133_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_1135 + + +* +name + + " +op_1135*# +val + + + + + +��� +reshape +x + +attn_weights_67 +shape + + +var_1135, +var_1136 +  + + +� +�* +name + + " +op_1136� +add +x + + +var_1136 +y + +attention_mask_13 +attn_weights_69 +  + + +� +�*% +name + +" +attn_weights_69d +const +var_1141 + + +* +name + + " +op_1141*" +val + + + +  +��� +reshape +x + +attn_weights_69 +shape + + +var_1141' + input_249 +  + +� +�* +name + + " + input_249{ +softmax +x + + input_249 +axis + + +var_10' + input_251 +  + +� +�* +name + + " + input_251y +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_251 +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_1146 + + +* +name + + " +op_1146*" +val + + + +  +�@� +reshape +x + +attn_output_67 +shape + + +var_11461 +attn_output_69 +  + + +� +@*$ +name + +" +attn_output_69x +const" +attn_output_perm_0 + + +*( +name  + +" +attn_output_perm_0*! +val + + + + + +d +const +var_1149 + + +* +name + + " +op_1149*" +val + + + +  +��� + transpose +x + +attn_output_69 +perm + +attn_output_perm_0. + transpose_0 +  + +� + +@*! +name + +" + transpose_0� +reshape +x + + transpose_0 +shape + + +var_1149' + input_253 +  + +� +�* +name + + " + input_253� +linear +x + + input_253E +weight; +9 +7model_model_decoder_layers_11_self_attn_out_proj_weightA +bias9 +7 +5model_model_decoder_layers_11_self_attn_out_proj_bias' + linear_69 +  + +� +�* +name + + " + linear_69~ +add +x + +hidden_states_67 +y + + linear_69' + input_257 +  + +� +�* +name + + " + input_257z +const +input_259_axes_0 + + +*& +name + +" +input_259_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_257 +axes + +input_259_axes_0F +gamma= +; +9model_model_decoder_layers_11_self_attn_layer_norm_weightC +beta; +9 +7model_model_decoder_layers_11_self_attn_layer_norm_bias +epsilon + + +var_19' + input_259 +  + +� +�* +name + + " + input_259� +linear +x + + input_2596 +weight, +* +(model_model_decoder_layers_11_fc1_weight2 +bias* +( +&model_model_decoder_layers_11_fc1_bias' + linear_70 +  + +� +� * +name + + " + linear_70e +const +input_263_mode_0 +*& +name + +" +input_263_mode_0* +val + + " +EXACT� +gelu +x + + linear_70 +mode + +input_263_mode_0' + input_263 +  + +� +� * +name + + " + input_263� +linear +x + + input_2636 +weight, +* +(model_model_decoder_layers_11_fc2_weight2 +bias* +( +&model_model_decoder_layers_11_fc2_bias' + linear_71 +  + +� +�* +name + + " + linear_71w +add +x + + input_259 +y + + linear_71' + input_269 +  + +� +�* +name + + " + input_269r +const + input_axes_0 + + +*" +name + +" + input_axes_0*' +val  + + + + + +���������� + +layer_norm +x + + input_269 +axes + + input_axes_0B +gamma9 +7 +5model_model_decoder_layers_11_final_layer_norm_weight? +beta7 +5 +3model_model_decoder_layers_11_final_layer_norm_bias +epsilon + + +var_19# +input +  + +� +�* +name + + " +input� +const" +linear_72_bias_0 +   +و*& +name + +" +linear_72_bias_0*? +val8 +   +و*& +@model_path/weights/weight.bin���� +linear +x + +input5 +weight+ +) +'model_model_decoder_embed_tokens_weight +bias + +linear_72_bias_0% +logits +  + +� +و* +name + + " + linear_72"� + buildInfo�" + + +|"z +6 +! + +" +coremltools-version + +" +7.0 +@ +) +! +" +coremltools-component-torch + + " +2.0.1 \ No newline at end of file