7767517 6974 8577 Input attention_mask 0 1 attention_mask Input position_ids 0 1 position_ids Split splitncnn_input2 1 2 position_ids position_ids_splitncnn_0 position_ids_splitncnn_1 Input past_key_values.0.key 0 1 past_key_values.0.key Split splitncnn_input3 1 2 past_key_values.0.key past_key_values.0.key_splitncnn_0 past_key_values.0.key_splitncnn_1 Input past_key_values.0.value 0 1 past_key_values.0.value Input past_key_values.1.key 0 1 past_key_values.1.key Input past_key_values.1.value 0 1 past_key_values.1.value Input past_key_values.2.key 0 1 past_key_values.2.key Input past_key_values.2.value 0 1 past_key_values.2.value Input past_key_values.3.key 0 1 past_key_values.3.key Input past_key_values.3.value 0 1 past_key_values.3.value Input past_key_values.4.key 0 1 past_key_values.4.key Input past_key_values.4.value 0 1 past_key_values.4.value Input past_key_values.5.key 0 1 past_key_values.5.key Input past_key_values.5.value 0 1 past_key_values.5.value Input past_key_values.6.key 0 1 past_key_values.6.key Input past_key_values.6.value 0 1 past_key_values.6.value Input past_key_values.7.key 0 1 past_key_values.7.key Input past_key_values.7.value 0 1 past_key_values.7.value Input past_key_values.8.key 0 1 past_key_values.8.key Input past_key_values.8.value 0 1 past_key_values.8.value Input past_key_values.9.key 0 1 past_key_values.9.key Input past_key_values.9.value 0 1 past_key_values.9.value Input past_key_values.10.key 0 1 past_key_values.10.key Input past_key_values.10.value 0 1 past_key_values.10.value Input past_key_values.11.key 0 1 past_key_values.11.key Input past_key_values.11.value 0 1 past_key_values.11.value Input past_key_values.12.key 0 1 past_key_values.12.key Input past_key_values.12.value 0 1 past_key_values.12.value Input past_key_values.13.key 0 1 past_key_values.13.key Input past_key_values.13.value 0 1 past_key_values.13.value Input past_key_values.14.key 0 1 past_key_values.14.key Input past_key_values.14.value 0 1 past_key_values.14.value Input past_key_values.15.key 0 1 past_key_values.15.key Input past_key_values.15.value 0 1 past_key_values.15.value Input past_key_values.16.key 0 1 past_key_values.16.key Input past_key_values.16.value 0 1 past_key_values.16.value Input past_key_values.17.key 0 1 past_key_values.17.key Input past_key_values.17.value 0 1 past_key_values.17.value Input past_key_values.18.key 0 1 past_key_values.18.key Input past_key_values.18.value 0 1 past_key_values.18.value Input past_key_values.19.key 0 1 past_key_values.19.key Input past_key_values.19.value 0 1 past_key_values.19.value Input past_key_values.20.key 0 1 past_key_values.20.key Input past_key_values.20.value 0 1 past_key_values.20.value Input past_key_values.21.key 0 1 past_key_values.21.key Input past_key_values.21.value 0 1 past_key_values.21.value Input past_key_values.22.key 0 1 past_key_values.22.key Input past_key_values.22.value 0 1 past_key_values.22.value Input past_key_values.23.key 0 1 past_key_values.23.key Input past_key_values.23.value 0 1 past_key_values.23.value MemoryData /model/Add_2_output_0 0 1 /model/Add_2_output_0 0=0 Split splitncnn_0 1 2 /model/Add_2_output_0 /model/Add_2_output_0_splitncnn_0 /model/Add_2_output_0_splitncnn_1 MemoryData /model/Constant_10_output_0 0 1 /model/Constant_10_output_0 0=1 MemoryData /model/Constant_11_output_0 0 1 /model/Constant_11_output_0 0=1 MemoryData /model/Constant_12_output_0 0 1 /model/Constant_12_output_0 0=1 MemoryData /model/Constant_13_output_0 0 1 /model/Constant_13_output_0 0=2 MemoryData /model/Constant_14_output_0 0 1 /model/Constant_14_output_0 0=2 MemoryData /model/Constant_15_output_0 0 1 /model/Constant_15_output_0 0=1 MemoryData /model/Constant_16_output_0 0 1 /model/Constant_16_output_0 0=1 MemoryData /model/Constant_17_output_0 0 1 /model/Constant_17_output_0 0=1 MemoryData /model/Constant_18_output_0 0 1 /model/Constant_18_output_0 0=0 MemoryData /model/Constant_21_output_0 0 1 /model/Constant_21_output_0 0=1 MemoryData /model/Constant_22_output_0 0 1 /model/Constant_22_output_0 0=1 MemoryData /model/Constant_23_output_0 0 1 /model/Constant_23_output_0 0=1 MemoryData /model/Constant_24_output_0 0 1 /model/Constant_24_output_0 0=1 MemoryData /model/Constant_26_output_0 0 1 /model/Constant_26_output_0 0=2 MemoryData /model/Constant_2_output_0 0 1 /model/Constant_2_output_0 0=2 MemoryData /model/Constant_5_output_0 0 1 /model/Constant_5_output_0 0=1 MemoryData /model/Constant_6_output_0 0 1 /model/Constant_6_output_0 0=1 MemoryData /model/Constant_7_output_0 0 1 /model/Constant_7_output_0 0=1 MemoryData /model/Constant_8_output_0 0 1 /model/Constant_8_output_0 0=2 MemoryData /model/Constant_9_output_0 0 1 /model/Constant_9_output_0 0=2 MemoryData /model/layers.0/input_layernorm/Constant_1_output_0 0 1 /model/layers.0/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.0/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.0/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_11_output_0 0 1 /model/layers.0/self_attn/Constant_11_output_0 0=2 MemoryData /model/layers.0/self_attn/Constant_14_output_0 0 1 /model/layers.0/self_attn/Constant_14_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_17_output_0 0 1 /model/layers.0/self_attn/Constant_17_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_21_output_0 0 1 /model/layers.0/self_attn/Constant_21_output_0 0=2 MemoryData /model/layers.0/self_attn/Constant_24_output_0 0 1 /model/layers.0/self_attn/Constant_24_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_27_output_0 0 1 /model/layers.0/self_attn/Constant_27_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_2_output_0 0 1 /model/layers.0/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_34_output_0 0 1 /model/layers.0/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_35_output_0 0 1 /model/layers.0/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_36_output_0 0 1 /model/layers.0/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_37_output_0 0 1 /model/layers.0/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_38_output_0 0 1 /model/layers.0/self_attn/Constant_38_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_39_output_0 0 1 /model/layers.0/self_attn/Constant_39_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_3_output_0 0 1 /model/layers.0/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_41_output_0 0 1 /model/layers.0/self_attn/Constant_41_output_0 0=2 MemoryData /model/layers.0/self_attn/Constant_42_output_0 0 1 /model/layers.0/self_attn/Constant_42_output_0 0=2 MemoryData /model/layers.0/self_attn/Constant_47_output_0 0 1 /model/layers.0/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_48_output_0 0 1 /model/layers.0/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_49_output_0 0 1 /model/layers.0/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_4_output_0 0 1 /model/layers.0/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_50_output_0 0 1 /model/layers.0/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_51_output_0 0 1 /model/layers.0/self_attn/Constant_51_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_52_output_0 0 1 /model/layers.0/self_attn/Constant_52_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_54_output_0 0 1 /model/layers.0/self_attn/Constant_54_output_0 0=2 MemoryData /model/layers.0/self_attn/Constant_55_output_0 0 1 /model/layers.0/self_attn/Constant_55_output_0 0=2 MemoryData /model/layers.0/self_attn/Constant_59_output_0 0 1 /model/layers.0/self_attn/Constant_59_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_5_output_0 0 1 /model/layers.0/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_63_output_0 0 1 /model/layers.0/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_64_output_0 0 1 /model/layers.0/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_65_output_0 0 1 /model/layers.0/self_attn/Constant_65_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_66_output_0 0 1 /model/layers.0/self_attn/Constant_66_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_6_output_0 0 1 /model/layers.0/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_7_output_0 0 1 /model/layers.0/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_8_output_0 0 1 /model/layers.0/self_attn/Constant_8_output_0 0=1 MemoryData /model/layers.0/self_attn/Constant_9_output_0 0 1 /model/layers.0/self_attn/Constant_9_output_0 0=1 MemoryData /model/layers.1/input_layernorm/Constant_1_output_0 0 1 /model/layers.1/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.1/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.1/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_12_output_0 0 1 /model/layers.1/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_15_output_0 0 1 /model/layers.1/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_19_output_0 0 1 /model/layers.1/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.1/self_attn/Constant_22_output_0 0 1 /model/layers.1/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_25_output_0 0 1 /model/layers.1/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_2_output_0 0 1 /model/layers.1/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_32_output_0 0 1 /model/layers.1/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_33_output_0 0 1 /model/layers.1/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_34_output_0 0 1 /model/layers.1/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_35_output_0 0 1 /model/layers.1/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_36_output_0 0 1 /model/layers.1/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_37_output_0 0 1 /model/layers.1/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_39_output_0 0 1 /model/layers.1/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.1/self_attn/Constant_3_output_0 0 1 /model/layers.1/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_40_output_0 0 1 /model/layers.1/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.1/self_attn/Constant_45_output_0 0 1 /model/layers.1/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_46_output_0 0 1 /model/layers.1/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_47_output_0 0 1 /model/layers.1/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_48_output_0 0 1 /model/layers.1/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_49_output_0 0 1 /model/layers.1/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_4_output_0 0 1 /model/layers.1/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_50_output_0 0 1 /model/layers.1/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_52_output_0 0 1 /model/layers.1/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.1/self_attn/Constant_53_output_0 0 1 /model/layers.1/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.1/self_attn/Constant_57_output_0 0 1 /model/layers.1/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_5_output_0 0 1 /model/layers.1/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_61_output_0 0 1 /model/layers.1/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_62_output_0 0 1 /model/layers.1/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_63_output_0 0 1 /model/layers.1/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_64_output_0 0 1 /model/layers.1/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_6_output_0 0 1 /model/layers.1/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_7_output_0 0 1 /model/layers.1/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.1/self_attn/Constant_9_output_0 0 1 /model/layers.1/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.10/input_layernorm/Constant_1_output_0 0 1 /model/layers.10/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.10/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.10/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_12_output_0 0 1 /model/layers.10/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_15_output_0 0 1 /model/layers.10/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_19_output_0 0 1 /model/layers.10/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.10/self_attn/Constant_22_output_0 0 1 /model/layers.10/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_25_output_0 0 1 /model/layers.10/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_2_output_0 0 1 /model/layers.10/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_32_output_0 0 1 /model/layers.10/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_33_output_0 0 1 /model/layers.10/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_34_output_0 0 1 /model/layers.10/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_35_output_0 0 1 /model/layers.10/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_36_output_0 0 1 /model/layers.10/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_37_output_0 0 1 /model/layers.10/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_39_output_0 0 1 /model/layers.10/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.10/self_attn/Constant_3_output_0 0 1 /model/layers.10/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_40_output_0 0 1 /model/layers.10/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.10/self_attn/Constant_45_output_0 0 1 /model/layers.10/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_46_output_0 0 1 /model/layers.10/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_47_output_0 0 1 /model/layers.10/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_48_output_0 0 1 /model/layers.10/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_49_output_0 0 1 /model/layers.10/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_4_output_0 0 1 /model/layers.10/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_50_output_0 0 1 /model/layers.10/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_52_output_0 0 1 /model/layers.10/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.10/self_attn/Constant_53_output_0 0 1 /model/layers.10/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.10/self_attn/Constant_57_output_0 0 1 /model/layers.10/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_5_output_0 0 1 /model/layers.10/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_61_output_0 0 1 /model/layers.10/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_62_output_0 0 1 /model/layers.10/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_63_output_0 0 1 /model/layers.10/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_64_output_0 0 1 /model/layers.10/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_6_output_0 0 1 /model/layers.10/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_7_output_0 0 1 /model/layers.10/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.10/self_attn/Constant_9_output_0 0 1 /model/layers.10/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.11/input_layernorm/Constant_1_output_0 0 1 /model/layers.11/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.11/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.11/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_12_output_0 0 1 /model/layers.11/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_15_output_0 0 1 /model/layers.11/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_19_output_0 0 1 /model/layers.11/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.11/self_attn/Constant_22_output_0 0 1 /model/layers.11/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_25_output_0 0 1 /model/layers.11/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_2_output_0 0 1 /model/layers.11/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_32_output_0 0 1 /model/layers.11/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_33_output_0 0 1 /model/layers.11/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_34_output_0 0 1 /model/layers.11/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_35_output_0 0 1 /model/layers.11/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_36_output_0 0 1 /model/layers.11/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_37_output_0 0 1 /model/layers.11/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_39_output_0 0 1 /model/layers.11/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.11/self_attn/Constant_3_output_0 0 1 /model/layers.11/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_40_output_0 0 1 /model/layers.11/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.11/self_attn/Constant_45_output_0 0 1 /model/layers.11/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_46_output_0 0 1 /model/layers.11/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_47_output_0 0 1 /model/layers.11/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_48_output_0 0 1 /model/layers.11/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_49_output_0 0 1 /model/layers.11/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_4_output_0 0 1 /model/layers.11/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_50_output_0 0 1 /model/layers.11/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_52_output_0 0 1 /model/layers.11/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.11/self_attn/Constant_53_output_0 0 1 /model/layers.11/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.11/self_attn/Constant_57_output_0 0 1 /model/layers.11/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_5_output_0 0 1 /model/layers.11/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_61_output_0 0 1 /model/layers.11/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_62_output_0 0 1 /model/layers.11/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_63_output_0 0 1 /model/layers.11/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_64_output_0 0 1 /model/layers.11/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_6_output_0 0 1 /model/layers.11/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_7_output_0 0 1 /model/layers.11/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.11/self_attn/Constant_9_output_0 0 1 /model/layers.11/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.12/input_layernorm/Constant_1_output_0 0 1 /model/layers.12/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.12/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.12/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_12_output_0 0 1 /model/layers.12/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_15_output_0 0 1 /model/layers.12/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_19_output_0 0 1 /model/layers.12/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.12/self_attn/Constant_22_output_0 0 1 /model/layers.12/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_25_output_0 0 1 /model/layers.12/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_2_output_0 0 1 /model/layers.12/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_32_output_0 0 1 /model/layers.12/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_33_output_0 0 1 /model/layers.12/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_34_output_0 0 1 /model/layers.12/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_35_output_0 0 1 /model/layers.12/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_36_output_0 0 1 /model/layers.12/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_37_output_0 0 1 /model/layers.12/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_39_output_0 0 1 /model/layers.12/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.12/self_attn/Constant_3_output_0 0 1 /model/layers.12/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_40_output_0 0 1 /model/layers.12/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.12/self_attn/Constant_45_output_0 0 1 /model/layers.12/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_46_output_0 0 1 /model/layers.12/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_47_output_0 0 1 /model/layers.12/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_48_output_0 0 1 /model/layers.12/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_49_output_0 0 1 /model/layers.12/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_4_output_0 0 1 /model/layers.12/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_50_output_0 0 1 /model/layers.12/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_52_output_0 0 1 /model/layers.12/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.12/self_attn/Constant_53_output_0 0 1 /model/layers.12/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.12/self_attn/Constant_57_output_0 0 1 /model/layers.12/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_5_output_0 0 1 /model/layers.12/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_61_output_0 0 1 /model/layers.12/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_62_output_0 0 1 /model/layers.12/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_63_output_0 0 1 /model/layers.12/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_64_output_0 0 1 /model/layers.12/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_6_output_0 0 1 /model/layers.12/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_7_output_0 0 1 /model/layers.12/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.12/self_attn/Constant_9_output_0 0 1 /model/layers.12/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.13/input_layernorm/Constant_1_output_0 0 1 /model/layers.13/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.13/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.13/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_12_output_0 0 1 /model/layers.13/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_15_output_0 0 1 /model/layers.13/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_19_output_0 0 1 /model/layers.13/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.13/self_attn/Constant_22_output_0 0 1 /model/layers.13/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_25_output_0 0 1 /model/layers.13/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_2_output_0 0 1 /model/layers.13/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_32_output_0 0 1 /model/layers.13/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_33_output_0 0 1 /model/layers.13/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_34_output_0 0 1 /model/layers.13/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_35_output_0 0 1 /model/layers.13/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_36_output_0 0 1 /model/layers.13/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_37_output_0 0 1 /model/layers.13/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_39_output_0 0 1 /model/layers.13/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.13/self_attn/Constant_3_output_0 0 1 /model/layers.13/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_40_output_0 0 1 /model/layers.13/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.13/self_attn/Constant_45_output_0 0 1 /model/layers.13/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_46_output_0 0 1 /model/layers.13/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_47_output_0 0 1 /model/layers.13/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_48_output_0 0 1 /model/layers.13/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_49_output_0 0 1 /model/layers.13/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_4_output_0 0 1 /model/layers.13/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_50_output_0 0 1 /model/layers.13/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_52_output_0 0 1 /model/layers.13/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.13/self_attn/Constant_53_output_0 0 1 /model/layers.13/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.13/self_attn/Constant_57_output_0 0 1 /model/layers.13/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_5_output_0 0 1 /model/layers.13/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_61_output_0 0 1 /model/layers.13/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_62_output_0 0 1 /model/layers.13/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_63_output_0 0 1 /model/layers.13/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_64_output_0 0 1 /model/layers.13/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_6_output_0 0 1 /model/layers.13/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_7_output_0 0 1 /model/layers.13/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.13/self_attn/Constant_9_output_0 0 1 /model/layers.13/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.14/input_layernorm/Constant_1_output_0 0 1 /model/layers.14/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.14/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.14/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_12_output_0 0 1 /model/layers.14/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_15_output_0 0 1 /model/layers.14/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_19_output_0 0 1 /model/layers.14/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.14/self_attn/Constant_22_output_0 0 1 /model/layers.14/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_25_output_0 0 1 /model/layers.14/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_2_output_0 0 1 /model/layers.14/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_32_output_0 0 1 /model/layers.14/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_33_output_0 0 1 /model/layers.14/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_34_output_0 0 1 /model/layers.14/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_35_output_0 0 1 /model/layers.14/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_36_output_0 0 1 /model/layers.14/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_37_output_0 0 1 /model/layers.14/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_39_output_0 0 1 /model/layers.14/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.14/self_attn/Constant_3_output_0 0 1 /model/layers.14/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_40_output_0 0 1 /model/layers.14/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.14/self_attn/Constant_45_output_0 0 1 /model/layers.14/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_46_output_0 0 1 /model/layers.14/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_47_output_0 0 1 /model/layers.14/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_48_output_0 0 1 /model/layers.14/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_49_output_0 0 1 /model/layers.14/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_4_output_0 0 1 /model/layers.14/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_50_output_0 0 1 /model/layers.14/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_52_output_0 0 1 /model/layers.14/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.14/self_attn/Constant_53_output_0 0 1 /model/layers.14/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.14/self_attn/Constant_57_output_0 0 1 /model/layers.14/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_5_output_0 0 1 /model/layers.14/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_61_output_0 0 1 /model/layers.14/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_62_output_0 0 1 /model/layers.14/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_63_output_0 0 1 /model/layers.14/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_64_output_0 0 1 /model/layers.14/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_6_output_0 0 1 /model/layers.14/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_7_output_0 0 1 /model/layers.14/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.14/self_attn/Constant_9_output_0 0 1 /model/layers.14/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.15/input_layernorm/Constant_1_output_0 0 1 /model/layers.15/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.15/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.15/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_12_output_0 0 1 /model/layers.15/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_15_output_0 0 1 /model/layers.15/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_19_output_0 0 1 /model/layers.15/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.15/self_attn/Constant_22_output_0 0 1 /model/layers.15/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_25_output_0 0 1 /model/layers.15/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_2_output_0 0 1 /model/layers.15/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_32_output_0 0 1 /model/layers.15/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_33_output_0 0 1 /model/layers.15/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_34_output_0 0 1 /model/layers.15/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_35_output_0 0 1 /model/layers.15/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_36_output_0 0 1 /model/layers.15/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_37_output_0 0 1 /model/layers.15/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_39_output_0 0 1 /model/layers.15/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.15/self_attn/Constant_3_output_0 0 1 /model/layers.15/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_40_output_0 0 1 /model/layers.15/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.15/self_attn/Constant_45_output_0 0 1 /model/layers.15/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_46_output_0 0 1 /model/layers.15/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_47_output_0 0 1 /model/layers.15/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_48_output_0 0 1 /model/layers.15/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_49_output_0 0 1 /model/layers.15/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_4_output_0 0 1 /model/layers.15/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_50_output_0 0 1 /model/layers.15/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_52_output_0 0 1 /model/layers.15/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.15/self_attn/Constant_53_output_0 0 1 /model/layers.15/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.15/self_attn/Constant_57_output_0 0 1 /model/layers.15/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_5_output_0 0 1 /model/layers.15/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_61_output_0 0 1 /model/layers.15/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_62_output_0 0 1 /model/layers.15/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_63_output_0 0 1 /model/layers.15/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_64_output_0 0 1 /model/layers.15/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_6_output_0 0 1 /model/layers.15/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_7_output_0 0 1 /model/layers.15/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.15/self_attn/Constant_9_output_0 0 1 /model/layers.15/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.16/input_layernorm/Constant_1_output_0 0 1 /model/layers.16/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.16/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.16/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_12_output_0 0 1 /model/layers.16/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_15_output_0 0 1 /model/layers.16/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_19_output_0 0 1 /model/layers.16/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.16/self_attn/Constant_22_output_0 0 1 /model/layers.16/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_25_output_0 0 1 /model/layers.16/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_2_output_0 0 1 /model/layers.16/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_32_output_0 0 1 /model/layers.16/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_33_output_0 0 1 /model/layers.16/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_34_output_0 0 1 /model/layers.16/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_35_output_0 0 1 /model/layers.16/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_36_output_0 0 1 /model/layers.16/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_37_output_0 0 1 /model/layers.16/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_39_output_0 0 1 /model/layers.16/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.16/self_attn/Constant_3_output_0 0 1 /model/layers.16/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_40_output_0 0 1 /model/layers.16/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.16/self_attn/Constant_45_output_0 0 1 /model/layers.16/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_46_output_0 0 1 /model/layers.16/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_47_output_0 0 1 /model/layers.16/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_48_output_0 0 1 /model/layers.16/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_49_output_0 0 1 /model/layers.16/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_4_output_0 0 1 /model/layers.16/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_50_output_0 0 1 /model/layers.16/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_52_output_0 0 1 /model/layers.16/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.16/self_attn/Constant_53_output_0 0 1 /model/layers.16/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.16/self_attn/Constant_57_output_0 0 1 /model/layers.16/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_5_output_0 0 1 /model/layers.16/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_61_output_0 0 1 /model/layers.16/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_62_output_0 0 1 /model/layers.16/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_63_output_0 0 1 /model/layers.16/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_64_output_0 0 1 /model/layers.16/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_6_output_0 0 1 /model/layers.16/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_7_output_0 0 1 /model/layers.16/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.16/self_attn/Constant_9_output_0 0 1 /model/layers.16/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.17/input_layernorm/Constant_1_output_0 0 1 /model/layers.17/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.17/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.17/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_12_output_0 0 1 /model/layers.17/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_15_output_0 0 1 /model/layers.17/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_19_output_0 0 1 /model/layers.17/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.17/self_attn/Constant_22_output_0 0 1 /model/layers.17/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_25_output_0 0 1 /model/layers.17/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_2_output_0 0 1 /model/layers.17/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_32_output_0 0 1 /model/layers.17/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_33_output_0 0 1 /model/layers.17/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_34_output_0 0 1 /model/layers.17/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_35_output_0 0 1 /model/layers.17/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_36_output_0 0 1 /model/layers.17/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_37_output_0 0 1 /model/layers.17/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_39_output_0 0 1 /model/layers.17/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.17/self_attn/Constant_3_output_0 0 1 /model/layers.17/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_40_output_0 0 1 /model/layers.17/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.17/self_attn/Constant_45_output_0 0 1 /model/layers.17/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_46_output_0 0 1 /model/layers.17/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_47_output_0 0 1 /model/layers.17/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_48_output_0 0 1 /model/layers.17/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_49_output_0 0 1 /model/layers.17/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_4_output_0 0 1 /model/layers.17/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_50_output_0 0 1 /model/layers.17/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_52_output_0 0 1 /model/layers.17/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.17/self_attn/Constant_53_output_0 0 1 /model/layers.17/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.17/self_attn/Constant_57_output_0 0 1 /model/layers.17/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_5_output_0 0 1 /model/layers.17/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_61_output_0 0 1 /model/layers.17/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_62_output_0 0 1 /model/layers.17/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_63_output_0 0 1 /model/layers.17/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_64_output_0 0 1 /model/layers.17/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_6_output_0 0 1 /model/layers.17/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_7_output_0 0 1 /model/layers.17/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.17/self_attn/Constant_9_output_0 0 1 /model/layers.17/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.18/input_layernorm/Constant_1_output_0 0 1 /model/layers.18/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.18/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.18/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_12_output_0 0 1 /model/layers.18/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_15_output_0 0 1 /model/layers.18/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_19_output_0 0 1 /model/layers.18/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.18/self_attn/Constant_22_output_0 0 1 /model/layers.18/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_25_output_0 0 1 /model/layers.18/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_2_output_0 0 1 /model/layers.18/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_32_output_0 0 1 /model/layers.18/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_33_output_0 0 1 /model/layers.18/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_34_output_0 0 1 /model/layers.18/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_35_output_0 0 1 /model/layers.18/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_36_output_0 0 1 /model/layers.18/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_37_output_0 0 1 /model/layers.18/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_39_output_0 0 1 /model/layers.18/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.18/self_attn/Constant_3_output_0 0 1 /model/layers.18/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_40_output_0 0 1 /model/layers.18/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.18/self_attn/Constant_45_output_0 0 1 /model/layers.18/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_46_output_0 0 1 /model/layers.18/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_47_output_0 0 1 /model/layers.18/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_48_output_0 0 1 /model/layers.18/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_49_output_0 0 1 /model/layers.18/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_4_output_0 0 1 /model/layers.18/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_50_output_0 0 1 /model/layers.18/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_52_output_0 0 1 /model/layers.18/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.18/self_attn/Constant_53_output_0 0 1 /model/layers.18/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.18/self_attn/Constant_57_output_0 0 1 /model/layers.18/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_5_output_0 0 1 /model/layers.18/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_61_output_0 0 1 /model/layers.18/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_62_output_0 0 1 /model/layers.18/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_63_output_0 0 1 /model/layers.18/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_64_output_0 0 1 /model/layers.18/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_6_output_0 0 1 /model/layers.18/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_7_output_0 0 1 /model/layers.18/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.18/self_attn/Constant_9_output_0 0 1 /model/layers.18/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.19/input_layernorm/Constant_1_output_0 0 1 /model/layers.19/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.19/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.19/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_12_output_0 0 1 /model/layers.19/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_15_output_0 0 1 /model/layers.19/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_19_output_0 0 1 /model/layers.19/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.19/self_attn/Constant_22_output_0 0 1 /model/layers.19/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_25_output_0 0 1 /model/layers.19/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_2_output_0 0 1 /model/layers.19/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_32_output_0 0 1 /model/layers.19/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_33_output_0 0 1 /model/layers.19/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_34_output_0 0 1 /model/layers.19/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_35_output_0 0 1 /model/layers.19/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_36_output_0 0 1 /model/layers.19/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_37_output_0 0 1 /model/layers.19/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_39_output_0 0 1 /model/layers.19/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.19/self_attn/Constant_3_output_0 0 1 /model/layers.19/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_40_output_0 0 1 /model/layers.19/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.19/self_attn/Constant_45_output_0 0 1 /model/layers.19/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_46_output_0 0 1 /model/layers.19/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_47_output_0 0 1 /model/layers.19/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_48_output_0 0 1 /model/layers.19/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_49_output_0 0 1 /model/layers.19/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_4_output_0 0 1 /model/layers.19/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_50_output_0 0 1 /model/layers.19/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_52_output_0 0 1 /model/layers.19/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.19/self_attn/Constant_53_output_0 0 1 /model/layers.19/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.19/self_attn/Constant_57_output_0 0 1 /model/layers.19/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_5_output_0 0 1 /model/layers.19/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_61_output_0 0 1 /model/layers.19/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_62_output_0 0 1 /model/layers.19/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_63_output_0 0 1 /model/layers.19/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_64_output_0 0 1 /model/layers.19/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_6_output_0 0 1 /model/layers.19/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_7_output_0 0 1 /model/layers.19/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.19/self_attn/Constant_9_output_0 0 1 /model/layers.19/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.2/input_layernorm/Constant_1_output_0 0 1 /model/layers.2/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.2/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.2/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_12_output_0 0 1 /model/layers.2/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_15_output_0 0 1 /model/layers.2/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_19_output_0 0 1 /model/layers.2/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.2/self_attn/Constant_22_output_0 0 1 /model/layers.2/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_25_output_0 0 1 /model/layers.2/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_2_output_0 0 1 /model/layers.2/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_32_output_0 0 1 /model/layers.2/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_33_output_0 0 1 /model/layers.2/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_34_output_0 0 1 /model/layers.2/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_35_output_0 0 1 /model/layers.2/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_36_output_0 0 1 /model/layers.2/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_37_output_0 0 1 /model/layers.2/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_39_output_0 0 1 /model/layers.2/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.2/self_attn/Constant_3_output_0 0 1 /model/layers.2/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_40_output_0 0 1 /model/layers.2/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.2/self_attn/Constant_45_output_0 0 1 /model/layers.2/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_46_output_0 0 1 /model/layers.2/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_47_output_0 0 1 /model/layers.2/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_48_output_0 0 1 /model/layers.2/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_49_output_0 0 1 /model/layers.2/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_4_output_0 0 1 /model/layers.2/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_50_output_0 0 1 /model/layers.2/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_52_output_0 0 1 /model/layers.2/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.2/self_attn/Constant_53_output_0 0 1 /model/layers.2/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.2/self_attn/Constant_57_output_0 0 1 /model/layers.2/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_5_output_0 0 1 /model/layers.2/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_61_output_0 0 1 /model/layers.2/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_62_output_0 0 1 /model/layers.2/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_63_output_0 0 1 /model/layers.2/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_64_output_0 0 1 /model/layers.2/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_6_output_0 0 1 /model/layers.2/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_7_output_0 0 1 /model/layers.2/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.2/self_attn/Constant_9_output_0 0 1 /model/layers.2/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.20/input_layernorm/Constant_1_output_0 0 1 /model/layers.20/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.20/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.20/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_12_output_0 0 1 /model/layers.20/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_15_output_0 0 1 /model/layers.20/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_19_output_0 0 1 /model/layers.20/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.20/self_attn/Constant_22_output_0 0 1 /model/layers.20/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_25_output_0 0 1 /model/layers.20/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_2_output_0 0 1 /model/layers.20/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_32_output_0 0 1 /model/layers.20/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_33_output_0 0 1 /model/layers.20/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_34_output_0 0 1 /model/layers.20/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_35_output_0 0 1 /model/layers.20/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_36_output_0 0 1 /model/layers.20/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_37_output_0 0 1 /model/layers.20/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_39_output_0 0 1 /model/layers.20/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.20/self_attn/Constant_3_output_0 0 1 /model/layers.20/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_40_output_0 0 1 /model/layers.20/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.20/self_attn/Constant_45_output_0 0 1 /model/layers.20/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_46_output_0 0 1 /model/layers.20/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_47_output_0 0 1 /model/layers.20/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_48_output_0 0 1 /model/layers.20/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_49_output_0 0 1 /model/layers.20/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_4_output_0 0 1 /model/layers.20/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_50_output_0 0 1 /model/layers.20/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_52_output_0 0 1 /model/layers.20/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.20/self_attn/Constant_53_output_0 0 1 /model/layers.20/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.20/self_attn/Constant_57_output_0 0 1 /model/layers.20/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_5_output_0 0 1 /model/layers.20/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_61_output_0 0 1 /model/layers.20/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_62_output_0 0 1 /model/layers.20/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_63_output_0 0 1 /model/layers.20/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_64_output_0 0 1 /model/layers.20/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_6_output_0 0 1 /model/layers.20/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_7_output_0 0 1 /model/layers.20/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.20/self_attn/Constant_9_output_0 0 1 /model/layers.20/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.21/input_layernorm/Constant_1_output_0 0 1 /model/layers.21/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.21/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.21/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_12_output_0 0 1 /model/layers.21/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_15_output_0 0 1 /model/layers.21/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_19_output_0 0 1 /model/layers.21/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.21/self_attn/Constant_22_output_0 0 1 /model/layers.21/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_25_output_0 0 1 /model/layers.21/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_2_output_0 0 1 /model/layers.21/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_32_output_0 0 1 /model/layers.21/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_33_output_0 0 1 /model/layers.21/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_34_output_0 0 1 /model/layers.21/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_35_output_0 0 1 /model/layers.21/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_36_output_0 0 1 /model/layers.21/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_37_output_0 0 1 /model/layers.21/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_39_output_0 0 1 /model/layers.21/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.21/self_attn/Constant_3_output_0 0 1 /model/layers.21/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_40_output_0 0 1 /model/layers.21/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.21/self_attn/Constant_45_output_0 0 1 /model/layers.21/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_46_output_0 0 1 /model/layers.21/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_47_output_0 0 1 /model/layers.21/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_48_output_0 0 1 /model/layers.21/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_49_output_0 0 1 /model/layers.21/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_4_output_0 0 1 /model/layers.21/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_50_output_0 0 1 /model/layers.21/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_52_output_0 0 1 /model/layers.21/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.21/self_attn/Constant_53_output_0 0 1 /model/layers.21/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.21/self_attn/Constant_57_output_0 0 1 /model/layers.21/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_5_output_0 0 1 /model/layers.21/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_61_output_0 0 1 /model/layers.21/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_62_output_0 0 1 /model/layers.21/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_63_output_0 0 1 /model/layers.21/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_64_output_0 0 1 /model/layers.21/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_6_output_0 0 1 /model/layers.21/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_7_output_0 0 1 /model/layers.21/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.21/self_attn/Constant_9_output_0 0 1 /model/layers.21/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.22/input_layernorm/Constant_1_output_0 0 1 /model/layers.22/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.22/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.22/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_12_output_0 0 1 /model/layers.22/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_15_output_0 0 1 /model/layers.22/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_19_output_0 0 1 /model/layers.22/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.22/self_attn/Constant_22_output_0 0 1 /model/layers.22/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_25_output_0 0 1 /model/layers.22/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_2_output_0 0 1 /model/layers.22/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_32_output_0 0 1 /model/layers.22/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_33_output_0 0 1 /model/layers.22/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_34_output_0 0 1 /model/layers.22/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_35_output_0 0 1 /model/layers.22/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_36_output_0 0 1 /model/layers.22/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_37_output_0 0 1 /model/layers.22/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_39_output_0 0 1 /model/layers.22/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.22/self_attn/Constant_3_output_0 0 1 /model/layers.22/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_40_output_0 0 1 /model/layers.22/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.22/self_attn/Constant_45_output_0 0 1 /model/layers.22/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_46_output_0 0 1 /model/layers.22/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_47_output_0 0 1 /model/layers.22/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_48_output_0 0 1 /model/layers.22/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_49_output_0 0 1 /model/layers.22/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_4_output_0 0 1 /model/layers.22/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_50_output_0 0 1 /model/layers.22/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_52_output_0 0 1 /model/layers.22/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.22/self_attn/Constant_53_output_0 0 1 /model/layers.22/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.22/self_attn/Constant_57_output_0 0 1 /model/layers.22/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_5_output_0 0 1 /model/layers.22/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_61_output_0 0 1 /model/layers.22/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_62_output_0 0 1 /model/layers.22/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_63_output_0 0 1 /model/layers.22/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_64_output_0 0 1 /model/layers.22/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_6_output_0 0 1 /model/layers.22/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_7_output_0 0 1 /model/layers.22/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.22/self_attn/Constant_9_output_0 0 1 /model/layers.22/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.23/input_layernorm/Constant_1_output_0 0 1 /model/layers.23/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.23/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.23/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_12_output_0 0 1 /model/layers.23/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_15_output_0 0 1 /model/layers.23/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_19_output_0 0 1 /model/layers.23/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.23/self_attn/Constant_22_output_0 0 1 /model/layers.23/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_25_output_0 0 1 /model/layers.23/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_2_output_0 0 1 /model/layers.23/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_32_output_0 0 1 /model/layers.23/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_33_output_0 0 1 /model/layers.23/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_34_output_0 0 1 /model/layers.23/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_35_output_0 0 1 /model/layers.23/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_36_output_0 0 1 /model/layers.23/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_37_output_0 0 1 /model/layers.23/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_39_output_0 0 1 /model/layers.23/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.23/self_attn/Constant_3_output_0 0 1 /model/layers.23/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_40_output_0 0 1 /model/layers.23/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.23/self_attn/Constant_45_output_0 0 1 /model/layers.23/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_46_output_0 0 1 /model/layers.23/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_47_output_0 0 1 /model/layers.23/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_48_output_0 0 1 /model/layers.23/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_49_output_0 0 1 /model/layers.23/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_4_output_0 0 1 /model/layers.23/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_50_output_0 0 1 /model/layers.23/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_52_output_0 0 1 /model/layers.23/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.23/self_attn/Constant_53_output_0 0 1 /model/layers.23/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.23/self_attn/Constant_57_output_0 0 1 /model/layers.23/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_5_output_0 0 1 /model/layers.23/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_61_output_0 0 1 /model/layers.23/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_62_output_0 0 1 /model/layers.23/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_63_output_0 0 1 /model/layers.23/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_64_output_0 0 1 /model/layers.23/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_6_output_0 0 1 /model/layers.23/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_7_output_0 0 1 /model/layers.23/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.23/self_attn/Constant_9_output_0 0 1 /model/layers.23/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.3/input_layernorm/Constant_1_output_0 0 1 /model/layers.3/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.3/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.3/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_12_output_0 0 1 /model/layers.3/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_15_output_0 0 1 /model/layers.3/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_19_output_0 0 1 /model/layers.3/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.3/self_attn/Constant_22_output_0 0 1 /model/layers.3/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_25_output_0 0 1 /model/layers.3/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_2_output_0 0 1 /model/layers.3/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_32_output_0 0 1 /model/layers.3/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_33_output_0 0 1 /model/layers.3/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_34_output_0 0 1 /model/layers.3/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_35_output_0 0 1 /model/layers.3/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_36_output_0 0 1 /model/layers.3/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_37_output_0 0 1 /model/layers.3/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_39_output_0 0 1 /model/layers.3/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.3/self_attn/Constant_3_output_0 0 1 /model/layers.3/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_40_output_0 0 1 /model/layers.3/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.3/self_attn/Constant_45_output_0 0 1 /model/layers.3/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_46_output_0 0 1 /model/layers.3/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_47_output_0 0 1 /model/layers.3/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_48_output_0 0 1 /model/layers.3/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_49_output_0 0 1 /model/layers.3/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_4_output_0 0 1 /model/layers.3/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_50_output_0 0 1 /model/layers.3/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_52_output_0 0 1 /model/layers.3/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.3/self_attn/Constant_53_output_0 0 1 /model/layers.3/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.3/self_attn/Constant_57_output_0 0 1 /model/layers.3/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_5_output_0 0 1 /model/layers.3/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_61_output_0 0 1 /model/layers.3/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_62_output_0 0 1 /model/layers.3/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_63_output_0 0 1 /model/layers.3/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_64_output_0 0 1 /model/layers.3/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_6_output_0 0 1 /model/layers.3/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_7_output_0 0 1 /model/layers.3/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.3/self_attn/Constant_9_output_0 0 1 /model/layers.3/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.4/input_layernorm/Constant_1_output_0 0 1 /model/layers.4/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.4/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.4/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_12_output_0 0 1 /model/layers.4/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_15_output_0 0 1 /model/layers.4/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_19_output_0 0 1 /model/layers.4/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.4/self_attn/Constant_22_output_0 0 1 /model/layers.4/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_25_output_0 0 1 /model/layers.4/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_2_output_0 0 1 /model/layers.4/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_32_output_0 0 1 /model/layers.4/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_33_output_0 0 1 /model/layers.4/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_34_output_0 0 1 /model/layers.4/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_35_output_0 0 1 /model/layers.4/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_36_output_0 0 1 /model/layers.4/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_37_output_0 0 1 /model/layers.4/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_39_output_0 0 1 /model/layers.4/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.4/self_attn/Constant_3_output_0 0 1 /model/layers.4/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_40_output_0 0 1 /model/layers.4/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.4/self_attn/Constant_45_output_0 0 1 /model/layers.4/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_46_output_0 0 1 /model/layers.4/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_47_output_0 0 1 /model/layers.4/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_48_output_0 0 1 /model/layers.4/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_49_output_0 0 1 /model/layers.4/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_4_output_0 0 1 /model/layers.4/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_50_output_0 0 1 /model/layers.4/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_52_output_0 0 1 /model/layers.4/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.4/self_attn/Constant_53_output_0 0 1 /model/layers.4/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.4/self_attn/Constant_57_output_0 0 1 /model/layers.4/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_5_output_0 0 1 /model/layers.4/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_61_output_0 0 1 /model/layers.4/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_62_output_0 0 1 /model/layers.4/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_63_output_0 0 1 /model/layers.4/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_64_output_0 0 1 /model/layers.4/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_6_output_0 0 1 /model/layers.4/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_7_output_0 0 1 /model/layers.4/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.4/self_attn/Constant_9_output_0 0 1 /model/layers.4/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.5/input_layernorm/Constant_1_output_0 0 1 /model/layers.5/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.5/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.5/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_12_output_0 0 1 /model/layers.5/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_15_output_0 0 1 /model/layers.5/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_19_output_0 0 1 /model/layers.5/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.5/self_attn/Constant_22_output_0 0 1 /model/layers.5/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_25_output_0 0 1 /model/layers.5/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_2_output_0 0 1 /model/layers.5/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_32_output_0 0 1 /model/layers.5/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_33_output_0 0 1 /model/layers.5/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_34_output_0 0 1 /model/layers.5/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_35_output_0 0 1 /model/layers.5/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_36_output_0 0 1 /model/layers.5/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_37_output_0 0 1 /model/layers.5/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_39_output_0 0 1 /model/layers.5/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.5/self_attn/Constant_3_output_0 0 1 /model/layers.5/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_40_output_0 0 1 /model/layers.5/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.5/self_attn/Constant_45_output_0 0 1 /model/layers.5/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_46_output_0 0 1 /model/layers.5/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_47_output_0 0 1 /model/layers.5/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_48_output_0 0 1 /model/layers.5/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_49_output_0 0 1 /model/layers.5/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_4_output_0 0 1 /model/layers.5/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_50_output_0 0 1 /model/layers.5/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_52_output_0 0 1 /model/layers.5/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.5/self_attn/Constant_53_output_0 0 1 /model/layers.5/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.5/self_attn/Constant_57_output_0 0 1 /model/layers.5/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_5_output_0 0 1 /model/layers.5/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_61_output_0 0 1 /model/layers.5/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_62_output_0 0 1 /model/layers.5/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_63_output_0 0 1 /model/layers.5/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_64_output_0 0 1 /model/layers.5/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_6_output_0 0 1 /model/layers.5/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_7_output_0 0 1 /model/layers.5/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.5/self_attn/Constant_9_output_0 0 1 /model/layers.5/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.6/input_layernorm/Constant_1_output_0 0 1 /model/layers.6/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.6/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.6/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_12_output_0 0 1 /model/layers.6/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_15_output_0 0 1 /model/layers.6/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_19_output_0 0 1 /model/layers.6/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.6/self_attn/Constant_22_output_0 0 1 /model/layers.6/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_25_output_0 0 1 /model/layers.6/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_2_output_0 0 1 /model/layers.6/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_32_output_0 0 1 /model/layers.6/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_33_output_0 0 1 /model/layers.6/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_34_output_0 0 1 /model/layers.6/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_35_output_0 0 1 /model/layers.6/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_36_output_0 0 1 /model/layers.6/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_37_output_0 0 1 /model/layers.6/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_39_output_0 0 1 /model/layers.6/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.6/self_attn/Constant_3_output_0 0 1 /model/layers.6/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_40_output_0 0 1 /model/layers.6/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.6/self_attn/Constant_45_output_0 0 1 /model/layers.6/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_46_output_0 0 1 /model/layers.6/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_47_output_0 0 1 /model/layers.6/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_48_output_0 0 1 /model/layers.6/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_49_output_0 0 1 /model/layers.6/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_4_output_0 0 1 /model/layers.6/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_50_output_0 0 1 /model/layers.6/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_52_output_0 0 1 /model/layers.6/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.6/self_attn/Constant_53_output_0 0 1 /model/layers.6/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.6/self_attn/Constant_57_output_0 0 1 /model/layers.6/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_5_output_0 0 1 /model/layers.6/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_61_output_0 0 1 /model/layers.6/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_62_output_0 0 1 /model/layers.6/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_63_output_0 0 1 /model/layers.6/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_64_output_0 0 1 /model/layers.6/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_6_output_0 0 1 /model/layers.6/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_7_output_0 0 1 /model/layers.6/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.6/self_attn/Constant_9_output_0 0 1 /model/layers.6/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.7/input_layernorm/Constant_1_output_0 0 1 /model/layers.7/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.7/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.7/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_12_output_0 0 1 /model/layers.7/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_15_output_0 0 1 /model/layers.7/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_19_output_0 0 1 /model/layers.7/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.7/self_attn/Constant_22_output_0 0 1 /model/layers.7/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_25_output_0 0 1 /model/layers.7/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_2_output_0 0 1 /model/layers.7/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_32_output_0 0 1 /model/layers.7/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_33_output_0 0 1 /model/layers.7/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_34_output_0 0 1 /model/layers.7/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_35_output_0 0 1 /model/layers.7/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_36_output_0 0 1 /model/layers.7/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_37_output_0 0 1 /model/layers.7/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_39_output_0 0 1 /model/layers.7/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.7/self_attn/Constant_3_output_0 0 1 /model/layers.7/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_40_output_0 0 1 /model/layers.7/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.7/self_attn/Constant_45_output_0 0 1 /model/layers.7/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_46_output_0 0 1 /model/layers.7/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_47_output_0 0 1 /model/layers.7/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_48_output_0 0 1 /model/layers.7/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_49_output_0 0 1 /model/layers.7/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_4_output_0 0 1 /model/layers.7/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_50_output_0 0 1 /model/layers.7/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_52_output_0 0 1 /model/layers.7/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.7/self_attn/Constant_53_output_0 0 1 /model/layers.7/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.7/self_attn/Constant_57_output_0 0 1 /model/layers.7/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_5_output_0 0 1 /model/layers.7/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_61_output_0 0 1 /model/layers.7/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_62_output_0 0 1 /model/layers.7/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_63_output_0 0 1 /model/layers.7/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_64_output_0 0 1 /model/layers.7/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_6_output_0 0 1 /model/layers.7/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_7_output_0 0 1 /model/layers.7/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.7/self_attn/Constant_9_output_0 0 1 /model/layers.7/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.8/input_layernorm/Constant_1_output_0 0 1 /model/layers.8/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.8/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.8/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_12_output_0 0 1 /model/layers.8/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_15_output_0 0 1 /model/layers.8/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_19_output_0 0 1 /model/layers.8/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.8/self_attn/Constant_22_output_0 0 1 /model/layers.8/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_25_output_0 0 1 /model/layers.8/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_2_output_0 0 1 /model/layers.8/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_32_output_0 0 1 /model/layers.8/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_33_output_0 0 1 /model/layers.8/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_34_output_0 0 1 /model/layers.8/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_35_output_0 0 1 /model/layers.8/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_36_output_0 0 1 /model/layers.8/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_37_output_0 0 1 /model/layers.8/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_39_output_0 0 1 /model/layers.8/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.8/self_attn/Constant_3_output_0 0 1 /model/layers.8/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_40_output_0 0 1 /model/layers.8/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.8/self_attn/Constant_45_output_0 0 1 /model/layers.8/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_46_output_0 0 1 /model/layers.8/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_47_output_0 0 1 /model/layers.8/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_48_output_0 0 1 /model/layers.8/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_49_output_0 0 1 /model/layers.8/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_4_output_0 0 1 /model/layers.8/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_50_output_0 0 1 /model/layers.8/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_52_output_0 0 1 /model/layers.8/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.8/self_attn/Constant_53_output_0 0 1 /model/layers.8/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.8/self_attn/Constant_57_output_0 0 1 /model/layers.8/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_5_output_0 0 1 /model/layers.8/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_61_output_0 0 1 /model/layers.8/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_62_output_0 0 1 /model/layers.8/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_63_output_0 0 1 /model/layers.8/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_64_output_0 0 1 /model/layers.8/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_6_output_0 0 1 /model/layers.8/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_7_output_0 0 1 /model/layers.8/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.8/self_attn/Constant_9_output_0 0 1 /model/layers.8/self_attn/Constant_9_output_0 0=2 MemoryData /model/layers.9/input_layernorm/Constant_1_output_0 0 1 /model/layers.9/input_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.9/post_attention_layernorm/Constant_1_output_0 0 1 /model/layers.9/post_attention_layernorm/Constant_1_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_12_output_0 0 1 /model/layers.9/self_attn/Constant_12_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_15_output_0 0 1 /model/layers.9/self_attn/Constant_15_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_19_output_0 0 1 /model/layers.9/self_attn/Constant_19_output_0 0=2 MemoryData /model/layers.9/self_attn/Constant_22_output_0 0 1 /model/layers.9/self_attn/Constant_22_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_25_output_0 0 1 /model/layers.9/self_attn/Constant_25_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_2_output_0 0 1 /model/layers.9/self_attn/Constant_2_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_32_output_0 0 1 /model/layers.9/self_attn/Constant_32_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_33_output_0 0 1 /model/layers.9/self_attn/Constant_33_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_34_output_0 0 1 /model/layers.9/self_attn/Constant_34_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_35_output_0 0 1 /model/layers.9/self_attn/Constant_35_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_36_output_0 0 1 /model/layers.9/self_attn/Constant_36_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_37_output_0 0 1 /model/layers.9/self_attn/Constant_37_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_39_output_0 0 1 /model/layers.9/self_attn/Constant_39_output_0 0=2 MemoryData /model/layers.9/self_attn/Constant_3_output_0 0 1 /model/layers.9/self_attn/Constant_3_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_40_output_0 0 1 /model/layers.9/self_attn/Constant_40_output_0 0=2 MemoryData /model/layers.9/self_attn/Constant_45_output_0 0 1 /model/layers.9/self_attn/Constant_45_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_46_output_0 0 1 /model/layers.9/self_attn/Constant_46_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_47_output_0 0 1 /model/layers.9/self_attn/Constant_47_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_48_output_0 0 1 /model/layers.9/self_attn/Constant_48_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_49_output_0 0 1 /model/layers.9/self_attn/Constant_49_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_4_output_0 0 1 /model/layers.9/self_attn/Constant_4_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_50_output_0 0 1 /model/layers.9/self_attn/Constant_50_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_52_output_0 0 1 /model/layers.9/self_attn/Constant_52_output_0 0=2 MemoryData /model/layers.9/self_attn/Constant_53_output_0 0 1 /model/layers.9/self_attn/Constant_53_output_0 0=2 MemoryData /model/layers.9/self_attn/Constant_57_output_0 0 1 /model/layers.9/self_attn/Constant_57_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_5_output_0 0 1 /model/layers.9/self_attn/Constant_5_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_61_output_0 0 1 /model/layers.9/self_attn/Constant_61_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_62_output_0 0 1 /model/layers.9/self_attn/Constant_62_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_63_output_0 0 1 /model/layers.9/self_attn/Constant_63_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_64_output_0 0 1 /model/layers.9/self_attn/Constant_64_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_6_output_0 0 1 /model/layers.9/self_attn/Constant_6_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_7_output_0 0 1 /model/layers.9/self_attn/Constant_7_output_0 0=1 MemoryData /model/layers.9/self_attn/Constant_9_output_0 0 1 /model/layers.9/self_attn/Constant_9_output_0 0=2 MemoryData /model/norm/Constant_1_output_0 0 1 /model/norm/Constant_1_output_0 0=1 MemoryData /model/rotary_emb/Constant_1_output_0 0 1 /model/rotary_emb/Constant_1_output_0 0=1 MemoryData /model/rotary_emb/Constant_2_output_0 0 1 /model/rotary_emb/Constant_2_output_0 0=1 MemoryData /model/rotary_emb/Constant_3_output_0 0 1 /model/rotary_emb/Constant_3_output_0 0=1 MemoryData /model/rotary_emb/Constant_5_output_0 0 1 /model/rotary_emb/Constant_5_output_0 0=2 MemoryData /model/rotary_emb/Constant_6_output_0 0 1 /model/rotary_emb/Constant_6_output_0 0=1 MemoryData input_ids 0 1 input_ids 0=0 MemoryData model.embed_tokens.weight_quantized 0 1 model.embed_tokens.weight_quantized 0=896 1=151936 MemoryData model.embed_tokens.weight_scale 0 1 model.embed_tokens.weight_scale 0=1 MemoryData model.embed_tokens.weight_zero_point 0 1 model.embed_tokens.weight_zero_point 0=0 MemoryData model.layers.0.input_layernorm.weight 0 1 model.layers.0.input_layernorm.weight 0=896 MemoryData model.layers.0.post_attention_layernorm.weight 0 1 model.layers.0.post_attention_layernorm.weight 0=896 MemoryData model.layers.0.self_attn.k_proj.bias 0 1 model.layers.0.self_attn.k_proj.bias 0=128 MemoryData model.layers.0.self_attn.q_proj.bias 0 1 model.layers.0.self_attn.q_proj.bias 0=896 MemoryData model.layers.0.self_attn.v_proj.bias 0 1 model.layers.0.self_attn.v_proj.bias 0=128 MemoryData model.layers.1.input_layernorm.weight 0 1 model.layers.1.input_layernorm.weight 0=896 MemoryData model.layers.1.post_attention_layernorm.weight 0 1 model.layers.1.post_attention_layernorm.weight 0=896 MemoryData model.layers.1.self_attn.k_proj.bias 0 1 model.layers.1.self_attn.k_proj.bias 0=128 MemoryData model.layers.1.self_attn.q_proj.bias 0 1 model.layers.1.self_attn.q_proj.bias 0=896 MemoryData model.layers.1.self_attn.v_proj.bias 0 1 model.layers.1.self_attn.v_proj.bias 0=128 MemoryData model.layers.10.input_layernorm.weight 0 1 model.layers.10.input_layernorm.weight 0=896 MemoryData model.layers.10.post_attention_layernorm.weight 0 1 model.layers.10.post_attention_layernorm.weight 0=896 MemoryData model.layers.10.self_attn.k_proj.bias 0 1 model.layers.10.self_attn.k_proj.bias 0=128 MemoryData model.layers.10.self_attn.q_proj.bias 0 1 model.layers.10.self_attn.q_proj.bias 0=896 MemoryData model.layers.10.self_attn.v_proj.bias 0 1 model.layers.10.self_attn.v_proj.bias 0=128 MemoryData model.layers.11.input_layernorm.weight 0 1 model.layers.11.input_layernorm.weight 0=896 MemoryData model.layers.11.post_attention_layernorm.weight 0 1 model.layers.11.post_attention_layernorm.weight 0=896 MemoryData model.layers.11.self_attn.k_proj.bias 0 1 model.layers.11.self_attn.k_proj.bias 0=128 MemoryData model.layers.11.self_attn.q_proj.bias 0 1 model.layers.11.self_attn.q_proj.bias 0=896 MemoryData model.layers.11.self_attn.v_proj.bias 0 1 model.layers.11.self_attn.v_proj.bias 0=128 MemoryData model.layers.12.input_layernorm.weight 0 1 model.layers.12.input_layernorm.weight 0=896 MemoryData model.layers.12.post_attention_layernorm.weight 0 1 model.layers.12.post_attention_layernorm.weight 0=896 MemoryData model.layers.12.self_attn.k_proj.bias 0 1 model.layers.12.self_attn.k_proj.bias 0=128 MemoryData model.layers.12.self_attn.q_proj.bias 0 1 model.layers.12.self_attn.q_proj.bias 0=896 MemoryData model.layers.12.self_attn.v_proj.bias 0 1 model.layers.12.self_attn.v_proj.bias 0=128 MemoryData model.layers.13.input_layernorm.weight 0 1 model.layers.13.input_layernorm.weight 0=896 MemoryData model.layers.13.post_attention_layernorm.weight 0 1 model.layers.13.post_attention_layernorm.weight 0=896 MemoryData model.layers.13.self_attn.k_proj.bias 0 1 model.layers.13.self_attn.k_proj.bias 0=128 MemoryData model.layers.13.self_attn.q_proj.bias 0 1 model.layers.13.self_attn.q_proj.bias 0=896 MemoryData model.layers.13.self_attn.v_proj.bias 0 1 model.layers.13.self_attn.v_proj.bias 0=128 MemoryData model.layers.14.input_layernorm.weight 0 1 model.layers.14.input_layernorm.weight 0=896 MemoryData model.layers.14.post_attention_layernorm.weight 0 1 model.layers.14.post_attention_layernorm.weight 0=896 MemoryData model.layers.14.self_attn.k_proj.bias 0 1 model.layers.14.self_attn.k_proj.bias 0=128 MemoryData model.layers.14.self_attn.q_proj.bias 0 1 model.layers.14.self_attn.q_proj.bias 0=896 MemoryData model.layers.14.self_attn.v_proj.bias 0 1 model.layers.14.self_attn.v_proj.bias 0=128 MemoryData model.layers.15.input_layernorm.weight 0 1 model.layers.15.input_layernorm.weight 0=896 MemoryData model.layers.15.post_attention_layernorm.weight 0 1 model.layers.15.post_attention_layernorm.weight 0=896 MemoryData model.layers.15.self_attn.k_proj.bias 0 1 model.layers.15.self_attn.k_proj.bias 0=128 MemoryData model.layers.15.self_attn.q_proj.bias 0 1 model.layers.15.self_attn.q_proj.bias 0=896 MemoryData model.layers.15.self_attn.v_proj.bias 0 1 model.layers.15.self_attn.v_proj.bias 0=128 MemoryData model.layers.16.input_layernorm.weight 0 1 model.layers.16.input_layernorm.weight 0=896 MemoryData model.layers.16.post_attention_layernorm.weight 0 1 model.layers.16.post_attention_layernorm.weight 0=896 MemoryData model.layers.16.self_attn.k_proj.bias 0 1 model.layers.16.self_attn.k_proj.bias 0=128 MemoryData model.layers.16.self_attn.q_proj.bias 0 1 model.layers.16.self_attn.q_proj.bias 0=896 MemoryData model.layers.16.self_attn.v_proj.bias 0 1 model.layers.16.self_attn.v_proj.bias 0=128 MemoryData model.layers.17.input_layernorm.weight 0 1 model.layers.17.input_layernorm.weight 0=896 MemoryData model.layers.17.post_attention_layernorm.weight 0 1 model.layers.17.post_attention_layernorm.weight 0=896 MemoryData model.layers.17.self_attn.k_proj.bias 0 1 model.layers.17.self_attn.k_proj.bias 0=128 MemoryData model.layers.17.self_attn.q_proj.bias 0 1 model.layers.17.self_attn.q_proj.bias 0=896 MemoryData model.layers.17.self_attn.v_proj.bias 0 1 model.layers.17.self_attn.v_proj.bias 0=128 MemoryData model.layers.18.input_layernorm.weight 0 1 model.layers.18.input_layernorm.weight 0=896 MemoryData model.layers.18.post_attention_layernorm.weight 0 1 model.layers.18.post_attention_layernorm.weight 0=896 MemoryData model.layers.18.self_attn.k_proj.bias 0 1 model.layers.18.self_attn.k_proj.bias 0=128 MemoryData model.layers.18.self_attn.q_proj.bias 0 1 model.layers.18.self_attn.q_proj.bias 0=896 MemoryData model.layers.18.self_attn.v_proj.bias 0 1 model.layers.18.self_attn.v_proj.bias 0=128 MemoryData model.layers.19.input_layernorm.weight 0 1 model.layers.19.input_layernorm.weight 0=896 MemoryData model.layers.19.post_attention_layernorm.weight 0 1 model.layers.19.post_attention_layernorm.weight 0=896 MemoryData model.layers.19.self_attn.k_proj.bias 0 1 model.layers.19.self_attn.k_proj.bias 0=128 MemoryData model.layers.19.self_attn.q_proj.bias 0 1 model.layers.19.self_attn.q_proj.bias 0=896 MemoryData model.layers.19.self_attn.v_proj.bias 0 1 model.layers.19.self_attn.v_proj.bias 0=128 MemoryData model.layers.2.input_layernorm.weight 0 1 model.layers.2.input_layernorm.weight 0=896 MemoryData model.layers.2.post_attention_layernorm.weight 0 1 model.layers.2.post_attention_layernorm.weight 0=896 MemoryData model.layers.2.self_attn.k_proj.bias 0 1 model.layers.2.self_attn.k_proj.bias 0=128 MemoryData model.layers.2.self_attn.q_proj.bias 0 1 model.layers.2.self_attn.q_proj.bias 0=896 MemoryData model.layers.2.self_attn.v_proj.bias 0 1 model.layers.2.self_attn.v_proj.bias 0=128 MemoryData model.layers.20.input_layernorm.weight 0 1 model.layers.20.input_layernorm.weight 0=896 MemoryData model.layers.20.post_attention_layernorm.weight 0 1 model.layers.20.post_attention_layernorm.weight 0=896 MemoryData model.layers.20.self_attn.k_proj.bias 0 1 model.layers.20.self_attn.k_proj.bias 0=128 MemoryData model.layers.20.self_attn.q_proj.bias 0 1 model.layers.20.self_attn.q_proj.bias 0=896 MemoryData model.layers.20.self_attn.v_proj.bias 0 1 model.layers.20.self_attn.v_proj.bias 0=128 MemoryData model.layers.21.input_layernorm.weight 0 1 model.layers.21.input_layernorm.weight 0=896 MemoryData model.layers.21.post_attention_layernorm.weight 0 1 model.layers.21.post_attention_layernorm.weight 0=896 MemoryData model.layers.21.self_attn.k_proj.bias 0 1 model.layers.21.self_attn.k_proj.bias 0=128 MemoryData model.layers.21.self_attn.q_proj.bias 0 1 model.layers.21.self_attn.q_proj.bias 0=896 MemoryData model.layers.21.self_attn.v_proj.bias 0 1 model.layers.21.self_attn.v_proj.bias 0=128 MemoryData model.layers.22.input_layernorm.weight 0 1 model.layers.22.input_layernorm.weight 0=896 MemoryData model.layers.22.post_attention_layernorm.weight 0 1 model.layers.22.post_attention_layernorm.weight 0=896 MemoryData model.layers.22.self_attn.k_proj.bias 0 1 model.layers.22.self_attn.k_proj.bias 0=128 MemoryData model.layers.22.self_attn.q_proj.bias 0 1 model.layers.22.self_attn.q_proj.bias 0=896 MemoryData model.layers.22.self_attn.v_proj.bias 0 1 model.layers.22.self_attn.v_proj.bias 0=128 MemoryData model.layers.23.input_layernorm.weight 0 1 model.layers.23.input_layernorm.weight 0=896 MemoryData model.layers.23.post_attention_layernorm.weight 0 1 model.layers.23.post_attention_layernorm.weight 0=896 MemoryData model.layers.23.self_attn.k_proj.bias 0 1 model.layers.23.self_attn.k_proj.bias 0=128 MemoryData model.layers.23.self_attn.q_proj.bias 0 1 model.layers.23.self_attn.q_proj.bias 0=896 MemoryData model.layers.23.self_attn.v_proj.bias 0 1 model.layers.23.self_attn.v_proj.bias 0=128 MemoryData model.layers.3.input_layernorm.weight 0 1 model.layers.3.input_layernorm.weight 0=896 MemoryData model.layers.3.post_attention_layernorm.weight 0 1 model.layers.3.post_attention_layernorm.weight 0=896 MemoryData model.layers.3.self_attn.k_proj.bias 0 1 model.layers.3.self_attn.k_proj.bias 0=128 MemoryData model.layers.3.self_attn.q_proj.bias 0 1 model.layers.3.self_attn.q_proj.bias 0=896 MemoryData model.layers.3.self_attn.v_proj.bias 0 1 model.layers.3.self_attn.v_proj.bias 0=128 MemoryData model.layers.4.input_layernorm.weight 0 1 model.layers.4.input_layernorm.weight 0=896 MemoryData model.layers.4.post_attention_layernorm.weight 0 1 model.layers.4.post_attention_layernorm.weight 0=896 MemoryData model.layers.4.self_attn.k_proj.bias 0 1 model.layers.4.self_attn.k_proj.bias 0=128 MemoryData model.layers.4.self_attn.q_proj.bias 0 1 model.layers.4.self_attn.q_proj.bias 0=896 MemoryData model.layers.4.self_attn.v_proj.bias 0 1 model.layers.4.self_attn.v_proj.bias 0=128 MemoryData model.layers.5.input_layernorm.weight 0 1 model.layers.5.input_layernorm.weight 0=896 MemoryData model.layers.5.post_attention_layernorm.weight 0 1 model.layers.5.post_attention_layernorm.weight 0=896 MemoryData model.layers.5.self_attn.k_proj.bias 0 1 model.layers.5.self_attn.k_proj.bias 0=128 MemoryData model.layers.5.self_attn.q_proj.bias 0 1 model.layers.5.self_attn.q_proj.bias 0=896 MemoryData model.layers.5.self_attn.v_proj.bias 0 1 model.layers.5.self_attn.v_proj.bias 0=128 MemoryData model.layers.6.input_layernorm.weight 0 1 model.layers.6.input_layernorm.weight 0=896 MemoryData model.layers.6.post_attention_layernorm.weight 0 1 model.layers.6.post_attention_layernorm.weight 0=896 MemoryData model.layers.6.self_attn.k_proj.bias 0 1 model.layers.6.self_attn.k_proj.bias 0=128 MemoryData model.layers.6.self_attn.q_proj.bias 0 1 model.layers.6.self_attn.q_proj.bias 0=896 MemoryData model.layers.6.self_attn.v_proj.bias 0 1 model.layers.6.self_attn.v_proj.bias 0=128 MemoryData model.layers.7.input_layernorm.weight 0 1 model.layers.7.input_layernorm.weight 0=896 MemoryData model.layers.7.post_attention_layernorm.weight 0 1 model.layers.7.post_attention_layernorm.weight 0=896 MemoryData model.layers.7.self_attn.k_proj.bias 0 1 model.layers.7.self_attn.k_proj.bias 0=128 MemoryData model.layers.7.self_attn.q_proj.bias 0 1 model.layers.7.self_attn.q_proj.bias 0=896 MemoryData model.layers.7.self_attn.v_proj.bias 0 1 model.layers.7.self_attn.v_proj.bias 0=128 MemoryData model.layers.8.input_layernorm.weight 0 1 model.layers.8.input_layernorm.weight 0=896 MemoryData model.layers.8.post_attention_layernorm.weight 0 1 model.layers.8.post_attention_layernorm.weight 0=896 MemoryData model.layers.8.self_attn.k_proj.bias 0 1 model.layers.8.self_attn.k_proj.bias 0=128 MemoryData model.layers.8.self_attn.q_proj.bias 0 1 model.layers.8.self_attn.q_proj.bias 0=896 MemoryData model.layers.8.self_attn.v_proj.bias 0 1 model.layers.8.self_attn.v_proj.bias 0=128 MemoryData model.layers.9.input_layernorm.weight 0 1 model.layers.9.input_layernorm.weight 0=896 MemoryData model.layers.9.post_attention_layernorm.weight 0 1 model.layers.9.post_attention_layernorm.weight 0=896 MemoryData model.layers.9.self_attn.k_proj.bias 0 1 model.layers.9.self_attn.k_proj.bias 0=128 MemoryData model.layers.9.self_attn.q_proj.bias 0 1 model.layers.9.self_attn.q_proj.bias 0=896 MemoryData model.layers.9.self_attn.v_proj.bias 0 1 model.layers.9.self_attn.v_proj.bias 0=128 MemoryData model.norm.weight 0 1 model.norm.weight 0=896 MemoryData onnx::Expand_431 0 1 onnx::Expand_431 0=1 1=32 MemoryData onnx::MatMul_7190_quantized 0 1 onnx::MatMul_7190_quantized 0=896 1=896 MemoryData onnx::MatMul_7190_zero_point 0 1 onnx::MatMul_7190_zero_point 0=0 MemoryData onnx::MatMul_7197_quantized 0 1 onnx::MatMul_7197_quantized 0=128 1=896 MemoryData onnx::MatMul_7197_zero_point 0 1 onnx::MatMul_7197_zero_point 0=0 MemoryData onnx::MatMul_7198_quantized 0 1 onnx::MatMul_7198_quantized 0=128 1=896 MemoryData onnx::MatMul_7198_zero_point 0 1 onnx::MatMul_7198_zero_point 0=0 MemoryData onnx::MatMul_7221_quantized 0 1 onnx::MatMul_7221_quantized 0=896 1=896 MemoryData onnx::MatMul_7221_zero_point 0 1 onnx::MatMul_7221_zero_point 0=0 MemoryData onnx::MatMul_7222_quantized 0 1 onnx::MatMul_7222_quantized 0=4864 1=896 MemoryData onnx::MatMul_7222_zero_point 0 1 onnx::MatMul_7222_zero_point 0=0 MemoryData onnx::MatMul_7223_quantized 0 1 onnx::MatMul_7223_quantized 0=4864 1=896 MemoryData onnx::MatMul_7223_zero_point 0 1 onnx::MatMul_7223_zero_point 0=0 MemoryData onnx::MatMul_7224_quantized 0 1 onnx::MatMul_7224_quantized 0=896 1=4864 MemoryData onnx::MatMul_7224_zero_point 0 1 onnx::MatMul_7224_zero_point 0=0 MemoryData onnx::MatMul_7225_quantized 0 1 onnx::MatMul_7225_quantized 0=896 1=896 MemoryData onnx::MatMul_7225_zero_point 0 1 onnx::MatMul_7225_zero_point 0=0 MemoryData onnx::MatMul_7232_quantized 0 1 onnx::MatMul_7232_quantized 0=128 1=896 MemoryData onnx::MatMul_7232_zero_point 0 1 onnx::MatMul_7232_zero_point 0=0 MemoryData onnx::MatMul_7233_quantized 0 1 onnx::MatMul_7233_quantized 0=128 1=896 MemoryData onnx::MatMul_7233_zero_point 0 1 onnx::MatMul_7233_zero_point 0=0 MemoryData onnx::MatMul_7256_quantized 0 1 onnx::MatMul_7256_quantized 0=896 1=896 MemoryData onnx::MatMul_7256_zero_point 0 1 onnx::MatMul_7256_zero_point 0=0 MemoryData onnx::MatMul_7257_quantized 0 1 onnx::MatMul_7257_quantized 0=4864 1=896 MemoryData onnx::MatMul_7257_zero_point 0 1 onnx::MatMul_7257_zero_point 0=0 MemoryData onnx::MatMul_7258_quantized 0 1 onnx::MatMul_7258_quantized 0=4864 1=896 MemoryData onnx::MatMul_7258_zero_point 0 1 onnx::MatMul_7258_zero_point 0=0 MemoryData onnx::MatMul_7259_quantized 0 1 onnx::MatMul_7259_quantized 0=896 1=4864 MemoryData onnx::MatMul_7259_zero_point 0 1 onnx::MatMul_7259_zero_point 0=0 MemoryData onnx::MatMul_7260_quantized 0 1 onnx::MatMul_7260_quantized 0=896 1=896 MemoryData onnx::MatMul_7260_zero_point 0 1 onnx::MatMul_7260_zero_point 0=0 MemoryData onnx::MatMul_7267_quantized 0 1 onnx::MatMul_7267_quantized 0=128 1=896 MemoryData onnx::MatMul_7267_zero_point 0 1 onnx::MatMul_7267_zero_point 0=0 MemoryData onnx::MatMul_7268_quantized 0 1 onnx::MatMul_7268_quantized 0=128 1=896 MemoryData onnx::MatMul_7268_zero_point 0 1 onnx::MatMul_7268_zero_point 0=0 MemoryData onnx::MatMul_7291_quantized 0 1 onnx::MatMul_7291_quantized 0=896 1=896 MemoryData onnx::MatMul_7291_zero_point 0 1 onnx::MatMul_7291_zero_point 0=0 MemoryData onnx::MatMul_7292_quantized 0 1 onnx::MatMul_7292_quantized 0=4864 1=896 MemoryData onnx::MatMul_7292_zero_point 0 1 onnx::MatMul_7292_zero_point 0=0 MemoryData onnx::MatMul_7293_quantized 0 1 onnx::MatMul_7293_quantized 0=4864 1=896 MemoryData onnx::MatMul_7293_zero_point 0 1 onnx::MatMul_7293_zero_point 0=0 MemoryData onnx::MatMul_7294_quantized 0 1 onnx::MatMul_7294_quantized 0=896 1=4864 MemoryData onnx::MatMul_7294_zero_point 0 1 onnx::MatMul_7294_zero_point 0=0 MemoryData onnx::MatMul_7295_quantized 0 1 onnx::MatMul_7295_quantized 0=896 1=896 MemoryData onnx::MatMul_7295_zero_point 0 1 onnx::MatMul_7295_zero_point 0=0 MemoryData onnx::MatMul_7302_quantized 0 1 onnx::MatMul_7302_quantized 0=128 1=896 MemoryData onnx::MatMul_7302_zero_point 0 1 onnx::MatMul_7302_zero_point 0=0 MemoryData onnx::MatMul_7303_quantized 0 1 onnx::MatMul_7303_quantized 0=128 1=896 MemoryData onnx::MatMul_7303_zero_point 0 1 onnx::MatMul_7303_zero_point 0=0 MemoryData onnx::MatMul_7326_quantized 0 1 onnx::MatMul_7326_quantized 0=896 1=896 MemoryData onnx::MatMul_7326_zero_point 0 1 onnx::MatMul_7326_zero_point 0=0 MemoryData onnx::MatMul_7327_quantized 0 1 onnx::MatMul_7327_quantized 0=4864 1=896 MemoryData onnx::MatMul_7327_zero_point 0 1 onnx::MatMul_7327_zero_point 0=0 MemoryData onnx::MatMul_7328_quantized 0 1 onnx::MatMul_7328_quantized 0=4864 1=896 MemoryData onnx::MatMul_7328_zero_point 0 1 onnx::MatMul_7328_zero_point 0=0 MemoryData onnx::MatMul_7329_quantized 0 1 onnx::MatMul_7329_quantized 0=896 1=4864 MemoryData onnx::MatMul_7329_zero_point 0 1 onnx::MatMul_7329_zero_point 0=0 MemoryData onnx::MatMul_7330_quantized 0 1 onnx::MatMul_7330_quantized 0=896 1=896 MemoryData onnx::MatMul_7330_zero_point 0 1 onnx::MatMul_7330_zero_point 0=0 MemoryData onnx::MatMul_7337_quantized 0 1 onnx::MatMul_7337_quantized 0=128 1=896 MemoryData onnx::MatMul_7337_zero_point 0 1 onnx::MatMul_7337_zero_point 0=0 MemoryData onnx::MatMul_7338_quantized 0 1 onnx::MatMul_7338_quantized 0=128 1=896 MemoryData onnx::MatMul_7338_zero_point 0 1 onnx::MatMul_7338_zero_point 0=0 MemoryData onnx::MatMul_7361_quantized 0 1 onnx::MatMul_7361_quantized 0=896 1=896 MemoryData onnx::MatMul_7361_zero_point 0 1 onnx::MatMul_7361_zero_point 0=0 MemoryData onnx::MatMul_7362_quantized 0 1 onnx::MatMul_7362_quantized 0=4864 1=896 MemoryData onnx::MatMul_7362_zero_point 0 1 onnx::MatMul_7362_zero_point 0=0 MemoryData onnx::MatMul_7363_quantized 0 1 onnx::MatMul_7363_quantized 0=4864 1=896 MemoryData onnx::MatMul_7363_zero_point 0 1 onnx::MatMul_7363_zero_point 0=0 MemoryData onnx::MatMul_7364_quantized 0 1 onnx::MatMul_7364_quantized 0=896 1=4864 MemoryData onnx::MatMul_7364_zero_point 0 1 onnx::MatMul_7364_zero_point 0=0 MemoryData onnx::MatMul_7365_quantized 0 1 onnx::MatMul_7365_quantized 0=896 1=896 MemoryData onnx::MatMul_7365_zero_point 0 1 onnx::MatMul_7365_zero_point 0=0 MemoryData onnx::MatMul_7372_quantized 0 1 onnx::MatMul_7372_quantized 0=128 1=896 MemoryData onnx::MatMul_7372_zero_point 0 1 onnx::MatMul_7372_zero_point 0=0 MemoryData onnx::MatMul_7373_quantized 0 1 onnx::MatMul_7373_quantized 0=128 1=896 MemoryData onnx::MatMul_7373_zero_point 0 1 onnx::MatMul_7373_zero_point 0=0 MemoryData onnx::MatMul_7396_quantized 0 1 onnx::MatMul_7396_quantized 0=896 1=896 MemoryData onnx::MatMul_7396_zero_point 0 1 onnx::MatMul_7396_zero_point 0=0 MemoryData onnx::MatMul_7397_quantized 0 1 onnx::MatMul_7397_quantized 0=4864 1=896 MemoryData onnx::MatMul_7397_zero_point 0 1 onnx::MatMul_7397_zero_point 0=0 MemoryData onnx::MatMul_7398_quantized 0 1 onnx::MatMul_7398_quantized 0=4864 1=896 MemoryData onnx::MatMul_7398_zero_point 0 1 onnx::MatMul_7398_zero_point 0=0 MemoryData onnx::MatMul_7399_quantized 0 1 onnx::MatMul_7399_quantized 0=896 1=4864 MemoryData onnx::MatMul_7399_zero_point 0 1 onnx::MatMul_7399_zero_point 0=0 MemoryData onnx::MatMul_7400_quantized 0 1 onnx::MatMul_7400_quantized 0=896 1=896 MemoryData onnx::MatMul_7400_zero_point 0 1 onnx::MatMul_7400_zero_point 0=0 MemoryData onnx::MatMul_7407_quantized 0 1 onnx::MatMul_7407_quantized 0=128 1=896 MemoryData onnx::MatMul_7407_zero_point 0 1 onnx::MatMul_7407_zero_point 0=0 MemoryData onnx::MatMul_7408_quantized 0 1 onnx::MatMul_7408_quantized 0=128 1=896 MemoryData onnx::MatMul_7408_zero_point 0 1 onnx::MatMul_7408_zero_point 0=0 MemoryData onnx::MatMul_7431_quantized 0 1 onnx::MatMul_7431_quantized 0=896 1=896 MemoryData onnx::MatMul_7431_zero_point 0 1 onnx::MatMul_7431_zero_point 0=0 MemoryData onnx::MatMul_7432_quantized 0 1 onnx::MatMul_7432_quantized 0=4864 1=896 MemoryData onnx::MatMul_7432_zero_point 0 1 onnx::MatMul_7432_zero_point 0=0 MemoryData onnx::MatMul_7433_quantized 0 1 onnx::MatMul_7433_quantized 0=4864 1=896 MemoryData onnx::MatMul_7433_zero_point 0 1 onnx::MatMul_7433_zero_point 0=0 MemoryData onnx::MatMul_7434_quantized 0 1 onnx::MatMul_7434_quantized 0=896 1=4864 MemoryData onnx::MatMul_7434_zero_point 0 1 onnx::MatMul_7434_zero_point 0=0 MemoryData onnx::MatMul_7435_quantized 0 1 onnx::MatMul_7435_quantized 0=896 1=896 MemoryData onnx::MatMul_7435_zero_point 0 1 onnx::MatMul_7435_zero_point 0=0 MemoryData onnx::MatMul_7442_quantized 0 1 onnx::MatMul_7442_quantized 0=128 1=896 MemoryData onnx::MatMul_7442_zero_point 0 1 onnx::MatMul_7442_zero_point 0=0 MemoryData onnx::MatMul_7443_quantized 0 1 onnx::MatMul_7443_quantized 0=128 1=896 MemoryData onnx::MatMul_7443_zero_point 0 1 onnx::MatMul_7443_zero_point 0=0 MemoryData onnx::MatMul_7466_quantized 0 1 onnx::MatMul_7466_quantized 0=896 1=896 MemoryData onnx::MatMul_7466_zero_point 0 1 onnx::MatMul_7466_zero_point 0=0 MemoryData onnx::MatMul_7467_quantized 0 1 onnx::MatMul_7467_quantized 0=4864 1=896 MemoryData onnx::MatMul_7467_zero_point 0 1 onnx::MatMul_7467_zero_point 0=0 MemoryData onnx::MatMul_7468_quantized 0 1 onnx::MatMul_7468_quantized 0=4864 1=896 MemoryData onnx::MatMul_7468_zero_point 0 1 onnx::MatMul_7468_zero_point 0=0 MemoryData onnx::MatMul_7469_quantized 0 1 onnx::MatMul_7469_quantized 0=896 1=4864 MemoryData onnx::MatMul_7469_zero_point 0 1 onnx::MatMul_7469_zero_point 0=0 MemoryData onnx::MatMul_7470_quantized 0 1 onnx::MatMul_7470_quantized 0=896 1=896 MemoryData onnx::MatMul_7470_zero_point 0 1 onnx::MatMul_7470_zero_point 0=0 MemoryData onnx::MatMul_7477_quantized 0 1 onnx::MatMul_7477_quantized 0=128 1=896 MemoryData onnx::MatMul_7477_zero_point 0 1 onnx::MatMul_7477_zero_point 0=0 MemoryData onnx::MatMul_7478_quantized 0 1 onnx::MatMul_7478_quantized 0=128 1=896 MemoryData onnx::MatMul_7478_zero_point 0 1 onnx::MatMul_7478_zero_point 0=0 MemoryData onnx::MatMul_7501_quantized 0 1 onnx::MatMul_7501_quantized 0=896 1=896 MemoryData onnx::MatMul_7501_zero_point 0 1 onnx::MatMul_7501_zero_point 0=0 MemoryData onnx::MatMul_7502_quantized 0 1 onnx::MatMul_7502_quantized 0=4864 1=896 MemoryData onnx::MatMul_7502_zero_point 0 1 onnx::MatMul_7502_zero_point 0=0 MemoryData onnx::MatMul_7503_quantized 0 1 onnx::MatMul_7503_quantized 0=4864 1=896 MemoryData onnx::MatMul_7503_zero_point 0 1 onnx::MatMul_7503_zero_point 0=0 MemoryData onnx::MatMul_7504_quantized 0 1 onnx::MatMul_7504_quantized 0=896 1=4864 MemoryData onnx::MatMul_7504_zero_point 0 1 onnx::MatMul_7504_zero_point 0=0 MemoryData onnx::MatMul_7505_quantized 0 1 onnx::MatMul_7505_quantized 0=896 1=896 MemoryData onnx::MatMul_7505_zero_point 0 1 onnx::MatMul_7505_zero_point 0=0 MemoryData onnx::MatMul_7512_quantized 0 1 onnx::MatMul_7512_quantized 0=128 1=896 MemoryData onnx::MatMul_7512_zero_point 0 1 onnx::MatMul_7512_zero_point 0=0 MemoryData onnx::MatMul_7513_quantized 0 1 onnx::MatMul_7513_quantized 0=128 1=896 MemoryData onnx::MatMul_7513_zero_point 0 1 onnx::MatMul_7513_zero_point 0=0 MemoryData onnx::MatMul_7536_quantized 0 1 onnx::MatMul_7536_quantized 0=896 1=896 MemoryData onnx::MatMul_7536_zero_point 0 1 onnx::MatMul_7536_zero_point 0=0 MemoryData onnx::MatMul_7537_quantized 0 1 onnx::MatMul_7537_quantized 0=4864 1=896 MemoryData onnx::MatMul_7537_zero_point 0 1 onnx::MatMul_7537_zero_point 0=0 MemoryData onnx::MatMul_7538_quantized 0 1 onnx::MatMul_7538_quantized 0=4864 1=896 MemoryData onnx::MatMul_7538_zero_point 0 1 onnx::MatMul_7538_zero_point 0=0 MemoryData onnx::MatMul_7539_quantized 0 1 onnx::MatMul_7539_quantized 0=896 1=4864 MemoryData onnx::MatMul_7539_zero_point 0 1 onnx::MatMul_7539_zero_point 0=0 MemoryData onnx::MatMul_7540_quantized 0 1 onnx::MatMul_7540_quantized 0=896 1=896 MemoryData onnx::MatMul_7540_zero_point 0 1 onnx::MatMul_7540_zero_point 0=0 MemoryData onnx::MatMul_7547_quantized 0 1 onnx::MatMul_7547_quantized 0=128 1=896 MemoryData onnx::MatMul_7547_zero_point 0 1 onnx::MatMul_7547_zero_point 0=0 MemoryData onnx::MatMul_7548_quantized 0 1 onnx::MatMul_7548_quantized 0=128 1=896 MemoryData onnx::MatMul_7548_zero_point 0 1 onnx::MatMul_7548_zero_point 0=0 MemoryData onnx::MatMul_7571_quantized 0 1 onnx::MatMul_7571_quantized 0=896 1=896 MemoryData onnx::MatMul_7571_zero_point 0 1 onnx::MatMul_7571_zero_point 0=0 MemoryData onnx::MatMul_7572_quantized 0 1 onnx::MatMul_7572_quantized 0=4864 1=896 MemoryData onnx::MatMul_7572_zero_point 0 1 onnx::MatMul_7572_zero_point 0=0 MemoryData onnx::MatMul_7573_quantized 0 1 onnx::MatMul_7573_quantized 0=4864 1=896 MemoryData onnx::MatMul_7573_zero_point 0 1 onnx::MatMul_7573_zero_point 0=0 MemoryData onnx::MatMul_7574_quantized 0 1 onnx::MatMul_7574_quantized 0=896 1=4864 MemoryData onnx::MatMul_7574_zero_point 0 1 onnx::MatMul_7574_zero_point 0=0 MemoryData onnx::MatMul_7575_quantized 0 1 onnx::MatMul_7575_quantized 0=896 1=896 MemoryData onnx::MatMul_7575_zero_point 0 1 onnx::MatMul_7575_zero_point 0=0 MemoryData onnx::MatMul_7582_quantized 0 1 onnx::MatMul_7582_quantized 0=128 1=896 MemoryData onnx::MatMul_7582_zero_point 0 1 onnx::MatMul_7582_zero_point 0=0 MemoryData onnx::MatMul_7583_quantized 0 1 onnx::MatMul_7583_quantized 0=128 1=896 MemoryData onnx::MatMul_7583_zero_point 0 1 onnx::MatMul_7583_zero_point 0=0 MemoryData onnx::MatMul_7606_quantized 0 1 onnx::MatMul_7606_quantized 0=896 1=896 MemoryData onnx::MatMul_7606_zero_point 0 1 onnx::MatMul_7606_zero_point 0=0 MemoryData onnx::MatMul_7607_quantized 0 1 onnx::MatMul_7607_quantized 0=4864 1=896 MemoryData onnx::MatMul_7607_zero_point 0 1 onnx::MatMul_7607_zero_point 0=0 MemoryData onnx::MatMul_7608_quantized 0 1 onnx::MatMul_7608_quantized 0=4864 1=896 MemoryData onnx::MatMul_7608_zero_point 0 1 onnx::MatMul_7608_zero_point 0=0 MemoryData onnx::MatMul_7609_quantized 0 1 onnx::MatMul_7609_quantized 0=896 1=4864 MemoryData onnx::MatMul_7609_zero_point 0 1 onnx::MatMul_7609_zero_point 0=0 MemoryData onnx::MatMul_7610_quantized 0 1 onnx::MatMul_7610_quantized 0=896 1=896 MemoryData onnx::MatMul_7610_zero_point 0 1 onnx::MatMul_7610_zero_point 0=0 MemoryData onnx::MatMul_7617_quantized 0 1 onnx::MatMul_7617_quantized 0=128 1=896 MemoryData onnx::MatMul_7617_zero_point 0 1 onnx::MatMul_7617_zero_point 0=0 MemoryData onnx::MatMul_7618_quantized 0 1 onnx::MatMul_7618_quantized 0=128 1=896 MemoryData onnx::MatMul_7618_zero_point 0 1 onnx::MatMul_7618_zero_point 0=0 MemoryData onnx::MatMul_7641_quantized 0 1 onnx::MatMul_7641_quantized 0=896 1=896 MemoryData onnx::MatMul_7641_zero_point 0 1 onnx::MatMul_7641_zero_point 0=0 MemoryData onnx::MatMul_7642_quantized 0 1 onnx::MatMul_7642_quantized 0=4864 1=896 MemoryData onnx::MatMul_7642_zero_point 0 1 onnx::MatMul_7642_zero_point 0=0 MemoryData onnx::MatMul_7643_quantized 0 1 onnx::MatMul_7643_quantized 0=4864 1=896 MemoryData onnx::MatMul_7643_zero_point 0 1 onnx::MatMul_7643_zero_point 0=0 MemoryData onnx::MatMul_7644_quantized 0 1 onnx::MatMul_7644_quantized 0=896 1=4864 MemoryData onnx::MatMul_7644_zero_point 0 1 onnx::MatMul_7644_zero_point 0=0 MemoryData onnx::MatMul_7645_quantized 0 1 onnx::MatMul_7645_quantized 0=896 1=896 MemoryData onnx::MatMul_7645_zero_point 0 1 onnx::MatMul_7645_zero_point 0=0 MemoryData onnx::MatMul_7652_quantized 0 1 onnx::MatMul_7652_quantized 0=128 1=896 MemoryData onnx::MatMul_7652_zero_point 0 1 onnx::MatMul_7652_zero_point 0=0 MemoryData onnx::MatMul_7653_quantized 0 1 onnx::MatMul_7653_quantized 0=128 1=896 MemoryData onnx::MatMul_7653_zero_point 0 1 onnx::MatMul_7653_zero_point 0=0 MemoryData onnx::MatMul_7676_quantized 0 1 onnx::MatMul_7676_quantized 0=896 1=896 MemoryData onnx::MatMul_7676_zero_point 0 1 onnx::MatMul_7676_zero_point 0=0 MemoryData onnx::MatMul_7677_quantized 0 1 onnx::MatMul_7677_quantized 0=4864 1=896 MemoryData onnx::MatMul_7677_zero_point 0 1 onnx::MatMul_7677_zero_point 0=0 MemoryData onnx::MatMul_7678_quantized 0 1 onnx::MatMul_7678_quantized 0=4864 1=896 MemoryData onnx::MatMul_7678_zero_point 0 1 onnx::MatMul_7678_zero_point 0=0 MemoryData onnx::MatMul_7679_quantized 0 1 onnx::MatMul_7679_quantized 0=896 1=4864 MemoryData onnx::MatMul_7679_zero_point 0 1 onnx::MatMul_7679_zero_point 0=0 MemoryData onnx::MatMul_7680_quantized 0 1 onnx::MatMul_7680_quantized 0=896 1=896 MemoryData onnx::MatMul_7680_zero_point 0 1 onnx::MatMul_7680_zero_point 0=0 MemoryData onnx::MatMul_7687_quantized 0 1 onnx::MatMul_7687_quantized 0=128 1=896 MemoryData onnx::MatMul_7687_zero_point 0 1 onnx::MatMul_7687_zero_point 0=0 MemoryData onnx::MatMul_7688_quantized 0 1 onnx::MatMul_7688_quantized 0=128 1=896 MemoryData onnx::MatMul_7688_zero_point 0 1 onnx::MatMul_7688_zero_point 0=0 MemoryData onnx::MatMul_7711_quantized 0 1 onnx::MatMul_7711_quantized 0=896 1=896 MemoryData onnx::MatMul_7711_zero_point 0 1 onnx::MatMul_7711_zero_point 0=0 MemoryData onnx::MatMul_7712_quantized 0 1 onnx::MatMul_7712_quantized 0=4864 1=896 MemoryData onnx::MatMul_7712_zero_point 0 1 onnx::MatMul_7712_zero_point 0=0 MemoryData onnx::MatMul_7713_quantized 0 1 onnx::MatMul_7713_quantized 0=4864 1=896 MemoryData onnx::MatMul_7713_zero_point 0 1 onnx::MatMul_7713_zero_point 0=0 MemoryData onnx::MatMul_7714_quantized 0 1 onnx::MatMul_7714_quantized 0=896 1=4864 MemoryData onnx::MatMul_7714_zero_point 0 1 onnx::MatMul_7714_zero_point 0=0 MemoryData onnx::MatMul_7715_quantized 0 1 onnx::MatMul_7715_quantized 0=896 1=896 MemoryData onnx::MatMul_7715_zero_point 0 1 onnx::MatMul_7715_zero_point 0=0 MemoryData onnx::MatMul_7722_quantized 0 1 onnx::MatMul_7722_quantized 0=128 1=896 MemoryData onnx::MatMul_7722_zero_point 0 1 onnx::MatMul_7722_zero_point 0=0 MemoryData onnx::MatMul_7723_quantized 0 1 onnx::MatMul_7723_quantized 0=128 1=896 MemoryData onnx::MatMul_7723_zero_point 0 1 onnx::MatMul_7723_zero_point 0=0 MemoryData onnx::MatMul_7746_quantized 0 1 onnx::MatMul_7746_quantized 0=896 1=896 MemoryData onnx::MatMul_7746_zero_point 0 1 onnx::MatMul_7746_zero_point 0=0 MemoryData onnx::MatMul_7747_quantized 0 1 onnx::MatMul_7747_quantized 0=4864 1=896 MemoryData onnx::MatMul_7747_zero_point 0 1 onnx::MatMul_7747_zero_point 0=0 MemoryData onnx::MatMul_7748_quantized 0 1 onnx::MatMul_7748_quantized 0=4864 1=896 MemoryData onnx::MatMul_7748_zero_point 0 1 onnx::MatMul_7748_zero_point 0=0 MemoryData onnx::MatMul_7749_quantized 0 1 onnx::MatMul_7749_quantized 0=896 1=4864 MemoryData onnx::MatMul_7749_zero_point 0 1 onnx::MatMul_7749_zero_point 0=0 MemoryData onnx::MatMul_7750_quantized 0 1 onnx::MatMul_7750_quantized 0=896 1=896 MemoryData onnx::MatMul_7750_zero_point 0 1 onnx::MatMul_7750_zero_point 0=0 MemoryData onnx::MatMul_7757_quantized 0 1 onnx::MatMul_7757_quantized 0=128 1=896 MemoryData onnx::MatMul_7757_zero_point 0 1 onnx::MatMul_7757_zero_point 0=0 MemoryData onnx::MatMul_7758_quantized 0 1 onnx::MatMul_7758_quantized 0=128 1=896 MemoryData onnx::MatMul_7758_zero_point 0 1 onnx::MatMul_7758_zero_point 0=0 MemoryData onnx::MatMul_7781_quantized 0 1 onnx::MatMul_7781_quantized 0=896 1=896 MemoryData onnx::MatMul_7781_zero_point 0 1 onnx::MatMul_7781_zero_point 0=0 MemoryData onnx::MatMul_7782_quantized 0 1 onnx::MatMul_7782_quantized 0=4864 1=896 MemoryData onnx::MatMul_7782_zero_point 0 1 onnx::MatMul_7782_zero_point 0=0 MemoryData onnx::MatMul_7783_quantized 0 1 onnx::MatMul_7783_quantized 0=4864 1=896 MemoryData onnx::MatMul_7783_zero_point 0 1 onnx::MatMul_7783_zero_point 0=0 MemoryData onnx::MatMul_7784_quantized 0 1 onnx::MatMul_7784_quantized 0=896 1=4864 MemoryData onnx::MatMul_7784_zero_point 0 1 onnx::MatMul_7784_zero_point 0=0 MemoryData onnx::MatMul_7785_quantized 0 1 onnx::MatMul_7785_quantized 0=896 1=896 MemoryData onnx::MatMul_7785_zero_point 0 1 onnx::MatMul_7785_zero_point 0=0 MemoryData onnx::MatMul_7792_quantized 0 1 onnx::MatMul_7792_quantized 0=128 1=896 MemoryData onnx::MatMul_7792_zero_point 0 1 onnx::MatMul_7792_zero_point 0=0 MemoryData onnx::MatMul_7793_quantized 0 1 onnx::MatMul_7793_quantized 0=128 1=896 MemoryData onnx::MatMul_7793_zero_point 0 1 onnx::MatMul_7793_zero_point 0=0 MemoryData onnx::MatMul_7816_quantized 0 1 onnx::MatMul_7816_quantized 0=896 1=896 MemoryData onnx::MatMul_7816_zero_point 0 1 onnx::MatMul_7816_zero_point 0=0 MemoryData onnx::MatMul_7817_quantized 0 1 onnx::MatMul_7817_quantized 0=4864 1=896 MemoryData onnx::MatMul_7817_zero_point 0 1 onnx::MatMul_7817_zero_point 0=0 MemoryData onnx::MatMul_7818_quantized 0 1 onnx::MatMul_7818_quantized 0=4864 1=896 MemoryData onnx::MatMul_7818_zero_point 0 1 onnx::MatMul_7818_zero_point 0=0 MemoryData onnx::MatMul_7819_quantized 0 1 onnx::MatMul_7819_quantized 0=896 1=4864 MemoryData onnx::MatMul_7819_zero_point 0 1 onnx::MatMul_7819_zero_point 0=0 MemoryData onnx::MatMul_7820_quantized 0 1 onnx::MatMul_7820_quantized 0=896 1=896 MemoryData onnx::MatMul_7820_zero_point 0 1 onnx::MatMul_7820_zero_point 0=0 MemoryData onnx::MatMul_7827_quantized 0 1 onnx::MatMul_7827_quantized 0=128 1=896 MemoryData onnx::MatMul_7827_zero_point 0 1 onnx::MatMul_7827_zero_point 0=0 MemoryData onnx::MatMul_7828_quantized 0 1 onnx::MatMul_7828_quantized 0=128 1=896 MemoryData onnx::MatMul_7828_zero_point 0 1 onnx::MatMul_7828_zero_point 0=0 MemoryData onnx::MatMul_7851_quantized 0 1 onnx::MatMul_7851_quantized 0=896 1=896 MemoryData onnx::MatMul_7851_zero_point 0 1 onnx::MatMul_7851_zero_point 0=0 MemoryData onnx::MatMul_7852_quantized 0 1 onnx::MatMul_7852_quantized 0=4864 1=896 MemoryData onnx::MatMul_7852_zero_point 0 1 onnx::MatMul_7852_zero_point 0=0 MemoryData onnx::MatMul_7853_quantized 0 1 onnx::MatMul_7853_quantized 0=4864 1=896 MemoryData onnx::MatMul_7853_zero_point 0 1 onnx::MatMul_7853_zero_point 0=0 MemoryData onnx::MatMul_7854_quantized 0 1 onnx::MatMul_7854_quantized 0=896 1=4864 MemoryData onnx::MatMul_7854_zero_point 0 1 onnx::MatMul_7854_zero_point 0=0 MemoryData onnx::MatMul_7855_quantized 0 1 onnx::MatMul_7855_quantized 0=896 1=896 MemoryData onnx::MatMul_7855_zero_point 0 1 onnx::MatMul_7855_zero_point 0=0 MemoryData onnx::MatMul_7862_quantized 0 1 onnx::MatMul_7862_quantized 0=128 1=896 MemoryData onnx::MatMul_7862_zero_point 0 1 onnx::MatMul_7862_zero_point 0=0 MemoryData onnx::MatMul_7863_quantized 0 1 onnx::MatMul_7863_quantized 0=128 1=896 MemoryData onnx::MatMul_7863_zero_point 0 1 onnx::MatMul_7863_zero_point 0=0 MemoryData onnx::MatMul_7886_quantized 0 1 onnx::MatMul_7886_quantized 0=896 1=896 MemoryData onnx::MatMul_7886_zero_point 0 1 onnx::MatMul_7886_zero_point 0=0 MemoryData onnx::MatMul_7887_quantized 0 1 onnx::MatMul_7887_quantized 0=4864 1=896 MemoryData onnx::MatMul_7887_zero_point 0 1 onnx::MatMul_7887_zero_point 0=0 MemoryData onnx::MatMul_7888_quantized 0 1 onnx::MatMul_7888_quantized 0=4864 1=896 MemoryData onnx::MatMul_7888_zero_point 0 1 onnx::MatMul_7888_zero_point 0=0 MemoryData onnx::MatMul_7889_quantized 0 1 onnx::MatMul_7889_quantized 0=896 1=4864 MemoryData onnx::MatMul_7889_zero_point 0 1 onnx::MatMul_7889_zero_point 0=0 MemoryData onnx::MatMul_7890_quantized 0 1 onnx::MatMul_7890_quantized 0=896 1=896 MemoryData onnx::MatMul_7890_zero_point 0 1 onnx::MatMul_7890_zero_point 0=0 MemoryData onnx::MatMul_7897_quantized 0 1 onnx::MatMul_7897_quantized 0=128 1=896 MemoryData onnx::MatMul_7897_zero_point 0 1 onnx::MatMul_7897_zero_point 0=0 MemoryData onnx::MatMul_7898_quantized 0 1 onnx::MatMul_7898_quantized 0=128 1=896 MemoryData onnx::MatMul_7898_zero_point 0 1 onnx::MatMul_7898_zero_point 0=0 MemoryData onnx::MatMul_7921_quantized 0 1 onnx::MatMul_7921_quantized 0=896 1=896 MemoryData onnx::MatMul_7921_zero_point 0 1 onnx::MatMul_7921_zero_point 0=0 MemoryData onnx::MatMul_7922_quantized 0 1 onnx::MatMul_7922_quantized 0=4864 1=896 MemoryData onnx::MatMul_7922_zero_point 0 1 onnx::MatMul_7922_zero_point 0=0 MemoryData onnx::MatMul_7923_quantized 0 1 onnx::MatMul_7923_quantized 0=4864 1=896 MemoryData onnx::MatMul_7923_zero_point 0 1 onnx::MatMul_7923_zero_point 0=0 MemoryData onnx::MatMul_7924_quantized 0 1 onnx::MatMul_7924_quantized 0=896 1=4864 MemoryData onnx::MatMul_7924_zero_point 0 1 onnx::MatMul_7924_zero_point 0=0 MemoryData onnx::MatMul_7925_quantized 0 1 onnx::MatMul_7925_quantized 0=896 1=896 MemoryData onnx::MatMul_7925_zero_point 0 1 onnx::MatMul_7925_zero_point 0=0 MemoryData onnx::MatMul_7932_quantized 0 1 onnx::MatMul_7932_quantized 0=128 1=896 MemoryData onnx::MatMul_7932_zero_point 0 1 onnx::MatMul_7932_zero_point 0=0 MemoryData onnx::MatMul_7933_quantized 0 1 onnx::MatMul_7933_quantized 0=128 1=896 MemoryData onnx::MatMul_7933_zero_point 0 1 onnx::MatMul_7933_zero_point 0=0 MemoryData onnx::MatMul_7956_quantized 0 1 onnx::MatMul_7956_quantized 0=896 1=896 MemoryData onnx::MatMul_7956_zero_point 0 1 onnx::MatMul_7956_zero_point 0=0 MemoryData onnx::MatMul_7957_quantized 0 1 onnx::MatMul_7957_quantized 0=4864 1=896 MemoryData onnx::MatMul_7957_zero_point 0 1 onnx::MatMul_7957_zero_point 0=0 MemoryData onnx::MatMul_7958_quantized 0 1 onnx::MatMul_7958_quantized 0=4864 1=896 MemoryData onnx::MatMul_7958_zero_point 0 1 onnx::MatMul_7958_zero_point 0=0 MemoryData onnx::MatMul_7959_quantized 0 1 onnx::MatMul_7959_quantized 0=896 1=4864 MemoryData onnx::MatMul_7959_zero_point 0 1 onnx::MatMul_7959_zero_point 0=0 MemoryData onnx::MatMul_7960_quantized 0 1 onnx::MatMul_7960_quantized 0=896 1=896 MemoryData onnx::MatMul_7960_zero_point 0 1 onnx::MatMul_7960_zero_point 0=0 MemoryData onnx::MatMul_7967_quantized 0 1 onnx::MatMul_7967_quantized 0=128 1=896 MemoryData onnx::MatMul_7967_zero_point 0 1 onnx::MatMul_7967_zero_point 0=0 MemoryData onnx::MatMul_7968_quantized 0 1 onnx::MatMul_7968_quantized 0=128 1=896 MemoryData onnx::MatMul_7968_zero_point 0 1 onnx::MatMul_7968_zero_point 0=0 MemoryData onnx::MatMul_7991_quantized 0 1 onnx::MatMul_7991_quantized 0=896 1=896 MemoryData onnx::MatMul_7991_zero_point 0 1 onnx::MatMul_7991_zero_point 0=0 MemoryData onnx::MatMul_7992_quantized 0 1 onnx::MatMul_7992_quantized 0=4864 1=896 MemoryData onnx::MatMul_7992_zero_point 0 1 onnx::MatMul_7992_zero_point 0=0 MemoryData onnx::MatMul_7993_quantized 0 1 onnx::MatMul_7993_quantized 0=4864 1=896 MemoryData onnx::MatMul_7993_zero_point 0 1 onnx::MatMul_7993_zero_point 0=0 MemoryData onnx::MatMul_7994_quantized 0 1 onnx::MatMul_7994_quantized 0=896 1=4864 MemoryData onnx::MatMul_7994_zero_point 0 1 onnx::MatMul_7994_zero_point 0=0 MemoryData onnx::MatMul_7995_quantized 0 1 onnx::MatMul_7995_quantized 0=896 1=896 MemoryData onnx::MatMul_7995_zero_point 0 1 onnx::MatMul_7995_zero_point 0=0 MemoryData onnx::MatMul_8002_quantized 0 1 onnx::MatMul_8002_quantized 0=128 1=896 MemoryData onnx::MatMul_8002_zero_point 0 1 onnx::MatMul_8002_zero_point 0=0 MemoryData onnx::MatMul_8003_quantized 0 1 onnx::MatMul_8003_quantized 0=128 1=896 MemoryData onnx::MatMul_8003_zero_point 0 1 onnx::MatMul_8003_zero_point 0=0 MemoryData onnx::MatMul_8026_quantized 0 1 onnx::MatMul_8026_quantized 0=896 1=896 MemoryData onnx::MatMul_8026_zero_point 0 1 onnx::MatMul_8026_zero_point 0=0 MemoryData onnx::MatMul_8027_quantized 0 1 onnx::MatMul_8027_quantized 0=4864 1=896 MemoryData onnx::MatMul_8027_zero_point 0 1 onnx::MatMul_8027_zero_point 0=0 MemoryData onnx::MatMul_8028_quantized 0 1 onnx::MatMul_8028_quantized 0=4864 1=896 MemoryData onnx::MatMul_8028_zero_point 0 1 onnx::MatMul_8028_zero_point 0=0 MemoryData onnx::MatMul_8029_quantized 0 1 onnx::MatMul_8029_quantized 0=896 1=4864 MemoryData onnx::MatMul_8029_zero_point 0 1 onnx::MatMul_8029_zero_point 0=0 MemoryData onnx::MatMul_8030_quantized 0 1 onnx::MatMul_8030_quantized 0=151936 1=896 MemoryData onnx::MatMul_8030_zero_point 0 1 onnx::MatMul_8030_zero_point 0=0 MemoryData onnx::Unsqueeze_1001 0 1 onnx::Unsqueeze_1001 0=1 MemoryData onnx::Unsqueeze_1055 0 1 onnx::Unsqueeze_1055 0=1 MemoryData onnx::Unsqueeze_1057 0 1 onnx::Unsqueeze_1057 0=1 MemoryData onnx::Unsqueeze_1064 0 1 onnx::Unsqueeze_1064 0=1 MemoryData onnx::Unsqueeze_1066 0 1 onnx::Unsqueeze_1066 0=1 MemoryData onnx::Unsqueeze_1073 0 1 onnx::Unsqueeze_1073 0=1 MemoryData onnx::Unsqueeze_1075 0 1 onnx::Unsqueeze_1075 0=1 MemoryData onnx::Unsqueeze_1192 0 1 onnx::Unsqueeze_1192 0=1 MemoryData onnx::Unsqueeze_1194 0 1 onnx::Unsqueeze_1194 0=1 MemoryData onnx::Unsqueeze_1196 0 1 onnx::Unsqueeze_1196 0=1 MemoryData onnx::Unsqueeze_1198 0 1 onnx::Unsqueeze_1198 0=1 MemoryData onnx::Unsqueeze_1238 0 1 onnx::Unsqueeze_1238 0=1 MemoryData onnx::Unsqueeze_1240 0 1 onnx::Unsqueeze_1240 0=1 MemoryData onnx::Unsqueeze_1242 0 1 onnx::Unsqueeze_1242 0=1 MemoryData onnx::Unsqueeze_1244 0 1 onnx::Unsqueeze_1244 0=1 MemoryData onnx::Unsqueeze_1278 0 1 onnx::Unsqueeze_1278 0=1 MemoryData onnx::Unsqueeze_1280 0 1 onnx::Unsqueeze_1280 0=1 MemoryData onnx::Unsqueeze_1334 0 1 onnx::Unsqueeze_1334 0=1 MemoryData onnx::Unsqueeze_1336 0 1 onnx::Unsqueeze_1336 0=1 MemoryData onnx::Unsqueeze_1343 0 1 onnx::Unsqueeze_1343 0=1 MemoryData onnx::Unsqueeze_1345 0 1 onnx::Unsqueeze_1345 0=1 MemoryData onnx::Unsqueeze_1352 0 1 onnx::Unsqueeze_1352 0=1 MemoryData onnx::Unsqueeze_1354 0 1 onnx::Unsqueeze_1354 0=1 MemoryData onnx::Unsqueeze_1471 0 1 onnx::Unsqueeze_1471 0=1 MemoryData onnx::Unsqueeze_1473 0 1 onnx::Unsqueeze_1473 0=1 MemoryData onnx::Unsqueeze_1475 0 1 onnx::Unsqueeze_1475 0=1 MemoryData onnx::Unsqueeze_1477 0 1 onnx::Unsqueeze_1477 0=1 MemoryData onnx::Unsqueeze_1517 0 1 onnx::Unsqueeze_1517 0=1 MemoryData onnx::Unsqueeze_1519 0 1 onnx::Unsqueeze_1519 0=1 MemoryData onnx::Unsqueeze_1521 0 1 onnx::Unsqueeze_1521 0=1 MemoryData onnx::Unsqueeze_1523 0 1 onnx::Unsqueeze_1523 0=1 MemoryData onnx::Unsqueeze_1557 0 1 onnx::Unsqueeze_1557 0=1 MemoryData onnx::Unsqueeze_1559 0 1 onnx::Unsqueeze_1559 0=1 MemoryData onnx::Unsqueeze_1613 0 1 onnx::Unsqueeze_1613 0=1 MemoryData onnx::Unsqueeze_1615 0 1 onnx::Unsqueeze_1615 0=1 MemoryData onnx::Unsqueeze_1622 0 1 onnx::Unsqueeze_1622 0=1 MemoryData onnx::Unsqueeze_1624 0 1 onnx::Unsqueeze_1624 0=1 MemoryData onnx::Unsqueeze_1631 0 1 onnx::Unsqueeze_1631 0=1 MemoryData onnx::Unsqueeze_1633 0 1 onnx::Unsqueeze_1633 0=1 MemoryData onnx::Unsqueeze_1750 0 1 onnx::Unsqueeze_1750 0=1 MemoryData onnx::Unsqueeze_1752 0 1 onnx::Unsqueeze_1752 0=1 MemoryData onnx::Unsqueeze_1754 0 1 onnx::Unsqueeze_1754 0=1 MemoryData onnx::Unsqueeze_1756 0 1 onnx::Unsqueeze_1756 0=1 MemoryData onnx::Unsqueeze_1796 0 1 onnx::Unsqueeze_1796 0=1 MemoryData onnx::Unsqueeze_1798 0 1 onnx::Unsqueeze_1798 0=1 MemoryData onnx::Unsqueeze_1800 0 1 onnx::Unsqueeze_1800 0=1 MemoryData onnx::Unsqueeze_1802 0 1 onnx::Unsqueeze_1802 0=1 MemoryData onnx::Unsqueeze_1836 0 1 onnx::Unsqueeze_1836 0=1 MemoryData onnx::Unsqueeze_1838 0 1 onnx::Unsqueeze_1838 0=1 MemoryData onnx::Unsqueeze_1892 0 1 onnx::Unsqueeze_1892 0=1 MemoryData onnx::Unsqueeze_1894 0 1 onnx::Unsqueeze_1894 0=1 MemoryData onnx::Unsqueeze_1901 0 1 onnx::Unsqueeze_1901 0=1 MemoryData onnx::Unsqueeze_1903 0 1 onnx::Unsqueeze_1903 0=1 MemoryData onnx::Unsqueeze_1910 0 1 onnx::Unsqueeze_1910 0=1 MemoryData onnx::Unsqueeze_1912 0 1 onnx::Unsqueeze_1912 0=1 MemoryData onnx::Unsqueeze_2029 0 1 onnx::Unsqueeze_2029 0=1 MemoryData onnx::Unsqueeze_2031 0 1 onnx::Unsqueeze_2031 0=1 MemoryData onnx::Unsqueeze_2033 0 1 onnx::Unsqueeze_2033 0=1 MemoryData onnx::Unsqueeze_2035 0 1 onnx::Unsqueeze_2035 0=1 MemoryData onnx::Unsqueeze_2075 0 1 onnx::Unsqueeze_2075 0=1 MemoryData onnx::Unsqueeze_2077 0 1 onnx::Unsqueeze_2077 0=1 MemoryData onnx::Unsqueeze_2079 0 1 onnx::Unsqueeze_2079 0=1 MemoryData onnx::Unsqueeze_2081 0 1 onnx::Unsqueeze_2081 0=1 MemoryData onnx::Unsqueeze_2115 0 1 onnx::Unsqueeze_2115 0=1 MemoryData onnx::Unsqueeze_2117 0 1 onnx::Unsqueeze_2117 0=1 MemoryData onnx::Unsqueeze_2171 0 1 onnx::Unsqueeze_2171 0=1 MemoryData onnx::Unsqueeze_2173 0 1 onnx::Unsqueeze_2173 0=1 MemoryData onnx::Unsqueeze_2180 0 1 onnx::Unsqueeze_2180 0=1 MemoryData onnx::Unsqueeze_2182 0 1 onnx::Unsqueeze_2182 0=1 MemoryData onnx::Unsqueeze_2189 0 1 onnx::Unsqueeze_2189 0=1 MemoryData onnx::Unsqueeze_2191 0 1 onnx::Unsqueeze_2191 0=1 MemoryData onnx::Unsqueeze_2308 0 1 onnx::Unsqueeze_2308 0=1 MemoryData onnx::Unsqueeze_2310 0 1 onnx::Unsqueeze_2310 0=1 MemoryData onnx::Unsqueeze_2312 0 1 onnx::Unsqueeze_2312 0=1 MemoryData onnx::Unsqueeze_2314 0 1 onnx::Unsqueeze_2314 0=1 MemoryData onnx::Unsqueeze_2354 0 1 onnx::Unsqueeze_2354 0=1 MemoryData onnx::Unsqueeze_2356 0 1 onnx::Unsqueeze_2356 0=1 MemoryData onnx::Unsqueeze_2358 0 1 onnx::Unsqueeze_2358 0=1 MemoryData onnx::Unsqueeze_2360 0 1 onnx::Unsqueeze_2360 0=1 MemoryData onnx::Unsqueeze_2394 0 1 onnx::Unsqueeze_2394 0=1 MemoryData onnx::Unsqueeze_2396 0 1 onnx::Unsqueeze_2396 0=1 MemoryData onnx::Unsqueeze_2450 0 1 onnx::Unsqueeze_2450 0=1 MemoryData onnx::Unsqueeze_2452 0 1 onnx::Unsqueeze_2452 0=1 MemoryData onnx::Unsqueeze_2459 0 1 onnx::Unsqueeze_2459 0=1 MemoryData onnx::Unsqueeze_2461 0 1 onnx::Unsqueeze_2461 0=1 MemoryData onnx::Unsqueeze_2468 0 1 onnx::Unsqueeze_2468 0=1 MemoryData onnx::Unsqueeze_2470 0 1 onnx::Unsqueeze_2470 0=1 MemoryData onnx::Unsqueeze_2587 0 1 onnx::Unsqueeze_2587 0=1 MemoryData onnx::Unsqueeze_2589 0 1 onnx::Unsqueeze_2589 0=1 MemoryData onnx::Unsqueeze_2591 0 1 onnx::Unsqueeze_2591 0=1 MemoryData onnx::Unsqueeze_2593 0 1 onnx::Unsqueeze_2593 0=1 MemoryData onnx::Unsqueeze_2633 0 1 onnx::Unsqueeze_2633 0=1 MemoryData onnx::Unsqueeze_2635 0 1 onnx::Unsqueeze_2635 0=1 MemoryData onnx::Unsqueeze_2637 0 1 onnx::Unsqueeze_2637 0=1 MemoryData onnx::Unsqueeze_2639 0 1 onnx::Unsqueeze_2639 0=1 MemoryData onnx::Unsqueeze_2673 0 1 onnx::Unsqueeze_2673 0=1 MemoryData onnx::Unsqueeze_2675 0 1 onnx::Unsqueeze_2675 0=1 MemoryData onnx::Unsqueeze_2729 0 1 onnx::Unsqueeze_2729 0=1 MemoryData onnx::Unsqueeze_2731 0 1 onnx::Unsqueeze_2731 0=1 MemoryData onnx::Unsqueeze_2738 0 1 onnx::Unsqueeze_2738 0=1 MemoryData onnx::Unsqueeze_2740 0 1 onnx::Unsqueeze_2740 0=1 MemoryData onnx::Unsqueeze_2747 0 1 onnx::Unsqueeze_2747 0=1 MemoryData onnx::Unsqueeze_2749 0 1 onnx::Unsqueeze_2749 0=1 MemoryData onnx::Unsqueeze_2866 0 1 onnx::Unsqueeze_2866 0=1 MemoryData onnx::Unsqueeze_2868 0 1 onnx::Unsqueeze_2868 0=1 MemoryData onnx::Unsqueeze_2870 0 1 onnx::Unsqueeze_2870 0=1 MemoryData onnx::Unsqueeze_2872 0 1 onnx::Unsqueeze_2872 0=1 MemoryData onnx::Unsqueeze_2912 0 1 onnx::Unsqueeze_2912 0=1 MemoryData onnx::Unsqueeze_2914 0 1 onnx::Unsqueeze_2914 0=1 MemoryData onnx::Unsqueeze_2916 0 1 onnx::Unsqueeze_2916 0=1 MemoryData onnx::Unsqueeze_2918 0 1 onnx::Unsqueeze_2918 0=1 MemoryData onnx::Unsqueeze_2952 0 1 onnx::Unsqueeze_2952 0=1 MemoryData onnx::Unsqueeze_2954 0 1 onnx::Unsqueeze_2954 0=1 MemoryData onnx::Unsqueeze_3008 0 1 onnx::Unsqueeze_3008 0=1 MemoryData onnx::Unsqueeze_3010 0 1 onnx::Unsqueeze_3010 0=1 MemoryData onnx::Unsqueeze_3017 0 1 onnx::Unsqueeze_3017 0=1 MemoryData onnx::Unsqueeze_3019 0 1 onnx::Unsqueeze_3019 0=1 MemoryData onnx::Unsqueeze_3026 0 1 onnx::Unsqueeze_3026 0=1 MemoryData onnx::Unsqueeze_3028 0 1 onnx::Unsqueeze_3028 0=1 MemoryData onnx::Unsqueeze_3145 0 1 onnx::Unsqueeze_3145 0=1 MemoryData onnx::Unsqueeze_3147 0 1 onnx::Unsqueeze_3147 0=1 MemoryData onnx::Unsqueeze_3149 0 1 onnx::Unsqueeze_3149 0=1 MemoryData onnx::Unsqueeze_3151 0 1 onnx::Unsqueeze_3151 0=1 MemoryData onnx::Unsqueeze_3191 0 1 onnx::Unsqueeze_3191 0=1 MemoryData onnx::Unsqueeze_3193 0 1 onnx::Unsqueeze_3193 0=1 MemoryData onnx::Unsqueeze_3195 0 1 onnx::Unsqueeze_3195 0=1 MemoryData onnx::Unsqueeze_3197 0 1 onnx::Unsqueeze_3197 0=1 MemoryData onnx::Unsqueeze_3231 0 1 onnx::Unsqueeze_3231 0=1 MemoryData onnx::Unsqueeze_3233 0 1 onnx::Unsqueeze_3233 0=1 MemoryData onnx::Unsqueeze_3287 0 1 onnx::Unsqueeze_3287 0=1 MemoryData onnx::Unsqueeze_3289 0 1 onnx::Unsqueeze_3289 0=1 MemoryData onnx::Unsqueeze_3296 0 1 onnx::Unsqueeze_3296 0=1 MemoryData onnx::Unsqueeze_3298 0 1 onnx::Unsqueeze_3298 0=1 MemoryData onnx::Unsqueeze_3305 0 1 onnx::Unsqueeze_3305 0=1 MemoryData onnx::Unsqueeze_3307 0 1 onnx::Unsqueeze_3307 0=1 MemoryData onnx::Unsqueeze_3424 0 1 onnx::Unsqueeze_3424 0=1 MemoryData onnx::Unsqueeze_3426 0 1 onnx::Unsqueeze_3426 0=1 MemoryData onnx::Unsqueeze_3428 0 1 onnx::Unsqueeze_3428 0=1 MemoryData onnx::Unsqueeze_3430 0 1 onnx::Unsqueeze_3430 0=1 MemoryData onnx::Unsqueeze_3470 0 1 onnx::Unsqueeze_3470 0=1 MemoryData onnx::Unsqueeze_3472 0 1 onnx::Unsqueeze_3472 0=1 MemoryData onnx::Unsqueeze_3474 0 1 onnx::Unsqueeze_3474 0=1 MemoryData onnx::Unsqueeze_3476 0 1 onnx::Unsqueeze_3476 0=1 MemoryData onnx::Unsqueeze_3510 0 1 onnx::Unsqueeze_3510 0=1 MemoryData onnx::Unsqueeze_3512 0 1 onnx::Unsqueeze_3512 0=1 MemoryData onnx::Unsqueeze_3566 0 1 onnx::Unsqueeze_3566 0=1 MemoryData onnx::Unsqueeze_3568 0 1 onnx::Unsqueeze_3568 0=1 MemoryData onnx::Unsqueeze_3575 0 1 onnx::Unsqueeze_3575 0=1 MemoryData onnx::Unsqueeze_3577 0 1 onnx::Unsqueeze_3577 0=1 MemoryData onnx::Unsqueeze_3584 0 1 onnx::Unsqueeze_3584 0=1 MemoryData onnx::Unsqueeze_3586 0 1 onnx::Unsqueeze_3586 0=1 MemoryData onnx::Unsqueeze_3703 0 1 onnx::Unsqueeze_3703 0=1 MemoryData onnx::Unsqueeze_3705 0 1 onnx::Unsqueeze_3705 0=1 MemoryData onnx::Unsqueeze_3707 0 1 onnx::Unsqueeze_3707 0=1 MemoryData onnx::Unsqueeze_3709 0 1 onnx::Unsqueeze_3709 0=1 MemoryData onnx::Unsqueeze_3749 0 1 onnx::Unsqueeze_3749 0=1 MemoryData onnx::Unsqueeze_3751 0 1 onnx::Unsqueeze_3751 0=1 MemoryData onnx::Unsqueeze_3753 0 1 onnx::Unsqueeze_3753 0=1 MemoryData onnx::Unsqueeze_3755 0 1 onnx::Unsqueeze_3755 0=1 MemoryData onnx::Unsqueeze_3789 0 1 onnx::Unsqueeze_3789 0=1 MemoryData onnx::Unsqueeze_3791 0 1 onnx::Unsqueeze_3791 0=1 MemoryData onnx::Unsqueeze_3845 0 1 onnx::Unsqueeze_3845 0=1 MemoryData onnx::Unsqueeze_3847 0 1 onnx::Unsqueeze_3847 0=1 MemoryData onnx::Unsqueeze_3854 0 1 onnx::Unsqueeze_3854 0=1 MemoryData onnx::Unsqueeze_3856 0 1 onnx::Unsqueeze_3856 0=1 MemoryData onnx::Unsqueeze_3863 0 1 onnx::Unsqueeze_3863 0=1 MemoryData onnx::Unsqueeze_3865 0 1 onnx::Unsqueeze_3865 0=1 MemoryData onnx::Unsqueeze_3982 0 1 onnx::Unsqueeze_3982 0=1 MemoryData onnx::Unsqueeze_3984 0 1 onnx::Unsqueeze_3984 0=1 MemoryData onnx::Unsqueeze_3986 0 1 onnx::Unsqueeze_3986 0=1 MemoryData onnx::Unsqueeze_3988 0 1 onnx::Unsqueeze_3988 0=1 MemoryData onnx::Unsqueeze_4028 0 1 onnx::Unsqueeze_4028 0=1 MemoryData onnx::Unsqueeze_4030 0 1 onnx::Unsqueeze_4030 0=1 MemoryData onnx::Unsqueeze_4032 0 1 onnx::Unsqueeze_4032 0=1 MemoryData onnx::Unsqueeze_4034 0 1 onnx::Unsqueeze_4034 0=1 MemoryData onnx::Unsqueeze_4068 0 1 onnx::Unsqueeze_4068 0=1 MemoryData onnx::Unsqueeze_4070 0 1 onnx::Unsqueeze_4070 0=1 MemoryData onnx::Unsqueeze_4124 0 1 onnx::Unsqueeze_4124 0=1 MemoryData onnx::Unsqueeze_4126 0 1 onnx::Unsqueeze_4126 0=1 MemoryData onnx::Unsqueeze_4133 0 1 onnx::Unsqueeze_4133 0=1 MemoryData onnx::Unsqueeze_4135 0 1 onnx::Unsqueeze_4135 0=1 MemoryData onnx::Unsqueeze_4142 0 1 onnx::Unsqueeze_4142 0=1 MemoryData onnx::Unsqueeze_4144 0 1 onnx::Unsqueeze_4144 0=1 MemoryData onnx::Unsqueeze_4261 0 1 onnx::Unsqueeze_4261 0=1 MemoryData onnx::Unsqueeze_4263 0 1 onnx::Unsqueeze_4263 0=1 MemoryData onnx::Unsqueeze_4265 0 1 onnx::Unsqueeze_4265 0=1 MemoryData onnx::Unsqueeze_4267 0 1 onnx::Unsqueeze_4267 0=1 MemoryData onnx::Unsqueeze_4307 0 1 onnx::Unsqueeze_4307 0=1 MemoryData onnx::Unsqueeze_4309 0 1 onnx::Unsqueeze_4309 0=1 MemoryData onnx::Unsqueeze_4311 0 1 onnx::Unsqueeze_4311 0=1 MemoryData onnx::Unsqueeze_4313 0 1 onnx::Unsqueeze_4313 0=1 MemoryData onnx::Unsqueeze_4347 0 1 onnx::Unsqueeze_4347 0=1 MemoryData onnx::Unsqueeze_4349 0 1 onnx::Unsqueeze_4349 0=1 MemoryData onnx::Unsqueeze_4403 0 1 onnx::Unsqueeze_4403 0=1 MemoryData onnx::Unsqueeze_4405 0 1 onnx::Unsqueeze_4405 0=1 MemoryData onnx::Unsqueeze_4412 0 1 onnx::Unsqueeze_4412 0=1 MemoryData onnx::Unsqueeze_4414 0 1 onnx::Unsqueeze_4414 0=1 MemoryData onnx::Unsqueeze_4421 0 1 onnx::Unsqueeze_4421 0=1 MemoryData onnx::Unsqueeze_4423 0 1 onnx::Unsqueeze_4423 0=1 MemoryData onnx::Unsqueeze_4540 0 1 onnx::Unsqueeze_4540 0=1 MemoryData onnx::Unsqueeze_4542 0 1 onnx::Unsqueeze_4542 0=1 MemoryData onnx::Unsqueeze_4544 0 1 onnx::Unsqueeze_4544 0=1 MemoryData onnx::Unsqueeze_4546 0 1 onnx::Unsqueeze_4546 0=1 MemoryData onnx::Unsqueeze_4586 0 1 onnx::Unsqueeze_4586 0=1 MemoryData onnx::Unsqueeze_4588 0 1 onnx::Unsqueeze_4588 0=1 MemoryData onnx::Unsqueeze_4590 0 1 onnx::Unsqueeze_4590 0=1 MemoryData onnx::Unsqueeze_4592 0 1 onnx::Unsqueeze_4592 0=1 MemoryData onnx::Unsqueeze_4626 0 1 onnx::Unsqueeze_4626 0=1 MemoryData onnx::Unsqueeze_4628 0 1 onnx::Unsqueeze_4628 0=1 MemoryData onnx::Unsqueeze_4682 0 1 onnx::Unsqueeze_4682 0=1 MemoryData onnx::Unsqueeze_4684 0 1 onnx::Unsqueeze_4684 0=1 MemoryData onnx::Unsqueeze_4691 0 1 onnx::Unsqueeze_4691 0=1 MemoryData onnx::Unsqueeze_4693 0 1 onnx::Unsqueeze_4693 0=1 MemoryData onnx::Unsqueeze_4700 0 1 onnx::Unsqueeze_4700 0=1 MemoryData onnx::Unsqueeze_4702 0 1 onnx::Unsqueeze_4702 0=1 MemoryData onnx::Unsqueeze_4819 0 1 onnx::Unsqueeze_4819 0=1 MemoryData onnx::Unsqueeze_4821 0 1 onnx::Unsqueeze_4821 0=1 MemoryData onnx::Unsqueeze_4823 0 1 onnx::Unsqueeze_4823 0=1 MemoryData onnx::Unsqueeze_4825 0 1 onnx::Unsqueeze_4825 0=1 MemoryData onnx::Unsqueeze_4865 0 1 onnx::Unsqueeze_4865 0=1 MemoryData onnx::Unsqueeze_4867 0 1 onnx::Unsqueeze_4867 0=1 MemoryData onnx::Unsqueeze_4869 0 1 onnx::Unsqueeze_4869 0=1 MemoryData onnx::Unsqueeze_4871 0 1 onnx::Unsqueeze_4871 0=1 MemoryData onnx::Unsqueeze_4905 0 1 onnx::Unsqueeze_4905 0=1 MemoryData onnx::Unsqueeze_4907 0 1 onnx::Unsqueeze_4907 0=1 MemoryData onnx::Unsqueeze_491 0 1 onnx::Unsqueeze_491 0=1 MemoryData onnx::Unsqueeze_493 0 1 onnx::Unsqueeze_493 0=1 MemoryData onnx::Unsqueeze_4961 0 1 onnx::Unsqueeze_4961 0=1 MemoryData onnx::Unsqueeze_4963 0 1 onnx::Unsqueeze_4963 0=1 MemoryData onnx::Unsqueeze_4970 0 1 onnx::Unsqueeze_4970 0=1 MemoryData onnx::Unsqueeze_4972 0 1 onnx::Unsqueeze_4972 0=1 MemoryData onnx::Unsqueeze_4979 0 1 onnx::Unsqueeze_4979 0=1 MemoryData onnx::Unsqueeze_4981 0 1 onnx::Unsqueeze_4981 0=1 MemoryData onnx::Unsqueeze_500 0 1 onnx::Unsqueeze_500 0=1 MemoryData onnx::Unsqueeze_502 0 1 onnx::Unsqueeze_502 0=1 MemoryData onnx::Unsqueeze_509 0 1 onnx::Unsqueeze_509 0=1 MemoryData onnx::Unsqueeze_5098 0 1 onnx::Unsqueeze_5098 0=1 MemoryData onnx::Unsqueeze_5100 0 1 onnx::Unsqueeze_5100 0=1 MemoryData onnx::Unsqueeze_5102 0 1 onnx::Unsqueeze_5102 0=1 MemoryData onnx::Unsqueeze_5104 0 1 onnx::Unsqueeze_5104 0=1 MemoryData onnx::Unsqueeze_511 0 1 onnx::Unsqueeze_511 0=1 MemoryData onnx::Unsqueeze_5144 0 1 onnx::Unsqueeze_5144 0=1 MemoryData onnx::Unsqueeze_5146 0 1 onnx::Unsqueeze_5146 0=1 MemoryData onnx::Unsqueeze_5148 0 1 onnx::Unsqueeze_5148 0=1 MemoryData onnx::Unsqueeze_5150 0 1 onnx::Unsqueeze_5150 0=1 MemoryData onnx::Unsqueeze_5184 0 1 onnx::Unsqueeze_5184 0=1 MemoryData onnx::Unsqueeze_5186 0 1 onnx::Unsqueeze_5186 0=1 MemoryData onnx::Unsqueeze_5240 0 1 onnx::Unsqueeze_5240 0=1 MemoryData onnx::Unsqueeze_5242 0 1 onnx::Unsqueeze_5242 0=1 MemoryData onnx::Unsqueeze_5249 0 1 onnx::Unsqueeze_5249 0=1 MemoryData onnx::Unsqueeze_5251 0 1 onnx::Unsqueeze_5251 0=1 MemoryData onnx::Unsqueeze_5258 0 1 onnx::Unsqueeze_5258 0=1 MemoryData onnx::Unsqueeze_5260 0 1 onnx::Unsqueeze_5260 0=1 MemoryData onnx::Unsqueeze_5377 0 1 onnx::Unsqueeze_5377 0=1 MemoryData onnx::Unsqueeze_5379 0 1 onnx::Unsqueeze_5379 0=1 MemoryData onnx::Unsqueeze_5381 0 1 onnx::Unsqueeze_5381 0=1 MemoryData onnx::Unsqueeze_5383 0 1 onnx::Unsqueeze_5383 0=1 MemoryData onnx::Unsqueeze_5423 0 1 onnx::Unsqueeze_5423 0=1 MemoryData onnx::Unsqueeze_5425 0 1 onnx::Unsqueeze_5425 0=1 MemoryData onnx::Unsqueeze_5427 0 1 onnx::Unsqueeze_5427 0=1 MemoryData onnx::Unsqueeze_5429 0 1 onnx::Unsqueeze_5429 0=1 MemoryData onnx::Unsqueeze_5463 0 1 onnx::Unsqueeze_5463 0=1 MemoryData onnx::Unsqueeze_5465 0 1 onnx::Unsqueeze_5465 0=1 MemoryData onnx::Unsqueeze_5519 0 1 onnx::Unsqueeze_5519 0=1 MemoryData onnx::Unsqueeze_5521 0 1 onnx::Unsqueeze_5521 0=1 MemoryData onnx::Unsqueeze_5528 0 1 onnx::Unsqueeze_5528 0=1 MemoryData onnx::Unsqueeze_5530 0 1 onnx::Unsqueeze_5530 0=1 MemoryData onnx::Unsqueeze_5537 0 1 onnx::Unsqueeze_5537 0=1 MemoryData onnx::Unsqueeze_5539 0 1 onnx::Unsqueeze_5539 0=1 MemoryData onnx::Unsqueeze_5656 0 1 onnx::Unsqueeze_5656 0=1 MemoryData onnx::Unsqueeze_5658 0 1 onnx::Unsqueeze_5658 0=1 MemoryData onnx::Unsqueeze_5660 0 1 onnx::Unsqueeze_5660 0=1 MemoryData onnx::Unsqueeze_5662 0 1 onnx::Unsqueeze_5662 0=1 MemoryData onnx::Unsqueeze_5702 0 1 onnx::Unsqueeze_5702 0=1 MemoryData onnx::Unsqueeze_5704 0 1 onnx::Unsqueeze_5704 0=1 MemoryData onnx::Unsqueeze_5706 0 1 onnx::Unsqueeze_5706 0=1 MemoryData onnx::Unsqueeze_5708 0 1 onnx::Unsqueeze_5708 0=1 MemoryData onnx::Unsqueeze_5742 0 1 onnx::Unsqueeze_5742 0=1 MemoryData onnx::Unsqueeze_5744 0 1 onnx::Unsqueeze_5744 0=1 MemoryData onnx::Unsqueeze_5798 0 1 onnx::Unsqueeze_5798 0=1 MemoryData onnx::Unsqueeze_5800 0 1 onnx::Unsqueeze_5800 0=1 MemoryData onnx::Unsqueeze_5807 0 1 onnx::Unsqueeze_5807 0=1 MemoryData onnx::Unsqueeze_5809 0 1 onnx::Unsqueeze_5809 0=1 MemoryData onnx::Unsqueeze_5816 0 1 onnx::Unsqueeze_5816 0=1 MemoryData onnx::Unsqueeze_5818 0 1 onnx::Unsqueeze_5818 0=1 MemoryData onnx::Unsqueeze_5935 0 1 onnx::Unsqueeze_5935 0=1 MemoryData onnx::Unsqueeze_5937 0 1 onnx::Unsqueeze_5937 0=1 MemoryData onnx::Unsqueeze_5939 0 1 onnx::Unsqueeze_5939 0=1 MemoryData onnx::Unsqueeze_5941 0 1 onnx::Unsqueeze_5941 0=1 MemoryData onnx::Unsqueeze_5981 0 1 onnx::Unsqueeze_5981 0=1 MemoryData onnx::Unsqueeze_5983 0 1 onnx::Unsqueeze_5983 0=1 MemoryData onnx::Unsqueeze_5985 0 1 onnx::Unsqueeze_5985 0=1 MemoryData onnx::Unsqueeze_5987 0 1 onnx::Unsqueeze_5987 0=1 MemoryData onnx::Unsqueeze_6021 0 1 onnx::Unsqueeze_6021 0=1 MemoryData onnx::Unsqueeze_6023 0 1 onnx::Unsqueeze_6023 0=1 MemoryData onnx::Unsqueeze_6077 0 1 onnx::Unsqueeze_6077 0=1 MemoryData onnx::Unsqueeze_6079 0 1 onnx::Unsqueeze_6079 0=1 MemoryData onnx::Unsqueeze_6086 0 1 onnx::Unsqueeze_6086 0=1 MemoryData onnx::Unsqueeze_6088 0 1 onnx::Unsqueeze_6088 0=1 MemoryData onnx::Unsqueeze_6095 0 1 onnx::Unsqueeze_6095 0=1 MemoryData onnx::Unsqueeze_6097 0 1 onnx::Unsqueeze_6097 0=1 MemoryData onnx::Unsqueeze_6214 0 1 onnx::Unsqueeze_6214 0=1 MemoryData onnx::Unsqueeze_6216 0 1 onnx::Unsqueeze_6216 0=1 MemoryData onnx::Unsqueeze_6218 0 1 onnx::Unsqueeze_6218 0=1 MemoryData onnx::Unsqueeze_6220 0 1 onnx::Unsqueeze_6220 0=1 MemoryData onnx::Unsqueeze_6260 0 1 onnx::Unsqueeze_6260 0=1 MemoryData onnx::Unsqueeze_6262 0 1 onnx::Unsqueeze_6262 0=1 MemoryData onnx::Unsqueeze_6264 0 1 onnx::Unsqueeze_6264 0=1 MemoryData onnx::Unsqueeze_6266 0 1 onnx::Unsqueeze_6266 0=1 MemoryData onnx::Unsqueeze_6300 0 1 onnx::Unsqueeze_6300 0=1 MemoryData onnx::Unsqueeze_6302 0 1 onnx::Unsqueeze_6302 0=1 MemoryData onnx::Unsqueeze_633 0 1 onnx::Unsqueeze_633 0=1 MemoryData onnx::Unsqueeze_635 0 1 onnx::Unsqueeze_635 0=1 MemoryData onnx::Unsqueeze_6356 0 1 onnx::Unsqueeze_6356 0=1 MemoryData onnx::Unsqueeze_6358 0 1 onnx::Unsqueeze_6358 0=1 MemoryData onnx::Unsqueeze_6365 0 1 onnx::Unsqueeze_6365 0=1 MemoryData onnx::Unsqueeze_6367 0 1 onnx::Unsqueeze_6367 0=1 MemoryData onnx::Unsqueeze_637 0 1 onnx::Unsqueeze_637 0=1 MemoryData onnx::Unsqueeze_6374 0 1 onnx::Unsqueeze_6374 0=1 MemoryData onnx::Unsqueeze_6376 0 1 onnx::Unsqueeze_6376 0=1 MemoryData onnx::Unsqueeze_639 0 1 onnx::Unsqueeze_639 0=1 MemoryData onnx::Unsqueeze_6493 0 1 onnx::Unsqueeze_6493 0=1 MemoryData onnx::Unsqueeze_6495 0 1 onnx::Unsqueeze_6495 0=1 MemoryData onnx::Unsqueeze_6497 0 1 onnx::Unsqueeze_6497 0=1 MemoryData onnx::Unsqueeze_6499 0 1 onnx::Unsqueeze_6499 0=1 MemoryData onnx::Unsqueeze_6539 0 1 onnx::Unsqueeze_6539 0=1 MemoryData onnx::Unsqueeze_6541 0 1 onnx::Unsqueeze_6541 0=1 MemoryData onnx::Unsqueeze_6543 0 1 onnx::Unsqueeze_6543 0=1 MemoryData onnx::Unsqueeze_6545 0 1 onnx::Unsqueeze_6545 0=1 MemoryData onnx::Unsqueeze_6579 0 1 onnx::Unsqueeze_6579 0=1 MemoryData onnx::Unsqueeze_6581 0 1 onnx::Unsqueeze_6581 0=1 MemoryData onnx::Unsqueeze_6635 0 1 onnx::Unsqueeze_6635 0=1 MemoryData onnx::Unsqueeze_6637 0 1 onnx::Unsqueeze_6637 0=1 MemoryData onnx::Unsqueeze_6644 0 1 onnx::Unsqueeze_6644 0=1 MemoryData onnx::Unsqueeze_6646 0 1 onnx::Unsqueeze_6646 0=1 MemoryData onnx::Unsqueeze_6653 0 1 onnx::Unsqueeze_6653 0=1 MemoryData onnx::Unsqueeze_6655 0 1 onnx::Unsqueeze_6655 0=1 MemoryData onnx::Unsqueeze_6772 0 1 onnx::Unsqueeze_6772 0=1 MemoryData onnx::Unsqueeze_6774 0 1 onnx::Unsqueeze_6774 0=1 MemoryData onnx::Unsqueeze_6776 0 1 onnx::Unsqueeze_6776 0=1 MemoryData onnx::Unsqueeze_6778 0 1 onnx::Unsqueeze_6778 0=1 MemoryData onnx::Unsqueeze_679 0 1 onnx::Unsqueeze_679 0=1 MemoryData onnx::Unsqueeze_681 0 1 onnx::Unsqueeze_681 0=1 MemoryData onnx::Unsqueeze_6818 0 1 onnx::Unsqueeze_6818 0=1 MemoryData onnx::Unsqueeze_6820 0 1 onnx::Unsqueeze_6820 0=1 MemoryData onnx::Unsqueeze_6822 0 1 onnx::Unsqueeze_6822 0=1 MemoryData onnx::Unsqueeze_6824 0 1 onnx::Unsqueeze_6824 0=1 MemoryData onnx::Unsqueeze_683 0 1 onnx::Unsqueeze_683 0=1 MemoryData onnx::Unsqueeze_685 0 1 onnx::Unsqueeze_685 0=1 MemoryData onnx::Unsqueeze_6858 0 1 onnx::Unsqueeze_6858 0=1 MemoryData onnx::Unsqueeze_6860 0 1 onnx::Unsqueeze_6860 0=1 MemoryData onnx::Unsqueeze_6914 0 1 onnx::Unsqueeze_6914 0=1 MemoryData onnx::Unsqueeze_6916 0 1 onnx::Unsqueeze_6916 0=1 MemoryData onnx::Unsqueeze_6923 0 1 onnx::Unsqueeze_6923 0=1 MemoryData onnx::Unsqueeze_6925 0 1 onnx::Unsqueeze_6925 0=1 MemoryData onnx::Unsqueeze_6932 0 1 onnx::Unsqueeze_6932 0=1 MemoryData onnx::Unsqueeze_6934 0 1 onnx::Unsqueeze_6934 0=1 MemoryData onnx::Unsqueeze_7051 0 1 onnx::Unsqueeze_7051 0=1 MemoryData onnx::Unsqueeze_7053 0 1 onnx::Unsqueeze_7053 0=1 MemoryData onnx::Unsqueeze_7055 0 1 onnx::Unsqueeze_7055 0=1 MemoryData onnx::Unsqueeze_7057 0 1 onnx::Unsqueeze_7057 0=1 MemoryData onnx::Unsqueeze_7097 0 1 onnx::Unsqueeze_7097 0=1 MemoryData onnx::Unsqueeze_7099 0 1 onnx::Unsqueeze_7099 0=1 MemoryData onnx::Unsqueeze_7101 0 1 onnx::Unsqueeze_7101 0=1 MemoryData onnx::Unsqueeze_7103 0 1 onnx::Unsqueeze_7103 0=1 MemoryData onnx::Unsqueeze_7137 0 1 onnx::Unsqueeze_7137 0=1 MemoryData onnx::Unsqueeze_7139 0 1 onnx::Unsqueeze_7139 0=1 MemoryData onnx::Unsqueeze_720 0 1 onnx::Unsqueeze_720 0=1 MemoryData onnx::Unsqueeze_722 0 1 onnx::Unsqueeze_722 0=1 MemoryData onnx::Unsqueeze_776 0 1 onnx::Unsqueeze_776 0=1 MemoryData onnx::Unsqueeze_778 0 1 onnx::Unsqueeze_778 0=1 MemoryData onnx::Unsqueeze_785 0 1 onnx::Unsqueeze_785 0=1 MemoryData onnx::Unsqueeze_787 0 1 onnx::Unsqueeze_787 0=1 MemoryData onnx::Unsqueeze_794 0 1 onnx::Unsqueeze_794 0=1 MemoryData onnx::Unsqueeze_796 0 1 onnx::Unsqueeze_796 0=1 MemoryData onnx::Unsqueeze_913 0 1 onnx::Unsqueeze_913 0=1 MemoryData onnx::Unsqueeze_915 0 1 onnx::Unsqueeze_915 0=1 MemoryData onnx::Unsqueeze_917 0 1 onnx::Unsqueeze_917 0=1 MemoryData onnx::Unsqueeze_919 0 1 onnx::Unsqueeze_919 0=1 MemoryData onnx::Unsqueeze_959 0 1 onnx::Unsqueeze_959 0=1 MemoryData onnx::Unsqueeze_961 0 1 onnx::Unsqueeze_961 0=1 MemoryData onnx::Unsqueeze_963 0 1 onnx::Unsqueeze_963 0=1 MemoryData onnx::Unsqueeze_965 0 1 onnx::Unsqueeze_965 0=1 MemoryData onnx::Unsqueeze_999 0 1 onnx::Unsqueeze_999 0=1 Cast /model/Cast_2 1 1 attention_mask /model/Cast_2_output_0 Split splitncnn_1 1 2 /model/Cast_2_output_0 /model/Cast_2_output_0_splitncnn_0 /model/Cast_2_output_0_splitncnn_1 Gather /model/embed_tokens/Gather 2 1 model.embed_tokens.weight_quantized input_ids /model/embed_tokens/Gather_output_0_quantized Shape /model/Shape 1 1 past_key_values.0.key_splitncnn_1 /model/Shape_output_0 Shape /model/rotary_emb/Shape 1 1 position_ids_splitncnn_1 /model/rotary_emb/Shape_output_0 ConstantOfShape /model/ConstantOfShape 1 1 /model/Constant_18_output_0 /model/ConstantOfShape_output_0 ExpandDims /model/rotary_emb/Unsqueeze_1 2 1 position_ids_splitncnn_0 /model/rotary_emb/Constant_6_output_0 /model/rotary_emb/Unsqueeze_1_output_0 -23303=0 Shape /model/Shape_4 1 1 /model/Cast_2_output_0_splitncnn_1 /model/Shape_4_output_0 Flatten /model/Flatten 1 1 /model/Cast_2_output_0_splitncnn_0 /model/Flatten_output_0 DequantizeLinear /model/embed_tokens/Gather_output_0_DequantizeLinear 3 1 /model/embed_tokens/Gather_output_0_quantized model.embed_tokens.weight_scale model.embed_tokens.weight_zero_point /model/embed_tokens/Gather_output_0 Split splitncnn_2 1 3 /model/embed_tokens/Gather_output_0 /model/embed_tokens/Gather_output_0_splitncnn_0 /model/embed_tokens/Gather_output_0_splitncnn_1 /model/embed_tokens/Gather_output_0_splitncnn_2 Crop /model/Gather 1 1 /model/Shape_output_0 /model/Gather_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_3 1 3 /model/Gather_output_0 /model/Gather_output_0_splitncnn_0 /model/Gather_output_0_splitncnn_1 /model/Gather_output_0_splitncnn_2 Crop /model/rotary_emb/Gather 1 1 /model/rotary_emb/Shape_output_0 /model/rotary_emb/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Cast /model/rotary_emb/Cast_1 1 1 /model/rotary_emb/Unsqueeze_1_output_0 /model/rotary_emb/Cast_1_output_0 Crop /model/Gather_4 1 1 /model/Shape_4_output_0 /model/Gather_4_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Shape /model/Shape_1 1 1 /model/embed_tokens/Gather_output_0_splitncnn_2 /model/Shape_1_output_0 Shape /model/Shape_3 1 1 /model/embed_tokens/Gather_output_0_splitncnn_1 /model/Shape_3_output_0 Cast /model/layers.0/input_layernorm/Cast 1 1 /model/embed_tokens/Gather_output_0_splitncnn_0 /model/layers.0/input_layernorm/Cast_output_0 Split splitncnn_4 1 3 /model/layers.0/input_layernorm/Cast_output_0 /model/layers.0/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.0/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.0/input_layernorm/Cast_output_0_splitncnn_2 Cast /model/Cast 1 1 /model/Gather_output_0_splitncnn_2 /model/Cast_output_0 ExpandDims /model/rotary_emb/Unsqueeze 2 1 /model/rotary_emb/Gather_output_0 /model/rotary_emb/Constant_1_output_0 /model/rotary_emb/Unsqueeze_output_0 -23303=0 Cast /model/rotary_emb/Cast_3 1 1 /model/rotary_emb/Cast_1_output_0 /model/rotary_emb/Cast_3_output_0 Crop /model/Gather_1 1 1 /model/Shape_1_output_0 /model/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Crop /model/Gather_3 1 1 /model/Shape_3_output_0 /model/Gather_3_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_5 1 2 /model/Gather_3_output_0 /model/Gather_3_output_0_splitncnn_0 /model/Gather_3_output_0_splitncnn_1 BinaryOp /model/layers.0/input_layernorm/Pow 1 1 /model/layers.0/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.0/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Concat /model/rotary_emb/Concat 3 1 /model/rotary_emb/Unsqueeze_output_0 /model/rotary_emb/Constant_2_output_0 /model/rotary_emb/Constant_3_output_0 /model/rotary_emb/Concat_output_0 0=0 BinaryOp /model/Add 2 1 /model/Gather_output_0_splitncnn_1 /model/Gather_1_output_0 /model/Add_output_0 0=0 Cast /model/Cast_3 1 1 /model/Gather_3_output_0_splitncnn_1 /model/Cast_3_output_0 ExpandDims /model/Unsqueeze_9 2 1 /model/Gather_3_output_0_splitncnn_0 /model/Constant_21_output_0 /model/Unsqueeze_9_output_0 -23303=0 Reduction /model/layers.0/input_layernorm/ReduceMean 2 1 /model/layers.0/input_layernorm/Pow_output_0 /model/layers.0/input_layernorm/Constant_1_output_0 /model/layers.0/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 Reshape /model/rotary_emb/Reshape 1 1 /model/rotary_emb/Concat_output_0 /model/rotary_emb/Reshape_output_0 0=-1 Split splitncnn_6 1 3 /model/rotary_emb/Reshape_output_0 /model/rotary_emb/Reshape_output_0_splitncnn_0 /model/rotary_emb/Reshape_output_0_splitncnn_1 /model/rotary_emb/Reshape_output_0_splitncnn_2 Cast /model/Cast_1 1 1 /model/Add_output_0 /model/Cast_1_output_0 Range /model/Range_1 3 1 /model/Constant_8_output_0 /model/Cast_3_output_0 /model/Constant_9_output_0 /model/Range_1_output_0 BinaryOp /model/layers.0/input_layernorm/Add 1 1 /model/layers.0/input_layernorm/ReduceMean_output_0 /model/layers.0/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 Shape /model/rotary_emb/Shape_1 1 1 /model/rotary_emb/Reshape_output_0_splitncnn_2 /model/rotary_emb/Shape_1_output_0 Range /model/Range 3 1 /model/Cast_output_0 /model/Cast_1_output_0 /model/Constant_2_output_0 /model/Range_output_0 Split splitncnn_7 1 2 /model/Range_output_0 /model/Range_output_0_splitncnn_0 /model/Range_output_0_splitncnn_1 ExpandDims /model/Unsqueeze_3 2 1 /model/Range_1_output_0 /model/Constant_10_output_0 /model/Unsqueeze_3_output_0 -23303=0 UnaryOp /model/layers.0/input_layernorm/Sqrt 1 1 /model/layers.0/input_layernorm/Add_output_0 /model/layers.0/input_layernorm/Sqrt_output_0 0=5 ConstantOfShape /model/rotary_emb/ConstantOfShape 1 1 /model/rotary_emb/Shape_1_output_0 /model/rotary_emb/ConstantOfShape_output_0 Split splitncnn_8 1 2 /model/rotary_emb/ConstantOfShape_output_0 /model/rotary_emb/ConstantOfShape_output_0_splitncnn_0 /model/rotary_emb/ConstantOfShape_output_0_splitncnn_1 Shape /model/Shape_2 1 1 /model/Range_output_0_splitncnn_1 /model/Shape_2_output_0 ExpandDims /model/Unsqueeze 2 1 /model/Range_output_0_splitncnn_0 /model/Constant_5_output_0 /model/Unsqueeze_output_0 -23303=0 ExpandDims /model/Unsqueeze_4 2 1 /model/Unsqueeze_3_output_0 /model/Constant_11_output_0 /model/Unsqueeze_4_output_0 -23303=0 BinaryOp /model/layers.0/input_layernorm/Div 1 1 /model/layers.0/input_layernorm/Sqrt_output_0 /model/layers.0/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/rotary_emb/Mul 2 1 /model/rotary_emb/ConstantOfShape_output_0_splitncnn_1 /model/rotary_emb/Constant_5_output_0 /model/rotary_emb/Mul_output_0 0=2 Crop /model/Gather_2 1 1 /model/Shape_2_output_0 /model/Gather_2_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_9 1 2 /model/Gather_2_output_0 /model/Gather_2_output_0_splitncnn_0 /model/Gather_2_output_0_splitncnn_1 ExpandDims /model/Unsqueeze_1 2 1 /model/Unsqueeze_output_0 /model/Constant_6_output_0 /model/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/Unsqueeze_5 2 1 /model/Unsqueeze_4_output_0 /model/Constant_12_output_0 /model/Unsqueeze_5_output_0 -23303=0 BinaryOp /model/layers.0/input_layernorm/Mul 2 1 /model/layers.0/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.0/input_layernorm/Div_output_0 /model/layers.0/input_layernorm/Mul_output_0 0=2 Equal /model/rotary_emb/Equal 2 1 /model/rotary_emb/Reshape_output_0_splitncnn_1 /model/rotary_emb/Mul_output_0 /model/rotary_emb/Equal_output_0 BinaryOp /model/Add_1 2 1 /model/Gather_output_0_splitncnn_0 /model/Gather_2_output_0_splitncnn_1 /model/Add_1_output_0 0=0 Split splitncnn_10 1 2 /model/Add_1_output_0 /model/Add_1_output_0_splitncnn_0 /model/Add_1_output_0_splitncnn_1 ExpandDims /model/Unsqueeze_10 2 1 /model/Gather_2_output_0_splitncnn_0 /model/Constant_23_output_0 /model/Unsqueeze_10_output_0 -23303=0 ExpandDims /model/Unsqueeze_2 2 1 /model/Unsqueeze_1_output_0 /model/Constant_7_output_0 /model/Unsqueeze_2_output_0 -23303=0 BinaryOp /model/Mul 2 1 /model/Unsqueeze_5_output_0 /model/Gather_4_output_0 /model/Mul_output_0 0=2 Cast /model/layers.0/input_layernorm/Cast_1 1 1 /model/layers.0/input_layernorm/Mul_output_0 /model/layers.0/input_layernorm/Cast_1_output_0 Where /model/rotary_emb/Where 3 1 /model/rotary_emb/Equal_output_0 /model/rotary_emb/ConstantOfShape_output_0_splitncnn_0 /model/rotary_emb/Reshape_output_0_splitncnn_0 /model/rotary_emb/Where_output_0 Cast /model/Cast_4 1 1 /model/Add_1_output_0_splitncnn_1 /model/Cast_4_output_0 ExpandDims /model/Unsqueeze_11 2 1 /model/Add_1_output_0_splitncnn_0 /model/Constant_24_output_0 /model/Unsqueeze_11_output_0 -23303=0 BinaryOp /model/layers.0/input_layernorm/Mul_1 2 1 model.layers.0.input_layernorm.weight /model/layers.0/input_layernorm/Cast_1_output_0 /model/layers.0/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_11 1 3 /model/layers.0/input_layernorm/Mul_1_output_0 /model/layers.0/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.0/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.0/input_layernorm/Mul_1_output_0_splitncnn_2 Expand /model/rotary_emb/Expand 2 1 onnx::Expand_431 /model/rotary_emb/Where_output_0 /model/rotary_emb/Expand_output_0 Range /model/Range_2 3 1 /model/Constant_13_output_0 /model/Cast_4_output_0 /model/Constant_14_output_0 /model/Range_2_output_0 Concat /model/Concat_1 4 1 /model/Unsqueeze_9_output_0 /model/Constant_22_output_0 /model/Unsqueeze_10_output_0 /model/Unsqueeze_11_output_0 /model/Concat_1_output_0 0=0 Shape /model/layers.0/self_attn/Shape 1 1 /model/layers.0/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.0/self_attn/Shape_output_0 Shape /model/layers.0/self_attn/Shape_1 1 1 /model/layers.0/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.0/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.0/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.0/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.0/input_layernorm/Mul_1_output_0_quantized /model/layers.0/input_layernorm/Mul_1_output_0_scale /model/layers.0/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_12 1 3 /model/layers.0/input_layernorm/Mul_1_output_0_quantized /model/layers.0/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.0/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.0/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_13 1 3 /model/layers.0/input_layernorm/Mul_1_output_0_scale /model/layers.0/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.0/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.0/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_14 1 3 /model/layers.0/input_layernorm/Mul_1_output_0_zero_point /model/layers.0/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.0/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.0/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Cast /model/rotary_emb/Cast 1 1 /model/rotary_emb/Expand_output_0 /model/rotary_emb/Cast_output_0 ExpandDims /model/Unsqueeze_6 2 1 /model/Range_2_output_0 /model/Constant_15_output_0 /model/Unsqueeze_6_output_0 -23303=0 Reshape /model/Reshape_2 1 1 /model/Concat_1_output_0 /model/Reshape_2_output_0 0=-1 Split splitncnn_15 1 3 /model/Reshape_2_output_0 /model/Reshape_2_output_0_splitncnn_0 /model/Reshape_2_output_0_splitncnn_1 /model/Reshape_2_output_0_splitncnn_2 Crop /model/layers.0/self_attn/Gather 1 1 /model/layers.0/self_attn/Shape_output_0 /model/layers.0/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_16 1 4 /model/layers.0/self_attn/Gather_output_0 /model/layers.0/self_attn/Gather_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_output_0_splitncnn_1 /model/layers.0/self_attn/Gather_output_0_splitncnn_2 /model/layers.0/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.0/self_attn/Gather_1 1 1 /model/layers.0/self_attn/Shape_1_output_0 /model/layers.0/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_17 1 4 /model/layers.0/self_attn/Gather_1_output_0 /model/layers.0/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.0/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.0/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.0/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.0/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.0/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=9.657972e-03 BinaryOp /model/layers.0/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.0/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.0/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=8.058563e-03 BinaryOp /model/layers.0/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.0/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.0/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=9.534941e-04 MatMulInteger /model/layers.0/self_attn/q_proj/MatMul_quant 4 1 /model/layers.0/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7190_quantized /model/layers.0/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7190_zero_point /model/layers.0/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.0/self_attn/k_proj/MatMul_quant 4 1 /model/layers.0/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7197_quantized /model/layers.0/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7197_zero_point /model/layers.0/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.0/self_attn/v_proj/MatMul_quant 4 1 /model/layers.0/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7198_quantized /model/layers.0/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7198_zero_point /model/layers.0/self_attn/v_proj/MatMul_output_0_output_quantized Cast /model/rotary_emb/Cast_2 1 1 /model/rotary_emb/Cast_output_0 /model/rotary_emb/Cast_2_output_0 ExpandDims /model/Unsqueeze_7 2 1 /model/Unsqueeze_6_output_0 /model/Constant_16_output_0 /model/Unsqueeze_7_output_0 -23303=0 Shape /model/Shape_6 1 1 /model/Reshape_2_output_0_splitncnn_2 /model/Shape_6_output_0 ExpandDims /model/layers.0/self_attn/Unsqueeze 2 1 /model/layers.0/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_491 /model/layers.0/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_2 2 1 /model/layers.0/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_500 /model/layers.0/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_4 2 1 /model/layers.0/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_509 /model/layers.0/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_31 2 1 /model/layers.0/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_720 /model/layers.0/self_attn/Unsqueeze_31_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_1 2 1 /model/layers.0/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_493 /model/layers.0/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_3 2 1 /model/layers.0/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_502 /model/layers.0/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_5 2 1 /model/layers.0/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_511 /model/layers.0/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_32 2 1 /model/layers.0/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_722 /model/layers.0/self_attn/Unsqueeze_32_output_0 -23303=0 Cast /model/layers.0/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.0/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.0/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.0/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.0/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.0/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.0/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.0/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.0/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Gemm /model/rotary_emb/MatMul 2 1 /model/rotary_emb/Cast_2_output_0 /model/rotary_emb/Cast_3_output_0 /model/rotary_emb/MatMul_output_0 ExpandDims /model/Unsqueeze_8 2 1 /model/Unsqueeze_7_output_0 /model/Constant_17_output_0 /model/Unsqueeze_8_output_0 -23303=0 Split splitncnn_18 1 2 /model/Unsqueeze_8_output_0 /model/Unsqueeze_8_output_0_splitncnn_0 /model/Unsqueeze_8_output_0_splitncnn_1 ConstantOfShape /model/ConstantOfShape_1 1 1 /model/Shape_6_output_0 /model/ConstantOfShape_1_output_0 Split splitncnn_19 1 2 /model/ConstantOfShape_1_output_0 /model/ConstantOfShape_1_output_0_splitncnn_0 /model/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.0/self_attn/Concat 4 1 /model/layers.0/self_attn/Unsqueeze_output_0 /model/layers.0/self_attn/Unsqueeze_1_output_0 /model/layers.0/self_attn/Constant_2_output_0 /model/layers.0/self_attn/Constant_3_output_0 /model/layers.0/self_attn/Concat_output_0 0=0 Concat /model/layers.0/self_attn/Concat_1 4 1 /model/layers.0/self_attn/Unsqueeze_2_output_0 /model/layers.0/self_attn/Unsqueeze_3_output_0 /model/layers.0/self_attn/Constant_4_output_0 /model/layers.0/self_attn/Constant_5_output_0 /model/layers.0/self_attn/Concat_1_output_0 0=0 Concat /model/layers.0/self_attn/Concat_2 4 1 /model/layers.0/self_attn/Unsqueeze_4_output_0 /model/layers.0/self_attn/Unsqueeze_5_output_0 /model/layers.0/self_attn/Constant_6_output_0 /model/layers.0/self_attn/Constant_7_output_0 /model/layers.0/self_attn/Concat_2_output_0 0=0 Concat /model/layers.0/self_attn/Concat_11 3 1 /model/layers.0/self_attn/Unsqueeze_31_output_0 /model/layers.0/self_attn/Unsqueeze_32_output_0 /model/layers.0/self_attn/Constant_66_output_0 /model/layers.0/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.0/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.0/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.0/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.0/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.0/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.0/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.0/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.0/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.0/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.0/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.0/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.0/self_attn/v_proj/MatMul_output_0 0=2 Permute /model/rotary_emb/Transpose 1 1 /model/rotary_emb/MatMul_output_0 /model/rotary_emb/Transpose_output_0 0=1 Split splitncnn_20 1 2 /model/rotary_emb/Transpose_output_0 /model/rotary_emb/Transpose_output_0_splitncnn_0 /model/rotary_emb/Transpose_output_0_splitncnn_1 LessOrEqual /model/LessOrEqual 2 1 /model/Unsqueeze_8_output_0_splitncnn_1 /model/Unsqueeze_2_output_0 /model/LessOrEqual_output_0 BinaryOp /model/Add_2 2 1 /model/Unsqueeze_8_output_0_splitncnn_0 /model/Mul_output_0 /model/Add_2_output_0 0=0 Split splitncnn_21 1 2 /model/Add_2_output_0 /model/Add_2_output_0_splitncnn_0 /model/Add_2_output_0_splitncnn_1 BinaryOp /model/Mul_1 2 1 /model/ConstantOfShape_1_output_0_splitncnn_1 /model/Constant_26_output_0 /model/Mul_1_output_0 0=2 BinaryOp /model/layers.0/self_attn/q_proj/Add 2 1 model.layers.0.self_attn.q_proj.bias /model/layers.0/self_attn/q_proj/MatMul_output_0 /model/layers.0/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.0/self_attn/k_proj/Add 2 1 model.layers.0.self_attn.k_proj.bias /model/layers.0/self_attn/k_proj/MatMul_output_0 /model/layers.0/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.0/self_attn/v_proj/Add 2 1 model.layers.0.self_attn.v_proj.bias /model/layers.0/self_attn/v_proj/MatMul_output_0 /model/layers.0/self_attn/v_proj/Add_output_0 0=0 Concat /model/rotary_emb/Concat_1 2 1 /model/rotary_emb/Transpose_output_0_splitncnn_1 /model/rotary_emb/Transpose_output_0_splitncnn_0 /model/rotary_emb/Concat_1_output_0 0=-1 Split splitncnn_22 1 2 /model/rotary_emb/Concat_1_output_0 /model/rotary_emb/Concat_1_output_0_splitncnn_0 /model/rotary_emb/Concat_1_output_0_splitncnn_1 Cast /model/Cast_5 1 1 /model/LessOrEqual_output_0 /model/Cast_5_output_0 Gather /model/Gather_5 2 1 /model/Flatten_output_0 /model/Add_2_output_0_splitncnn_1 /model/Gather_5_output_0 Shape /model/Shape_5 1 1 /model/Add_2_output_0_splitncnn_0 /model/Shape_5_output_0 Equal /model/Equal 2 1 /model/Reshape_2_output_0_splitncnn_1 /model/Mul_1_output_0 /model/Equal_output_0 Reshape /model/layers.0/self_attn/Reshape 2 1 /model/layers.0/self_attn/q_proj/Add_output_0 /model/layers.0/self_attn/Concat_output_0 /model/layers.0/self_attn/Reshape_output_0 Reshape /model/layers.0/self_attn/Reshape_1 2 1 /model/layers.0/self_attn/k_proj/Add_output_0 /model/layers.0/self_attn/Concat_1_output_0 /model/layers.0/self_attn/Reshape_1_output_0 Reshape /model/layers.0/self_attn/Reshape_2 2 1 /model/layers.0/self_attn/v_proj/Add_output_0 /model/layers.0/self_attn/Concat_2_output_0 /model/layers.0/self_attn/Reshape_2_output_0 UnaryOp /model/rotary_emb/Cos 1 1 /model/rotary_emb/Concat_1_output_0_splitncnn_1 /model/rotary_emb/Cos_output_0 0=10 UnaryOp /model/rotary_emb/Sin 1 1 /model/rotary_emb/Concat_1_output_0_splitncnn_0 /model/rotary_emb/Sin_output_0 0=9 And /model/And 2 1 /model/ConstantOfShape_output_0 /model/Cast_5_output_0 /model/And_output_0 Reshape /model/Reshape 1 1 /model/Gather_5_output_0 /model/Reshape_output_0 0=-1 Concat /model/Concat 1 1 /model/Shape_5_output_0 /model/Concat_output_0 0=0 Where /model/Where 3 1 /model/Equal_output_0 /model/ConstantOfShape_1_output_0_splitncnn_0 /model/Reshape_2_output_0_splitncnn_0 /model/Where_output_0 Permute /model/layers.0/self_attn/Transpose 1 1 /model/layers.0/self_attn/Reshape_output_0 /model/layers.0/self_attn/Transpose_output_0 0=2 Split splitncnn_23 1 4 /model/layers.0/self_attn/Transpose_output_0 /model/layers.0/self_attn/Transpose_output_0_splitncnn_0 /model/layers.0/self_attn/Transpose_output_0_splitncnn_1 /model/layers.0/self_attn/Transpose_output_0_splitncnn_2 /model/layers.0/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.0/self_attn/Transpose_1 1 1 /model/layers.0/self_attn/Reshape_1_output_0 /model/layers.0/self_attn/Transpose_1_output_0 0=2 Split splitncnn_24 1 4 /model/layers.0/self_attn/Transpose_1_output_0 /model/layers.0/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.0/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.0/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.0/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.0/self_attn/Transpose_2 1 1 /model/layers.0/self_attn/Reshape_2_output_0 /model/layers.0/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/rotary_emb/Mul_1 1 1 /model/rotary_emb/Cos_output_0 /model/rotary_emb/Mul_1_output_0 0=2 1=1 2=1.000000e+00 BinaryOp /model/rotary_emb/Mul_2 1 1 /model/rotary_emb/Sin_output_0 /model/rotary_emb/Mul_2_output_0 0=2 1=1 2=1.000000e+00 Reshape /model/Reshape_1 2 1 /model/Reshape_output_0 /model/Concat_output_0 /model/Reshape_1_output_0 Shape /model/layers.0/self_attn/Shape_2 1 1 /model/layers.0/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Shape_2_output_0 Shape /model/layers.0/self_attn/Shape_3 1 1 /model/layers.0/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Shape_3_output_0 Concat /model/layers.0/self_attn/Concat_6 2 1 past_key_values.0.value /model/layers.0/self_attn/Transpose_2_output_0 present.0.value 0=-2 Split splitncnn_25 1 5 present.0.value present.0.value_splitncnn_0 present.0.value_splitncnn_1 present.0.value_splitncnn_2 present.0.value_splitncnn_3 present.0.value_splitncnn_4 Cast /model/rotary_emb/Cast_4 1 1 /model/rotary_emb/Mul_1_output_0 /model/rotary_emb/Cast_4_output_0 Cast /model/rotary_emb/Cast_5 1 1 /model/rotary_emb/Mul_2_output_0 /model/rotary_emb/Cast_5_output_0 Cast /model/Cast_6 1 1 /model/Reshape_1_output_0 /model/Cast_6_output_0 Crop /model/layers.0/self_attn/Gather_2 1 1 /model/layers.0/self_attn/Shape_2_output_0 /model/layers.0/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.0/self_attn/Gather_3 1 1 /model/layers.0/self_attn/Shape_3_output_0 /model/layers.0/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.0/self_attn/Shape_9 1 1 present.0.value_splitncnn_4 /model/layers.0/self_attn/Shape_9_output_0 Shape /model/layers.0/self_attn/Shape_10 1 1 present.0.value_splitncnn_3 /model/layers.0/self_attn/Shape_10_output_0 Shape /model/layers.0/self_attn/Shape_11 1 1 present.0.value_splitncnn_2 /model/layers.0/self_attn/Shape_11_output_0 Shape /model/layers.0/self_attn/Shape_12 1 1 present.0.value_splitncnn_1 /model/layers.0/self_attn/Shape_12_output_0 ExpandDims /model/layers.0/self_attn/Unsqueeze_21 2 1 present.0.value_splitncnn_0 /model/layers.0/self_attn/Constant_47_output_0 /model/layers.0/self_attn/Unsqueeze_21_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_6 2 1 /model/rotary_emb/Cast_4_output_0 /model/layers.0/self_attn/Constant_8_output_0 /model/layers.0/self_attn/Unsqueeze_6_output_0 -23303=0 Split splitncnn_26 1 48 /model/layers.0/self_attn/Unsqueeze_6_output_0 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_0 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_1 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_2 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_4 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_5 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_6 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_7 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_8 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_9 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_10 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_11 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_12 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_13 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_14 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_15 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_16 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_17 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_18 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_19 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_20 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_21 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_22 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_23 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_24 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_25 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_26 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_27 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_28 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_29 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_30 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_31 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_32 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_33 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_34 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_35 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_36 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_37 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_38 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_39 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_40 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_41 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_42 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_43 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_44 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_45 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_46 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_47 ExpandDims /model/layers.0/self_attn/Unsqueeze_7 2 1 /model/rotary_emb/Cast_5_output_0 /model/layers.0/self_attn/Constant_9_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0 -23303=0 Split splitncnn_27 1 48 /model/layers.0/self_attn/Unsqueeze_7_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_1 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_2 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_4 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_5 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_6 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_7 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_8 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_9 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_10 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_11 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_12 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_13 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_14 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_15 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_16 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_17 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_18 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_19 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_20 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_21 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_22 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_23 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_24 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_25 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_26 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_27 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_28 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_29 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_30 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_31 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_32 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_33 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_34 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_35 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_36 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_37 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_38 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_39 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_40 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_41 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_42 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_43 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_44 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_45 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_46 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_47 And /model/And_1 2 1 /model/And_output_0 /model/Cast_6_output_0 /model/And_1_output_0 BinaryOp /model/layers.0/self_attn/Div 2 1 /model/layers.0/self_attn/Gather_2_output_0 /model/layers.0/self_attn/Constant_11_output_0 /model/layers.0/self_attn/Div_output_0 0=3 BinaryOp /model/layers.0/self_attn/Div_1 2 1 /model/layers.0/self_attn/Gather_3_output_0 /model/layers.0/self_attn/Constant_21_output_0 /model/layers.0/self_attn/Div_1_output_0 0=3 Crop /model/layers.0/self_attn/Gather_8 1 1 /model/layers.0/self_attn/Shape_9_output_0 /model/layers.0/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_28 1 2 /model/layers.0/self_attn/Gather_8_output_0 /model/layers.0/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.0/self_attn/Gather_9 1 1 /model/layers.0/self_attn/Shape_10_output_0 /model/layers.0/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_29 1 2 /model/layers.0/self_attn/Gather_9_output_0 /model/layers.0/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.0/self_attn/Gather_10 1 1 /model/layers.0/self_attn/Shape_11_output_0 /model/layers.0/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_30 1 2 /model/layers.0/self_attn/Gather_10_output_0 /model/layers.0/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.0/self_attn/Gather_11 1 1 /model/layers.0/self_attn/Shape_12_output_0 /model/layers.0/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_31 1 2 /model/layers.0/self_attn/Gather_11_output_0 /model/layers.0/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_11_output_0_splitncnn_1 BinaryOp /model/layers.0/self_attn/Mul 2 1 /model/layers.0/self_attn/Transpose_output_0_splitncnn_2 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_47 /model/layers.0/self_attn/Mul_output_0 0=2 BinaryOp /model/layers.0/self_attn/Mul_2 2 1 /model/layers.0/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_46 /model/layers.0/self_attn/Mul_2_output_0 0=2 Expand /model/Expand 2 1 /model/And_1_output_0 /model/Where_output_0 /model/Expand_output_0 Split splitncnn_32 1 24 /model/Expand_output_0 /model/Expand_output_0_splitncnn_0 /model/Expand_output_0_splitncnn_1 /model/Expand_output_0_splitncnn_2 /model/Expand_output_0_splitncnn_3 /model/Expand_output_0_splitncnn_4 /model/Expand_output_0_splitncnn_5 /model/Expand_output_0_splitncnn_6 /model/Expand_output_0_splitncnn_7 /model/Expand_output_0_splitncnn_8 /model/Expand_output_0_splitncnn_9 /model/Expand_output_0_splitncnn_10 /model/Expand_output_0_splitncnn_11 /model/Expand_output_0_splitncnn_12 /model/Expand_output_0_splitncnn_13 /model/Expand_output_0_splitncnn_14 /model/Expand_output_0_splitncnn_15 /model/Expand_output_0_splitncnn_16 /model/Expand_output_0_splitncnn_17 /model/Expand_output_0_splitncnn_18 /model/Expand_output_0_splitncnn_19 /model/Expand_output_0_splitncnn_20 /model/Expand_output_0_splitncnn_21 /model/Expand_output_0_splitncnn_22 /model/Expand_output_0_splitncnn_23 Cast /model/layers.0/self_attn/Cast 1 1 /model/layers.0/self_attn/Div_output_0 /model/layers.0/self_attn/Cast_output_0 Cast /model/layers.0/self_attn/Cast_2 1 1 /model/layers.0/self_attn/Div_1_output_0 /model/layers.0/self_attn/Cast_2_output_0 ExpandDims /model/layers.0/self_attn/Unsqueeze_22 2 1 /model/layers.0/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_48_output_0 /model/layers.0/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_26 2 1 /model/layers.0/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_679 /model/layers.0/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_23 2 1 /model/layers.0/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_49_output_0 /model/layers.0/self_attn/Unsqueeze_23_output_0 -23303=0 BinaryOp /model/layers.0/self_attn/Mul_7 2 1 /model/layers.0/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.0/self_attn/Constant_55_output_0 /model/layers.0/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.0/self_attn/Unsqueeze_24 2 1 /model/layers.0/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_51_output_0 /model/layers.0/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_28 2 1 /model/layers.0/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_683 /model/layers.0/self_attn/Unsqueeze_28_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_25 2 1 /model/layers.0/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_52_output_0 /model/layers.0/self_attn/Unsqueeze_25_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_29 2 1 /model/layers.0/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_685 /model/layers.0/self_attn/Unsqueeze_29_output_0 -23303=0 Cast /model/layers.0/self_attn/Cast_1 1 1 /model/layers.0/self_attn/Cast_output_0 /model/layers.0/self_attn/Cast_1_output_0 Split splitncnn_33 1 2 /model/layers.0/self_attn/Cast_1_output_0 /model/layers.0/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.0/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.0/self_attn/Cast_3 1 1 /model/layers.0/self_attn/Cast_2_output_0 /model/layers.0/self_attn/Cast_3_output_0 Split splitncnn_34 1 2 /model/layers.0/self_attn/Cast_3_output_0 /model/layers.0/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.0/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.0/self_attn/Unsqueeze_27 2 1 /model/layers.0/self_attn/Mul_7_output_0 onnx::Unsqueeze_681 /model/layers.0/self_attn/Unsqueeze_27_output_0 -23303=0 Concat /model/layers.0/self_attn/Concat_9 5 1 /model/layers.0/self_attn/Unsqueeze_22_output_0 /model/layers.0/self_attn/Unsqueeze_23_output_0 /model/layers.0/self_attn/Constant_50_output_0 /model/layers.0/self_attn/Unsqueeze_24_output_0 /model/layers.0/self_attn/Unsqueeze_25_output_0 /model/layers.0/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_8 2 1 /model/layers.0/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_14_output_0 /model/layers.0/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_9 2 1 /model/layers.0/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.0/self_attn/Constant_17_output_0 /model/layers.0/self_attn/Unsqueeze_9_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_10 2 1 /model/layers.0/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_24_output_0 /model/layers.0/self_attn/Unsqueeze_10_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_11 2 1 /model/layers.0/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.0/self_attn/Constant_27_output_0 /model/layers.0/self_attn/Unsqueeze_11_output_0 -23303=0 Concat /model/layers.0/self_attn/Concat_10 4 1 /model/layers.0/self_attn/Unsqueeze_26_output_0 /model/layers.0/self_attn/Unsqueeze_27_output_0 /model/layers.0/self_attn/Unsqueeze_28_output_0 /model/layers.0/self_attn/Unsqueeze_29_output_0 /model/layers.0/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.0/self_attn/Reshape_5 1 1 /model/layers.0/self_attn/Concat_9_output_0 /model/layers.0/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_35 1 3 /model/layers.0/self_attn/Reshape_5_output_0 /model/layers.0/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.0/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.0/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.0/self_attn/Slice 2 1 /model/layers.0/self_attn/Transpose_output_0_splitncnn_1 /model/layers.0/self_attn/Unsqueeze_8_output_0 /model/layers.0/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.0/self_attn/Slice_1 2 1 /model/layers.0/self_attn/Transpose_output_0_splitncnn_0 /model/layers.0/self_attn/Unsqueeze_9_output_0 /model/layers.0/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.0/self_attn/Slice_2 2 1 /model/layers.0/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.0/self_attn/Unsqueeze_10_output_0 /model/layers.0/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.0/self_attn/Slice_3 2 1 /model/layers.0/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.0/self_attn/Unsqueeze_11_output_0 /model/layers.0/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.0/self_attn/Shape_13 1 1 /model/layers.0/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.0/self_attn/Shape_13_output_0 UnaryOp /model/layers.0/self_attn/Neg 1 1 /model/layers.0/self_attn/Slice_1_output_0 /model/layers.0/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.0/self_attn/Neg_1 1 1 /model/layers.0/self_attn/Slice_3_output_0 /model/layers.0/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.0/self_attn/ConstantOfShape_1 1 1 /model/layers.0/self_attn/Shape_13_output_0 /model/layers.0/self_attn/ConstantOfShape_1_output_0 Split splitncnn_36 1 2 /model/layers.0/self_attn/ConstantOfShape_1_output_0 /model/layers.0/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.0/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.0/self_attn/Concat_3 2 1 /model/layers.0/self_attn/Neg_output_0 /model/layers.0/self_attn/Slice_output_0 /model/layers.0/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.0/self_attn/Concat_4 2 1 /model/layers.0/self_attn/Neg_1_output_0 /model/layers.0/self_attn/Slice_2_output_0 /model/layers.0/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.0/self_attn/Mul_6 2 1 /model/layers.0/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_54_output_0 /model/layers.0/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.0/self_attn/Mul_1 2 1 /model/layers.0/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_47 /model/layers.0/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.0/self_attn/Mul_3 2 1 /model/layers.0/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_46 /model/layers.0/self_attn/Mul_3_output_0 0=2 Equal /model/layers.0/self_attn/Equal_1 2 1 /model/layers.0/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.0/self_attn/Mul_6_output_0 /model/layers.0/self_attn/Equal_1_output_0 BinaryOp /model/layers.0/self_attn/Add 2 1 /model/layers.0/self_attn/Mul_output_0 /model/layers.0/self_attn/Mul_1_output_0 /model/layers.0/self_attn/Add_output_0 0=0 BinaryOp /model/layers.0/self_attn/Add_1 2 1 /model/layers.0/self_attn/Mul_2_output_0 /model/layers.0/self_attn/Mul_3_output_0 /model/layers.0/self_attn/Add_1_output_0 0=0 Where /model/layers.0/self_attn/Where_1 3 1 /model/layers.0/self_attn/Equal_1_output_0 /model/layers.0/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.0/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.0/self_attn/Where_1_output_0 BinaryOp /model/layers.0/self_attn/Mul_8 1 1 /model/layers.0/self_attn/Add_output_0 /model/layers.0/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.0/self_attn/Concat_5 2 1 past_key_values.0.key_splitncnn_0 /model/layers.0/self_attn/Add_1_output_0 present.0.key 0=-2 Split splitncnn_37 1 5 present.0.key present.0.key_splitncnn_0 present.0.key_splitncnn_1 present.0.key_splitncnn_2 present.0.key_splitncnn_3 present.0.key_splitncnn_4 Expand /model/layers.0/self_attn/Expand_1 2 1 /model/layers.0/self_attn/Unsqueeze_21_output_0 /model/layers.0/self_attn/Where_1_output_0 /model/layers.0/self_attn/Expand_1_output_0 Shape /model/layers.0/self_attn/Shape_4 1 1 present.0.key_splitncnn_4 /model/layers.0/self_attn/Shape_4_output_0 Shape /model/layers.0/self_attn/Shape_5 1 1 present.0.key_splitncnn_3 /model/layers.0/self_attn/Shape_5_output_0 Shape /model/layers.0/self_attn/Shape_6 1 1 present.0.key_splitncnn_2 /model/layers.0/self_attn/Shape_6_output_0 Shape /model/layers.0/self_attn/Shape_7 1 1 present.0.key_splitncnn_1 /model/layers.0/self_attn/Shape_7_output_0 ExpandDims /model/layers.0/self_attn/Unsqueeze_12 2 1 present.0.key_splitncnn_0 /model/layers.0/self_attn/Constant_34_output_0 /model/layers.0/self_attn/Unsqueeze_12_output_0 -23303=0 Reshape /model/layers.0/self_attn/Reshape_6 2 1 /model/layers.0/self_attn/Expand_1_output_0 /model/layers.0/self_attn/Concat_10_output_0 /model/layers.0/self_attn/Reshape_6_output_0 Crop /model/layers.0/self_attn/Gather_4 1 1 /model/layers.0/self_attn/Shape_4_output_0 /model/layers.0/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_38 1 2 /model/layers.0/self_attn/Gather_4_output_0 /model/layers.0/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.0/self_attn/Gather_5 1 1 /model/layers.0/self_attn/Shape_5_output_0 /model/layers.0/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_39 1 2 /model/layers.0/self_attn/Gather_5_output_0 /model/layers.0/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.0/self_attn/Gather_6 1 1 /model/layers.0/self_attn/Shape_6_output_0 /model/layers.0/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_40 1 2 /model/layers.0/self_attn/Gather_6_output_0 /model/layers.0/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.0/self_attn/Gather_7 1 1 /model/layers.0/self_attn/Shape_7_output_0 /model/layers.0/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_41 1 2 /model/layers.0/self_attn/Gather_7_output_0 /model/layers.0/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.0/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.0/self_attn/Unsqueeze_13 2 1 /model/layers.0/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_35_output_0 /model/layers.0/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_17 2 1 /model/layers.0/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_633 /model/layers.0/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_14 2 1 /model/layers.0/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_36_output_0 /model/layers.0/self_attn/Unsqueeze_14_output_0 -23303=0 BinaryOp /model/layers.0/self_attn/Mul_5 2 1 /model/layers.0/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.0/self_attn/Constant_42_output_0 /model/layers.0/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.0/self_attn/Unsqueeze_15 2 1 /model/layers.0/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_38_output_0 /model/layers.0/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_19 2 1 /model/layers.0/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_637 /model/layers.0/self_attn/Unsqueeze_19_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_16 2 1 /model/layers.0/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_39_output_0 /model/layers.0/self_attn/Unsqueeze_16_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_20 2 1 /model/layers.0/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_639 /model/layers.0/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.0/self_attn/Unsqueeze_18 2 1 /model/layers.0/self_attn/Mul_5_output_0 onnx::Unsqueeze_635 /model/layers.0/self_attn/Unsqueeze_18_output_0 -23303=0 Concat /model/layers.0/self_attn/Concat_7 5 1 /model/layers.0/self_attn/Unsqueeze_13_output_0 /model/layers.0/self_attn/Unsqueeze_14_output_0 /model/layers.0/self_attn/Constant_37_output_0 /model/layers.0/self_attn/Unsqueeze_15_output_0 /model/layers.0/self_attn/Unsqueeze_16_output_0 /model/layers.0/self_attn/Concat_7_output_0 0=0 Concat /model/layers.0/self_attn/Concat_8 4 1 /model/layers.0/self_attn/Unsqueeze_17_output_0 /model/layers.0/self_attn/Unsqueeze_18_output_0 /model/layers.0/self_attn/Unsqueeze_19_output_0 /model/layers.0/self_attn/Unsqueeze_20_output_0 /model/layers.0/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.0/self_attn/Reshape_3 1 1 /model/layers.0/self_attn/Concat_7_output_0 /model/layers.0/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_42 1 3 /model/layers.0/self_attn/Reshape_3_output_0 /model/layers.0/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.0/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.0/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.0/self_attn/Shape_8 1 1 /model/layers.0/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.0/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.0/self_attn/ConstantOfShape 1 1 /model/layers.0/self_attn/Shape_8_output_0 /model/layers.0/self_attn/ConstantOfShape_output_0 Split splitncnn_43 1 2 /model/layers.0/self_attn/ConstantOfShape_output_0 /model/layers.0/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.0/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.0/self_attn/Mul_4 2 1 /model/layers.0/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.0/self_attn/Constant_41_output_0 /model/layers.0/self_attn/Mul_4_output_0 0=2 Equal /model/layers.0/self_attn/Equal 2 1 /model/layers.0/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.0/self_attn/Mul_4_output_0 /model/layers.0/self_attn/Equal_output_0 Where /model/layers.0/self_attn/Where 3 1 /model/layers.0/self_attn/Equal_output_0 /model/layers.0/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.0/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.0/self_attn/Where_output_0 Expand /model/layers.0/self_attn/Expand 2 1 /model/layers.0/self_attn/Unsqueeze_12_output_0 /model/layers.0/self_attn/Where_output_0 /model/layers.0/self_attn/Expand_output_0 Reshape /model/layers.0/self_attn/Reshape_4 2 1 /model/layers.0/self_attn/Expand_output_0 /model/layers.0/self_attn/Concat_8_output_0 /model/layers.0/self_attn/Reshape_4_output_0 Split splitncnn_44 1 2 /model/layers.0/self_attn/Reshape_4_output_0 /model/layers.0/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.0/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.0/self_attn/Shape_14 1 1 /model/layers.0/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.0/self_attn/Shape_14_output_0 Permute /model/layers.0/self_attn/Transpose_3 1 1 /model/layers.0/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.0/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.0/self_attn/Gather_12 1 1 /model/layers.0/self_attn/Shape_14_output_0 /model/layers.0/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.0/self_attn/Mul_9 1 1 /model/layers.0/self_attn/Transpose_3_output_0 /model/layers.0/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.0/self_attn/Unsqueeze_30 2 1 /model/layers.0/self_attn/Gather_12_output_0 /model/layers.0/self_attn/Constant_59_output_0 /model/layers.0/self_attn/Unsqueeze_30_output_0 -23303=0 Gemm /model/layers.0/self_attn/MatMul 2 1 /model/layers.0/self_attn/Mul_8_output_0 /model/layers.0/self_attn/Mul_9_output_0 /model/layers.0/self_attn/MatMul_output_0 Crop /model/layers.0/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_23 /model/layers.0/self_attn/Unsqueeze_30_output_0 /model/layers.0/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.0/self_attn/Where_2 3 1 /model/layers.0/self_attn/Slice_4_output_0 /model/layers.0/self_attn/Constant_63_output_0 /model/layers.0/self_attn/Constant_64_output_0 /model/layers.0/self_attn/Where_2_output_0 BinaryOp /model/layers.0/self_attn/Add_2 2 1 /model/layers.0/self_attn/MatMul_output_0 /model/layers.0/self_attn/Where_2_output_0 /model/layers.0/self_attn/Add_2_output_0 0=0 Softmax /model/layers.0/self_attn/Softmax 1 1 /model/layers.0/self_attn/Add_2_output_0 /model/layers.0/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_45 1 2 /model/layers.0/self_attn/Softmax_output_0 /model/layers.0/self_attn/Softmax_output_0_splitncnn_0 /model/layers.0/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.0/self_attn/IsNaN 1 1 /model/layers.0/self_attn/Softmax_output_0_splitncnn_1 /model/layers.0/self_attn/IsNaN_output_0 Where /model/layers.0/self_attn/Where_3 3 1 /model/layers.0/self_attn/IsNaN_output_0 /model/layers.0/self_attn/Constant_65_output_0 /model/layers.0/self_attn/Softmax_output_0_splitncnn_0 /model/layers.0/self_attn/Where_3_output_0 Gemm /model/layers.0/self_attn/MatMul_1 2 1 /model/layers.0/self_attn/Where_3_output_0 /model/layers.0/self_attn/Reshape_6_output_0 /model/layers.0/self_attn/MatMul_1_output_0 Permute /model/layers.0/self_attn/Transpose_4 1 1 /model/layers.0/self_attn/MatMul_1_output_0 /model/layers.0/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.0/self_attn/Reshape_7 2 1 /model/layers.0/self_attn/Transpose_4_output_0 /model/layers.0/self_attn/Concat_11_output_0 /model/layers.0/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.0/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.0/self_attn/Reshape_7_output_0 /model/layers.0/self_attn/Reshape_7_output_0_quantized /model/layers.0/self_attn/Reshape_7_output_0_scale /model/layers.0/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.0/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.0/self_attn/Reshape_7_output_0_scale /model/layers.0/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.429134e-03 MatMulInteger /model/layers.0/self_attn/o_proj/MatMul_quant 4 1 /model/layers.0/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7221_quantized /model/layers.0/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7221_zero_point /model/layers.0/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.0/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.0/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.0/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.0/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.0/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.0/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.0/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.0/Add 2 1 /model/layers.0/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.0/self_attn/o_proj/MatMul_output_0 /model/layers.0/Add_output_0 0=0 Cast /model/layers.0/post_attention_layernorm/Cast 1 1 /model/layers.0/Add_output_0 /model/layers.0/post_attention_layernorm/Cast_output_0 Split splitncnn_46 1 3 /model/layers.0/post_attention_layernorm/Cast_output_0 /model/layers.0/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.0/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.0/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.0/post_attention_layernorm/Pow 1 1 /model/layers.0/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.0/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.0/post_attention_layernorm/ReduceMean 2 1 /model/layers.0/post_attention_layernorm/Pow_output_0 /model/layers.0/post_attention_layernorm/Constant_1_output_0 /model/layers.0/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.0/post_attention_layernorm/Add 1 1 /model/layers.0/post_attention_layernorm/ReduceMean_output_0 /model/layers.0/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.0/post_attention_layernorm/Sqrt 1 1 /model/layers.0/post_attention_layernorm/Add_output_0 /model/layers.0/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.0/post_attention_layernorm/Div 1 1 /model/layers.0/post_attention_layernorm/Sqrt_output_0 /model/layers.0/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.0/post_attention_layernorm/Mul 2 1 /model/layers.0/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.0/post_attention_layernorm/Div_output_0 /model/layers.0/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.0/post_attention_layernorm/Cast_1 1 1 /model/layers.0/post_attention_layernorm/Mul_output_0 /model/layers.0/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.0/post_attention_layernorm/Mul_1 2 1 model.layers.0.post_attention_layernorm.weight /model/layers.0/post_attention_layernorm/Cast_1_output_0 /model/layers.0/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.0/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.0/post_attention_layernorm/Mul_1_output_0 /model/layers.0/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.0/post_attention_layernorm/Mul_1_output_0_scale /model/layers.0/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_47 1 2 /model/layers.0/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.0/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.0/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_48 1 2 /model/layers.0/post_attention_layernorm/Mul_1_output_0_scale /model/layers.0/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.0/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_49 1 2 /model/layers.0/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.0/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.0/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.0/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.0/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.0/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.875492e-03 BinaryOp /model/layers.0/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.0/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.0/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.199188e-03 MatMulInteger /model/layers.0/mlp/gate_proj/MatMul_quant 4 1 /model/layers.0/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7222_quantized /model/layers.0/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7222_zero_point /model/layers.0/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.0/mlp/up_proj/MatMul_quant 4 1 /model/layers.0/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7223_quantized /model/layers.0/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7223_zero_point /model/layers.0/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.0/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.0/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.0/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.0/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.0/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.0/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.0/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.0/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.0/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.0/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.0/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.0/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.0/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.0/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.0/mlp/act_fn/Mul 1 1 /model/layers.0/mlp/gate_proj/MatMul_output_0 /model/layers.0/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.0/mlp/Mul 2 1 /model/layers.0/mlp/act_fn/Mul_output_0 /model/layers.0/mlp/up_proj/MatMul_output_0 /model/layers.0/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.0/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.0/mlp/Mul_output_0 /model/layers.0/mlp/Mul_output_0_quantized /model/layers.0/mlp/Mul_output_0_scale /model/layers.0/mlp/Mul_output_0_zero_point BinaryOp /model/layers.0/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.0/mlp/Mul_output_0_scale /model/layers.0/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.399114e-03 MatMulInteger /model/layers.0/mlp/down_proj/MatMul_quant 4 1 /model/layers.0/mlp/Mul_output_0_quantized onnx::MatMul_7224_quantized /model/layers.0/mlp/Mul_output_0_zero_point onnx::MatMul_7224_zero_point /model/layers.0/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.0/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.0/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.0/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.0/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.0/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.0/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.0/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.0/Add_1 2 1 /model/layers.0/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.0/mlp/down_proj/MatMul_output_0 /model/layers.0/Add_1_output_0 0=0 Cast /model/layers.1/input_layernorm/Cast 1 1 /model/layers.0/Add_1_output_0 /model/layers.1/input_layernorm/Cast_output_0 Split splitncnn_50 1 3 /model/layers.1/input_layernorm/Cast_output_0 /model/layers.1/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.1/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.1/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.1/input_layernorm/Pow 1 1 /model/layers.1/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.1/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.1/input_layernorm/ReduceMean 2 1 /model/layers.1/input_layernorm/Pow_output_0 /model/layers.1/input_layernorm/Constant_1_output_0 /model/layers.1/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.1/input_layernorm/Add 1 1 /model/layers.1/input_layernorm/ReduceMean_output_0 /model/layers.1/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.1/input_layernorm/Sqrt 1 1 /model/layers.1/input_layernorm/Add_output_0 /model/layers.1/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.1/input_layernorm/Div 1 1 /model/layers.1/input_layernorm/Sqrt_output_0 /model/layers.1/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.1/input_layernorm/Mul 2 1 /model/layers.1/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.1/input_layernorm/Div_output_0 /model/layers.1/input_layernorm/Mul_output_0 0=2 Cast /model/layers.1/input_layernorm/Cast_1 1 1 /model/layers.1/input_layernorm/Mul_output_0 /model/layers.1/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.1/input_layernorm/Mul_1 2 1 model.layers.1.input_layernorm.weight /model/layers.1/input_layernorm/Cast_1_output_0 /model/layers.1/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_51 1 3 /model/layers.1/input_layernorm/Mul_1_output_0 /model/layers.1/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.1/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.1/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.1/self_attn/Shape 1 1 /model/layers.1/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.1/self_attn/Shape_output_0 Shape /model/layers.1/self_attn/Shape_1 1 1 /model/layers.1/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.1/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.1/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.1/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.1/input_layernorm/Mul_1_output_0_quantized /model/layers.1/input_layernorm/Mul_1_output_0_scale /model/layers.1/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_52 1 3 /model/layers.1/input_layernorm/Mul_1_output_0_quantized /model/layers.1/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.1/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.1/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_53 1 3 /model/layers.1/input_layernorm/Mul_1_output_0_scale /model/layers.1/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.1/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.1/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_54 1 3 /model/layers.1/input_layernorm/Mul_1_output_0_zero_point /model/layers.1/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.1/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.1/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.1/self_attn/Gather 1 1 /model/layers.1/self_attn/Shape_output_0 /model/layers.1/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_55 1 4 /model/layers.1/self_attn/Gather_output_0 /model/layers.1/self_attn/Gather_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_output_0_splitncnn_1 /model/layers.1/self_attn/Gather_output_0_splitncnn_2 /model/layers.1/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.1/self_attn/Gather_1 1 1 /model/layers.1/self_attn/Shape_1_output_0 /model/layers.1/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_56 1 4 /model/layers.1/self_attn/Gather_1_output_0 /model/layers.1/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.1/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.1/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.1/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.1/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.1/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.722072e-03 BinaryOp /model/layers.1/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.1/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.1/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.245325e-03 BinaryOp /model/layers.1/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.1/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.1/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=6.497601e-04 MatMulInteger /model/layers.1/self_attn/q_proj/MatMul_quant 4 1 /model/layers.1/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7225_quantized /model/layers.1/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7225_zero_point /model/layers.1/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.1/self_attn/k_proj/MatMul_quant 4 1 /model/layers.1/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7232_quantized /model/layers.1/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7232_zero_point /model/layers.1/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.1/self_attn/v_proj/MatMul_quant 4 1 /model/layers.1/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7233_quantized /model/layers.1/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7233_zero_point /model/layers.1/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.1/self_attn/Unsqueeze 2 1 /model/layers.1/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_776 /model/layers.1/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_2 2 1 /model/layers.1/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_785 /model/layers.1/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_4 2 1 /model/layers.1/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_794 /model/layers.1/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_29 2 1 /model/layers.1/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_999 /model/layers.1/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_1 2 1 /model/layers.1/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_778 /model/layers.1/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_3 2 1 /model/layers.1/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_787 /model/layers.1/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_5 2 1 /model/layers.1/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_796 /model/layers.1/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_30 2 1 /model/layers.1/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_1001 /model/layers.1/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.1/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.1/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.1/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.1/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.1/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.1/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.1/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.1/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.1/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.1/self_attn/Concat 4 1 /model/layers.1/self_attn/Unsqueeze_output_0 /model/layers.1/self_attn/Unsqueeze_1_output_0 /model/layers.1/self_attn/Constant_2_output_0 /model/layers.1/self_attn/Constant_3_output_0 /model/layers.1/self_attn/Concat_output_0 0=0 Concat /model/layers.1/self_attn/Concat_1 4 1 /model/layers.1/self_attn/Unsqueeze_2_output_0 /model/layers.1/self_attn/Unsqueeze_3_output_0 /model/layers.1/self_attn/Constant_4_output_0 /model/layers.1/self_attn/Constant_5_output_0 /model/layers.1/self_attn/Concat_1_output_0 0=0 Concat /model/layers.1/self_attn/Concat_2 4 1 /model/layers.1/self_attn/Unsqueeze_4_output_0 /model/layers.1/self_attn/Unsqueeze_5_output_0 /model/layers.1/self_attn/Constant_6_output_0 /model/layers.1/self_attn/Constant_7_output_0 /model/layers.1/self_attn/Concat_2_output_0 0=0 Concat /model/layers.1/self_attn/Concat_11 3 1 /model/layers.1/self_attn/Unsqueeze_29_output_0 /model/layers.1/self_attn/Unsqueeze_30_output_0 /model/layers.1/self_attn/Constant_64_output_0 /model/layers.1/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.1/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.1/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.1/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.1/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.1/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.1/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.1/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.1/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.1/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.1/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.1/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.1/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.1/self_attn/q_proj/Add 2 1 model.layers.1.self_attn.q_proj.bias /model/layers.1/self_attn/q_proj/MatMul_output_0 /model/layers.1/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.1/self_attn/k_proj/Add 2 1 model.layers.1.self_attn.k_proj.bias /model/layers.1/self_attn/k_proj/MatMul_output_0 /model/layers.1/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.1/self_attn/v_proj/Add 2 1 model.layers.1.self_attn.v_proj.bias /model/layers.1/self_attn/v_proj/MatMul_output_0 /model/layers.1/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.1/self_attn/Reshape 2 1 /model/layers.1/self_attn/q_proj/Add_output_0 /model/layers.1/self_attn/Concat_output_0 /model/layers.1/self_attn/Reshape_output_0 Reshape /model/layers.1/self_attn/Reshape_1 2 1 /model/layers.1/self_attn/k_proj/Add_output_0 /model/layers.1/self_attn/Concat_1_output_0 /model/layers.1/self_attn/Reshape_1_output_0 Reshape /model/layers.1/self_attn/Reshape_2 2 1 /model/layers.1/self_attn/v_proj/Add_output_0 /model/layers.1/self_attn/Concat_2_output_0 /model/layers.1/self_attn/Reshape_2_output_0 Permute /model/layers.1/self_attn/Transpose 1 1 /model/layers.1/self_attn/Reshape_output_0 /model/layers.1/self_attn/Transpose_output_0 0=2 Split splitncnn_57 1 4 /model/layers.1/self_attn/Transpose_output_0 /model/layers.1/self_attn/Transpose_output_0_splitncnn_0 /model/layers.1/self_attn/Transpose_output_0_splitncnn_1 /model/layers.1/self_attn/Transpose_output_0_splitncnn_2 /model/layers.1/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.1/self_attn/Transpose_1 1 1 /model/layers.1/self_attn/Reshape_1_output_0 /model/layers.1/self_attn/Transpose_1_output_0 0=2 Split splitncnn_58 1 4 /model/layers.1/self_attn/Transpose_1_output_0 /model/layers.1/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.1/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.1/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.1/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.1/self_attn/Transpose_2 1 1 /model/layers.1/self_attn/Reshape_2_output_0 /model/layers.1/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.1/self_attn/Mul 2 1 /model/layers.1/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_45 /model/layers.1/self_attn/Mul_output_0 0=2 Shape /model/layers.1/self_attn/Shape_2 1 1 /model/layers.1/self_attn/Transpose_output_0_splitncnn_2 /model/layers.1/self_attn/Shape_2_output_0 BinaryOp /model/layers.1/self_attn/Mul_2 2 1 /model/layers.1/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_44 /model/layers.1/self_attn/Mul_2_output_0 0=2 Shape /model/layers.1/self_attn/Shape_3 1 1 /model/layers.1/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.1/self_attn/Shape_3_output_0 Concat /model/layers.1/self_attn/Concat_6 2 1 past_key_values.1.value /model/layers.1/self_attn/Transpose_2_output_0 present.1.value 0=-2 Split splitncnn_59 1 5 present.1.value present.1.value_splitncnn_0 present.1.value_splitncnn_1 present.1.value_splitncnn_2 present.1.value_splitncnn_3 present.1.value_splitncnn_4 Crop /model/layers.1/self_attn/Gather_2 1 1 /model/layers.1/self_attn/Shape_2_output_0 /model/layers.1/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.1/self_attn/Gather_3 1 1 /model/layers.1/self_attn/Shape_3_output_0 /model/layers.1/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.1/self_attn/Shape_9 1 1 present.1.value_splitncnn_4 /model/layers.1/self_attn/Shape_9_output_0 Shape /model/layers.1/self_attn/Shape_10 1 1 present.1.value_splitncnn_3 /model/layers.1/self_attn/Shape_10_output_0 Shape /model/layers.1/self_attn/Shape_11 1 1 present.1.value_splitncnn_2 /model/layers.1/self_attn/Shape_11_output_0 Shape /model/layers.1/self_attn/Shape_12 1 1 present.1.value_splitncnn_1 /model/layers.1/self_attn/Shape_12_output_0 ExpandDims /model/layers.1/self_attn/Unsqueeze_19 2 1 present.1.value_splitncnn_0 /model/layers.1/self_attn/Constant_45_output_0 /model/layers.1/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.1/self_attn/Div 2 1 /model/layers.1/self_attn/Gather_2_output_0 /model/layers.1/self_attn/Constant_9_output_0 /model/layers.1/self_attn/Div_output_0 0=3 BinaryOp /model/layers.1/self_attn/Div_1 2 1 /model/layers.1/self_attn/Gather_3_output_0 /model/layers.1/self_attn/Constant_19_output_0 /model/layers.1/self_attn/Div_1_output_0 0=3 Crop /model/layers.1/self_attn/Gather_8 1 1 /model/layers.1/self_attn/Shape_9_output_0 /model/layers.1/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_60 1 2 /model/layers.1/self_attn/Gather_8_output_0 /model/layers.1/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.1/self_attn/Gather_9 1 1 /model/layers.1/self_attn/Shape_10_output_0 /model/layers.1/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_61 1 2 /model/layers.1/self_attn/Gather_9_output_0 /model/layers.1/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.1/self_attn/Gather_10 1 1 /model/layers.1/self_attn/Shape_11_output_0 /model/layers.1/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_62 1 2 /model/layers.1/self_attn/Gather_10_output_0 /model/layers.1/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.1/self_attn/Gather_11 1 1 /model/layers.1/self_attn/Shape_12_output_0 /model/layers.1/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_63 1 2 /model/layers.1/self_attn/Gather_11_output_0 /model/layers.1/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.1/self_attn/Cast 1 1 /model/layers.1/self_attn/Div_output_0 /model/layers.1/self_attn/Cast_output_0 Cast /model/layers.1/self_attn/Cast_2 1 1 /model/layers.1/self_attn/Div_1_output_0 /model/layers.1/self_attn/Cast_2_output_0 ExpandDims /model/layers.1/self_attn/Unsqueeze_20 2 1 /model/layers.1/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_46_output_0 /model/layers.1/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_24 2 1 /model/layers.1/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_959 /model/layers.1/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_21 2 1 /model/layers.1/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_47_output_0 /model/layers.1/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.1/self_attn/Mul_7 2 1 /model/layers.1/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.1/self_attn/Constant_53_output_0 /model/layers.1/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.1/self_attn/Unsqueeze_22 2 1 /model/layers.1/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_49_output_0 /model/layers.1/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_26 2 1 /model/layers.1/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_963 /model/layers.1/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_23 2 1 /model/layers.1/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_50_output_0 /model/layers.1/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_27 2 1 /model/layers.1/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_965 /model/layers.1/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.1/self_attn/Cast_1 1 1 /model/layers.1/self_attn/Cast_output_0 /model/layers.1/self_attn/Cast_1_output_0 Split splitncnn_64 1 2 /model/layers.1/self_attn/Cast_1_output_0 /model/layers.1/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.1/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.1/self_attn/Cast_3 1 1 /model/layers.1/self_attn/Cast_2_output_0 /model/layers.1/self_attn/Cast_3_output_0 Split splitncnn_65 1 2 /model/layers.1/self_attn/Cast_3_output_0 /model/layers.1/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.1/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.1/self_attn/Unsqueeze_25 2 1 /model/layers.1/self_attn/Mul_7_output_0 onnx::Unsqueeze_961 /model/layers.1/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.1/self_attn/Concat_9 5 1 /model/layers.1/self_attn/Unsqueeze_20_output_0 /model/layers.1/self_attn/Unsqueeze_21_output_0 /model/layers.1/self_attn/Constant_48_output_0 /model/layers.1/self_attn/Unsqueeze_22_output_0 /model/layers.1/self_attn/Unsqueeze_23_output_0 /model/layers.1/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_6 2 1 /model/layers.1/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_12_output_0 /model/layers.1/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_7 2 1 /model/layers.1/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.1/self_attn/Constant_15_output_0 /model/layers.1/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_8 2 1 /model/layers.1/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_22_output_0 /model/layers.1/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_9 2 1 /model/layers.1/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.1/self_attn/Constant_25_output_0 /model/layers.1/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.1/self_attn/Concat_10 4 1 /model/layers.1/self_attn/Unsqueeze_24_output_0 /model/layers.1/self_attn/Unsqueeze_25_output_0 /model/layers.1/self_attn/Unsqueeze_26_output_0 /model/layers.1/self_attn/Unsqueeze_27_output_0 /model/layers.1/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.1/self_attn/Reshape_5 1 1 /model/layers.1/self_attn/Concat_9_output_0 /model/layers.1/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_66 1 3 /model/layers.1/self_attn/Reshape_5_output_0 /model/layers.1/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.1/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.1/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.1/self_attn/Slice 2 1 /model/layers.1/self_attn/Transpose_output_0_splitncnn_1 /model/layers.1/self_attn/Unsqueeze_6_output_0 /model/layers.1/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.1/self_attn/Slice_1 2 1 /model/layers.1/self_attn/Transpose_output_0_splitncnn_0 /model/layers.1/self_attn/Unsqueeze_7_output_0 /model/layers.1/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.1/self_attn/Slice_2 2 1 /model/layers.1/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.1/self_attn/Unsqueeze_8_output_0 /model/layers.1/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.1/self_attn/Slice_3 2 1 /model/layers.1/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.1/self_attn/Unsqueeze_9_output_0 /model/layers.1/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.1/self_attn/Shape_13 1 1 /model/layers.1/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.1/self_attn/Shape_13_output_0 UnaryOp /model/layers.1/self_attn/Neg 1 1 /model/layers.1/self_attn/Slice_1_output_0 /model/layers.1/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.1/self_attn/Neg_1 1 1 /model/layers.1/self_attn/Slice_3_output_0 /model/layers.1/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.1/self_attn/ConstantOfShape_1 1 1 /model/layers.1/self_attn/Shape_13_output_0 /model/layers.1/self_attn/ConstantOfShape_1_output_0 Split splitncnn_67 1 2 /model/layers.1/self_attn/ConstantOfShape_1_output_0 /model/layers.1/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.1/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.1/self_attn/Concat_3 2 1 /model/layers.1/self_attn/Neg_output_0 /model/layers.1/self_attn/Slice_output_0 /model/layers.1/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.1/self_attn/Concat_4 2 1 /model/layers.1/self_attn/Neg_1_output_0 /model/layers.1/self_attn/Slice_2_output_0 /model/layers.1/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.1/self_attn/Mul_6 2 1 /model/layers.1/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_52_output_0 /model/layers.1/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.1/self_attn/Mul_1 2 1 /model/layers.1/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_45 /model/layers.1/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.1/self_attn/Mul_3 2 1 /model/layers.1/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_44 /model/layers.1/self_attn/Mul_3_output_0 0=2 Equal /model/layers.1/self_attn/Equal_1 2 1 /model/layers.1/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.1/self_attn/Mul_6_output_0 /model/layers.1/self_attn/Equal_1_output_0 BinaryOp /model/layers.1/self_attn/Add 2 1 /model/layers.1/self_attn/Mul_output_0 /model/layers.1/self_attn/Mul_1_output_0 /model/layers.1/self_attn/Add_output_0 0=0 BinaryOp /model/layers.1/self_attn/Add_1 2 1 /model/layers.1/self_attn/Mul_2_output_0 /model/layers.1/self_attn/Mul_3_output_0 /model/layers.1/self_attn/Add_1_output_0 0=0 Where /model/layers.1/self_attn/Where_1 3 1 /model/layers.1/self_attn/Equal_1_output_0 /model/layers.1/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.1/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.1/self_attn/Where_1_output_0 BinaryOp /model/layers.1/self_attn/Mul_8 1 1 /model/layers.1/self_attn/Add_output_0 /model/layers.1/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.1/self_attn/Concat_5 2 1 past_key_values.1.key /model/layers.1/self_attn/Add_1_output_0 present.1.key 0=-2 Split splitncnn_68 1 5 present.1.key present.1.key_splitncnn_0 present.1.key_splitncnn_1 present.1.key_splitncnn_2 present.1.key_splitncnn_3 present.1.key_splitncnn_4 Expand /model/layers.1/self_attn/Expand_1 2 1 /model/layers.1/self_attn/Unsqueeze_19_output_0 /model/layers.1/self_attn/Where_1_output_0 /model/layers.1/self_attn/Expand_1_output_0 Shape /model/layers.1/self_attn/Shape_4 1 1 present.1.key_splitncnn_4 /model/layers.1/self_attn/Shape_4_output_0 Shape /model/layers.1/self_attn/Shape_5 1 1 present.1.key_splitncnn_3 /model/layers.1/self_attn/Shape_5_output_0 Shape /model/layers.1/self_attn/Shape_6 1 1 present.1.key_splitncnn_2 /model/layers.1/self_attn/Shape_6_output_0 Shape /model/layers.1/self_attn/Shape_7 1 1 present.1.key_splitncnn_1 /model/layers.1/self_attn/Shape_7_output_0 ExpandDims /model/layers.1/self_attn/Unsqueeze_10 2 1 present.1.key_splitncnn_0 /model/layers.1/self_attn/Constant_32_output_0 /model/layers.1/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.1/self_attn/Reshape_6 2 1 /model/layers.1/self_attn/Expand_1_output_0 /model/layers.1/self_attn/Concat_10_output_0 /model/layers.1/self_attn/Reshape_6_output_0 Crop /model/layers.1/self_attn/Gather_4 1 1 /model/layers.1/self_attn/Shape_4_output_0 /model/layers.1/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_69 1 2 /model/layers.1/self_attn/Gather_4_output_0 /model/layers.1/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.1/self_attn/Gather_5 1 1 /model/layers.1/self_attn/Shape_5_output_0 /model/layers.1/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_70 1 2 /model/layers.1/self_attn/Gather_5_output_0 /model/layers.1/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.1/self_attn/Gather_6 1 1 /model/layers.1/self_attn/Shape_6_output_0 /model/layers.1/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_71 1 2 /model/layers.1/self_attn/Gather_6_output_0 /model/layers.1/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.1/self_attn/Gather_7 1 1 /model/layers.1/self_attn/Shape_7_output_0 /model/layers.1/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_72 1 2 /model/layers.1/self_attn/Gather_7_output_0 /model/layers.1/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.1/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.1/self_attn/Unsqueeze_11 2 1 /model/layers.1/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_33_output_0 /model/layers.1/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_15 2 1 /model/layers.1/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_913 /model/layers.1/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_12 2 1 /model/layers.1/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_34_output_0 /model/layers.1/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.1/self_attn/Mul_5 2 1 /model/layers.1/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.1/self_attn/Constant_40_output_0 /model/layers.1/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.1/self_attn/Unsqueeze_13 2 1 /model/layers.1/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_36_output_0 /model/layers.1/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_17 2 1 /model/layers.1/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_917 /model/layers.1/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_14 2 1 /model/layers.1/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_37_output_0 /model/layers.1/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_18 2 1 /model/layers.1/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_919 /model/layers.1/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.1/self_attn/Unsqueeze_16 2 1 /model/layers.1/self_attn/Mul_5_output_0 onnx::Unsqueeze_915 /model/layers.1/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.1/self_attn/Concat_7 5 1 /model/layers.1/self_attn/Unsqueeze_11_output_0 /model/layers.1/self_attn/Unsqueeze_12_output_0 /model/layers.1/self_attn/Constant_35_output_0 /model/layers.1/self_attn/Unsqueeze_13_output_0 /model/layers.1/self_attn/Unsqueeze_14_output_0 /model/layers.1/self_attn/Concat_7_output_0 0=0 Concat /model/layers.1/self_attn/Concat_8 4 1 /model/layers.1/self_attn/Unsqueeze_15_output_0 /model/layers.1/self_attn/Unsqueeze_16_output_0 /model/layers.1/self_attn/Unsqueeze_17_output_0 /model/layers.1/self_attn/Unsqueeze_18_output_0 /model/layers.1/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.1/self_attn/Reshape_3 1 1 /model/layers.1/self_attn/Concat_7_output_0 /model/layers.1/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_73 1 3 /model/layers.1/self_attn/Reshape_3_output_0 /model/layers.1/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.1/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.1/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.1/self_attn/Shape_8 1 1 /model/layers.1/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.1/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.1/self_attn/ConstantOfShape 1 1 /model/layers.1/self_attn/Shape_8_output_0 /model/layers.1/self_attn/ConstantOfShape_output_0 Split splitncnn_74 1 2 /model/layers.1/self_attn/ConstantOfShape_output_0 /model/layers.1/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.1/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.1/self_attn/Mul_4 2 1 /model/layers.1/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.1/self_attn/Constant_39_output_0 /model/layers.1/self_attn/Mul_4_output_0 0=2 Equal /model/layers.1/self_attn/Equal 2 1 /model/layers.1/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.1/self_attn/Mul_4_output_0 /model/layers.1/self_attn/Equal_output_0 Where /model/layers.1/self_attn/Where 3 1 /model/layers.1/self_attn/Equal_output_0 /model/layers.1/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.1/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.1/self_attn/Where_output_0 Expand /model/layers.1/self_attn/Expand 2 1 /model/layers.1/self_attn/Unsqueeze_10_output_0 /model/layers.1/self_attn/Where_output_0 /model/layers.1/self_attn/Expand_output_0 Reshape /model/layers.1/self_attn/Reshape_4 2 1 /model/layers.1/self_attn/Expand_output_0 /model/layers.1/self_attn/Concat_8_output_0 /model/layers.1/self_attn/Reshape_4_output_0 Split splitncnn_75 1 2 /model/layers.1/self_attn/Reshape_4_output_0 /model/layers.1/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.1/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.1/self_attn/Shape_14 1 1 /model/layers.1/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.1/self_attn/Shape_14_output_0 Permute /model/layers.1/self_attn/Transpose_3 1 1 /model/layers.1/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.1/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.1/self_attn/Gather_12 1 1 /model/layers.1/self_attn/Shape_14_output_0 /model/layers.1/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.1/self_attn/Mul_9 1 1 /model/layers.1/self_attn/Transpose_3_output_0 /model/layers.1/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.1/self_attn/Unsqueeze_28 2 1 /model/layers.1/self_attn/Gather_12_output_0 /model/layers.1/self_attn/Constant_57_output_0 /model/layers.1/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.1/self_attn/MatMul 2 1 /model/layers.1/self_attn/Mul_8_output_0 /model/layers.1/self_attn/Mul_9_output_0 /model/layers.1/self_attn/MatMul_output_0 Crop /model/layers.1/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_22 /model/layers.1/self_attn/Unsqueeze_28_output_0 /model/layers.1/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.1/self_attn/Where_2 3 1 /model/layers.1/self_attn/Slice_4_output_0 /model/layers.1/self_attn/Constant_61_output_0 /model/layers.1/self_attn/Constant_62_output_0 /model/layers.1/self_attn/Where_2_output_0 BinaryOp /model/layers.1/self_attn/Add_2 2 1 /model/layers.1/self_attn/MatMul_output_0 /model/layers.1/self_attn/Where_2_output_0 /model/layers.1/self_attn/Add_2_output_0 0=0 Softmax /model/layers.1/self_attn/Softmax 1 1 /model/layers.1/self_attn/Add_2_output_0 /model/layers.1/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_76 1 2 /model/layers.1/self_attn/Softmax_output_0 /model/layers.1/self_attn/Softmax_output_0_splitncnn_0 /model/layers.1/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.1/self_attn/IsNaN 1 1 /model/layers.1/self_attn/Softmax_output_0_splitncnn_1 /model/layers.1/self_attn/IsNaN_output_0 Where /model/layers.1/self_attn/Where_3 3 1 /model/layers.1/self_attn/IsNaN_output_0 /model/layers.1/self_attn/Constant_63_output_0 /model/layers.1/self_attn/Softmax_output_0_splitncnn_0 /model/layers.1/self_attn/Where_3_output_0 Gemm /model/layers.1/self_attn/MatMul_1 2 1 /model/layers.1/self_attn/Where_3_output_0 /model/layers.1/self_attn/Reshape_6_output_0 /model/layers.1/self_attn/MatMul_1_output_0 Permute /model/layers.1/self_attn/Transpose_4 1 1 /model/layers.1/self_attn/MatMul_1_output_0 /model/layers.1/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.1/self_attn/Reshape_7 2 1 /model/layers.1/self_attn/Transpose_4_output_0 /model/layers.1/self_attn/Concat_11_output_0 /model/layers.1/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.1/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.1/self_attn/Reshape_7_output_0 /model/layers.1/self_attn/Reshape_7_output_0_quantized /model/layers.1/self_attn/Reshape_7_output_0_scale /model/layers.1/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.1/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.1/self_attn/Reshape_7_output_0_scale /model/layers.1/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.622478e-03 MatMulInteger /model/layers.1/self_attn/o_proj/MatMul_quant 4 1 /model/layers.1/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7256_quantized /model/layers.1/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7256_zero_point /model/layers.1/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.1/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.1/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.1/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.1/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.1/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.1/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.1/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.1/Add 2 1 /model/layers.1/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.1/self_attn/o_proj/MatMul_output_0 /model/layers.1/Add_output_0 0=0 Cast /model/layers.1/post_attention_layernorm/Cast 1 1 /model/layers.1/Add_output_0 /model/layers.1/post_attention_layernorm/Cast_output_0 Split splitncnn_77 1 3 /model/layers.1/post_attention_layernorm/Cast_output_0 /model/layers.1/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.1/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.1/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.1/post_attention_layernorm/Pow 1 1 /model/layers.1/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.1/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.1/post_attention_layernorm/ReduceMean 2 1 /model/layers.1/post_attention_layernorm/Pow_output_0 /model/layers.1/post_attention_layernorm/Constant_1_output_0 /model/layers.1/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.1/post_attention_layernorm/Add 1 1 /model/layers.1/post_attention_layernorm/ReduceMean_output_0 /model/layers.1/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.1/post_attention_layernorm/Sqrt 1 1 /model/layers.1/post_attention_layernorm/Add_output_0 /model/layers.1/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.1/post_attention_layernorm/Div 1 1 /model/layers.1/post_attention_layernorm/Sqrt_output_0 /model/layers.1/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.1/post_attention_layernorm/Mul 2 1 /model/layers.1/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.1/post_attention_layernorm/Div_output_0 /model/layers.1/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.1/post_attention_layernorm/Cast_1 1 1 /model/layers.1/post_attention_layernorm/Mul_output_0 /model/layers.1/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.1/post_attention_layernorm/Mul_1 2 1 model.layers.1.post_attention_layernorm.weight /model/layers.1/post_attention_layernorm/Cast_1_output_0 /model/layers.1/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.1/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.1/post_attention_layernorm/Mul_1_output_0 /model/layers.1/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.1/post_attention_layernorm/Mul_1_output_0_scale /model/layers.1/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_78 1 2 /model/layers.1/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.1/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.1/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_79 1 2 /model/layers.1/post_attention_layernorm/Mul_1_output_0_scale /model/layers.1/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.1/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_80 1 2 /model/layers.1/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.1/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.1/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.1/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.1/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.1/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.244956e-03 BinaryOp /model/layers.1/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.1/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.1/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.153051e-03 MatMulInteger /model/layers.1/mlp/gate_proj/MatMul_quant 4 1 /model/layers.1/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7257_quantized /model/layers.1/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7257_zero_point /model/layers.1/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.1/mlp/up_proj/MatMul_quant 4 1 /model/layers.1/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7258_quantized /model/layers.1/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7258_zero_point /model/layers.1/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.1/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.1/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.1/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.1/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.1/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.1/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.1/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.1/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.1/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.1/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.1/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.1/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.1/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.1/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.1/mlp/act_fn/Mul 1 1 /model/layers.1/mlp/gate_proj/MatMul_output_0 /model/layers.1/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.1/mlp/Mul 2 1 /model/layers.1/mlp/act_fn/Mul_output_0 /model/layers.1/mlp/up_proj/MatMul_output_0 /model/layers.1/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.1/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.1/mlp/Mul_output_0 /model/layers.1/mlp/Mul_output_0_quantized /model/layers.1/mlp/Mul_output_0_scale /model/layers.1/mlp/Mul_output_0_zero_point BinaryOp /model/layers.1/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.1/mlp/Mul_output_0_scale /model/layers.1/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.829724e-03 MatMulInteger /model/layers.1/mlp/down_proj/MatMul_quant 4 1 /model/layers.1/mlp/Mul_output_0_quantized onnx::MatMul_7259_quantized /model/layers.1/mlp/Mul_output_0_zero_point onnx::MatMul_7259_zero_point /model/layers.1/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.1/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.1/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.1/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.1/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.1/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.1/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.1/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.1/Add_1 2 1 /model/layers.1/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.1/mlp/down_proj/MatMul_output_0 /model/layers.1/Add_1_output_0 0=0 Cast /model/layers.2/input_layernorm/Cast 1 1 /model/layers.1/Add_1_output_0 /model/layers.2/input_layernorm/Cast_output_0 Split splitncnn_81 1 3 /model/layers.2/input_layernorm/Cast_output_0 /model/layers.2/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.2/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.2/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.2/input_layernorm/Pow 1 1 /model/layers.2/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.2/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.2/input_layernorm/ReduceMean 2 1 /model/layers.2/input_layernorm/Pow_output_0 /model/layers.2/input_layernorm/Constant_1_output_0 /model/layers.2/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.2/input_layernorm/Add 1 1 /model/layers.2/input_layernorm/ReduceMean_output_0 /model/layers.2/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.2/input_layernorm/Sqrt 1 1 /model/layers.2/input_layernorm/Add_output_0 /model/layers.2/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.2/input_layernorm/Div 1 1 /model/layers.2/input_layernorm/Sqrt_output_0 /model/layers.2/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.2/input_layernorm/Mul 2 1 /model/layers.2/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.2/input_layernorm/Div_output_0 /model/layers.2/input_layernorm/Mul_output_0 0=2 Cast /model/layers.2/input_layernorm/Cast_1 1 1 /model/layers.2/input_layernorm/Mul_output_0 /model/layers.2/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.2/input_layernorm/Mul_1 2 1 model.layers.2.input_layernorm.weight /model/layers.2/input_layernorm/Cast_1_output_0 /model/layers.2/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_82 1 3 /model/layers.2/input_layernorm/Mul_1_output_0 /model/layers.2/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.2/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.2/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.2/self_attn/Shape 1 1 /model/layers.2/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.2/self_attn/Shape_output_0 Shape /model/layers.2/self_attn/Shape_1 1 1 /model/layers.2/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.2/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.2/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.2/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.2/input_layernorm/Mul_1_output_0_quantized /model/layers.2/input_layernorm/Mul_1_output_0_scale /model/layers.2/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_83 1 3 /model/layers.2/input_layernorm/Mul_1_output_0_quantized /model/layers.2/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.2/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.2/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_84 1 3 /model/layers.2/input_layernorm/Mul_1_output_0_scale /model/layers.2/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.2/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.2/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_85 1 3 /model/layers.2/input_layernorm/Mul_1_output_0_zero_point /model/layers.2/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.2/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.2/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.2/self_attn/Gather 1 1 /model/layers.2/self_attn/Shape_output_0 /model/layers.2/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_86 1 4 /model/layers.2/self_attn/Gather_output_0 /model/layers.2/self_attn/Gather_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_output_0_splitncnn_1 /model/layers.2/self_attn/Gather_output_0_splitncnn_2 /model/layers.2/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.2/self_attn/Gather_1 1 1 /model/layers.2/self_attn/Shape_1_output_0 /model/layers.2/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_87 1 4 /model/layers.2/self_attn/Gather_1_output_0 /model/layers.2/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.2/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.2/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.2/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.2/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.2/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.737451e-03 BinaryOp /model/layers.2/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.2/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.2/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.045399e-03 BinaryOp /model/layers.2/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.2/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.2/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=7.535679e-04 MatMulInteger /model/layers.2/self_attn/q_proj/MatMul_quant 4 1 /model/layers.2/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7260_quantized /model/layers.2/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7260_zero_point /model/layers.2/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.2/self_attn/k_proj/MatMul_quant 4 1 /model/layers.2/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7267_quantized /model/layers.2/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7267_zero_point /model/layers.2/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.2/self_attn/v_proj/MatMul_quant 4 1 /model/layers.2/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7268_quantized /model/layers.2/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7268_zero_point /model/layers.2/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.2/self_attn/Unsqueeze 2 1 /model/layers.2/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_1055 /model/layers.2/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_2 2 1 /model/layers.2/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_1064 /model/layers.2/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_4 2 1 /model/layers.2/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_1073 /model/layers.2/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_29 2 1 /model/layers.2/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_1278 /model/layers.2/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_1 2 1 /model/layers.2/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_1057 /model/layers.2/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_3 2 1 /model/layers.2/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_1066 /model/layers.2/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_5 2 1 /model/layers.2/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_1075 /model/layers.2/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_30 2 1 /model/layers.2/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_1280 /model/layers.2/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.2/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.2/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.2/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.2/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.2/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.2/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.2/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.2/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.2/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.2/self_attn/Concat 4 1 /model/layers.2/self_attn/Unsqueeze_output_0 /model/layers.2/self_attn/Unsqueeze_1_output_0 /model/layers.2/self_attn/Constant_2_output_0 /model/layers.2/self_attn/Constant_3_output_0 /model/layers.2/self_attn/Concat_output_0 0=0 Concat /model/layers.2/self_attn/Concat_1 4 1 /model/layers.2/self_attn/Unsqueeze_2_output_0 /model/layers.2/self_attn/Unsqueeze_3_output_0 /model/layers.2/self_attn/Constant_4_output_0 /model/layers.2/self_attn/Constant_5_output_0 /model/layers.2/self_attn/Concat_1_output_0 0=0 Concat /model/layers.2/self_attn/Concat_2 4 1 /model/layers.2/self_attn/Unsqueeze_4_output_0 /model/layers.2/self_attn/Unsqueeze_5_output_0 /model/layers.2/self_attn/Constant_6_output_0 /model/layers.2/self_attn/Constant_7_output_0 /model/layers.2/self_attn/Concat_2_output_0 0=0 Concat /model/layers.2/self_attn/Concat_11 3 1 /model/layers.2/self_attn/Unsqueeze_29_output_0 /model/layers.2/self_attn/Unsqueeze_30_output_0 /model/layers.2/self_attn/Constant_64_output_0 /model/layers.2/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.2/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.2/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.2/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.2/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.2/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.2/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.2/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.2/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.2/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.2/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.2/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.2/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.2/self_attn/q_proj/Add 2 1 model.layers.2.self_attn.q_proj.bias /model/layers.2/self_attn/q_proj/MatMul_output_0 /model/layers.2/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.2/self_attn/k_proj/Add 2 1 model.layers.2.self_attn.k_proj.bias /model/layers.2/self_attn/k_proj/MatMul_output_0 /model/layers.2/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.2/self_attn/v_proj/Add 2 1 model.layers.2.self_attn.v_proj.bias /model/layers.2/self_attn/v_proj/MatMul_output_0 /model/layers.2/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.2/self_attn/Reshape 2 1 /model/layers.2/self_attn/q_proj/Add_output_0 /model/layers.2/self_attn/Concat_output_0 /model/layers.2/self_attn/Reshape_output_0 Reshape /model/layers.2/self_attn/Reshape_1 2 1 /model/layers.2/self_attn/k_proj/Add_output_0 /model/layers.2/self_attn/Concat_1_output_0 /model/layers.2/self_attn/Reshape_1_output_0 Reshape /model/layers.2/self_attn/Reshape_2 2 1 /model/layers.2/self_attn/v_proj/Add_output_0 /model/layers.2/self_attn/Concat_2_output_0 /model/layers.2/self_attn/Reshape_2_output_0 Permute /model/layers.2/self_attn/Transpose 1 1 /model/layers.2/self_attn/Reshape_output_0 /model/layers.2/self_attn/Transpose_output_0 0=2 Split splitncnn_88 1 4 /model/layers.2/self_attn/Transpose_output_0 /model/layers.2/self_attn/Transpose_output_0_splitncnn_0 /model/layers.2/self_attn/Transpose_output_0_splitncnn_1 /model/layers.2/self_attn/Transpose_output_0_splitncnn_2 /model/layers.2/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.2/self_attn/Transpose_1 1 1 /model/layers.2/self_attn/Reshape_1_output_0 /model/layers.2/self_attn/Transpose_1_output_0 0=2 Split splitncnn_89 1 4 /model/layers.2/self_attn/Transpose_1_output_0 /model/layers.2/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.2/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.2/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.2/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.2/self_attn/Transpose_2 1 1 /model/layers.2/self_attn/Reshape_2_output_0 /model/layers.2/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.2/self_attn/Mul 2 1 /model/layers.2/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_43 /model/layers.2/self_attn/Mul_output_0 0=2 Shape /model/layers.2/self_attn/Shape_2 1 1 /model/layers.2/self_attn/Transpose_output_0_splitncnn_2 /model/layers.2/self_attn/Shape_2_output_0 BinaryOp /model/layers.2/self_attn/Mul_2 2 1 /model/layers.2/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_42 /model/layers.2/self_attn/Mul_2_output_0 0=2 Shape /model/layers.2/self_attn/Shape_3 1 1 /model/layers.2/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.2/self_attn/Shape_3_output_0 Concat /model/layers.2/self_attn/Concat_6 2 1 past_key_values.2.value /model/layers.2/self_attn/Transpose_2_output_0 present.2.value 0=-2 Split splitncnn_90 1 5 present.2.value present.2.value_splitncnn_0 present.2.value_splitncnn_1 present.2.value_splitncnn_2 present.2.value_splitncnn_3 present.2.value_splitncnn_4 Crop /model/layers.2/self_attn/Gather_2 1 1 /model/layers.2/self_attn/Shape_2_output_0 /model/layers.2/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.2/self_attn/Gather_3 1 1 /model/layers.2/self_attn/Shape_3_output_0 /model/layers.2/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.2/self_attn/Shape_9 1 1 present.2.value_splitncnn_4 /model/layers.2/self_attn/Shape_9_output_0 Shape /model/layers.2/self_attn/Shape_10 1 1 present.2.value_splitncnn_3 /model/layers.2/self_attn/Shape_10_output_0 Shape /model/layers.2/self_attn/Shape_11 1 1 present.2.value_splitncnn_2 /model/layers.2/self_attn/Shape_11_output_0 Shape /model/layers.2/self_attn/Shape_12 1 1 present.2.value_splitncnn_1 /model/layers.2/self_attn/Shape_12_output_0 ExpandDims /model/layers.2/self_attn/Unsqueeze_19 2 1 present.2.value_splitncnn_0 /model/layers.2/self_attn/Constant_45_output_0 /model/layers.2/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.2/self_attn/Div 2 1 /model/layers.2/self_attn/Gather_2_output_0 /model/layers.2/self_attn/Constant_9_output_0 /model/layers.2/self_attn/Div_output_0 0=3 BinaryOp /model/layers.2/self_attn/Div_1 2 1 /model/layers.2/self_attn/Gather_3_output_0 /model/layers.2/self_attn/Constant_19_output_0 /model/layers.2/self_attn/Div_1_output_0 0=3 Crop /model/layers.2/self_attn/Gather_8 1 1 /model/layers.2/self_attn/Shape_9_output_0 /model/layers.2/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_91 1 2 /model/layers.2/self_attn/Gather_8_output_0 /model/layers.2/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.2/self_attn/Gather_9 1 1 /model/layers.2/self_attn/Shape_10_output_0 /model/layers.2/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_92 1 2 /model/layers.2/self_attn/Gather_9_output_0 /model/layers.2/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.2/self_attn/Gather_10 1 1 /model/layers.2/self_attn/Shape_11_output_0 /model/layers.2/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_93 1 2 /model/layers.2/self_attn/Gather_10_output_0 /model/layers.2/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.2/self_attn/Gather_11 1 1 /model/layers.2/self_attn/Shape_12_output_0 /model/layers.2/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_94 1 2 /model/layers.2/self_attn/Gather_11_output_0 /model/layers.2/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.2/self_attn/Cast 1 1 /model/layers.2/self_attn/Div_output_0 /model/layers.2/self_attn/Cast_output_0 Cast /model/layers.2/self_attn/Cast_2 1 1 /model/layers.2/self_attn/Div_1_output_0 /model/layers.2/self_attn/Cast_2_output_0 ExpandDims /model/layers.2/self_attn/Unsqueeze_20 2 1 /model/layers.2/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_46_output_0 /model/layers.2/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_24 2 1 /model/layers.2/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_1238 /model/layers.2/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_21 2 1 /model/layers.2/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_47_output_0 /model/layers.2/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.2/self_attn/Mul_7 2 1 /model/layers.2/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.2/self_attn/Constant_53_output_0 /model/layers.2/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.2/self_attn/Unsqueeze_22 2 1 /model/layers.2/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_49_output_0 /model/layers.2/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_26 2 1 /model/layers.2/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_1242 /model/layers.2/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_23 2 1 /model/layers.2/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_50_output_0 /model/layers.2/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_27 2 1 /model/layers.2/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_1244 /model/layers.2/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.2/self_attn/Cast_1 1 1 /model/layers.2/self_attn/Cast_output_0 /model/layers.2/self_attn/Cast_1_output_0 Split splitncnn_95 1 2 /model/layers.2/self_attn/Cast_1_output_0 /model/layers.2/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.2/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.2/self_attn/Cast_3 1 1 /model/layers.2/self_attn/Cast_2_output_0 /model/layers.2/self_attn/Cast_3_output_0 Split splitncnn_96 1 2 /model/layers.2/self_attn/Cast_3_output_0 /model/layers.2/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.2/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.2/self_attn/Unsqueeze_25 2 1 /model/layers.2/self_attn/Mul_7_output_0 onnx::Unsqueeze_1240 /model/layers.2/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.2/self_attn/Concat_9 5 1 /model/layers.2/self_attn/Unsqueeze_20_output_0 /model/layers.2/self_attn/Unsqueeze_21_output_0 /model/layers.2/self_attn/Constant_48_output_0 /model/layers.2/self_attn/Unsqueeze_22_output_0 /model/layers.2/self_attn/Unsqueeze_23_output_0 /model/layers.2/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_6 2 1 /model/layers.2/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_12_output_0 /model/layers.2/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_7 2 1 /model/layers.2/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.2/self_attn/Constant_15_output_0 /model/layers.2/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_8 2 1 /model/layers.2/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_22_output_0 /model/layers.2/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_9 2 1 /model/layers.2/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.2/self_attn/Constant_25_output_0 /model/layers.2/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.2/self_attn/Concat_10 4 1 /model/layers.2/self_attn/Unsqueeze_24_output_0 /model/layers.2/self_attn/Unsqueeze_25_output_0 /model/layers.2/self_attn/Unsqueeze_26_output_0 /model/layers.2/self_attn/Unsqueeze_27_output_0 /model/layers.2/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.2/self_attn/Reshape_5 1 1 /model/layers.2/self_attn/Concat_9_output_0 /model/layers.2/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_97 1 3 /model/layers.2/self_attn/Reshape_5_output_0 /model/layers.2/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.2/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.2/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.2/self_attn/Slice 2 1 /model/layers.2/self_attn/Transpose_output_0_splitncnn_1 /model/layers.2/self_attn/Unsqueeze_6_output_0 /model/layers.2/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.2/self_attn/Slice_1 2 1 /model/layers.2/self_attn/Transpose_output_0_splitncnn_0 /model/layers.2/self_attn/Unsqueeze_7_output_0 /model/layers.2/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.2/self_attn/Slice_2 2 1 /model/layers.2/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.2/self_attn/Unsqueeze_8_output_0 /model/layers.2/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.2/self_attn/Slice_3 2 1 /model/layers.2/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.2/self_attn/Unsqueeze_9_output_0 /model/layers.2/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.2/self_attn/Shape_13 1 1 /model/layers.2/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.2/self_attn/Shape_13_output_0 UnaryOp /model/layers.2/self_attn/Neg 1 1 /model/layers.2/self_attn/Slice_1_output_0 /model/layers.2/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.2/self_attn/Neg_1 1 1 /model/layers.2/self_attn/Slice_3_output_0 /model/layers.2/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.2/self_attn/ConstantOfShape_1 1 1 /model/layers.2/self_attn/Shape_13_output_0 /model/layers.2/self_attn/ConstantOfShape_1_output_0 Split splitncnn_98 1 2 /model/layers.2/self_attn/ConstantOfShape_1_output_0 /model/layers.2/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.2/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.2/self_attn/Concat_3 2 1 /model/layers.2/self_attn/Neg_output_0 /model/layers.2/self_attn/Slice_output_0 /model/layers.2/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.2/self_attn/Concat_4 2 1 /model/layers.2/self_attn/Neg_1_output_0 /model/layers.2/self_attn/Slice_2_output_0 /model/layers.2/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.2/self_attn/Mul_6 2 1 /model/layers.2/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_52_output_0 /model/layers.2/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.2/self_attn/Mul_1 2 1 /model/layers.2/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_43 /model/layers.2/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.2/self_attn/Mul_3 2 1 /model/layers.2/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_42 /model/layers.2/self_attn/Mul_3_output_0 0=2 Equal /model/layers.2/self_attn/Equal_1 2 1 /model/layers.2/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.2/self_attn/Mul_6_output_0 /model/layers.2/self_attn/Equal_1_output_0 BinaryOp /model/layers.2/self_attn/Add 2 1 /model/layers.2/self_attn/Mul_output_0 /model/layers.2/self_attn/Mul_1_output_0 /model/layers.2/self_attn/Add_output_0 0=0 BinaryOp /model/layers.2/self_attn/Add_1 2 1 /model/layers.2/self_attn/Mul_2_output_0 /model/layers.2/self_attn/Mul_3_output_0 /model/layers.2/self_attn/Add_1_output_0 0=0 Where /model/layers.2/self_attn/Where_1 3 1 /model/layers.2/self_attn/Equal_1_output_0 /model/layers.2/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.2/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.2/self_attn/Where_1_output_0 BinaryOp /model/layers.2/self_attn/Mul_8 1 1 /model/layers.2/self_attn/Add_output_0 /model/layers.2/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.2/self_attn/Concat_5 2 1 past_key_values.2.key /model/layers.2/self_attn/Add_1_output_0 present.2.key 0=-2 Split splitncnn_99 1 5 present.2.key present.2.key_splitncnn_0 present.2.key_splitncnn_1 present.2.key_splitncnn_2 present.2.key_splitncnn_3 present.2.key_splitncnn_4 Expand /model/layers.2/self_attn/Expand_1 2 1 /model/layers.2/self_attn/Unsqueeze_19_output_0 /model/layers.2/self_attn/Where_1_output_0 /model/layers.2/self_attn/Expand_1_output_0 Shape /model/layers.2/self_attn/Shape_4 1 1 present.2.key_splitncnn_4 /model/layers.2/self_attn/Shape_4_output_0 Shape /model/layers.2/self_attn/Shape_5 1 1 present.2.key_splitncnn_3 /model/layers.2/self_attn/Shape_5_output_0 Shape /model/layers.2/self_attn/Shape_6 1 1 present.2.key_splitncnn_2 /model/layers.2/self_attn/Shape_6_output_0 Shape /model/layers.2/self_attn/Shape_7 1 1 present.2.key_splitncnn_1 /model/layers.2/self_attn/Shape_7_output_0 ExpandDims /model/layers.2/self_attn/Unsqueeze_10 2 1 present.2.key_splitncnn_0 /model/layers.2/self_attn/Constant_32_output_0 /model/layers.2/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.2/self_attn/Reshape_6 2 1 /model/layers.2/self_attn/Expand_1_output_0 /model/layers.2/self_attn/Concat_10_output_0 /model/layers.2/self_attn/Reshape_6_output_0 Crop /model/layers.2/self_attn/Gather_4 1 1 /model/layers.2/self_attn/Shape_4_output_0 /model/layers.2/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_100 1 2 /model/layers.2/self_attn/Gather_4_output_0 /model/layers.2/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.2/self_attn/Gather_5 1 1 /model/layers.2/self_attn/Shape_5_output_0 /model/layers.2/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_101 1 2 /model/layers.2/self_attn/Gather_5_output_0 /model/layers.2/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.2/self_attn/Gather_6 1 1 /model/layers.2/self_attn/Shape_6_output_0 /model/layers.2/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_102 1 2 /model/layers.2/self_attn/Gather_6_output_0 /model/layers.2/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.2/self_attn/Gather_7 1 1 /model/layers.2/self_attn/Shape_7_output_0 /model/layers.2/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_103 1 2 /model/layers.2/self_attn/Gather_7_output_0 /model/layers.2/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.2/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.2/self_attn/Unsqueeze_11 2 1 /model/layers.2/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_33_output_0 /model/layers.2/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_15 2 1 /model/layers.2/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_1192 /model/layers.2/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_12 2 1 /model/layers.2/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_34_output_0 /model/layers.2/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.2/self_attn/Mul_5 2 1 /model/layers.2/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.2/self_attn/Constant_40_output_0 /model/layers.2/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.2/self_attn/Unsqueeze_13 2 1 /model/layers.2/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_36_output_0 /model/layers.2/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_17 2 1 /model/layers.2/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_1196 /model/layers.2/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_14 2 1 /model/layers.2/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_37_output_0 /model/layers.2/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_18 2 1 /model/layers.2/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_1198 /model/layers.2/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.2/self_attn/Unsqueeze_16 2 1 /model/layers.2/self_attn/Mul_5_output_0 onnx::Unsqueeze_1194 /model/layers.2/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.2/self_attn/Concat_7 5 1 /model/layers.2/self_attn/Unsqueeze_11_output_0 /model/layers.2/self_attn/Unsqueeze_12_output_0 /model/layers.2/self_attn/Constant_35_output_0 /model/layers.2/self_attn/Unsqueeze_13_output_0 /model/layers.2/self_attn/Unsqueeze_14_output_0 /model/layers.2/self_attn/Concat_7_output_0 0=0 Concat /model/layers.2/self_attn/Concat_8 4 1 /model/layers.2/self_attn/Unsqueeze_15_output_0 /model/layers.2/self_attn/Unsqueeze_16_output_0 /model/layers.2/self_attn/Unsqueeze_17_output_0 /model/layers.2/self_attn/Unsqueeze_18_output_0 /model/layers.2/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.2/self_attn/Reshape_3 1 1 /model/layers.2/self_attn/Concat_7_output_0 /model/layers.2/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_104 1 3 /model/layers.2/self_attn/Reshape_3_output_0 /model/layers.2/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.2/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.2/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.2/self_attn/Shape_8 1 1 /model/layers.2/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.2/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.2/self_attn/ConstantOfShape 1 1 /model/layers.2/self_attn/Shape_8_output_0 /model/layers.2/self_attn/ConstantOfShape_output_0 Split splitncnn_105 1 2 /model/layers.2/self_attn/ConstantOfShape_output_0 /model/layers.2/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.2/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.2/self_attn/Mul_4 2 1 /model/layers.2/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.2/self_attn/Constant_39_output_0 /model/layers.2/self_attn/Mul_4_output_0 0=2 Equal /model/layers.2/self_attn/Equal 2 1 /model/layers.2/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.2/self_attn/Mul_4_output_0 /model/layers.2/self_attn/Equal_output_0 Where /model/layers.2/self_attn/Where 3 1 /model/layers.2/self_attn/Equal_output_0 /model/layers.2/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.2/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.2/self_attn/Where_output_0 Expand /model/layers.2/self_attn/Expand 2 1 /model/layers.2/self_attn/Unsqueeze_10_output_0 /model/layers.2/self_attn/Where_output_0 /model/layers.2/self_attn/Expand_output_0 Reshape /model/layers.2/self_attn/Reshape_4 2 1 /model/layers.2/self_attn/Expand_output_0 /model/layers.2/self_attn/Concat_8_output_0 /model/layers.2/self_attn/Reshape_4_output_0 Split splitncnn_106 1 2 /model/layers.2/self_attn/Reshape_4_output_0 /model/layers.2/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.2/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.2/self_attn/Shape_14 1 1 /model/layers.2/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.2/self_attn/Shape_14_output_0 Permute /model/layers.2/self_attn/Transpose_3 1 1 /model/layers.2/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.2/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.2/self_attn/Gather_12 1 1 /model/layers.2/self_attn/Shape_14_output_0 /model/layers.2/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.2/self_attn/Mul_9 1 1 /model/layers.2/self_attn/Transpose_3_output_0 /model/layers.2/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.2/self_attn/Unsqueeze_28 2 1 /model/layers.2/self_attn/Gather_12_output_0 /model/layers.2/self_attn/Constant_57_output_0 /model/layers.2/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.2/self_attn/MatMul 2 1 /model/layers.2/self_attn/Mul_8_output_0 /model/layers.2/self_attn/Mul_9_output_0 /model/layers.2/self_attn/MatMul_output_0 Crop /model/layers.2/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_21 /model/layers.2/self_attn/Unsqueeze_28_output_0 /model/layers.2/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.2/self_attn/Where_2 3 1 /model/layers.2/self_attn/Slice_4_output_0 /model/layers.2/self_attn/Constant_61_output_0 /model/layers.2/self_attn/Constant_62_output_0 /model/layers.2/self_attn/Where_2_output_0 BinaryOp /model/layers.2/self_attn/Add_2 2 1 /model/layers.2/self_attn/MatMul_output_0 /model/layers.2/self_attn/Where_2_output_0 /model/layers.2/self_attn/Add_2_output_0 0=0 Softmax /model/layers.2/self_attn/Softmax 1 1 /model/layers.2/self_attn/Add_2_output_0 /model/layers.2/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_107 1 2 /model/layers.2/self_attn/Softmax_output_0 /model/layers.2/self_attn/Softmax_output_0_splitncnn_0 /model/layers.2/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.2/self_attn/IsNaN 1 1 /model/layers.2/self_attn/Softmax_output_0_splitncnn_1 /model/layers.2/self_attn/IsNaN_output_0 Where /model/layers.2/self_attn/Where_3 3 1 /model/layers.2/self_attn/IsNaN_output_0 /model/layers.2/self_attn/Constant_63_output_0 /model/layers.2/self_attn/Softmax_output_0_splitncnn_0 /model/layers.2/self_attn/Where_3_output_0 Gemm /model/layers.2/self_attn/MatMul_1 2 1 /model/layers.2/self_attn/Where_3_output_0 /model/layers.2/self_attn/Reshape_6_output_0 /model/layers.2/self_attn/MatMul_1_output_0 Permute /model/layers.2/self_attn/Transpose_4 1 1 /model/layers.2/self_attn/MatMul_1_output_0 /model/layers.2/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.2/self_attn/Reshape_7 2 1 /model/layers.2/self_attn/Transpose_4_output_0 /model/layers.2/self_attn/Concat_11_output_0 /model/layers.2/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.2/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.2/self_attn/Reshape_7_output_0 /model/layers.2/self_attn/Reshape_7_output_0_quantized /model/layers.2/self_attn/Reshape_7_output_0_scale /model/layers.2/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.2/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.2/self_attn/Reshape_7_output_0_scale /model/layers.2/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.244586e-03 MatMulInteger /model/layers.2/self_attn/o_proj/MatMul_quant 4 1 /model/layers.2/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7291_quantized /model/layers.2/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7291_zero_point /model/layers.2/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.2/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.2/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.2/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.2/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.2/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.2/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.2/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.2/Add 2 1 /model/layers.2/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.2/self_attn/o_proj/MatMul_output_0 /model/layers.2/Add_output_0 0=0 Cast /model/layers.2/post_attention_layernorm/Cast 1 1 /model/layers.2/Add_output_0 /model/layers.2/post_attention_layernorm/Cast_output_0 Split splitncnn_108 1 3 /model/layers.2/post_attention_layernorm/Cast_output_0 /model/layers.2/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.2/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.2/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.2/post_attention_layernorm/Pow 1 1 /model/layers.2/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.2/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.2/post_attention_layernorm/ReduceMean 2 1 /model/layers.2/post_attention_layernorm/Pow_output_0 /model/layers.2/post_attention_layernorm/Constant_1_output_0 /model/layers.2/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.2/post_attention_layernorm/Add 1 1 /model/layers.2/post_attention_layernorm/ReduceMean_output_0 /model/layers.2/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.2/post_attention_layernorm/Sqrt 1 1 /model/layers.2/post_attention_layernorm/Add_output_0 /model/layers.2/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.2/post_attention_layernorm/Div 1 1 /model/layers.2/post_attention_layernorm/Sqrt_output_0 /model/layers.2/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.2/post_attention_layernorm/Mul 2 1 /model/layers.2/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.2/post_attention_layernorm/Div_output_0 /model/layers.2/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.2/post_attention_layernorm/Cast_1 1 1 /model/layers.2/post_attention_layernorm/Mul_output_0 /model/layers.2/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.2/post_attention_layernorm/Mul_1 2 1 model.layers.2.post_attention_layernorm.weight /model/layers.2/post_attention_layernorm/Cast_1_output_0 /model/layers.2/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.2/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.2/post_attention_layernorm/Mul_1_output_0 /model/layers.2/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.2/post_attention_layernorm/Mul_1_output_0_scale /model/layers.2/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_109 1 2 /model/layers.2/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.2/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.2/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_110 1 2 /model/layers.2/post_attention_layernorm/Mul_1_output_0_scale /model/layers.2/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.2/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_111 1 2 /model/layers.2/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.2/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.2/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.2/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.2/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.2/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.567913e-03 BinaryOp /model/layers.2/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.2/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.2/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=5.013533e-03 MatMulInteger /model/layers.2/mlp/gate_proj/MatMul_quant 4 1 /model/layers.2/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7292_quantized /model/layers.2/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7292_zero_point /model/layers.2/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.2/mlp/up_proj/MatMul_quant 4 1 /model/layers.2/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7293_quantized /model/layers.2/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7293_zero_point /model/layers.2/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.2/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.2/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.2/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.2/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.2/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.2/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.2/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.2/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.2/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.2/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.2/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.2/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.2/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.2/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.2/mlp/act_fn/Mul 1 1 /model/layers.2/mlp/gate_proj/MatMul_output_0 /model/layers.2/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.2/mlp/Mul 2 1 /model/layers.2/mlp/act_fn/Mul_output_0 /model/layers.2/mlp/up_proj/MatMul_output_0 /model/layers.2/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.2/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.2/mlp/Mul_output_0 /model/layers.2/mlp/Mul_output_0_quantized /model/layers.2/mlp/Mul_output_0_scale /model/layers.2/mlp/Mul_output_0_zero_point BinaryOp /model/layers.2/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.2/mlp/Mul_output_0_scale /model/layers.2/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.475640e-03 MatMulInteger /model/layers.2/mlp/down_proj/MatMul_quant 4 1 /model/layers.2/mlp/Mul_output_0_quantized onnx::MatMul_7294_quantized /model/layers.2/mlp/Mul_output_0_zero_point onnx::MatMul_7294_zero_point /model/layers.2/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.2/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.2/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.2/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.2/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.2/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.2/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.2/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.2/Add_1 2 1 /model/layers.2/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.2/mlp/down_proj/MatMul_output_0 /model/layers.2/Add_1_output_0 0=0 Cast /model/layers.3/input_layernorm/Cast 1 1 /model/layers.2/Add_1_output_0 /model/layers.3/input_layernorm/Cast_output_0 Split splitncnn_112 1 3 /model/layers.3/input_layernorm/Cast_output_0 /model/layers.3/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.3/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.3/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.3/input_layernorm/Pow 1 1 /model/layers.3/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.3/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.3/input_layernorm/ReduceMean 2 1 /model/layers.3/input_layernorm/Pow_output_0 /model/layers.3/input_layernorm/Constant_1_output_0 /model/layers.3/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.3/input_layernorm/Add 1 1 /model/layers.3/input_layernorm/ReduceMean_output_0 /model/layers.3/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.3/input_layernorm/Sqrt 1 1 /model/layers.3/input_layernorm/Add_output_0 /model/layers.3/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.3/input_layernorm/Div 1 1 /model/layers.3/input_layernorm/Sqrt_output_0 /model/layers.3/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.3/input_layernorm/Mul 2 1 /model/layers.3/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.3/input_layernorm/Div_output_0 /model/layers.3/input_layernorm/Mul_output_0 0=2 Cast /model/layers.3/input_layernorm/Cast_1 1 1 /model/layers.3/input_layernorm/Mul_output_0 /model/layers.3/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.3/input_layernorm/Mul_1 2 1 model.layers.3.input_layernorm.weight /model/layers.3/input_layernorm/Cast_1_output_0 /model/layers.3/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_113 1 3 /model/layers.3/input_layernorm/Mul_1_output_0 /model/layers.3/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.3/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.3/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.3/self_attn/Shape 1 1 /model/layers.3/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.3/self_attn/Shape_output_0 Shape /model/layers.3/self_attn/Shape_1 1 1 /model/layers.3/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.3/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.3/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.3/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.3/input_layernorm/Mul_1_output_0_quantized /model/layers.3/input_layernorm/Mul_1_output_0_scale /model/layers.3/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_114 1 3 /model/layers.3/input_layernorm/Mul_1_output_0_quantized /model/layers.3/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.3/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.3/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_115 1 3 /model/layers.3/input_layernorm/Mul_1_output_0_scale /model/layers.3/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.3/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.3/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_116 1 3 /model/layers.3/input_layernorm/Mul_1_output_0_zero_point /model/layers.3/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.3/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.3/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.3/self_attn/Gather 1 1 /model/layers.3/self_attn/Shape_output_0 /model/layers.3/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_117 1 4 /model/layers.3/self_attn/Gather_output_0 /model/layers.3/self_attn/Gather_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_output_0_splitncnn_1 /model/layers.3/self_attn/Gather_output_0_splitncnn_2 /model/layers.3/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.3/self_attn/Gather_1 1 1 /model/layers.3/self_attn/Shape_1_output_0 /model/layers.3/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_118 1 4 /model/layers.3/self_attn/Gather_1_output_0 /model/layers.3/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.3/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.3/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.3/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.3/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.3/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.030020e-03 BinaryOp /model/layers.3/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.3/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.3/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.522515e-03 BinaryOp /model/layers.3/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.3/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.3/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=6.805180e-04 MatMulInteger /model/layers.3/self_attn/q_proj/MatMul_quant 4 1 /model/layers.3/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7295_quantized /model/layers.3/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7295_zero_point /model/layers.3/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.3/self_attn/k_proj/MatMul_quant 4 1 /model/layers.3/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7302_quantized /model/layers.3/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7302_zero_point /model/layers.3/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.3/self_attn/v_proj/MatMul_quant 4 1 /model/layers.3/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7303_quantized /model/layers.3/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7303_zero_point /model/layers.3/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.3/self_attn/Unsqueeze 2 1 /model/layers.3/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_1334 /model/layers.3/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_2 2 1 /model/layers.3/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_1343 /model/layers.3/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_4 2 1 /model/layers.3/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_1352 /model/layers.3/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_29 2 1 /model/layers.3/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_1557 /model/layers.3/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_1 2 1 /model/layers.3/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_1336 /model/layers.3/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_3 2 1 /model/layers.3/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_1345 /model/layers.3/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_5 2 1 /model/layers.3/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_1354 /model/layers.3/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_30 2 1 /model/layers.3/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_1559 /model/layers.3/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.3/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.3/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.3/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.3/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.3/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.3/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.3/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.3/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.3/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.3/self_attn/Concat 4 1 /model/layers.3/self_attn/Unsqueeze_output_0 /model/layers.3/self_attn/Unsqueeze_1_output_0 /model/layers.3/self_attn/Constant_2_output_0 /model/layers.3/self_attn/Constant_3_output_0 /model/layers.3/self_attn/Concat_output_0 0=0 Concat /model/layers.3/self_attn/Concat_1 4 1 /model/layers.3/self_attn/Unsqueeze_2_output_0 /model/layers.3/self_attn/Unsqueeze_3_output_0 /model/layers.3/self_attn/Constant_4_output_0 /model/layers.3/self_attn/Constant_5_output_0 /model/layers.3/self_attn/Concat_1_output_0 0=0 Concat /model/layers.3/self_attn/Concat_2 4 1 /model/layers.3/self_attn/Unsqueeze_4_output_0 /model/layers.3/self_attn/Unsqueeze_5_output_0 /model/layers.3/self_attn/Constant_6_output_0 /model/layers.3/self_attn/Constant_7_output_0 /model/layers.3/self_attn/Concat_2_output_0 0=0 Concat /model/layers.3/self_attn/Concat_11 3 1 /model/layers.3/self_attn/Unsqueeze_29_output_0 /model/layers.3/self_attn/Unsqueeze_30_output_0 /model/layers.3/self_attn/Constant_64_output_0 /model/layers.3/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.3/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.3/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.3/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.3/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.3/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.3/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.3/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.3/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.3/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.3/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.3/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.3/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.3/self_attn/q_proj/Add 2 1 model.layers.3.self_attn.q_proj.bias /model/layers.3/self_attn/q_proj/MatMul_output_0 /model/layers.3/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.3/self_attn/k_proj/Add 2 1 model.layers.3.self_attn.k_proj.bias /model/layers.3/self_attn/k_proj/MatMul_output_0 /model/layers.3/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.3/self_attn/v_proj/Add 2 1 model.layers.3.self_attn.v_proj.bias /model/layers.3/self_attn/v_proj/MatMul_output_0 /model/layers.3/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.3/self_attn/Reshape 2 1 /model/layers.3/self_attn/q_proj/Add_output_0 /model/layers.3/self_attn/Concat_output_0 /model/layers.3/self_attn/Reshape_output_0 Reshape /model/layers.3/self_attn/Reshape_1 2 1 /model/layers.3/self_attn/k_proj/Add_output_0 /model/layers.3/self_attn/Concat_1_output_0 /model/layers.3/self_attn/Reshape_1_output_0 Reshape /model/layers.3/self_attn/Reshape_2 2 1 /model/layers.3/self_attn/v_proj/Add_output_0 /model/layers.3/self_attn/Concat_2_output_0 /model/layers.3/self_attn/Reshape_2_output_0 Permute /model/layers.3/self_attn/Transpose 1 1 /model/layers.3/self_attn/Reshape_output_0 /model/layers.3/self_attn/Transpose_output_0 0=2 Split splitncnn_119 1 4 /model/layers.3/self_attn/Transpose_output_0 /model/layers.3/self_attn/Transpose_output_0_splitncnn_0 /model/layers.3/self_attn/Transpose_output_0_splitncnn_1 /model/layers.3/self_attn/Transpose_output_0_splitncnn_2 /model/layers.3/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.3/self_attn/Transpose_1 1 1 /model/layers.3/self_attn/Reshape_1_output_0 /model/layers.3/self_attn/Transpose_1_output_0 0=2 Split splitncnn_120 1 4 /model/layers.3/self_attn/Transpose_1_output_0 /model/layers.3/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.3/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.3/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.3/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.3/self_attn/Transpose_2 1 1 /model/layers.3/self_attn/Reshape_2_output_0 /model/layers.3/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.3/self_attn/Mul 2 1 /model/layers.3/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_41 /model/layers.3/self_attn/Mul_output_0 0=2 Shape /model/layers.3/self_attn/Shape_2 1 1 /model/layers.3/self_attn/Transpose_output_0_splitncnn_2 /model/layers.3/self_attn/Shape_2_output_0 BinaryOp /model/layers.3/self_attn/Mul_2 2 1 /model/layers.3/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_40 /model/layers.3/self_attn/Mul_2_output_0 0=2 Shape /model/layers.3/self_attn/Shape_3 1 1 /model/layers.3/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.3/self_attn/Shape_3_output_0 Concat /model/layers.3/self_attn/Concat_6 2 1 past_key_values.3.value /model/layers.3/self_attn/Transpose_2_output_0 present.3.value 0=-2 Split splitncnn_121 1 5 present.3.value present.3.value_splitncnn_0 present.3.value_splitncnn_1 present.3.value_splitncnn_2 present.3.value_splitncnn_3 present.3.value_splitncnn_4 Crop /model/layers.3/self_attn/Gather_2 1 1 /model/layers.3/self_attn/Shape_2_output_0 /model/layers.3/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.3/self_attn/Gather_3 1 1 /model/layers.3/self_attn/Shape_3_output_0 /model/layers.3/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.3/self_attn/Shape_9 1 1 present.3.value_splitncnn_4 /model/layers.3/self_attn/Shape_9_output_0 Shape /model/layers.3/self_attn/Shape_10 1 1 present.3.value_splitncnn_3 /model/layers.3/self_attn/Shape_10_output_0 Shape /model/layers.3/self_attn/Shape_11 1 1 present.3.value_splitncnn_2 /model/layers.3/self_attn/Shape_11_output_0 Shape /model/layers.3/self_attn/Shape_12 1 1 present.3.value_splitncnn_1 /model/layers.3/self_attn/Shape_12_output_0 ExpandDims /model/layers.3/self_attn/Unsqueeze_19 2 1 present.3.value_splitncnn_0 /model/layers.3/self_attn/Constant_45_output_0 /model/layers.3/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.3/self_attn/Div 2 1 /model/layers.3/self_attn/Gather_2_output_0 /model/layers.3/self_attn/Constant_9_output_0 /model/layers.3/self_attn/Div_output_0 0=3 BinaryOp /model/layers.3/self_attn/Div_1 2 1 /model/layers.3/self_attn/Gather_3_output_0 /model/layers.3/self_attn/Constant_19_output_0 /model/layers.3/self_attn/Div_1_output_0 0=3 Crop /model/layers.3/self_attn/Gather_8 1 1 /model/layers.3/self_attn/Shape_9_output_0 /model/layers.3/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_122 1 2 /model/layers.3/self_attn/Gather_8_output_0 /model/layers.3/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.3/self_attn/Gather_9 1 1 /model/layers.3/self_attn/Shape_10_output_0 /model/layers.3/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_123 1 2 /model/layers.3/self_attn/Gather_9_output_0 /model/layers.3/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.3/self_attn/Gather_10 1 1 /model/layers.3/self_attn/Shape_11_output_0 /model/layers.3/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_124 1 2 /model/layers.3/self_attn/Gather_10_output_0 /model/layers.3/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.3/self_attn/Gather_11 1 1 /model/layers.3/self_attn/Shape_12_output_0 /model/layers.3/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_125 1 2 /model/layers.3/self_attn/Gather_11_output_0 /model/layers.3/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.3/self_attn/Cast 1 1 /model/layers.3/self_attn/Div_output_0 /model/layers.3/self_attn/Cast_output_0 Cast /model/layers.3/self_attn/Cast_2 1 1 /model/layers.3/self_attn/Div_1_output_0 /model/layers.3/self_attn/Cast_2_output_0 ExpandDims /model/layers.3/self_attn/Unsqueeze_20 2 1 /model/layers.3/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_46_output_0 /model/layers.3/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_24 2 1 /model/layers.3/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_1517 /model/layers.3/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_21 2 1 /model/layers.3/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_47_output_0 /model/layers.3/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.3/self_attn/Mul_7 2 1 /model/layers.3/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.3/self_attn/Constant_53_output_0 /model/layers.3/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.3/self_attn/Unsqueeze_22 2 1 /model/layers.3/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_49_output_0 /model/layers.3/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_26 2 1 /model/layers.3/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_1521 /model/layers.3/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_23 2 1 /model/layers.3/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_50_output_0 /model/layers.3/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_27 2 1 /model/layers.3/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_1523 /model/layers.3/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.3/self_attn/Cast_1 1 1 /model/layers.3/self_attn/Cast_output_0 /model/layers.3/self_attn/Cast_1_output_0 Split splitncnn_126 1 2 /model/layers.3/self_attn/Cast_1_output_0 /model/layers.3/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.3/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.3/self_attn/Cast_3 1 1 /model/layers.3/self_attn/Cast_2_output_0 /model/layers.3/self_attn/Cast_3_output_0 Split splitncnn_127 1 2 /model/layers.3/self_attn/Cast_3_output_0 /model/layers.3/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.3/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.3/self_attn/Unsqueeze_25 2 1 /model/layers.3/self_attn/Mul_7_output_0 onnx::Unsqueeze_1519 /model/layers.3/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.3/self_attn/Concat_9 5 1 /model/layers.3/self_attn/Unsqueeze_20_output_0 /model/layers.3/self_attn/Unsqueeze_21_output_0 /model/layers.3/self_attn/Constant_48_output_0 /model/layers.3/self_attn/Unsqueeze_22_output_0 /model/layers.3/self_attn/Unsqueeze_23_output_0 /model/layers.3/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_6 2 1 /model/layers.3/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_12_output_0 /model/layers.3/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_7 2 1 /model/layers.3/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.3/self_attn/Constant_15_output_0 /model/layers.3/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_8 2 1 /model/layers.3/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_22_output_0 /model/layers.3/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_9 2 1 /model/layers.3/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.3/self_attn/Constant_25_output_0 /model/layers.3/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.3/self_attn/Concat_10 4 1 /model/layers.3/self_attn/Unsqueeze_24_output_0 /model/layers.3/self_attn/Unsqueeze_25_output_0 /model/layers.3/self_attn/Unsqueeze_26_output_0 /model/layers.3/self_attn/Unsqueeze_27_output_0 /model/layers.3/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.3/self_attn/Reshape_5 1 1 /model/layers.3/self_attn/Concat_9_output_0 /model/layers.3/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_128 1 3 /model/layers.3/self_attn/Reshape_5_output_0 /model/layers.3/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.3/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.3/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.3/self_attn/Slice 2 1 /model/layers.3/self_attn/Transpose_output_0_splitncnn_1 /model/layers.3/self_attn/Unsqueeze_6_output_0 /model/layers.3/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.3/self_attn/Slice_1 2 1 /model/layers.3/self_attn/Transpose_output_0_splitncnn_0 /model/layers.3/self_attn/Unsqueeze_7_output_0 /model/layers.3/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.3/self_attn/Slice_2 2 1 /model/layers.3/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.3/self_attn/Unsqueeze_8_output_0 /model/layers.3/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.3/self_attn/Slice_3 2 1 /model/layers.3/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.3/self_attn/Unsqueeze_9_output_0 /model/layers.3/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.3/self_attn/Shape_13 1 1 /model/layers.3/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.3/self_attn/Shape_13_output_0 UnaryOp /model/layers.3/self_attn/Neg 1 1 /model/layers.3/self_attn/Slice_1_output_0 /model/layers.3/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.3/self_attn/Neg_1 1 1 /model/layers.3/self_attn/Slice_3_output_0 /model/layers.3/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.3/self_attn/ConstantOfShape_1 1 1 /model/layers.3/self_attn/Shape_13_output_0 /model/layers.3/self_attn/ConstantOfShape_1_output_0 Split splitncnn_129 1 2 /model/layers.3/self_attn/ConstantOfShape_1_output_0 /model/layers.3/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.3/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.3/self_attn/Concat_3 2 1 /model/layers.3/self_attn/Neg_output_0 /model/layers.3/self_attn/Slice_output_0 /model/layers.3/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.3/self_attn/Concat_4 2 1 /model/layers.3/self_attn/Neg_1_output_0 /model/layers.3/self_attn/Slice_2_output_0 /model/layers.3/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.3/self_attn/Mul_6 2 1 /model/layers.3/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_52_output_0 /model/layers.3/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.3/self_attn/Mul_1 2 1 /model/layers.3/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_41 /model/layers.3/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.3/self_attn/Mul_3 2 1 /model/layers.3/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_40 /model/layers.3/self_attn/Mul_3_output_0 0=2 Equal /model/layers.3/self_attn/Equal_1 2 1 /model/layers.3/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.3/self_attn/Mul_6_output_0 /model/layers.3/self_attn/Equal_1_output_0 BinaryOp /model/layers.3/self_attn/Add 2 1 /model/layers.3/self_attn/Mul_output_0 /model/layers.3/self_attn/Mul_1_output_0 /model/layers.3/self_attn/Add_output_0 0=0 BinaryOp /model/layers.3/self_attn/Add_1 2 1 /model/layers.3/self_attn/Mul_2_output_0 /model/layers.3/self_attn/Mul_3_output_0 /model/layers.3/self_attn/Add_1_output_0 0=0 Where /model/layers.3/self_attn/Where_1 3 1 /model/layers.3/self_attn/Equal_1_output_0 /model/layers.3/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.3/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.3/self_attn/Where_1_output_0 BinaryOp /model/layers.3/self_attn/Mul_8 1 1 /model/layers.3/self_attn/Add_output_0 /model/layers.3/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.3/self_attn/Concat_5 2 1 past_key_values.3.key /model/layers.3/self_attn/Add_1_output_0 present.3.key 0=-2 Split splitncnn_130 1 5 present.3.key present.3.key_splitncnn_0 present.3.key_splitncnn_1 present.3.key_splitncnn_2 present.3.key_splitncnn_3 present.3.key_splitncnn_4 Expand /model/layers.3/self_attn/Expand_1 2 1 /model/layers.3/self_attn/Unsqueeze_19_output_0 /model/layers.3/self_attn/Where_1_output_0 /model/layers.3/self_attn/Expand_1_output_0 Shape /model/layers.3/self_attn/Shape_4 1 1 present.3.key_splitncnn_4 /model/layers.3/self_attn/Shape_4_output_0 Shape /model/layers.3/self_attn/Shape_5 1 1 present.3.key_splitncnn_3 /model/layers.3/self_attn/Shape_5_output_0 Shape /model/layers.3/self_attn/Shape_6 1 1 present.3.key_splitncnn_2 /model/layers.3/self_attn/Shape_6_output_0 Shape /model/layers.3/self_attn/Shape_7 1 1 present.3.key_splitncnn_1 /model/layers.3/self_attn/Shape_7_output_0 ExpandDims /model/layers.3/self_attn/Unsqueeze_10 2 1 present.3.key_splitncnn_0 /model/layers.3/self_attn/Constant_32_output_0 /model/layers.3/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.3/self_attn/Reshape_6 2 1 /model/layers.3/self_attn/Expand_1_output_0 /model/layers.3/self_attn/Concat_10_output_0 /model/layers.3/self_attn/Reshape_6_output_0 Crop /model/layers.3/self_attn/Gather_4 1 1 /model/layers.3/self_attn/Shape_4_output_0 /model/layers.3/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_131 1 2 /model/layers.3/self_attn/Gather_4_output_0 /model/layers.3/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.3/self_attn/Gather_5 1 1 /model/layers.3/self_attn/Shape_5_output_0 /model/layers.3/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_132 1 2 /model/layers.3/self_attn/Gather_5_output_0 /model/layers.3/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.3/self_attn/Gather_6 1 1 /model/layers.3/self_attn/Shape_6_output_0 /model/layers.3/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_133 1 2 /model/layers.3/self_attn/Gather_6_output_0 /model/layers.3/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.3/self_attn/Gather_7 1 1 /model/layers.3/self_attn/Shape_7_output_0 /model/layers.3/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_134 1 2 /model/layers.3/self_attn/Gather_7_output_0 /model/layers.3/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.3/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.3/self_attn/Unsqueeze_11 2 1 /model/layers.3/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_33_output_0 /model/layers.3/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_15 2 1 /model/layers.3/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_1471 /model/layers.3/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_12 2 1 /model/layers.3/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_34_output_0 /model/layers.3/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.3/self_attn/Mul_5 2 1 /model/layers.3/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.3/self_attn/Constant_40_output_0 /model/layers.3/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.3/self_attn/Unsqueeze_13 2 1 /model/layers.3/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_36_output_0 /model/layers.3/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_17 2 1 /model/layers.3/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_1475 /model/layers.3/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_14 2 1 /model/layers.3/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_37_output_0 /model/layers.3/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_18 2 1 /model/layers.3/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_1477 /model/layers.3/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.3/self_attn/Unsqueeze_16 2 1 /model/layers.3/self_attn/Mul_5_output_0 onnx::Unsqueeze_1473 /model/layers.3/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.3/self_attn/Concat_7 5 1 /model/layers.3/self_attn/Unsqueeze_11_output_0 /model/layers.3/self_attn/Unsqueeze_12_output_0 /model/layers.3/self_attn/Constant_35_output_0 /model/layers.3/self_attn/Unsqueeze_13_output_0 /model/layers.3/self_attn/Unsqueeze_14_output_0 /model/layers.3/self_attn/Concat_7_output_0 0=0 Concat /model/layers.3/self_attn/Concat_8 4 1 /model/layers.3/self_attn/Unsqueeze_15_output_0 /model/layers.3/self_attn/Unsqueeze_16_output_0 /model/layers.3/self_attn/Unsqueeze_17_output_0 /model/layers.3/self_attn/Unsqueeze_18_output_0 /model/layers.3/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.3/self_attn/Reshape_3 1 1 /model/layers.3/self_attn/Concat_7_output_0 /model/layers.3/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_135 1 3 /model/layers.3/self_attn/Reshape_3_output_0 /model/layers.3/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.3/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.3/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.3/self_attn/Shape_8 1 1 /model/layers.3/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.3/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.3/self_attn/ConstantOfShape 1 1 /model/layers.3/self_attn/Shape_8_output_0 /model/layers.3/self_attn/ConstantOfShape_output_0 Split splitncnn_136 1 2 /model/layers.3/self_attn/ConstantOfShape_output_0 /model/layers.3/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.3/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.3/self_attn/Mul_4 2 1 /model/layers.3/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.3/self_attn/Constant_39_output_0 /model/layers.3/self_attn/Mul_4_output_0 0=2 Equal /model/layers.3/self_attn/Equal 2 1 /model/layers.3/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.3/self_attn/Mul_4_output_0 /model/layers.3/self_attn/Equal_output_0 Where /model/layers.3/self_attn/Where 3 1 /model/layers.3/self_attn/Equal_output_0 /model/layers.3/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.3/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.3/self_attn/Where_output_0 Expand /model/layers.3/self_attn/Expand 2 1 /model/layers.3/self_attn/Unsqueeze_10_output_0 /model/layers.3/self_attn/Where_output_0 /model/layers.3/self_attn/Expand_output_0 Reshape /model/layers.3/self_attn/Reshape_4 2 1 /model/layers.3/self_attn/Expand_output_0 /model/layers.3/self_attn/Concat_8_output_0 /model/layers.3/self_attn/Reshape_4_output_0 Split splitncnn_137 1 2 /model/layers.3/self_attn/Reshape_4_output_0 /model/layers.3/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.3/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.3/self_attn/Shape_14 1 1 /model/layers.3/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.3/self_attn/Shape_14_output_0 Permute /model/layers.3/self_attn/Transpose_3 1 1 /model/layers.3/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.3/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.3/self_attn/Gather_12 1 1 /model/layers.3/self_attn/Shape_14_output_0 /model/layers.3/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.3/self_attn/Mul_9 1 1 /model/layers.3/self_attn/Transpose_3_output_0 /model/layers.3/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.3/self_attn/Unsqueeze_28 2 1 /model/layers.3/self_attn/Gather_12_output_0 /model/layers.3/self_attn/Constant_57_output_0 /model/layers.3/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.3/self_attn/MatMul 2 1 /model/layers.3/self_attn/Mul_8_output_0 /model/layers.3/self_attn/Mul_9_output_0 /model/layers.3/self_attn/MatMul_output_0 Crop /model/layers.3/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_20 /model/layers.3/self_attn/Unsqueeze_28_output_0 /model/layers.3/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.3/self_attn/Where_2 3 1 /model/layers.3/self_attn/Slice_4_output_0 /model/layers.3/self_attn/Constant_61_output_0 /model/layers.3/self_attn/Constant_62_output_0 /model/layers.3/self_attn/Where_2_output_0 BinaryOp /model/layers.3/self_attn/Add_2 2 1 /model/layers.3/self_attn/MatMul_output_0 /model/layers.3/self_attn/Where_2_output_0 /model/layers.3/self_attn/Add_2_output_0 0=0 Softmax /model/layers.3/self_attn/Softmax 1 1 /model/layers.3/self_attn/Add_2_output_0 /model/layers.3/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_138 1 2 /model/layers.3/self_attn/Softmax_output_0 /model/layers.3/self_attn/Softmax_output_0_splitncnn_0 /model/layers.3/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.3/self_attn/IsNaN 1 1 /model/layers.3/self_attn/Softmax_output_0_splitncnn_1 /model/layers.3/self_attn/IsNaN_output_0 Where /model/layers.3/self_attn/Where_3 3 1 /model/layers.3/self_attn/IsNaN_output_0 /model/layers.3/self_attn/Constant_63_output_0 /model/layers.3/self_attn/Softmax_output_0_splitncnn_0 /model/layers.3/self_attn/Where_3_output_0 Gemm /model/layers.3/self_attn/MatMul_1 2 1 /model/layers.3/self_attn/Where_3_output_0 /model/layers.3/self_attn/Reshape_6_output_0 /model/layers.3/self_attn/MatMul_1_output_0 Permute /model/layers.3/self_attn/Transpose_4 1 1 /model/layers.3/self_attn/MatMul_1_output_0 /model/layers.3/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.3/self_attn/Reshape_7 2 1 /model/layers.3/self_attn/Transpose_4_output_0 /model/layers.3/self_attn/Concat_11_output_0 /model/layers.3/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.3/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.3/self_attn/Reshape_7_output_0 /model/layers.3/self_attn/Reshape_7_output_0_quantized /model/layers.3/self_attn/Reshape_7_output_0_scale /model/layers.3/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.3/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.3/self_attn/Reshape_7_output_0_scale /model/layers.3/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.383735e-03 MatMulInteger /model/layers.3/self_attn/o_proj/MatMul_quant 4 1 /model/layers.3/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7326_quantized /model/layers.3/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7326_zero_point /model/layers.3/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.3/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.3/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.3/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.3/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.3/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.3/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.3/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.3/Add 2 1 /model/layers.3/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.3/self_attn/o_proj/MatMul_output_0 /model/layers.3/Add_output_0 0=0 Cast /model/layers.3/post_attention_layernorm/Cast 1 1 /model/layers.3/Add_output_0 /model/layers.3/post_attention_layernorm/Cast_output_0 Split splitncnn_139 1 3 /model/layers.3/post_attention_layernorm/Cast_output_0 /model/layers.3/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.3/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.3/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.3/post_attention_layernorm/Pow 1 1 /model/layers.3/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.3/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.3/post_attention_layernorm/ReduceMean 2 1 /model/layers.3/post_attention_layernorm/Pow_output_0 /model/layers.3/post_attention_layernorm/Constant_1_output_0 /model/layers.3/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.3/post_attention_layernorm/Add 1 1 /model/layers.3/post_attention_layernorm/ReduceMean_output_0 /model/layers.3/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.3/post_attention_layernorm/Sqrt 1 1 /model/layers.3/post_attention_layernorm/Add_output_0 /model/layers.3/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.3/post_attention_layernorm/Div 1 1 /model/layers.3/post_attention_layernorm/Sqrt_output_0 /model/layers.3/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.3/post_attention_layernorm/Mul 2 1 /model/layers.3/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.3/post_attention_layernorm/Div_output_0 /model/layers.3/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.3/post_attention_layernorm/Cast_1 1 1 /model/layers.3/post_attention_layernorm/Mul_output_0 /model/layers.3/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.3/post_attention_layernorm/Mul_1 2 1 model.layers.3.post_attention_layernorm.weight /model/layers.3/post_attention_layernorm/Cast_1_output_0 /model/layers.3/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.3/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.3/post_attention_layernorm/Mul_1_output_0 /model/layers.3/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.3/post_attention_layernorm/Mul_1_output_0_scale /model/layers.3/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_140 1 2 /model/layers.3/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.3/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.3/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_141 1 2 /model/layers.3/post_attention_layernorm/Mul_1_output_0_scale /model/layers.3/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.3/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_142 1 2 /model/layers.3/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.3/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.3/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.3/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.3/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.3/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.460630e-03 BinaryOp /model/layers.3/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.3/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.3/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.537525e-03 MatMulInteger /model/layers.3/mlp/gate_proj/MatMul_quant 4 1 /model/layers.3/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7327_quantized /model/layers.3/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7327_zero_point /model/layers.3/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.3/mlp/up_proj/MatMul_quant 4 1 /model/layers.3/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7328_quantized /model/layers.3/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7328_zero_point /model/layers.3/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.3/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.3/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.3/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.3/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.3/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.3/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.3/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.3/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.3/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.3/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.3/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.3/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.3/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.3/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.3/mlp/act_fn/Mul 1 1 /model/layers.3/mlp/gate_proj/MatMul_output_0 /model/layers.3/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.3/mlp/Mul 2 1 /model/layers.3/mlp/act_fn/Mul_output_0 /model/layers.3/mlp/up_proj/MatMul_output_0 /model/layers.3/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.3/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.3/mlp/Mul_output_0 /model/layers.3/mlp/Mul_output_0_quantized /model/layers.3/mlp/Mul_output_0_scale /model/layers.3/mlp/Mul_output_0_zero_point BinaryOp /model/layers.3/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.3/mlp/Mul_output_0_scale /model/layers.3/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.029281e-03 MatMulInteger /model/layers.3/mlp/down_proj/MatMul_quant 4 1 /model/layers.3/mlp/Mul_output_0_quantized onnx::MatMul_7329_quantized /model/layers.3/mlp/Mul_output_0_zero_point onnx::MatMul_7329_zero_point /model/layers.3/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.3/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.3/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.3/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.3/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.3/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.3/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.3/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.3/Add_1 2 1 /model/layers.3/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.3/mlp/down_proj/MatMul_output_0 /model/layers.3/Add_1_output_0 0=0 Cast /model/layers.4/input_layernorm/Cast 1 1 /model/layers.3/Add_1_output_0 /model/layers.4/input_layernorm/Cast_output_0 Split splitncnn_143 1 3 /model/layers.4/input_layernorm/Cast_output_0 /model/layers.4/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.4/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.4/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.4/input_layernorm/Pow 1 1 /model/layers.4/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.4/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.4/input_layernorm/ReduceMean 2 1 /model/layers.4/input_layernorm/Pow_output_0 /model/layers.4/input_layernorm/Constant_1_output_0 /model/layers.4/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.4/input_layernorm/Add 1 1 /model/layers.4/input_layernorm/ReduceMean_output_0 /model/layers.4/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.4/input_layernorm/Sqrt 1 1 /model/layers.4/input_layernorm/Add_output_0 /model/layers.4/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.4/input_layernorm/Div 1 1 /model/layers.4/input_layernorm/Sqrt_output_0 /model/layers.4/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.4/input_layernorm/Mul 2 1 /model/layers.4/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.4/input_layernorm/Div_output_0 /model/layers.4/input_layernorm/Mul_output_0 0=2 Cast /model/layers.4/input_layernorm/Cast_1 1 1 /model/layers.4/input_layernorm/Mul_output_0 /model/layers.4/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.4/input_layernorm/Mul_1 2 1 model.layers.4.input_layernorm.weight /model/layers.4/input_layernorm/Cast_1_output_0 /model/layers.4/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_144 1 3 /model/layers.4/input_layernorm/Mul_1_output_0 /model/layers.4/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.4/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.4/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.4/self_attn/Shape 1 1 /model/layers.4/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.4/self_attn/Shape_output_0 Shape /model/layers.4/self_attn/Shape_1 1 1 /model/layers.4/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.4/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.4/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.4/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.4/input_layernorm/Mul_1_output_0_quantized /model/layers.4/input_layernorm/Mul_1_output_0_scale /model/layers.4/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_145 1 3 /model/layers.4/input_layernorm/Mul_1_output_0_quantized /model/layers.4/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.4/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.4/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_146 1 3 /model/layers.4/input_layernorm/Mul_1_output_0_scale /model/layers.4/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.4/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.4/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_147 1 3 /model/layers.4/input_layernorm/Mul_1_output_0_zero_point /model/layers.4/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.4/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.4/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.4/self_attn/Gather 1 1 /model/layers.4/self_attn/Shape_output_0 /model/layers.4/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_148 1 4 /model/layers.4/self_attn/Gather_output_0 /model/layers.4/self_attn/Gather_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_output_0_splitncnn_1 /model/layers.4/self_attn/Gather_output_0_splitncnn_2 /model/layers.4/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.4/self_attn/Gather_1 1 1 /model/layers.4/self_attn/Shape_1_output_0 /model/layers.4/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_149 1 4 /model/layers.4/self_attn/Gather_1_output_0 /model/layers.4/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.4/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.4/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.4/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.4/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.4/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.676304e-03 BinaryOp /model/layers.4/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.4/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.4/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.430241e-03 BinaryOp /model/layers.4/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.4/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.4/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=8.689099e-04 MatMulInteger /model/layers.4/self_attn/q_proj/MatMul_quant 4 1 /model/layers.4/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7330_quantized /model/layers.4/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7330_zero_point /model/layers.4/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.4/self_attn/k_proj/MatMul_quant 4 1 /model/layers.4/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7337_quantized /model/layers.4/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7337_zero_point /model/layers.4/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.4/self_attn/v_proj/MatMul_quant 4 1 /model/layers.4/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7338_quantized /model/layers.4/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7338_zero_point /model/layers.4/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.4/self_attn/Unsqueeze 2 1 /model/layers.4/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_1613 /model/layers.4/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_2 2 1 /model/layers.4/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_1622 /model/layers.4/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_4 2 1 /model/layers.4/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_1631 /model/layers.4/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_29 2 1 /model/layers.4/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_1836 /model/layers.4/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_1 2 1 /model/layers.4/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_1615 /model/layers.4/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_3 2 1 /model/layers.4/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_1624 /model/layers.4/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_5 2 1 /model/layers.4/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_1633 /model/layers.4/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_30 2 1 /model/layers.4/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_1838 /model/layers.4/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.4/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.4/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.4/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.4/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.4/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.4/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.4/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.4/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.4/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.4/self_attn/Concat 4 1 /model/layers.4/self_attn/Unsqueeze_output_0 /model/layers.4/self_attn/Unsqueeze_1_output_0 /model/layers.4/self_attn/Constant_2_output_0 /model/layers.4/self_attn/Constant_3_output_0 /model/layers.4/self_attn/Concat_output_0 0=0 Concat /model/layers.4/self_attn/Concat_1 4 1 /model/layers.4/self_attn/Unsqueeze_2_output_0 /model/layers.4/self_attn/Unsqueeze_3_output_0 /model/layers.4/self_attn/Constant_4_output_0 /model/layers.4/self_attn/Constant_5_output_0 /model/layers.4/self_attn/Concat_1_output_0 0=0 Concat /model/layers.4/self_attn/Concat_2 4 1 /model/layers.4/self_attn/Unsqueeze_4_output_0 /model/layers.4/self_attn/Unsqueeze_5_output_0 /model/layers.4/self_attn/Constant_6_output_0 /model/layers.4/self_attn/Constant_7_output_0 /model/layers.4/self_attn/Concat_2_output_0 0=0 Concat /model/layers.4/self_attn/Concat_11 3 1 /model/layers.4/self_attn/Unsqueeze_29_output_0 /model/layers.4/self_attn/Unsqueeze_30_output_0 /model/layers.4/self_attn/Constant_64_output_0 /model/layers.4/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.4/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.4/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.4/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.4/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.4/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.4/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.4/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.4/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.4/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.4/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.4/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.4/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.4/self_attn/q_proj/Add 2 1 model.layers.4.self_attn.q_proj.bias /model/layers.4/self_attn/q_proj/MatMul_output_0 /model/layers.4/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.4/self_attn/k_proj/Add 2 1 model.layers.4.self_attn.k_proj.bias /model/layers.4/self_attn/k_proj/MatMul_output_0 /model/layers.4/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.4/self_attn/v_proj/Add 2 1 model.layers.4.self_attn.v_proj.bias /model/layers.4/self_attn/v_proj/MatMul_output_0 /model/layers.4/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.4/self_attn/Reshape 2 1 /model/layers.4/self_attn/q_proj/Add_output_0 /model/layers.4/self_attn/Concat_output_0 /model/layers.4/self_attn/Reshape_output_0 Reshape /model/layers.4/self_attn/Reshape_1 2 1 /model/layers.4/self_attn/k_proj/Add_output_0 /model/layers.4/self_attn/Concat_1_output_0 /model/layers.4/self_attn/Reshape_1_output_0 Reshape /model/layers.4/self_attn/Reshape_2 2 1 /model/layers.4/self_attn/v_proj/Add_output_0 /model/layers.4/self_attn/Concat_2_output_0 /model/layers.4/self_attn/Reshape_2_output_0 Permute /model/layers.4/self_attn/Transpose 1 1 /model/layers.4/self_attn/Reshape_output_0 /model/layers.4/self_attn/Transpose_output_0 0=2 Split splitncnn_150 1 4 /model/layers.4/self_attn/Transpose_output_0 /model/layers.4/self_attn/Transpose_output_0_splitncnn_0 /model/layers.4/self_attn/Transpose_output_0_splitncnn_1 /model/layers.4/self_attn/Transpose_output_0_splitncnn_2 /model/layers.4/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.4/self_attn/Transpose_1 1 1 /model/layers.4/self_attn/Reshape_1_output_0 /model/layers.4/self_attn/Transpose_1_output_0 0=2 Split splitncnn_151 1 4 /model/layers.4/self_attn/Transpose_1_output_0 /model/layers.4/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.4/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.4/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.4/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.4/self_attn/Transpose_2 1 1 /model/layers.4/self_attn/Reshape_2_output_0 /model/layers.4/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.4/self_attn/Mul 2 1 /model/layers.4/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_39 /model/layers.4/self_attn/Mul_output_0 0=2 Shape /model/layers.4/self_attn/Shape_2 1 1 /model/layers.4/self_attn/Transpose_output_0_splitncnn_2 /model/layers.4/self_attn/Shape_2_output_0 BinaryOp /model/layers.4/self_attn/Mul_2 2 1 /model/layers.4/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_38 /model/layers.4/self_attn/Mul_2_output_0 0=2 Shape /model/layers.4/self_attn/Shape_3 1 1 /model/layers.4/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.4/self_attn/Shape_3_output_0 Concat /model/layers.4/self_attn/Concat_6 2 1 past_key_values.4.value /model/layers.4/self_attn/Transpose_2_output_0 present.4.value 0=-2 Split splitncnn_152 1 5 present.4.value present.4.value_splitncnn_0 present.4.value_splitncnn_1 present.4.value_splitncnn_2 present.4.value_splitncnn_3 present.4.value_splitncnn_4 Crop /model/layers.4/self_attn/Gather_2 1 1 /model/layers.4/self_attn/Shape_2_output_0 /model/layers.4/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.4/self_attn/Gather_3 1 1 /model/layers.4/self_attn/Shape_3_output_0 /model/layers.4/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.4/self_attn/Shape_9 1 1 present.4.value_splitncnn_4 /model/layers.4/self_attn/Shape_9_output_0 Shape /model/layers.4/self_attn/Shape_10 1 1 present.4.value_splitncnn_3 /model/layers.4/self_attn/Shape_10_output_0 Shape /model/layers.4/self_attn/Shape_11 1 1 present.4.value_splitncnn_2 /model/layers.4/self_attn/Shape_11_output_0 Shape /model/layers.4/self_attn/Shape_12 1 1 present.4.value_splitncnn_1 /model/layers.4/self_attn/Shape_12_output_0 ExpandDims /model/layers.4/self_attn/Unsqueeze_19 2 1 present.4.value_splitncnn_0 /model/layers.4/self_attn/Constant_45_output_0 /model/layers.4/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.4/self_attn/Div 2 1 /model/layers.4/self_attn/Gather_2_output_0 /model/layers.4/self_attn/Constant_9_output_0 /model/layers.4/self_attn/Div_output_0 0=3 BinaryOp /model/layers.4/self_attn/Div_1 2 1 /model/layers.4/self_attn/Gather_3_output_0 /model/layers.4/self_attn/Constant_19_output_0 /model/layers.4/self_attn/Div_1_output_0 0=3 Crop /model/layers.4/self_attn/Gather_8 1 1 /model/layers.4/self_attn/Shape_9_output_0 /model/layers.4/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_153 1 2 /model/layers.4/self_attn/Gather_8_output_0 /model/layers.4/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.4/self_attn/Gather_9 1 1 /model/layers.4/self_attn/Shape_10_output_0 /model/layers.4/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_154 1 2 /model/layers.4/self_attn/Gather_9_output_0 /model/layers.4/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.4/self_attn/Gather_10 1 1 /model/layers.4/self_attn/Shape_11_output_0 /model/layers.4/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_155 1 2 /model/layers.4/self_attn/Gather_10_output_0 /model/layers.4/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.4/self_attn/Gather_11 1 1 /model/layers.4/self_attn/Shape_12_output_0 /model/layers.4/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_156 1 2 /model/layers.4/self_attn/Gather_11_output_0 /model/layers.4/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.4/self_attn/Cast 1 1 /model/layers.4/self_attn/Div_output_0 /model/layers.4/self_attn/Cast_output_0 Cast /model/layers.4/self_attn/Cast_2 1 1 /model/layers.4/self_attn/Div_1_output_0 /model/layers.4/self_attn/Cast_2_output_0 ExpandDims /model/layers.4/self_attn/Unsqueeze_20 2 1 /model/layers.4/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_46_output_0 /model/layers.4/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_24 2 1 /model/layers.4/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_1796 /model/layers.4/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_21 2 1 /model/layers.4/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_47_output_0 /model/layers.4/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.4/self_attn/Mul_7 2 1 /model/layers.4/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.4/self_attn/Constant_53_output_0 /model/layers.4/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.4/self_attn/Unsqueeze_22 2 1 /model/layers.4/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_49_output_0 /model/layers.4/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_26 2 1 /model/layers.4/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_1800 /model/layers.4/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_23 2 1 /model/layers.4/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_50_output_0 /model/layers.4/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_27 2 1 /model/layers.4/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_1802 /model/layers.4/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.4/self_attn/Cast_1 1 1 /model/layers.4/self_attn/Cast_output_0 /model/layers.4/self_attn/Cast_1_output_0 Split splitncnn_157 1 2 /model/layers.4/self_attn/Cast_1_output_0 /model/layers.4/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.4/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.4/self_attn/Cast_3 1 1 /model/layers.4/self_attn/Cast_2_output_0 /model/layers.4/self_attn/Cast_3_output_0 Split splitncnn_158 1 2 /model/layers.4/self_attn/Cast_3_output_0 /model/layers.4/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.4/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.4/self_attn/Unsqueeze_25 2 1 /model/layers.4/self_attn/Mul_7_output_0 onnx::Unsqueeze_1798 /model/layers.4/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.4/self_attn/Concat_9 5 1 /model/layers.4/self_attn/Unsqueeze_20_output_0 /model/layers.4/self_attn/Unsqueeze_21_output_0 /model/layers.4/self_attn/Constant_48_output_0 /model/layers.4/self_attn/Unsqueeze_22_output_0 /model/layers.4/self_attn/Unsqueeze_23_output_0 /model/layers.4/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_6 2 1 /model/layers.4/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_12_output_0 /model/layers.4/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_7 2 1 /model/layers.4/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.4/self_attn/Constant_15_output_0 /model/layers.4/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_8 2 1 /model/layers.4/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_22_output_0 /model/layers.4/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_9 2 1 /model/layers.4/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.4/self_attn/Constant_25_output_0 /model/layers.4/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.4/self_attn/Concat_10 4 1 /model/layers.4/self_attn/Unsqueeze_24_output_0 /model/layers.4/self_attn/Unsqueeze_25_output_0 /model/layers.4/self_attn/Unsqueeze_26_output_0 /model/layers.4/self_attn/Unsqueeze_27_output_0 /model/layers.4/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.4/self_attn/Reshape_5 1 1 /model/layers.4/self_attn/Concat_9_output_0 /model/layers.4/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_159 1 3 /model/layers.4/self_attn/Reshape_5_output_0 /model/layers.4/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.4/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.4/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.4/self_attn/Slice 2 1 /model/layers.4/self_attn/Transpose_output_0_splitncnn_1 /model/layers.4/self_attn/Unsqueeze_6_output_0 /model/layers.4/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.4/self_attn/Slice_1 2 1 /model/layers.4/self_attn/Transpose_output_0_splitncnn_0 /model/layers.4/self_attn/Unsqueeze_7_output_0 /model/layers.4/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.4/self_attn/Slice_2 2 1 /model/layers.4/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.4/self_attn/Unsqueeze_8_output_0 /model/layers.4/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.4/self_attn/Slice_3 2 1 /model/layers.4/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.4/self_attn/Unsqueeze_9_output_0 /model/layers.4/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.4/self_attn/Shape_13 1 1 /model/layers.4/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.4/self_attn/Shape_13_output_0 UnaryOp /model/layers.4/self_attn/Neg 1 1 /model/layers.4/self_attn/Slice_1_output_0 /model/layers.4/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.4/self_attn/Neg_1 1 1 /model/layers.4/self_attn/Slice_3_output_0 /model/layers.4/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.4/self_attn/ConstantOfShape_1 1 1 /model/layers.4/self_attn/Shape_13_output_0 /model/layers.4/self_attn/ConstantOfShape_1_output_0 Split splitncnn_160 1 2 /model/layers.4/self_attn/ConstantOfShape_1_output_0 /model/layers.4/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.4/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.4/self_attn/Concat_3 2 1 /model/layers.4/self_attn/Neg_output_0 /model/layers.4/self_attn/Slice_output_0 /model/layers.4/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.4/self_attn/Concat_4 2 1 /model/layers.4/self_attn/Neg_1_output_0 /model/layers.4/self_attn/Slice_2_output_0 /model/layers.4/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.4/self_attn/Mul_6 2 1 /model/layers.4/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_52_output_0 /model/layers.4/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.4/self_attn/Mul_1 2 1 /model/layers.4/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_39 /model/layers.4/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.4/self_attn/Mul_3 2 1 /model/layers.4/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_38 /model/layers.4/self_attn/Mul_3_output_0 0=2 Equal /model/layers.4/self_attn/Equal_1 2 1 /model/layers.4/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.4/self_attn/Mul_6_output_0 /model/layers.4/self_attn/Equal_1_output_0 BinaryOp /model/layers.4/self_attn/Add 2 1 /model/layers.4/self_attn/Mul_output_0 /model/layers.4/self_attn/Mul_1_output_0 /model/layers.4/self_attn/Add_output_0 0=0 BinaryOp /model/layers.4/self_attn/Add_1 2 1 /model/layers.4/self_attn/Mul_2_output_0 /model/layers.4/self_attn/Mul_3_output_0 /model/layers.4/self_attn/Add_1_output_0 0=0 Where /model/layers.4/self_attn/Where_1 3 1 /model/layers.4/self_attn/Equal_1_output_0 /model/layers.4/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.4/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.4/self_attn/Where_1_output_0 BinaryOp /model/layers.4/self_attn/Mul_8 1 1 /model/layers.4/self_attn/Add_output_0 /model/layers.4/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.4/self_attn/Concat_5 2 1 past_key_values.4.key /model/layers.4/self_attn/Add_1_output_0 present.4.key 0=-2 Split splitncnn_161 1 5 present.4.key present.4.key_splitncnn_0 present.4.key_splitncnn_1 present.4.key_splitncnn_2 present.4.key_splitncnn_3 present.4.key_splitncnn_4 Expand /model/layers.4/self_attn/Expand_1 2 1 /model/layers.4/self_attn/Unsqueeze_19_output_0 /model/layers.4/self_attn/Where_1_output_0 /model/layers.4/self_attn/Expand_1_output_0 Shape /model/layers.4/self_attn/Shape_4 1 1 present.4.key_splitncnn_4 /model/layers.4/self_attn/Shape_4_output_0 Shape /model/layers.4/self_attn/Shape_5 1 1 present.4.key_splitncnn_3 /model/layers.4/self_attn/Shape_5_output_0 Shape /model/layers.4/self_attn/Shape_6 1 1 present.4.key_splitncnn_2 /model/layers.4/self_attn/Shape_6_output_0 Shape /model/layers.4/self_attn/Shape_7 1 1 present.4.key_splitncnn_1 /model/layers.4/self_attn/Shape_7_output_0 ExpandDims /model/layers.4/self_attn/Unsqueeze_10 2 1 present.4.key_splitncnn_0 /model/layers.4/self_attn/Constant_32_output_0 /model/layers.4/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.4/self_attn/Reshape_6 2 1 /model/layers.4/self_attn/Expand_1_output_0 /model/layers.4/self_attn/Concat_10_output_0 /model/layers.4/self_attn/Reshape_6_output_0 Crop /model/layers.4/self_attn/Gather_4 1 1 /model/layers.4/self_attn/Shape_4_output_0 /model/layers.4/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_162 1 2 /model/layers.4/self_attn/Gather_4_output_0 /model/layers.4/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.4/self_attn/Gather_5 1 1 /model/layers.4/self_attn/Shape_5_output_0 /model/layers.4/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_163 1 2 /model/layers.4/self_attn/Gather_5_output_0 /model/layers.4/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.4/self_attn/Gather_6 1 1 /model/layers.4/self_attn/Shape_6_output_0 /model/layers.4/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_164 1 2 /model/layers.4/self_attn/Gather_6_output_0 /model/layers.4/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.4/self_attn/Gather_7 1 1 /model/layers.4/self_attn/Shape_7_output_0 /model/layers.4/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_165 1 2 /model/layers.4/self_attn/Gather_7_output_0 /model/layers.4/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.4/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.4/self_attn/Unsqueeze_11 2 1 /model/layers.4/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_33_output_0 /model/layers.4/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_15 2 1 /model/layers.4/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_1750 /model/layers.4/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_12 2 1 /model/layers.4/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_34_output_0 /model/layers.4/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.4/self_attn/Mul_5 2 1 /model/layers.4/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.4/self_attn/Constant_40_output_0 /model/layers.4/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.4/self_attn/Unsqueeze_13 2 1 /model/layers.4/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_36_output_0 /model/layers.4/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_17 2 1 /model/layers.4/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_1754 /model/layers.4/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_14 2 1 /model/layers.4/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_37_output_0 /model/layers.4/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_18 2 1 /model/layers.4/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_1756 /model/layers.4/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.4/self_attn/Unsqueeze_16 2 1 /model/layers.4/self_attn/Mul_5_output_0 onnx::Unsqueeze_1752 /model/layers.4/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.4/self_attn/Concat_7 5 1 /model/layers.4/self_attn/Unsqueeze_11_output_0 /model/layers.4/self_attn/Unsqueeze_12_output_0 /model/layers.4/self_attn/Constant_35_output_0 /model/layers.4/self_attn/Unsqueeze_13_output_0 /model/layers.4/self_attn/Unsqueeze_14_output_0 /model/layers.4/self_attn/Concat_7_output_0 0=0 Concat /model/layers.4/self_attn/Concat_8 4 1 /model/layers.4/self_attn/Unsqueeze_15_output_0 /model/layers.4/self_attn/Unsqueeze_16_output_0 /model/layers.4/self_attn/Unsqueeze_17_output_0 /model/layers.4/self_attn/Unsqueeze_18_output_0 /model/layers.4/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.4/self_attn/Reshape_3 1 1 /model/layers.4/self_attn/Concat_7_output_0 /model/layers.4/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_166 1 3 /model/layers.4/self_attn/Reshape_3_output_0 /model/layers.4/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.4/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.4/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.4/self_attn/Shape_8 1 1 /model/layers.4/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.4/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.4/self_attn/ConstantOfShape 1 1 /model/layers.4/self_attn/Shape_8_output_0 /model/layers.4/self_attn/ConstantOfShape_output_0 Split splitncnn_167 1 2 /model/layers.4/self_attn/ConstantOfShape_output_0 /model/layers.4/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.4/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.4/self_attn/Mul_4 2 1 /model/layers.4/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.4/self_attn/Constant_39_output_0 /model/layers.4/self_attn/Mul_4_output_0 0=2 Equal /model/layers.4/self_attn/Equal 2 1 /model/layers.4/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.4/self_attn/Mul_4_output_0 /model/layers.4/self_attn/Equal_output_0 Where /model/layers.4/self_attn/Where 3 1 /model/layers.4/self_attn/Equal_output_0 /model/layers.4/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.4/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.4/self_attn/Where_output_0 Expand /model/layers.4/self_attn/Expand 2 1 /model/layers.4/self_attn/Unsqueeze_10_output_0 /model/layers.4/self_attn/Where_output_0 /model/layers.4/self_attn/Expand_output_0 Reshape /model/layers.4/self_attn/Reshape_4 2 1 /model/layers.4/self_attn/Expand_output_0 /model/layers.4/self_attn/Concat_8_output_0 /model/layers.4/self_attn/Reshape_4_output_0 Split splitncnn_168 1 2 /model/layers.4/self_attn/Reshape_4_output_0 /model/layers.4/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.4/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.4/self_attn/Shape_14 1 1 /model/layers.4/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.4/self_attn/Shape_14_output_0 Permute /model/layers.4/self_attn/Transpose_3 1 1 /model/layers.4/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.4/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.4/self_attn/Gather_12 1 1 /model/layers.4/self_attn/Shape_14_output_0 /model/layers.4/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.4/self_attn/Mul_9 1 1 /model/layers.4/self_attn/Transpose_3_output_0 /model/layers.4/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.4/self_attn/Unsqueeze_28 2 1 /model/layers.4/self_attn/Gather_12_output_0 /model/layers.4/self_attn/Constant_57_output_0 /model/layers.4/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.4/self_attn/MatMul 2 1 /model/layers.4/self_attn/Mul_8_output_0 /model/layers.4/self_attn/Mul_9_output_0 /model/layers.4/self_attn/MatMul_output_0 Crop /model/layers.4/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_19 /model/layers.4/self_attn/Unsqueeze_28_output_0 /model/layers.4/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.4/self_attn/Where_2 3 1 /model/layers.4/self_attn/Slice_4_output_0 /model/layers.4/self_attn/Constant_61_output_0 /model/layers.4/self_attn/Constant_62_output_0 /model/layers.4/self_attn/Where_2_output_0 BinaryOp /model/layers.4/self_attn/Add_2 2 1 /model/layers.4/self_attn/MatMul_output_0 /model/layers.4/self_attn/Where_2_output_0 /model/layers.4/self_attn/Add_2_output_0 0=0 Softmax /model/layers.4/self_attn/Softmax 1 1 /model/layers.4/self_attn/Add_2_output_0 /model/layers.4/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_169 1 2 /model/layers.4/self_attn/Softmax_output_0 /model/layers.4/self_attn/Softmax_output_0_splitncnn_0 /model/layers.4/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.4/self_attn/IsNaN 1 1 /model/layers.4/self_attn/Softmax_output_0_splitncnn_1 /model/layers.4/self_attn/IsNaN_output_0 Where /model/layers.4/self_attn/Where_3 3 1 /model/layers.4/self_attn/IsNaN_output_0 /model/layers.4/self_attn/Constant_63_output_0 /model/layers.4/self_attn/Softmax_output_0_splitncnn_0 /model/layers.4/self_attn/Where_3_output_0 Gemm /model/layers.4/self_attn/MatMul_1 2 1 /model/layers.4/self_attn/Where_3_output_0 /model/layers.4/self_attn/Reshape_6_output_0 /model/layers.4/self_attn/MatMul_1_output_0 Permute /model/layers.4/self_attn/Transpose_4 1 1 /model/layers.4/self_attn/MatMul_1_output_0 /model/layers.4/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.4/self_attn/Reshape_7 2 1 /model/layers.4/self_attn/Transpose_4_output_0 /model/layers.4/self_attn/Concat_11_output_0 /model/layers.4/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.4/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.4/self_attn/Reshape_7_output_0 /model/layers.4/self_attn/Reshape_7_output_0_quantized /model/layers.4/self_attn/Reshape_7_output_0_scale /model/layers.4/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.4/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.4/self_attn/Reshape_7_output_0_scale /model/layers.4/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.522146e-03 MatMulInteger /model/layers.4/self_attn/o_proj/MatMul_quant 4 1 /model/layers.4/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7361_quantized /model/layers.4/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7361_zero_point /model/layers.4/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.4/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.4/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.4/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.4/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.4/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.4/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.4/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.4/Add 2 1 /model/layers.4/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.4/self_attn/o_proj/MatMul_output_0 /model/layers.4/Add_output_0 0=0 Cast /model/layers.4/post_attention_layernorm/Cast 1 1 /model/layers.4/Add_output_0 /model/layers.4/post_attention_layernorm/Cast_output_0 Split splitncnn_170 1 3 /model/layers.4/post_attention_layernorm/Cast_output_0 /model/layers.4/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.4/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.4/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.4/post_attention_layernorm/Pow 1 1 /model/layers.4/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.4/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.4/post_attention_layernorm/ReduceMean 2 1 /model/layers.4/post_attention_layernorm/Pow_output_0 /model/layers.4/post_attention_layernorm/Constant_1_output_0 /model/layers.4/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.4/post_attention_layernorm/Add 1 1 /model/layers.4/post_attention_layernorm/ReduceMean_output_0 /model/layers.4/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.4/post_attention_layernorm/Sqrt 1 1 /model/layers.4/post_attention_layernorm/Add_output_0 /model/layers.4/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.4/post_attention_layernorm/Div 1 1 /model/layers.4/post_attention_layernorm/Sqrt_output_0 /model/layers.4/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.4/post_attention_layernorm/Mul 2 1 /model/layers.4/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.4/post_attention_layernorm/Div_output_0 /model/layers.4/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.4/post_attention_layernorm/Cast_1 1 1 /model/layers.4/post_attention_layernorm/Mul_output_0 /model/layers.4/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.4/post_attention_layernorm/Mul_1 2 1 model.layers.4.post_attention_layernorm.weight /model/layers.4/post_attention_layernorm/Cast_1_output_0 /model/layers.4/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.4/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.4/post_attention_layernorm/Mul_1_output_0 /model/layers.4/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.4/post_attention_layernorm/Mul_1_output_0_scale /model/layers.4/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_171 1 2 /model/layers.4/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.4/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.4/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_172 1 2 /model/layers.4/post_attention_layernorm/Mul_1_output_0_scale /model/layers.4/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.4/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_173 1 2 /model/layers.4/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.4/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.4/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.4/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.4/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.4/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.152682e-03 BinaryOp /model/layers.4/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.4/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.4/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.183071e-03 MatMulInteger /model/layers.4/mlp/gate_proj/MatMul_quant 4 1 /model/layers.4/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7362_quantized /model/layers.4/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7362_zero_point /model/layers.4/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.4/mlp/up_proj/MatMul_quant 4 1 /model/layers.4/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7363_quantized /model/layers.4/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7363_zero_point /model/layers.4/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.4/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.4/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.4/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.4/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.4/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.4/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.4/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.4/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.4/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.4/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.4/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.4/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.4/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.4/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.4/mlp/act_fn/Mul 1 1 /model/layers.4/mlp/gate_proj/MatMul_output_0 /model/layers.4/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.4/mlp/Mul 2 1 /model/layers.4/mlp/act_fn/Mul_output_0 /model/layers.4/mlp/up_proj/MatMul_output_0 /model/layers.4/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.4/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.4/mlp/Mul_output_0 /model/layers.4/mlp/Mul_output_0_quantized /model/layers.4/mlp/Mul_output_0_scale /model/layers.4/mlp/Mul_output_0_zero_point BinaryOp /model/layers.4/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.4/mlp/Mul_output_0_scale /model/layers.4/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.706324e-03 MatMulInteger /model/layers.4/mlp/down_proj/MatMul_quant 4 1 /model/layers.4/mlp/Mul_output_0_quantized onnx::MatMul_7364_quantized /model/layers.4/mlp/Mul_output_0_zero_point onnx::MatMul_7364_zero_point /model/layers.4/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.4/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.4/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.4/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.4/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.4/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.4/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.4/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.4/Add_1 2 1 /model/layers.4/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.4/mlp/down_proj/MatMul_output_0 /model/layers.4/Add_1_output_0 0=0 Cast /model/layers.5/input_layernorm/Cast 1 1 /model/layers.4/Add_1_output_0 /model/layers.5/input_layernorm/Cast_output_0 Split splitncnn_174 1 3 /model/layers.5/input_layernorm/Cast_output_0 /model/layers.5/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.5/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.5/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.5/input_layernorm/Pow 1 1 /model/layers.5/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.5/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.5/input_layernorm/ReduceMean 2 1 /model/layers.5/input_layernorm/Pow_output_0 /model/layers.5/input_layernorm/Constant_1_output_0 /model/layers.5/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.5/input_layernorm/Add 1 1 /model/layers.5/input_layernorm/ReduceMean_output_0 /model/layers.5/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.5/input_layernorm/Sqrt 1 1 /model/layers.5/input_layernorm/Add_output_0 /model/layers.5/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.5/input_layernorm/Div 1 1 /model/layers.5/input_layernorm/Sqrt_output_0 /model/layers.5/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.5/input_layernorm/Mul 2 1 /model/layers.5/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.5/input_layernorm/Div_output_0 /model/layers.5/input_layernorm/Mul_output_0 0=2 Cast /model/layers.5/input_layernorm/Cast_1 1 1 /model/layers.5/input_layernorm/Mul_output_0 /model/layers.5/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.5/input_layernorm/Mul_1 2 1 model.layers.5.input_layernorm.weight /model/layers.5/input_layernorm/Cast_1_output_0 /model/layers.5/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_175 1 3 /model/layers.5/input_layernorm/Mul_1_output_0 /model/layers.5/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.5/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.5/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.5/self_attn/Shape 1 1 /model/layers.5/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.5/self_attn/Shape_output_0 Shape /model/layers.5/self_attn/Shape_1 1 1 /model/layers.5/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.5/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.5/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.5/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.5/input_layernorm/Mul_1_output_0_quantized /model/layers.5/input_layernorm/Mul_1_output_0_scale /model/layers.5/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_176 1 3 /model/layers.5/input_layernorm/Mul_1_output_0_quantized /model/layers.5/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.5/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.5/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_177 1 3 /model/layers.5/input_layernorm/Mul_1_output_0_scale /model/layers.5/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.5/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.5/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_178 1 3 /model/layers.5/input_layernorm/Mul_1_output_0_zero_point /model/layers.5/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.5/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.5/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.5/self_attn/Gather 1 1 /model/layers.5/self_attn/Shape_output_0 /model/layers.5/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_179 1 4 /model/layers.5/self_attn/Gather_output_0 /model/layers.5/self_attn/Gather_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_output_0_splitncnn_1 /model/layers.5/self_attn/Gather_output_0_splitncnn_2 /model/layers.5/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.5/self_attn/Gather_1 1 1 /model/layers.5/self_attn/Shape_1_output_0 /model/layers.5/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_180 1 4 /model/layers.5/self_attn/Gather_1_output_0 /model/layers.5/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.5/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.5/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.5/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.5/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.5/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.999262e-03 BinaryOp /model/layers.5/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.5/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.5/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.653236e-03 BinaryOp /model/layers.5/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.5/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.5/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=7.420337e-04 MatMulInteger /model/layers.5/self_attn/q_proj/MatMul_quant 4 1 /model/layers.5/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7365_quantized /model/layers.5/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7365_zero_point /model/layers.5/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.5/self_attn/k_proj/MatMul_quant 4 1 /model/layers.5/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7372_quantized /model/layers.5/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7372_zero_point /model/layers.5/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.5/self_attn/v_proj/MatMul_quant 4 1 /model/layers.5/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7373_quantized /model/layers.5/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7373_zero_point /model/layers.5/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.5/self_attn/Unsqueeze 2 1 /model/layers.5/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_1892 /model/layers.5/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_2 2 1 /model/layers.5/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_1901 /model/layers.5/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_4 2 1 /model/layers.5/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_1910 /model/layers.5/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_29 2 1 /model/layers.5/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_2115 /model/layers.5/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_1 2 1 /model/layers.5/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_1894 /model/layers.5/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_3 2 1 /model/layers.5/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_1903 /model/layers.5/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_5 2 1 /model/layers.5/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_1912 /model/layers.5/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_30 2 1 /model/layers.5/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_2117 /model/layers.5/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.5/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.5/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.5/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.5/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.5/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.5/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.5/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.5/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.5/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.5/self_attn/Concat 4 1 /model/layers.5/self_attn/Unsqueeze_output_0 /model/layers.5/self_attn/Unsqueeze_1_output_0 /model/layers.5/self_attn/Constant_2_output_0 /model/layers.5/self_attn/Constant_3_output_0 /model/layers.5/self_attn/Concat_output_0 0=0 Concat /model/layers.5/self_attn/Concat_1 4 1 /model/layers.5/self_attn/Unsqueeze_2_output_0 /model/layers.5/self_attn/Unsqueeze_3_output_0 /model/layers.5/self_attn/Constant_4_output_0 /model/layers.5/self_attn/Constant_5_output_0 /model/layers.5/self_attn/Concat_1_output_0 0=0 Concat /model/layers.5/self_attn/Concat_2 4 1 /model/layers.5/self_attn/Unsqueeze_4_output_0 /model/layers.5/self_attn/Unsqueeze_5_output_0 /model/layers.5/self_attn/Constant_6_output_0 /model/layers.5/self_attn/Constant_7_output_0 /model/layers.5/self_attn/Concat_2_output_0 0=0 Concat /model/layers.5/self_attn/Concat_11 3 1 /model/layers.5/self_attn/Unsqueeze_29_output_0 /model/layers.5/self_attn/Unsqueeze_30_output_0 /model/layers.5/self_attn/Constant_64_output_0 /model/layers.5/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.5/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.5/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.5/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.5/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.5/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.5/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.5/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.5/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.5/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.5/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.5/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.5/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.5/self_attn/q_proj/Add 2 1 model.layers.5.self_attn.q_proj.bias /model/layers.5/self_attn/q_proj/MatMul_output_0 /model/layers.5/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.5/self_attn/k_proj/Add 2 1 model.layers.5.self_attn.k_proj.bias /model/layers.5/self_attn/k_proj/MatMul_output_0 /model/layers.5/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.5/self_attn/v_proj/Add 2 1 model.layers.5.self_attn.v_proj.bias /model/layers.5/self_attn/v_proj/MatMul_output_0 /model/layers.5/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.5/self_attn/Reshape 2 1 /model/layers.5/self_attn/q_proj/Add_output_0 /model/layers.5/self_attn/Concat_output_0 /model/layers.5/self_attn/Reshape_output_0 Reshape /model/layers.5/self_attn/Reshape_1 2 1 /model/layers.5/self_attn/k_proj/Add_output_0 /model/layers.5/self_attn/Concat_1_output_0 /model/layers.5/self_attn/Reshape_1_output_0 Reshape /model/layers.5/self_attn/Reshape_2 2 1 /model/layers.5/self_attn/v_proj/Add_output_0 /model/layers.5/self_attn/Concat_2_output_0 /model/layers.5/self_attn/Reshape_2_output_0 Permute /model/layers.5/self_attn/Transpose 1 1 /model/layers.5/self_attn/Reshape_output_0 /model/layers.5/self_attn/Transpose_output_0 0=2 Split splitncnn_181 1 4 /model/layers.5/self_attn/Transpose_output_0 /model/layers.5/self_attn/Transpose_output_0_splitncnn_0 /model/layers.5/self_attn/Transpose_output_0_splitncnn_1 /model/layers.5/self_attn/Transpose_output_0_splitncnn_2 /model/layers.5/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.5/self_attn/Transpose_1 1 1 /model/layers.5/self_attn/Reshape_1_output_0 /model/layers.5/self_attn/Transpose_1_output_0 0=2 Split splitncnn_182 1 4 /model/layers.5/self_attn/Transpose_1_output_0 /model/layers.5/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.5/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.5/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.5/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.5/self_attn/Transpose_2 1 1 /model/layers.5/self_attn/Reshape_2_output_0 /model/layers.5/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.5/self_attn/Mul 2 1 /model/layers.5/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_37 /model/layers.5/self_attn/Mul_output_0 0=2 Shape /model/layers.5/self_attn/Shape_2 1 1 /model/layers.5/self_attn/Transpose_output_0_splitncnn_2 /model/layers.5/self_attn/Shape_2_output_0 BinaryOp /model/layers.5/self_attn/Mul_2 2 1 /model/layers.5/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_36 /model/layers.5/self_attn/Mul_2_output_0 0=2 Shape /model/layers.5/self_attn/Shape_3 1 1 /model/layers.5/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.5/self_attn/Shape_3_output_0 Concat /model/layers.5/self_attn/Concat_6 2 1 past_key_values.5.value /model/layers.5/self_attn/Transpose_2_output_0 present.5.value 0=-2 Split splitncnn_183 1 5 present.5.value present.5.value_splitncnn_0 present.5.value_splitncnn_1 present.5.value_splitncnn_2 present.5.value_splitncnn_3 present.5.value_splitncnn_4 Crop /model/layers.5/self_attn/Gather_2 1 1 /model/layers.5/self_attn/Shape_2_output_0 /model/layers.5/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.5/self_attn/Gather_3 1 1 /model/layers.5/self_attn/Shape_3_output_0 /model/layers.5/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.5/self_attn/Shape_9 1 1 present.5.value_splitncnn_4 /model/layers.5/self_attn/Shape_9_output_0 Shape /model/layers.5/self_attn/Shape_10 1 1 present.5.value_splitncnn_3 /model/layers.5/self_attn/Shape_10_output_0 Shape /model/layers.5/self_attn/Shape_11 1 1 present.5.value_splitncnn_2 /model/layers.5/self_attn/Shape_11_output_0 Shape /model/layers.5/self_attn/Shape_12 1 1 present.5.value_splitncnn_1 /model/layers.5/self_attn/Shape_12_output_0 ExpandDims /model/layers.5/self_attn/Unsqueeze_19 2 1 present.5.value_splitncnn_0 /model/layers.5/self_attn/Constant_45_output_0 /model/layers.5/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.5/self_attn/Div 2 1 /model/layers.5/self_attn/Gather_2_output_0 /model/layers.5/self_attn/Constant_9_output_0 /model/layers.5/self_attn/Div_output_0 0=3 BinaryOp /model/layers.5/self_attn/Div_1 2 1 /model/layers.5/self_attn/Gather_3_output_0 /model/layers.5/self_attn/Constant_19_output_0 /model/layers.5/self_attn/Div_1_output_0 0=3 Crop /model/layers.5/self_attn/Gather_8 1 1 /model/layers.5/self_attn/Shape_9_output_0 /model/layers.5/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_184 1 2 /model/layers.5/self_attn/Gather_8_output_0 /model/layers.5/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.5/self_attn/Gather_9 1 1 /model/layers.5/self_attn/Shape_10_output_0 /model/layers.5/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_185 1 2 /model/layers.5/self_attn/Gather_9_output_0 /model/layers.5/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.5/self_attn/Gather_10 1 1 /model/layers.5/self_attn/Shape_11_output_0 /model/layers.5/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_186 1 2 /model/layers.5/self_attn/Gather_10_output_0 /model/layers.5/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.5/self_attn/Gather_11 1 1 /model/layers.5/self_attn/Shape_12_output_0 /model/layers.5/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_187 1 2 /model/layers.5/self_attn/Gather_11_output_0 /model/layers.5/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.5/self_attn/Cast 1 1 /model/layers.5/self_attn/Div_output_0 /model/layers.5/self_attn/Cast_output_0 Cast /model/layers.5/self_attn/Cast_2 1 1 /model/layers.5/self_attn/Div_1_output_0 /model/layers.5/self_attn/Cast_2_output_0 ExpandDims /model/layers.5/self_attn/Unsqueeze_20 2 1 /model/layers.5/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_46_output_0 /model/layers.5/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_24 2 1 /model/layers.5/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_2075 /model/layers.5/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_21 2 1 /model/layers.5/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_47_output_0 /model/layers.5/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.5/self_attn/Mul_7 2 1 /model/layers.5/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.5/self_attn/Constant_53_output_0 /model/layers.5/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.5/self_attn/Unsqueeze_22 2 1 /model/layers.5/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_49_output_0 /model/layers.5/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_26 2 1 /model/layers.5/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_2079 /model/layers.5/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_23 2 1 /model/layers.5/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_50_output_0 /model/layers.5/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_27 2 1 /model/layers.5/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_2081 /model/layers.5/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.5/self_attn/Cast_1 1 1 /model/layers.5/self_attn/Cast_output_0 /model/layers.5/self_attn/Cast_1_output_0 Split splitncnn_188 1 2 /model/layers.5/self_attn/Cast_1_output_0 /model/layers.5/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.5/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.5/self_attn/Cast_3 1 1 /model/layers.5/self_attn/Cast_2_output_0 /model/layers.5/self_attn/Cast_3_output_0 Split splitncnn_189 1 2 /model/layers.5/self_attn/Cast_3_output_0 /model/layers.5/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.5/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.5/self_attn/Unsqueeze_25 2 1 /model/layers.5/self_attn/Mul_7_output_0 onnx::Unsqueeze_2077 /model/layers.5/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.5/self_attn/Concat_9 5 1 /model/layers.5/self_attn/Unsqueeze_20_output_0 /model/layers.5/self_attn/Unsqueeze_21_output_0 /model/layers.5/self_attn/Constant_48_output_0 /model/layers.5/self_attn/Unsqueeze_22_output_0 /model/layers.5/self_attn/Unsqueeze_23_output_0 /model/layers.5/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_6 2 1 /model/layers.5/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_12_output_0 /model/layers.5/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_7 2 1 /model/layers.5/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.5/self_attn/Constant_15_output_0 /model/layers.5/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_8 2 1 /model/layers.5/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_22_output_0 /model/layers.5/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_9 2 1 /model/layers.5/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.5/self_attn/Constant_25_output_0 /model/layers.5/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.5/self_attn/Concat_10 4 1 /model/layers.5/self_attn/Unsqueeze_24_output_0 /model/layers.5/self_attn/Unsqueeze_25_output_0 /model/layers.5/self_attn/Unsqueeze_26_output_0 /model/layers.5/self_attn/Unsqueeze_27_output_0 /model/layers.5/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.5/self_attn/Reshape_5 1 1 /model/layers.5/self_attn/Concat_9_output_0 /model/layers.5/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_190 1 3 /model/layers.5/self_attn/Reshape_5_output_0 /model/layers.5/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.5/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.5/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.5/self_attn/Slice 2 1 /model/layers.5/self_attn/Transpose_output_0_splitncnn_1 /model/layers.5/self_attn/Unsqueeze_6_output_0 /model/layers.5/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.5/self_attn/Slice_1 2 1 /model/layers.5/self_attn/Transpose_output_0_splitncnn_0 /model/layers.5/self_attn/Unsqueeze_7_output_0 /model/layers.5/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.5/self_attn/Slice_2 2 1 /model/layers.5/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.5/self_attn/Unsqueeze_8_output_0 /model/layers.5/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.5/self_attn/Slice_3 2 1 /model/layers.5/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.5/self_attn/Unsqueeze_9_output_0 /model/layers.5/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.5/self_attn/Shape_13 1 1 /model/layers.5/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.5/self_attn/Shape_13_output_0 UnaryOp /model/layers.5/self_attn/Neg 1 1 /model/layers.5/self_attn/Slice_1_output_0 /model/layers.5/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.5/self_attn/Neg_1 1 1 /model/layers.5/self_attn/Slice_3_output_0 /model/layers.5/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.5/self_attn/ConstantOfShape_1 1 1 /model/layers.5/self_attn/Shape_13_output_0 /model/layers.5/self_attn/ConstantOfShape_1_output_0 Split splitncnn_191 1 2 /model/layers.5/self_attn/ConstantOfShape_1_output_0 /model/layers.5/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.5/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.5/self_attn/Concat_3 2 1 /model/layers.5/self_attn/Neg_output_0 /model/layers.5/self_attn/Slice_output_0 /model/layers.5/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.5/self_attn/Concat_4 2 1 /model/layers.5/self_attn/Neg_1_output_0 /model/layers.5/self_attn/Slice_2_output_0 /model/layers.5/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.5/self_attn/Mul_6 2 1 /model/layers.5/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_52_output_0 /model/layers.5/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.5/self_attn/Mul_1 2 1 /model/layers.5/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_37 /model/layers.5/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.5/self_attn/Mul_3 2 1 /model/layers.5/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_36 /model/layers.5/self_attn/Mul_3_output_0 0=2 Equal /model/layers.5/self_attn/Equal_1 2 1 /model/layers.5/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.5/self_attn/Mul_6_output_0 /model/layers.5/self_attn/Equal_1_output_0 BinaryOp /model/layers.5/self_attn/Add 2 1 /model/layers.5/self_attn/Mul_output_0 /model/layers.5/self_attn/Mul_1_output_0 /model/layers.5/self_attn/Add_output_0 0=0 BinaryOp /model/layers.5/self_attn/Add_1 2 1 /model/layers.5/self_attn/Mul_2_output_0 /model/layers.5/self_attn/Mul_3_output_0 /model/layers.5/self_attn/Add_1_output_0 0=0 Where /model/layers.5/self_attn/Where_1 3 1 /model/layers.5/self_attn/Equal_1_output_0 /model/layers.5/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.5/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.5/self_attn/Where_1_output_0 BinaryOp /model/layers.5/self_attn/Mul_8 1 1 /model/layers.5/self_attn/Add_output_0 /model/layers.5/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.5/self_attn/Concat_5 2 1 past_key_values.5.key /model/layers.5/self_attn/Add_1_output_0 present.5.key 0=-2 Split splitncnn_192 1 5 present.5.key present.5.key_splitncnn_0 present.5.key_splitncnn_1 present.5.key_splitncnn_2 present.5.key_splitncnn_3 present.5.key_splitncnn_4 Expand /model/layers.5/self_attn/Expand_1 2 1 /model/layers.5/self_attn/Unsqueeze_19_output_0 /model/layers.5/self_attn/Where_1_output_0 /model/layers.5/self_attn/Expand_1_output_0 Shape /model/layers.5/self_attn/Shape_4 1 1 present.5.key_splitncnn_4 /model/layers.5/self_attn/Shape_4_output_0 Shape /model/layers.5/self_attn/Shape_5 1 1 present.5.key_splitncnn_3 /model/layers.5/self_attn/Shape_5_output_0 Shape /model/layers.5/self_attn/Shape_6 1 1 present.5.key_splitncnn_2 /model/layers.5/self_attn/Shape_6_output_0 Shape /model/layers.5/self_attn/Shape_7 1 1 present.5.key_splitncnn_1 /model/layers.5/self_attn/Shape_7_output_0 ExpandDims /model/layers.5/self_attn/Unsqueeze_10 2 1 present.5.key_splitncnn_0 /model/layers.5/self_attn/Constant_32_output_0 /model/layers.5/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.5/self_attn/Reshape_6 2 1 /model/layers.5/self_attn/Expand_1_output_0 /model/layers.5/self_attn/Concat_10_output_0 /model/layers.5/self_attn/Reshape_6_output_0 Crop /model/layers.5/self_attn/Gather_4 1 1 /model/layers.5/self_attn/Shape_4_output_0 /model/layers.5/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_193 1 2 /model/layers.5/self_attn/Gather_4_output_0 /model/layers.5/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.5/self_attn/Gather_5 1 1 /model/layers.5/self_attn/Shape_5_output_0 /model/layers.5/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_194 1 2 /model/layers.5/self_attn/Gather_5_output_0 /model/layers.5/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.5/self_attn/Gather_6 1 1 /model/layers.5/self_attn/Shape_6_output_0 /model/layers.5/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_195 1 2 /model/layers.5/self_attn/Gather_6_output_0 /model/layers.5/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.5/self_attn/Gather_7 1 1 /model/layers.5/self_attn/Shape_7_output_0 /model/layers.5/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_196 1 2 /model/layers.5/self_attn/Gather_7_output_0 /model/layers.5/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.5/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.5/self_attn/Unsqueeze_11 2 1 /model/layers.5/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_33_output_0 /model/layers.5/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_15 2 1 /model/layers.5/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_2029 /model/layers.5/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_12 2 1 /model/layers.5/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_34_output_0 /model/layers.5/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.5/self_attn/Mul_5 2 1 /model/layers.5/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.5/self_attn/Constant_40_output_0 /model/layers.5/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.5/self_attn/Unsqueeze_13 2 1 /model/layers.5/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_36_output_0 /model/layers.5/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_17 2 1 /model/layers.5/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_2033 /model/layers.5/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_14 2 1 /model/layers.5/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_37_output_0 /model/layers.5/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_18 2 1 /model/layers.5/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_2035 /model/layers.5/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.5/self_attn/Unsqueeze_16 2 1 /model/layers.5/self_attn/Mul_5_output_0 onnx::Unsqueeze_2031 /model/layers.5/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.5/self_attn/Concat_7 5 1 /model/layers.5/self_attn/Unsqueeze_11_output_0 /model/layers.5/self_attn/Unsqueeze_12_output_0 /model/layers.5/self_attn/Constant_35_output_0 /model/layers.5/self_attn/Unsqueeze_13_output_0 /model/layers.5/self_attn/Unsqueeze_14_output_0 /model/layers.5/self_attn/Concat_7_output_0 0=0 Concat /model/layers.5/self_attn/Concat_8 4 1 /model/layers.5/self_attn/Unsqueeze_15_output_0 /model/layers.5/self_attn/Unsqueeze_16_output_0 /model/layers.5/self_attn/Unsqueeze_17_output_0 /model/layers.5/self_attn/Unsqueeze_18_output_0 /model/layers.5/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.5/self_attn/Reshape_3 1 1 /model/layers.5/self_attn/Concat_7_output_0 /model/layers.5/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_197 1 3 /model/layers.5/self_attn/Reshape_3_output_0 /model/layers.5/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.5/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.5/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.5/self_attn/Shape_8 1 1 /model/layers.5/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.5/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.5/self_attn/ConstantOfShape 1 1 /model/layers.5/self_attn/Shape_8_output_0 /model/layers.5/self_attn/ConstantOfShape_output_0 Split splitncnn_198 1 2 /model/layers.5/self_attn/ConstantOfShape_output_0 /model/layers.5/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.5/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.5/self_attn/Mul_4 2 1 /model/layers.5/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.5/self_attn/Constant_39_output_0 /model/layers.5/self_attn/Mul_4_output_0 0=2 Equal /model/layers.5/self_attn/Equal 2 1 /model/layers.5/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.5/self_attn/Mul_4_output_0 /model/layers.5/self_attn/Equal_output_0 Where /model/layers.5/self_attn/Where 3 1 /model/layers.5/self_attn/Equal_output_0 /model/layers.5/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.5/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.5/self_attn/Where_output_0 Expand /model/layers.5/self_attn/Expand 2 1 /model/layers.5/self_attn/Unsqueeze_10_output_0 /model/layers.5/self_attn/Where_output_0 /model/layers.5/self_attn/Expand_output_0 Reshape /model/layers.5/self_attn/Reshape_4 2 1 /model/layers.5/self_attn/Expand_output_0 /model/layers.5/self_attn/Concat_8_output_0 /model/layers.5/self_attn/Reshape_4_output_0 Split splitncnn_199 1 2 /model/layers.5/self_attn/Reshape_4_output_0 /model/layers.5/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.5/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.5/self_attn/Shape_14 1 1 /model/layers.5/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.5/self_attn/Shape_14_output_0 Permute /model/layers.5/self_attn/Transpose_3 1 1 /model/layers.5/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.5/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.5/self_attn/Gather_12 1 1 /model/layers.5/self_attn/Shape_14_output_0 /model/layers.5/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.5/self_attn/Mul_9 1 1 /model/layers.5/self_attn/Transpose_3_output_0 /model/layers.5/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.5/self_attn/Unsqueeze_28 2 1 /model/layers.5/self_attn/Gather_12_output_0 /model/layers.5/self_attn/Constant_57_output_0 /model/layers.5/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.5/self_attn/MatMul 2 1 /model/layers.5/self_attn/Mul_8_output_0 /model/layers.5/self_attn/Mul_9_output_0 /model/layers.5/self_attn/MatMul_output_0 Crop /model/layers.5/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_18 /model/layers.5/self_attn/Unsqueeze_28_output_0 /model/layers.5/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.5/self_attn/Where_2 3 1 /model/layers.5/self_attn/Slice_4_output_0 /model/layers.5/self_attn/Constant_61_output_0 /model/layers.5/self_attn/Constant_62_output_0 /model/layers.5/self_attn/Where_2_output_0 BinaryOp /model/layers.5/self_attn/Add_2 2 1 /model/layers.5/self_attn/MatMul_output_0 /model/layers.5/self_attn/Where_2_output_0 /model/layers.5/self_attn/Add_2_output_0 0=0 Softmax /model/layers.5/self_attn/Softmax 1 1 /model/layers.5/self_attn/Add_2_output_0 /model/layers.5/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_200 1 2 /model/layers.5/self_attn/Softmax_output_0 /model/layers.5/self_attn/Softmax_output_0_splitncnn_0 /model/layers.5/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.5/self_attn/IsNaN 1 1 /model/layers.5/self_attn/Softmax_output_0_splitncnn_1 /model/layers.5/self_attn/IsNaN_output_0 Where /model/layers.5/self_attn/Where_3 3 1 /model/layers.5/self_attn/IsNaN_output_0 /model/layers.5/self_attn/Constant_63_output_0 /model/layers.5/self_attn/Softmax_output_0_splitncnn_0 /model/layers.5/self_attn/Where_3_output_0 Gemm /model/layers.5/self_attn/MatMul_1 2 1 /model/layers.5/self_attn/Where_3_output_0 /model/layers.5/self_attn/Reshape_6_output_0 /model/layers.5/self_attn/MatMul_1_output_0 Permute /model/layers.5/self_attn/Transpose_4 1 1 /model/layers.5/self_attn/MatMul_1_output_0 /model/layers.5/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.5/self_attn/Reshape_7 2 1 /model/layers.5/self_attn/Transpose_4_output_0 /model/layers.5/self_attn/Concat_11_output_0 /model/layers.5/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.5/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.5/self_attn/Reshape_7_output_0 /model/layers.5/self_attn/Reshape_7_output_0_quantized /model/layers.5/self_attn/Reshape_7_output_0_scale /model/layers.5/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.5/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.5/self_attn/Reshape_7_output_0_scale /model/layers.5/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.368356e-03 MatMulInteger /model/layers.5/self_attn/o_proj/MatMul_quant 4 1 /model/layers.5/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7396_quantized /model/layers.5/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7396_zero_point /model/layers.5/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.5/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.5/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.5/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.5/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.5/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.5/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.5/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.5/Add 2 1 /model/layers.5/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.5/self_attn/o_proj/MatMul_output_0 /model/layers.5/Add_output_0 0=0 Cast /model/layers.5/post_attention_layernorm/Cast 1 1 /model/layers.5/Add_output_0 /model/layers.5/post_attention_layernorm/Cast_output_0 Split splitncnn_201 1 3 /model/layers.5/post_attention_layernorm/Cast_output_0 /model/layers.5/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.5/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.5/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.5/post_attention_layernorm/Pow 1 1 /model/layers.5/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.5/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.5/post_attention_layernorm/ReduceMean 2 1 /model/layers.5/post_attention_layernorm/Pow_output_0 /model/layers.5/post_attention_layernorm/Constant_1_output_0 /model/layers.5/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.5/post_attention_layernorm/Add 1 1 /model/layers.5/post_attention_layernorm/ReduceMean_output_0 /model/layers.5/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.5/post_attention_layernorm/Sqrt 1 1 /model/layers.5/post_attention_layernorm/Add_output_0 /model/layers.5/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.5/post_attention_layernorm/Div 1 1 /model/layers.5/post_attention_layernorm/Sqrt_output_0 /model/layers.5/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.5/post_attention_layernorm/Mul 2 1 /model/layers.5/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.5/post_attention_layernorm/Div_output_0 /model/layers.5/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.5/post_attention_layernorm/Cast_1 1 1 /model/layers.5/post_attention_layernorm/Mul_output_0 /model/layers.5/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.5/post_attention_layernorm/Mul_1 2 1 model.layers.5.post_attention_layernorm.weight /model/layers.5/post_attention_layernorm/Cast_1_output_0 /model/layers.5/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.5/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.5/post_attention_layernorm/Mul_1_output_0 /model/layers.5/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.5/post_attention_layernorm/Mul_1_output_0_scale /model/layers.5/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_202 1 2 /model/layers.5/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.5/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.5/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_203 1 2 /model/layers.5/post_attention_layernorm/Mul_1_output_0_scale /model/layers.5/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.5/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_204 1 2 /model/layers.5/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.5/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.5/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.5/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.5/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.5/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.368356e-03 BinaryOp /model/layers.5/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.5/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.5/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=6.182333e-03 MatMulInteger /model/layers.5/mlp/gate_proj/MatMul_quant 4 1 /model/layers.5/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7397_quantized /model/layers.5/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7397_zero_point /model/layers.5/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.5/mlp/up_proj/MatMul_quant 4 1 /model/layers.5/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7398_quantized /model/layers.5/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7398_zero_point /model/layers.5/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.5/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.5/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.5/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.5/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.5/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.5/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.5/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.5/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.5/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.5/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.5/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.5/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.5/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.5/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.5/mlp/act_fn/Mul 1 1 /model/layers.5/mlp/gate_proj/MatMul_output_0 /model/layers.5/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.5/mlp/Mul 2 1 /model/layers.5/mlp/act_fn/Mul_output_0 /model/layers.5/mlp/up_proj/MatMul_output_0 /model/layers.5/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.5/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.5/mlp/Mul_output_0 /model/layers.5/mlp/Mul_output_0_quantized /model/layers.5/mlp/Mul_output_0_scale /model/layers.5/mlp/Mul_output_0_zero_point BinaryOp /model/layers.5/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.5/mlp/Mul_output_0_scale /model/layers.5/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.336860e-03 MatMulInteger /model/layers.5/mlp/down_proj/MatMul_quant 4 1 /model/layers.5/mlp/Mul_output_0_quantized onnx::MatMul_7399_quantized /model/layers.5/mlp/Mul_output_0_zero_point onnx::MatMul_7399_zero_point /model/layers.5/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.5/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.5/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.5/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.5/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.5/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.5/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.5/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.5/Add_1 2 1 /model/layers.5/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.5/mlp/down_proj/MatMul_output_0 /model/layers.5/Add_1_output_0 0=0 Cast /model/layers.6/input_layernorm/Cast 1 1 /model/layers.5/Add_1_output_0 /model/layers.6/input_layernorm/Cast_output_0 Split splitncnn_205 1 3 /model/layers.6/input_layernorm/Cast_output_0 /model/layers.6/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.6/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.6/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.6/input_layernorm/Pow 1 1 /model/layers.6/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.6/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.6/input_layernorm/ReduceMean 2 1 /model/layers.6/input_layernorm/Pow_output_0 /model/layers.6/input_layernorm/Constant_1_output_0 /model/layers.6/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.6/input_layernorm/Add 1 1 /model/layers.6/input_layernorm/ReduceMean_output_0 /model/layers.6/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.6/input_layernorm/Sqrt 1 1 /model/layers.6/input_layernorm/Add_output_0 /model/layers.6/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.6/input_layernorm/Div 1 1 /model/layers.6/input_layernorm/Sqrt_output_0 /model/layers.6/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.6/input_layernorm/Mul 2 1 /model/layers.6/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.6/input_layernorm/Div_output_0 /model/layers.6/input_layernorm/Mul_output_0 0=2 Cast /model/layers.6/input_layernorm/Cast_1 1 1 /model/layers.6/input_layernorm/Mul_output_0 /model/layers.6/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.6/input_layernorm/Mul_1 2 1 model.layers.6.input_layernorm.weight /model/layers.6/input_layernorm/Cast_1_output_0 /model/layers.6/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_206 1 3 /model/layers.6/input_layernorm/Mul_1_output_0 /model/layers.6/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.6/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.6/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.6/self_attn/Shape 1 1 /model/layers.6/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.6/self_attn/Shape_output_0 Shape /model/layers.6/self_attn/Shape_1 1 1 /model/layers.6/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.6/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.6/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.6/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.6/input_layernorm/Mul_1_output_0_quantized /model/layers.6/input_layernorm/Mul_1_output_0_scale /model/layers.6/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_207 1 3 /model/layers.6/input_layernorm/Mul_1_output_0_quantized /model/layers.6/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.6/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.6/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_208 1 3 /model/layers.6/input_layernorm/Mul_1_output_0_scale /model/layers.6/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.6/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.6/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_209 1 3 /model/layers.6/input_layernorm/Mul_1_output_0_zero_point /model/layers.6/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.6/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.6/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.6/self_attn/Gather 1 1 /model/layers.6/self_attn/Shape_output_0 /model/layers.6/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_210 1 4 /model/layers.6/self_attn/Gather_output_0 /model/layers.6/self_attn/Gather_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_output_0_splitncnn_1 /model/layers.6/self_attn/Gather_output_0_splitncnn_2 /model/layers.6/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.6/self_attn/Gather_1 1 1 /model/layers.6/self_attn/Shape_1_output_0 /model/layers.6/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_211 1 4 /model/layers.6/self_attn/Gather_1_output_0 /model/layers.6/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.6/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.6/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.6/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.6/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.6/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.014272e-03 BinaryOp /model/layers.6/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.6/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.6/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.514825e-03 BinaryOp /model/layers.6/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.6/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.6/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=6.843627e-04 MatMulInteger /model/layers.6/self_attn/q_proj/MatMul_quant 4 1 /model/layers.6/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7400_quantized /model/layers.6/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7400_zero_point /model/layers.6/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.6/self_attn/k_proj/MatMul_quant 4 1 /model/layers.6/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7407_quantized /model/layers.6/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7407_zero_point /model/layers.6/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.6/self_attn/v_proj/MatMul_quant 4 1 /model/layers.6/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7408_quantized /model/layers.6/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7408_zero_point /model/layers.6/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.6/self_attn/Unsqueeze 2 1 /model/layers.6/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_2171 /model/layers.6/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_2 2 1 /model/layers.6/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_2180 /model/layers.6/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_4 2 1 /model/layers.6/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_2189 /model/layers.6/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_29 2 1 /model/layers.6/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_2394 /model/layers.6/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_1 2 1 /model/layers.6/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_2173 /model/layers.6/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_3 2 1 /model/layers.6/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_2182 /model/layers.6/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_5 2 1 /model/layers.6/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_2191 /model/layers.6/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_30 2 1 /model/layers.6/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_2396 /model/layers.6/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.6/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.6/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.6/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.6/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.6/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.6/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.6/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.6/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.6/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.6/self_attn/Concat 4 1 /model/layers.6/self_attn/Unsqueeze_output_0 /model/layers.6/self_attn/Unsqueeze_1_output_0 /model/layers.6/self_attn/Constant_2_output_0 /model/layers.6/self_attn/Constant_3_output_0 /model/layers.6/self_attn/Concat_output_0 0=0 Concat /model/layers.6/self_attn/Concat_1 4 1 /model/layers.6/self_attn/Unsqueeze_2_output_0 /model/layers.6/self_attn/Unsqueeze_3_output_0 /model/layers.6/self_attn/Constant_4_output_0 /model/layers.6/self_attn/Constant_5_output_0 /model/layers.6/self_attn/Concat_1_output_0 0=0 Concat /model/layers.6/self_attn/Concat_2 4 1 /model/layers.6/self_attn/Unsqueeze_4_output_0 /model/layers.6/self_attn/Unsqueeze_5_output_0 /model/layers.6/self_attn/Constant_6_output_0 /model/layers.6/self_attn/Constant_7_output_0 /model/layers.6/self_attn/Concat_2_output_0 0=0 Concat /model/layers.6/self_attn/Concat_11 3 1 /model/layers.6/self_attn/Unsqueeze_29_output_0 /model/layers.6/self_attn/Unsqueeze_30_output_0 /model/layers.6/self_attn/Constant_64_output_0 /model/layers.6/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.6/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.6/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.6/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.6/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.6/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.6/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.6/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.6/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.6/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.6/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.6/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.6/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.6/self_attn/q_proj/Add 2 1 model.layers.6.self_attn.q_proj.bias /model/layers.6/self_attn/q_proj/MatMul_output_0 /model/layers.6/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.6/self_attn/k_proj/Add 2 1 model.layers.6.self_attn.k_proj.bias /model/layers.6/self_attn/k_proj/MatMul_output_0 /model/layers.6/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.6/self_attn/v_proj/Add 2 1 model.layers.6.self_attn.v_proj.bias /model/layers.6/self_attn/v_proj/MatMul_output_0 /model/layers.6/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.6/self_attn/Reshape 2 1 /model/layers.6/self_attn/q_proj/Add_output_0 /model/layers.6/self_attn/Concat_output_0 /model/layers.6/self_attn/Reshape_output_0 Reshape /model/layers.6/self_attn/Reshape_1 2 1 /model/layers.6/self_attn/k_proj/Add_output_0 /model/layers.6/self_attn/Concat_1_output_0 /model/layers.6/self_attn/Reshape_1_output_0 Reshape /model/layers.6/self_attn/Reshape_2 2 1 /model/layers.6/self_attn/v_proj/Add_output_0 /model/layers.6/self_attn/Concat_2_output_0 /model/layers.6/self_attn/Reshape_2_output_0 Permute /model/layers.6/self_attn/Transpose 1 1 /model/layers.6/self_attn/Reshape_output_0 /model/layers.6/self_attn/Transpose_output_0 0=2 Split splitncnn_212 1 4 /model/layers.6/self_attn/Transpose_output_0 /model/layers.6/self_attn/Transpose_output_0_splitncnn_0 /model/layers.6/self_attn/Transpose_output_0_splitncnn_1 /model/layers.6/self_attn/Transpose_output_0_splitncnn_2 /model/layers.6/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.6/self_attn/Transpose_1 1 1 /model/layers.6/self_attn/Reshape_1_output_0 /model/layers.6/self_attn/Transpose_1_output_0 0=2 Split splitncnn_213 1 4 /model/layers.6/self_attn/Transpose_1_output_0 /model/layers.6/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.6/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.6/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.6/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.6/self_attn/Transpose_2 1 1 /model/layers.6/self_attn/Reshape_2_output_0 /model/layers.6/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.6/self_attn/Mul 2 1 /model/layers.6/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_35 /model/layers.6/self_attn/Mul_output_0 0=2 Shape /model/layers.6/self_attn/Shape_2 1 1 /model/layers.6/self_attn/Transpose_output_0_splitncnn_2 /model/layers.6/self_attn/Shape_2_output_0 BinaryOp /model/layers.6/self_attn/Mul_2 2 1 /model/layers.6/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_34 /model/layers.6/self_attn/Mul_2_output_0 0=2 Shape /model/layers.6/self_attn/Shape_3 1 1 /model/layers.6/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.6/self_attn/Shape_3_output_0 Concat /model/layers.6/self_attn/Concat_6 2 1 past_key_values.6.value /model/layers.6/self_attn/Transpose_2_output_0 present.6.value 0=-2 Split splitncnn_214 1 5 present.6.value present.6.value_splitncnn_0 present.6.value_splitncnn_1 present.6.value_splitncnn_2 present.6.value_splitncnn_3 present.6.value_splitncnn_4 Crop /model/layers.6/self_attn/Gather_2 1 1 /model/layers.6/self_attn/Shape_2_output_0 /model/layers.6/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.6/self_attn/Gather_3 1 1 /model/layers.6/self_attn/Shape_3_output_0 /model/layers.6/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.6/self_attn/Shape_9 1 1 present.6.value_splitncnn_4 /model/layers.6/self_attn/Shape_9_output_0 Shape /model/layers.6/self_attn/Shape_10 1 1 present.6.value_splitncnn_3 /model/layers.6/self_attn/Shape_10_output_0 Shape /model/layers.6/self_attn/Shape_11 1 1 present.6.value_splitncnn_2 /model/layers.6/self_attn/Shape_11_output_0 Shape /model/layers.6/self_attn/Shape_12 1 1 present.6.value_splitncnn_1 /model/layers.6/self_attn/Shape_12_output_0 ExpandDims /model/layers.6/self_attn/Unsqueeze_19 2 1 present.6.value_splitncnn_0 /model/layers.6/self_attn/Constant_45_output_0 /model/layers.6/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.6/self_attn/Div 2 1 /model/layers.6/self_attn/Gather_2_output_0 /model/layers.6/self_attn/Constant_9_output_0 /model/layers.6/self_attn/Div_output_0 0=3 BinaryOp /model/layers.6/self_attn/Div_1 2 1 /model/layers.6/self_attn/Gather_3_output_0 /model/layers.6/self_attn/Constant_19_output_0 /model/layers.6/self_attn/Div_1_output_0 0=3 Crop /model/layers.6/self_attn/Gather_8 1 1 /model/layers.6/self_attn/Shape_9_output_0 /model/layers.6/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_215 1 2 /model/layers.6/self_attn/Gather_8_output_0 /model/layers.6/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.6/self_attn/Gather_9 1 1 /model/layers.6/self_attn/Shape_10_output_0 /model/layers.6/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_216 1 2 /model/layers.6/self_attn/Gather_9_output_0 /model/layers.6/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.6/self_attn/Gather_10 1 1 /model/layers.6/self_attn/Shape_11_output_0 /model/layers.6/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_217 1 2 /model/layers.6/self_attn/Gather_10_output_0 /model/layers.6/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.6/self_attn/Gather_11 1 1 /model/layers.6/self_attn/Shape_12_output_0 /model/layers.6/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_218 1 2 /model/layers.6/self_attn/Gather_11_output_0 /model/layers.6/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.6/self_attn/Cast 1 1 /model/layers.6/self_attn/Div_output_0 /model/layers.6/self_attn/Cast_output_0 Cast /model/layers.6/self_attn/Cast_2 1 1 /model/layers.6/self_attn/Div_1_output_0 /model/layers.6/self_attn/Cast_2_output_0 ExpandDims /model/layers.6/self_attn/Unsqueeze_20 2 1 /model/layers.6/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_46_output_0 /model/layers.6/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_24 2 1 /model/layers.6/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_2354 /model/layers.6/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_21 2 1 /model/layers.6/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_47_output_0 /model/layers.6/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.6/self_attn/Mul_7 2 1 /model/layers.6/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.6/self_attn/Constant_53_output_0 /model/layers.6/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.6/self_attn/Unsqueeze_22 2 1 /model/layers.6/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_49_output_0 /model/layers.6/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_26 2 1 /model/layers.6/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_2358 /model/layers.6/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_23 2 1 /model/layers.6/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_50_output_0 /model/layers.6/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_27 2 1 /model/layers.6/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_2360 /model/layers.6/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.6/self_attn/Cast_1 1 1 /model/layers.6/self_attn/Cast_output_0 /model/layers.6/self_attn/Cast_1_output_0 Split splitncnn_219 1 2 /model/layers.6/self_attn/Cast_1_output_0 /model/layers.6/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.6/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.6/self_attn/Cast_3 1 1 /model/layers.6/self_attn/Cast_2_output_0 /model/layers.6/self_attn/Cast_3_output_0 Split splitncnn_220 1 2 /model/layers.6/self_attn/Cast_3_output_0 /model/layers.6/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.6/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.6/self_attn/Unsqueeze_25 2 1 /model/layers.6/self_attn/Mul_7_output_0 onnx::Unsqueeze_2356 /model/layers.6/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.6/self_attn/Concat_9 5 1 /model/layers.6/self_attn/Unsqueeze_20_output_0 /model/layers.6/self_attn/Unsqueeze_21_output_0 /model/layers.6/self_attn/Constant_48_output_0 /model/layers.6/self_attn/Unsqueeze_22_output_0 /model/layers.6/self_attn/Unsqueeze_23_output_0 /model/layers.6/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_6 2 1 /model/layers.6/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_12_output_0 /model/layers.6/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_7 2 1 /model/layers.6/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.6/self_attn/Constant_15_output_0 /model/layers.6/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_8 2 1 /model/layers.6/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_22_output_0 /model/layers.6/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_9 2 1 /model/layers.6/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.6/self_attn/Constant_25_output_0 /model/layers.6/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.6/self_attn/Concat_10 4 1 /model/layers.6/self_attn/Unsqueeze_24_output_0 /model/layers.6/self_attn/Unsqueeze_25_output_0 /model/layers.6/self_attn/Unsqueeze_26_output_0 /model/layers.6/self_attn/Unsqueeze_27_output_0 /model/layers.6/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.6/self_attn/Reshape_5 1 1 /model/layers.6/self_attn/Concat_9_output_0 /model/layers.6/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_221 1 3 /model/layers.6/self_attn/Reshape_5_output_0 /model/layers.6/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.6/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.6/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.6/self_attn/Slice 2 1 /model/layers.6/self_attn/Transpose_output_0_splitncnn_1 /model/layers.6/self_attn/Unsqueeze_6_output_0 /model/layers.6/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.6/self_attn/Slice_1 2 1 /model/layers.6/self_attn/Transpose_output_0_splitncnn_0 /model/layers.6/self_attn/Unsqueeze_7_output_0 /model/layers.6/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.6/self_attn/Slice_2 2 1 /model/layers.6/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.6/self_attn/Unsqueeze_8_output_0 /model/layers.6/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.6/self_attn/Slice_3 2 1 /model/layers.6/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.6/self_attn/Unsqueeze_9_output_0 /model/layers.6/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.6/self_attn/Shape_13 1 1 /model/layers.6/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.6/self_attn/Shape_13_output_0 UnaryOp /model/layers.6/self_attn/Neg 1 1 /model/layers.6/self_attn/Slice_1_output_0 /model/layers.6/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.6/self_attn/Neg_1 1 1 /model/layers.6/self_attn/Slice_3_output_0 /model/layers.6/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.6/self_attn/ConstantOfShape_1 1 1 /model/layers.6/self_attn/Shape_13_output_0 /model/layers.6/self_attn/ConstantOfShape_1_output_0 Split splitncnn_222 1 2 /model/layers.6/self_attn/ConstantOfShape_1_output_0 /model/layers.6/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.6/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.6/self_attn/Concat_3 2 1 /model/layers.6/self_attn/Neg_output_0 /model/layers.6/self_attn/Slice_output_0 /model/layers.6/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.6/self_attn/Concat_4 2 1 /model/layers.6/self_attn/Neg_1_output_0 /model/layers.6/self_attn/Slice_2_output_0 /model/layers.6/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.6/self_attn/Mul_6 2 1 /model/layers.6/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_52_output_0 /model/layers.6/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.6/self_attn/Mul_1 2 1 /model/layers.6/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_35 /model/layers.6/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.6/self_attn/Mul_3 2 1 /model/layers.6/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_34 /model/layers.6/self_attn/Mul_3_output_0 0=2 Equal /model/layers.6/self_attn/Equal_1 2 1 /model/layers.6/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.6/self_attn/Mul_6_output_0 /model/layers.6/self_attn/Equal_1_output_0 BinaryOp /model/layers.6/self_attn/Add 2 1 /model/layers.6/self_attn/Mul_output_0 /model/layers.6/self_attn/Mul_1_output_0 /model/layers.6/self_attn/Add_output_0 0=0 BinaryOp /model/layers.6/self_attn/Add_1 2 1 /model/layers.6/self_attn/Mul_2_output_0 /model/layers.6/self_attn/Mul_3_output_0 /model/layers.6/self_attn/Add_1_output_0 0=0 Where /model/layers.6/self_attn/Where_1 3 1 /model/layers.6/self_attn/Equal_1_output_0 /model/layers.6/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.6/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.6/self_attn/Where_1_output_0 BinaryOp /model/layers.6/self_attn/Mul_8 1 1 /model/layers.6/self_attn/Add_output_0 /model/layers.6/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.6/self_attn/Concat_5 2 1 past_key_values.6.key /model/layers.6/self_attn/Add_1_output_0 present.6.key 0=-2 Split splitncnn_223 1 5 present.6.key present.6.key_splitncnn_0 present.6.key_splitncnn_1 present.6.key_splitncnn_2 present.6.key_splitncnn_3 present.6.key_splitncnn_4 Expand /model/layers.6/self_attn/Expand_1 2 1 /model/layers.6/self_attn/Unsqueeze_19_output_0 /model/layers.6/self_attn/Where_1_output_0 /model/layers.6/self_attn/Expand_1_output_0 Shape /model/layers.6/self_attn/Shape_4 1 1 present.6.key_splitncnn_4 /model/layers.6/self_attn/Shape_4_output_0 Shape /model/layers.6/self_attn/Shape_5 1 1 present.6.key_splitncnn_3 /model/layers.6/self_attn/Shape_5_output_0 Shape /model/layers.6/self_attn/Shape_6 1 1 present.6.key_splitncnn_2 /model/layers.6/self_attn/Shape_6_output_0 Shape /model/layers.6/self_attn/Shape_7 1 1 present.6.key_splitncnn_1 /model/layers.6/self_attn/Shape_7_output_0 ExpandDims /model/layers.6/self_attn/Unsqueeze_10 2 1 present.6.key_splitncnn_0 /model/layers.6/self_attn/Constant_32_output_0 /model/layers.6/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.6/self_attn/Reshape_6 2 1 /model/layers.6/self_attn/Expand_1_output_0 /model/layers.6/self_attn/Concat_10_output_0 /model/layers.6/self_attn/Reshape_6_output_0 Crop /model/layers.6/self_attn/Gather_4 1 1 /model/layers.6/self_attn/Shape_4_output_0 /model/layers.6/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_224 1 2 /model/layers.6/self_attn/Gather_4_output_0 /model/layers.6/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.6/self_attn/Gather_5 1 1 /model/layers.6/self_attn/Shape_5_output_0 /model/layers.6/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_225 1 2 /model/layers.6/self_attn/Gather_5_output_0 /model/layers.6/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.6/self_attn/Gather_6 1 1 /model/layers.6/self_attn/Shape_6_output_0 /model/layers.6/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_226 1 2 /model/layers.6/self_attn/Gather_6_output_0 /model/layers.6/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.6/self_attn/Gather_7 1 1 /model/layers.6/self_attn/Shape_7_output_0 /model/layers.6/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_227 1 2 /model/layers.6/self_attn/Gather_7_output_0 /model/layers.6/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.6/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.6/self_attn/Unsqueeze_11 2 1 /model/layers.6/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_33_output_0 /model/layers.6/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_15 2 1 /model/layers.6/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_2308 /model/layers.6/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_12 2 1 /model/layers.6/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_34_output_0 /model/layers.6/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.6/self_attn/Mul_5 2 1 /model/layers.6/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.6/self_attn/Constant_40_output_0 /model/layers.6/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.6/self_attn/Unsqueeze_13 2 1 /model/layers.6/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_36_output_0 /model/layers.6/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_17 2 1 /model/layers.6/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_2312 /model/layers.6/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_14 2 1 /model/layers.6/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_37_output_0 /model/layers.6/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_18 2 1 /model/layers.6/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_2314 /model/layers.6/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.6/self_attn/Unsqueeze_16 2 1 /model/layers.6/self_attn/Mul_5_output_0 onnx::Unsqueeze_2310 /model/layers.6/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.6/self_attn/Concat_7 5 1 /model/layers.6/self_attn/Unsqueeze_11_output_0 /model/layers.6/self_attn/Unsqueeze_12_output_0 /model/layers.6/self_attn/Constant_35_output_0 /model/layers.6/self_attn/Unsqueeze_13_output_0 /model/layers.6/self_attn/Unsqueeze_14_output_0 /model/layers.6/self_attn/Concat_7_output_0 0=0 Concat /model/layers.6/self_attn/Concat_8 4 1 /model/layers.6/self_attn/Unsqueeze_15_output_0 /model/layers.6/self_attn/Unsqueeze_16_output_0 /model/layers.6/self_attn/Unsqueeze_17_output_0 /model/layers.6/self_attn/Unsqueeze_18_output_0 /model/layers.6/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.6/self_attn/Reshape_3 1 1 /model/layers.6/self_attn/Concat_7_output_0 /model/layers.6/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_228 1 3 /model/layers.6/self_attn/Reshape_3_output_0 /model/layers.6/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.6/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.6/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.6/self_attn/Shape_8 1 1 /model/layers.6/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.6/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.6/self_attn/ConstantOfShape 1 1 /model/layers.6/self_attn/Shape_8_output_0 /model/layers.6/self_attn/ConstantOfShape_output_0 Split splitncnn_229 1 2 /model/layers.6/self_attn/ConstantOfShape_output_0 /model/layers.6/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.6/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.6/self_attn/Mul_4 2 1 /model/layers.6/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.6/self_attn/Constant_39_output_0 /model/layers.6/self_attn/Mul_4_output_0 0=2 Equal /model/layers.6/self_attn/Equal 2 1 /model/layers.6/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.6/self_attn/Mul_4_output_0 /model/layers.6/self_attn/Equal_output_0 Where /model/layers.6/self_attn/Where 3 1 /model/layers.6/self_attn/Equal_output_0 /model/layers.6/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.6/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.6/self_attn/Where_output_0 Expand /model/layers.6/self_attn/Expand 2 1 /model/layers.6/self_attn/Unsqueeze_10_output_0 /model/layers.6/self_attn/Where_output_0 /model/layers.6/self_attn/Expand_output_0 Reshape /model/layers.6/self_attn/Reshape_4 2 1 /model/layers.6/self_attn/Expand_output_0 /model/layers.6/self_attn/Concat_8_output_0 /model/layers.6/self_attn/Reshape_4_output_0 Split splitncnn_230 1 2 /model/layers.6/self_attn/Reshape_4_output_0 /model/layers.6/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.6/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.6/self_attn/Shape_14 1 1 /model/layers.6/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.6/self_attn/Shape_14_output_0 Permute /model/layers.6/self_attn/Transpose_3 1 1 /model/layers.6/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.6/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.6/self_attn/Gather_12 1 1 /model/layers.6/self_attn/Shape_14_output_0 /model/layers.6/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.6/self_attn/Mul_9 1 1 /model/layers.6/self_attn/Transpose_3_output_0 /model/layers.6/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.6/self_attn/Unsqueeze_28 2 1 /model/layers.6/self_attn/Gather_12_output_0 /model/layers.6/self_attn/Constant_57_output_0 /model/layers.6/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.6/self_attn/MatMul 2 1 /model/layers.6/self_attn/Mul_8_output_0 /model/layers.6/self_attn/Mul_9_output_0 /model/layers.6/self_attn/MatMul_output_0 Crop /model/layers.6/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_17 /model/layers.6/self_attn/Unsqueeze_28_output_0 /model/layers.6/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.6/self_attn/Where_2 3 1 /model/layers.6/self_attn/Slice_4_output_0 /model/layers.6/self_attn/Constant_61_output_0 /model/layers.6/self_attn/Constant_62_output_0 /model/layers.6/self_attn/Where_2_output_0 BinaryOp /model/layers.6/self_attn/Add_2 2 1 /model/layers.6/self_attn/MatMul_output_0 /model/layers.6/self_attn/Where_2_output_0 /model/layers.6/self_attn/Add_2_output_0 0=0 Softmax /model/layers.6/self_attn/Softmax 1 1 /model/layers.6/self_attn/Add_2_output_0 /model/layers.6/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_231 1 2 /model/layers.6/self_attn/Softmax_output_0 /model/layers.6/self_attn/Softmax_output_0_splitncnn_0 /model/layers.6/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.6/self_attn/IsNaN 1 1 /model/layers.6/self_attn/Softmax_output_0_splitncnn_1 /model/layers.6/self_attn/IsNaN_output_0 Where /model/layers.6/self_attn/Where_3 3 1 /model/layers.6/self_attn/IsNaN_output_0 /model/layers.6/self_attn/Constant_63_output_0 /model/layers.6/self_attn/Softmax_output_0_splitncnn_0 /model/layers.6/self_attn/Where_3_output_0 Gemm /model/layers.6/self_attn/MatMul_1 2 1 /model/layers.6/self_attn/Where_3_output_0 /model/layers.6/self_attn/Reshape_6_output_0 /model/layers.6/self_attn/MatMul_1_output_0 Permute /model/layers.6/self_attn/Transpose_4 1 1 /model/layers.6/self_attn/MatMul_1_output_0 /model/layers.6/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.6/self_attn/Reshape_7 2 1 /model/layers.6/self_attn/Transpose_4_output_0 /model/layers.6/self_attn/Concat_11_output_0 /model/layers.6/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.6/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.6/self_attn/Reshape_7_output_0 /model/layers.6/self_attn/Reshape_7_output_0_quantized /model/layers.6/self_attn/Reshape_7_output_0_scale /model/layers.6/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.6/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.6/self_attn/Reshape_7_output_0_scale /model/layers.6/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.367987e-03 MatMulInteger /model/layers.6/self_attn/o_proj/MatMul_quant 4 1 /model/layers.6/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7431_quantized /model/layers.6/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7431_zero_point /model/layers.6/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.6/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.6/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.6/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.6/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.6/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.6/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.6/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.6/Add 2 1 /model/layers.6/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.6/self_attn/o_proj/MatMul_output_0 /model/layers.6/Add_output_0 0=0 Cast /model/layers.6/post_attention_layernorm/Cast 1 1 /model/layers.6/Add_output_0 /model/layers.6/post_attention_layernorm/Cast_output_0 Split splitncnn_232 1 3 /model/layers.6/post_attention_layernorm/Cast_output_0 /model/layers.6/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.6/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.6/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.6/post_attention_layernorm/Pow 1 1 /model/layers.6/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.6/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.6/post_attention_layernorm/ReduceMean 2 1 /model/layers.6/post_attention_layernorm/Pow_output_0 /model/layers.6/post_attention_layernorm/Constant_1_output_0 /model/layers.6/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.6/post_attention_layernorm/Add 1 1 /model/layers.6/post_attention_layernorm/ReduceMean_output_0 /model/layers.6/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.6/post_attention_layernorm/Sqrt 1 1 /model/layers.6/post_attention_layernorm/Add_output_0 /model/layers.6/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.6/post_attention_layernorm/Div 1 1 /model/layers.6/post_attention_layernorm/Sqrt_output_0 /model/layers.6/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.6/post_attention_layernorm/Mul 2 1 /model/layers.6/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.6/post_attention_layernorm/Div_output_0 /model/layers.6/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.6/post_attention_layernorm/Cast_1 1 1 /model/layers.6/post_attention_layernorm/Mul_output_0 /model/layers.6/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.6/post_attention_layernorm/Mul_1 2 1 model.layers.6.post_attention_layernorm.weight /model/layers.6/post_attention_layernorm/Cast_1_output_0 /model/layers.6/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.6/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.6/post_attention_layernorm/Mul_1_output_0 /model/layers.6/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.6/post_attention_layernorm/Mul_1_output_0_scale /model/layers.6/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_233 1 2 /model/layers.6/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.6/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.6/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_234 1 2 /model/layers.6/post_attention_layernorm/Mul_1_output_0_scale /model/layers.6/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.6/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_235 1 2 /model/layers.6/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.6/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.6/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.6/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.6/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.6/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.106545e-03 BinaryOp /model/layers.6/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.6/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.6/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.968135e-03 MatMulInteger /model/layers.6/mlp/gate_proj/MatMul_quant 4 1 /model/layers.6/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7432_quantized /model/layers.6/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7432_zero_point /model/layers.6/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.6/mlp/up_proj/MatMul_quant 4 1 /model/layers.6/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7433_quantized /model/layers.6/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7433_zero_point /model/layers.6/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.6/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.6/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.6/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.6/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.6/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.6/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.6/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.6/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.6/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.6/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.6/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.6/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.6/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.6/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.6/mlp/act_fn/Mul 1 1 /model/layers.6/mlp/gate_proj/MatMul_output_0 /model/layers.6/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.6/mlp/Mul 2 1 /model/layers.6/mlp/act_fn/Mul_output_0 /model/layers.6/mlp/up_proj/MatMul_output_0 /model/layers.6/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.6/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.6/mlp/Mul_output_0 /model/layers.6/mlp/Mul_output_0_quantized /model/layers.6/mlp/Mul_output_0_scale /model/layers.6/mlp/Mul_output_0_zero_point BinaryOp /model/layers.6/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.6/mlp/Mul_output_0_scale /model/layers.6/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.337229e-03 MatMulInteger /model/layers.6/mlp/down_proj/MatMul_quant 4 1 /model/layers.6/mlp/Mul_output_0_quantized onnx::MatMul_7434_quantized /model/layers.6/mlp/Mul_output_0_zero_point onnx::MatMul_7434_zero_point /model/layers.6/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.6/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.6/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.6/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.6/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.6/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.6/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.6/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.6/Add_1 2 1 /model/layers.6/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.6/mlp/down_proj/MatMul_output_0 /model/layers.6/Add_1_output_0 0=0 Cast /model/layers.7/input_layernorm/Cast 1 1 /model/layers.6/Add_1_output_0 /model/layers.7/input_layernorm/Cast_output_0 Split splitncnn_236 1 3 /model/layers.7/input_layernorm/Cast_output_0 /model/layers.7/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.7/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.7/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.7/input_layernorm/Pow 1 1 /model/layers.7/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.7/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.7/input_layernorm/ReduceMean 2 1 /model/layers.7/input_layernorm/Pow_output_0 /model/layers.7/input_layernorm/Constant_1_output_0 /model/layers.7/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.7/input_layernorm/Add 1 1 /model/layers.7/input_layernorm/ReduceMean_output_0 /model/layers.7/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.7/input_layernorm/Sqrt 1 1 /model/layers.7/input_layernorm/Add_output_0 /model/layers.7/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.7/input_layernorm/Div 1 1 /model/layers.7/input_layernorm/Sqrt_output_0 /model/layers.7/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.7/input_layernorm/Mul 2 1 /model/layers.7/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.7/input_layernorm/Div_output_0 /model/layers.7/input_layernorm/Mul_output_0 0=2 Cast /model/layers.7/input_layernorm/Cast_1 1 1 /model/layers.7/input_layernorm/Mul_output_0 /model/layers.7/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.7/input_layernorm/Mul_1 2 1 model.layers.7.input_layernorm.weight /model/layers.7/input_layernorm/Cast_1_output_0 /model/layers.7/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_237 1 3 /model/layers.7/input_layernorm/Mul_1_output_0 /model/layers.7/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.7/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.7/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.7/self_attn/Shape 1 1 /model/layers.7/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.7/self_attn/Shape_output_0 Shape /model/layers.7/self_attn/Shape_1 1 1 /model/layers.7/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.7/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.7/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.7/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.7/input_layernorm/Mul_1_output_0_quantized /model/layers.7/input_layernorm/Mul_1_output_0_scale /model/layers.7/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_238 1 3 /model/layers.7/input_layernorm/Mul_1_output_0_quantized /model/layers.7/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.7/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.7/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_239 1 3 /model/layers.7/input_layernorm/Mul_1_output_0_scale /model/layers.7/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.7/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.7/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_240 1 3 /model/layers.7/input_layernorm/Mul_1_output_0_zero_point /model/layers.7/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.7/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.7/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.7/self_attn/Gather 1 1 /model/layers.7/self_attn/Shape_output_0 /model/layers.7/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_241 1 4 /model/layers.7/self_attn/Gather_output_0 /model/layers.7/self_attn/Gather_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_output_0_splitncnn_1 /model/layers.7/self_attn/Gather_output_0_splitncnn_2 /model/layers.7/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.7/self_attn/Gather_1 1 1 /model/layers.7/self_attn/Shape_1_output_0 /model/layers.7/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_242 1 4 /model/layers.7/self_attn/Gather_1_output_0 /model/layers.7/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.7/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.7/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.7/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.7/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.7/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.014641e-03 BinaryOp /model/layers.7/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.7/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.7/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.945436e-03 BinaryOp /model/layers.7/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.7/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.7/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=8.112389e-04 MatMulInteger /model/layers.7/self_attn/q_proj/MatMul_quant 4 1 /model/layers.7/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7435_quantized /model/layers.7/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7435_zero_point /model/layers.7/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.7/self_attn/k_proj/MatMul_quant 4 1 /model/layers.7/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7442_quantized /model/layers.7/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7442_zero_point /model/layers.7/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.7/self_attn/v_proj/MatMul_quant 4 1 /model/layers.7/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7443_quantized /model/layers.7/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7443_zero_point /model/layers.7/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.7/self_attn/Unsqueeze 2 1 /model/layers.7/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_2450 /model/layers.7/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_2 2 1 /model/layers.7/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_2459 /model/layers.7/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_4 2 1 /model/layers.7/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_2468 /model/layers.7/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_29 2 1 /model/layers.7/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_2673 /model/layers.7/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_1 2 1 /model/layers.7/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_2452 /model/layers.7/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_3 2 1 /model/layers.7/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_2461 /model/layers.7/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_5 2 1 /model/layers.7/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_2470 /model/layers.7/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_30 2 1 /model/layers.7/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_2675 /model/layers.7/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.7/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.7/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.7/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.7/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.7/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.7/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.7/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.7/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.7/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.7/self_attn/Concat 4 1 /model/layers.7/self_attn/Unsqueeze_output_0 /model/layers.7/self_attn/Unsqueeze_1_output_0 /model/layers.7/self_attn/Constant_2_output_0 /model/layers.7/self_attn/Constant_3_output_0 /model/layers.7/self_attn/Concat_output_0 0=0 Concat /model/layers.7/self_attn/Concat_1 4 1 /model/layers.7/self_attn/Unsqueeze_2_output_0 /model/layers.7/self_attn/Unsqueeze_3_output_0 /model/layers.7/self_attn/Constant_4_output_0 /model/layers.7/self_attn/Constant_5_output_0 /model/layers.7/self_attn/Concat_1_output_0 0=0 Concat /model/layers.7/self_attn/Concat_2 4 1 /model/layers.7/self_attn/Unsqueeze_4_output_0 /model/layers.7/self_attn/Unsqueeze_5_output_0 /model/layers.7/self_attn/Constant_6_output_0 /model/layers.7/self_attn/Constant_7_output_0 /model/layers.7/self_attn/Concat_2_output_0 0=0 Concat /model/layers.7/self_attn/Concat_11 3 1 /model/layers.7/self_attn/Unsqueeze_29_output_0 /model/layers.7/self_attn/Unsqueeze_30_output_0 /model/layers.7/self_attn/Constant_64_output_0 /model/layers.7/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.7/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.7/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.7/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.7/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.7/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.7/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.7/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.7/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.7/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.7/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.7/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.7/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.7/self_attn/q_proj/Add 2 1 model.layers.7.self_attn.q_proj.bias /model/layers.7/self_attn/q_proj/MatMul_output_0 /model/layers.7/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.7/self_attn/k_proj/Add 2 1 model.layers.7.self_attn.k_proj.bias /model/layers.7/self_attn/k_proj/MatMul_output_0 /model/layers.7/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.7/self_attn/v_proj/Add 2 1 model.layers.7.self_attn.v_proj.bias /model/layers.7/self_attn/v_proj/MatMul_output_0 /model/layers.7/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.7/self_attn/Reshape 2 1 /model/layers.7/self_attn/q_proj/Add_output_0 /model/layers.7/self_attn/Concat_output_0 /model/layers.7/self_attn/Reshape_output_0 Reshape /model/layers.7/self_attn/Reshape_1 2 1 /model/layers.7/self_attn/k_proj/Add_output_0 /model/layers.7/self_attn/Concat_1_output_0 /model/layers.7/self_attn/Reshape_1_output_0 Reshape /model/layers.7/self_attn/Reshape_2 2 1 /model/layers.7/self_attn/v_proj/Add_output_0 /model/layers.7/self_attn/Concat_2_output_0 /model/layers.7/self_attn/Reshape_2_output_0 Permute /model/layers.7/self_attn/Transpose 1 1 /model/layers.7/self_attn/Reshape_output_0 /model/layers.7/self_attn/Transpose_output_0 0=2 Split splitncnn_243 1 4 /model/layers.7/self_attn/Transpose_output_0 /model/layers.7/self_attn/Transpose_output_0_splitncnn_0 /model/layers.7/self_attn/Transpose_output_0_splitncnn_1 /model/layers.7/self_attn/Transpose_output_0_splitncnn_2 /model/layers.7/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.7/self_attn/Transpose_1 1 1 /model/layers.7/self_attn/Reshape_1_output_0 /model/layers.7/self_attn/Transpose_1_output_0 0=2 Split splitncnn_244 1 4 /model/layers.7/self_attn/Transpose_1_output_0 /model/layers.7/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.7/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.7/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.7/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.7/self_attn/Transpose_2 1 1 /model/layers.7/self_attn/Reshape_2_output_0 /model/layers.7/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.7/self_attn/Mul 2 1 /model/layers.7/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_33 /model/layers.7/self_attn/Mul_output_0 0=2 Shape /model/layers.7/self_attn/Shape_2 1 1 /model/layers.7/self_attn/Transpose_output_0_splitncnn_2 /model/layers.7/self_attn/Shape_2_output_0 BinaryOp /model/layers.7/self_attn/Mul_2 2 1 /model/layers.7/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_32 /model/layers.7/self_attn/Mul_2_output_0 0=2 Shape /model/layers.7/self_attn/Shape_3 1 1 /model/layers.7/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.7/self_attn/Shape_3_output_0 Concat /model/layers.7/self_attn/Concat_6 2 1 past_key_values.7.value /model/layers.7/self_attn/Transpose_2_output_0 present.7.value 0=-2 Split splitncnn_245 1 5 present.7.value present.7.value_splitncnn_0 present.7.value_splitncnn_1 present.7.value_splitncnn_2 present.7.value_splitncnn_3 present.7.value_splitncnn_4 Crop /model/layers.7/self_attn/Gather_2 1 1 /model/layers.7/self_attn/Shape_2_output_0 /model/layers.7/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.7/self_attn/Gather_3 1 1 /model/layers.7/self_attn/Shape_3_output_0 /model/layers.7/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.7/self_attn/Shape_9 1 1 present.7.value_splitncnn_4 /model/layers.7/self_attn/Shape_9_output_0 Shape /model/layers.7/self_attn/Shape_10 1 1 present.7.value_splitncnn_3 /model/layers.7/self_attn/Shape_10_output_0 Shape /model/layers.7/self_attn/Shape_11 1 1 present.7.value_splitncnn_2 /model/layers.7/self_attn/Shape_11_output_0 Shape /model/layers.7/self_attn/Shape_12 1 1 present.7.value_splitncnn_1 /model/layers.7/self_attn/Shape_12_output_0 ExpandDims /model/layers.7/self_attn/Unsqueeze_19 2 1 present.7.value_splitncnn_0 /model/layers.7/self_attn/Constant_45_output_0 /model/layers.7/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.7/self_attn/Div 2 1 /model/layers.7/self_attn/Gather_2_output_0 /model/layers.7/self_attn/Constant_9_output_0 /model/layers.7/self_attn/Div_output_0 0=3 BinaryOp /model/layers.7/self_attn/Div_1 2 1 /model/layers.7/self_attn/Gather_3_output_0 /model/layers.7/self_attn/Constant_19_output_0 /model/layers.7/self_attn/Div_1_output_0 0=3 Crop /model/layers.7/self_attn/Gather_8 1 1 /model/layers.7/self_attn/Shape_9_output_0 /model/layers.7/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_246 1 2 /model/layers.7/self_attn/Gather_8_output_0 /model/layers.7/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.7/self_attn/Gather_9 1 1 /model/layers.7/self_attn/Shape_10_output_0 /model/layers.7/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_247 1 2 /model/layers.7/self_attn/Gather_9_output_0 /model/layers.7/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.7/self_attn/Gather_10 1 1 /model/layers.7/self_attn/Shape_11_output_0 /model/layers.7/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_248 1 2 /model/layers.7/self_attn/Gather_10_output_0 /model/layers.7/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.7/self_attn/Gather_11 1 1 /model/layers.7/self_attn/Shape_12_output_0 /model/layers.7/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_249 1 2 /model/layers.7/self_attn/Gather_11_output_0 /model/layers.7/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.7/self_attn/Cast 1 1 /model/layers.7/self_attn/Div_output_0 /model/layers.7/self_attn/Cast_output_0 Cast /model/layers.7/self_attn/Cast_2 1 1 /model/layers.7/self_attn/Div_1_output_0 /model/layers.7/self_attn/Cast_2_output_0 ExpandDims /model/layers.7/self_attn/Unsqueeze_20 2 1 /model/layers.7/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_46_output_0 /model/layers.7/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_24 2 1 /model/layers.7/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_2633 /model/layers.7/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_21 2 1 /model/layers.7/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_47_output_0 /model/layers.7/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.7/self_attn/Mul_7 2 1 /model/layers.7/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.7/self_attn/Constant_53_output_0 /model/layers.7/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.7/self_attn/Unsqueeze_22 2 1 /model/layers.7/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_49_output_0 /model/layers.7/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_26 2 1 /model/layers.7/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_2637 /model/layers.7/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_23 2 1 /model/layers.7/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_50_output_0 /model/layers.7/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_27 2 1 /model/layers.7/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_2639 /model/layers.7/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.7/self_attn/Cast_1 1 1 /model/layers.7/self_attn/Cast_output_0 /model/layers.7/self_attn/Cast_1_output_0 Split splitncnn_250 1 2 /model/layers.7/self_attn/Cast_1_output_0 /model/layers.7/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.7/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.7/self_attn/Cast_3 1 1 /model/layers.7/self_attn/Cast_2_output_0 /model/layers.7/self_attn/Cast_3_output_0 Split splitncnn_251 1 2 /model/layers.7/self_attn/Cast_3_output_0 /model/layers.7/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.7/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.7/self_attn/Unsqueeze_25 2 1 /model/layers.7/self_attn/Mul_7_output_0 onnx::Unsqueeze_2635 /model/layers.7/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.7/self_attn/Concat_9 5 1 /model/layers.7/self_attn/Unsqueeze_20_output_0 /model/layers.7/self_attn/Unsqueeze_21_output_0 /model/layers.7/self_attn/Constant_48_output_0 /model/layers.7/self_attn/Unsqueeze_22_output_0 /model/layers.7/self_attn/Unsqueeze_23_output_0 /model/layers.7/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_6 2 1 /model/layers.7/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_12_output_0 /model/layers.7/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_7 2 1 /model/layers.7/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.7/self_attn/Constant_15_output_0 /model/layers.7/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_8 2 1 /model/layers.7/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_22_output_0 /model/layers.7/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_9 2 1 /model/layers.7/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.7/self_attn/Constant_25_output_0 /model/layers.7/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.7/self_attn/Concat_10 4 1 /model/layers.7/self_attn/Unsqueeze_24_output_0 /model/layers.7/self_attn/Unsqueeze_25_output_0 /model/layers.7/self_attn/Unsqueeze_26_output_0 /model/layers.7/self_attn/Unsqueeze_27_output_0 /model/layers.7/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.7/self_attn/Reshape_5 1 1 /model/layers.7/self_attn/Concat_9_output_0 /model/layers.7/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_252 1 3 /model/layers.7/self_attn/Reshape_5_output_0 /model/layers.7/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.7/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.7/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.7/self_attn/Slice 2 1 /model/layers.7/self_attn/Transpose_output_0_splitncnn_1 /model/layers.7/self_attn/Unsqueeze_6_output_0 /model/layers.7/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.7/self_attn/Slice_1 2 1 /model/layers.7/self_attn/Transpose_output_0_splitncnn_0 /model/layers.7/self_attn/Unsqueeze_7_output_0 /model/layers.7/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.7/self_attn/Slice_2 2 1 /model/layers.7/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.7/self_attn/Unsqueeze_8_output_0 /model/layers.7/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.7/self_attn/Slice_3 2 1 /model/layers.7/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.7/self_attn/Unsqueeze_9_output_0 /model/layers.7/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.7/self_attn/Shape_13 1 1 /model/layers.7/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.7/self_attn/Shape_13_output_0 UnaryOp /model/layers.7/self_attn/Neg 1 1 /model/layers.7/self_attn/Slice_1_output_0 /model/layers.7/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.7/self_attn/Neg_1 1 1 /model/layers.7/self_attn/Slice_3_output_0 /model/layers.7/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.7/self_attn/ConstantOfShape_1 1 1 /model/layers.7/self_attn/Shape_13_output_0 /model/layers.7/self_attn/ConstantOfShape_1_output_0 Split splitncnn_253 1 2 /model/layers.7/self_attn/ConstantOfShape_1_output_0 /model/layers.7/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.7/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.7/self_attn/Concat_3 2 1 /model/layers.7/self_attn/Neg_output_0 /model/layers.7/self_attn/Slice_output_0 /model/layers.7/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.7/self_attn/Concat_4 2 1 /model/layers.7/self_attn/Neg_1_output_0 /model/layers.7/self_attn/Slice_2_output_0 /model/layers.7/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.7/self_attn/Mul_6 2 1 /model/layers.7/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_52_output_0 /model/layers.7/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.7/self_attn/Mul_1 2 1 /model/layers.7/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_33 /model/layers.7/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.7/self_attn/Mul_3 2 1 /model/layers.7/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_32 /model/layers.7/self_attn/Mul_3_output_0 0=2 Equal /model/layers.7/self_attn/Equal_1 2 1 /model/layers.7/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.7/self_attn/Mul_6_output_0 /model/layers.7/self_attn/Equal_1_output_0 BinaryOp /model/layers.7/self_attn/Add 2 1 /model/layers.7/self_attn/Mul_output_0 /model/layers.7/self_attn/Mul_1_output_0 /model/layers.7/self_attn/Add_output_0 0=0 BinaryOp /model/layers.7/self_attn/Add_1 2 1 /model/layers.7/self_attn/Mul_2_output_0 /model/layers.7/self_attn/Mul_3_output_0 /model/layers.7/self_attn/Add_1_output_0 0=0 Where /model/layers.7/self_attn/Where_1 3 1 /model/layers.7/self_attn/Equal_1_output_0 /model/layers.7/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.7/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.7/self_attn/Where_1_output_0 BinaryOp /model/layers.7/self_attn/Mul_8 1 1 /model/layers.7/self_attn/Add_output_0 /model/layers.7/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.7/self_attn/Concat_5 2 1 past_key_values.7.key /model/layers.7/self_attn/Add_1_output_0 present.7.key 0=-2 Split splitncnn_254 1 5 present.7.key present.7.key_splitncnn_0 present.7.key_splitncnn_1 present.7.key_splitncnn_2 present.7.key_splitncnn_3 present.7.key_splitncnn_4 Expand /model/layers.7/self_attn/Expand_1 2 1 /model/layers.7/self_attn/Unsqueeze_19_output_0 /model/layers.7/self_attn/Where_1_output_0 /model/layers.7/self_attn/Expand_1_output_0 Shape /model/layers.7/self_attn/Shape_4 1 1 present.7.key_splitncnn_4 /model/layers.7/self_attn/Shape_4_output_0 Shape /model/layers.7/self_attn/Shape_5 1 1 present.7.key_splitncnn_3 /model/layers.7/self_attn/Shape_5_output_0 Shape /model/layers.7/self_attn/Shape_6 1 1 present.7.key_splitncnn_2 /model/layers.7/self_attn/Shape_6_output_0 Shape /model/layers.7/self_attn/Shape_7 1 1 present.7.key_splitncnn_1 /model/layers.7/self_attn/Shape_7_output_0 ExpandDims /model/layers.7/self_attn/Unsqueeze_10 2 1 present.7.key_splitncnn_0 /model/layers.7/self_attn/Constant_32_output_0 /model/layers.7/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.7/self_attn/Reshape_6 2 1 /model/layers.7/self_attn/Expand_1_output_0 /model/layers.7/self_attn/Concat_10_output_0 /model/layers.7/self_attn/Reshape_6_output_0 Crop /model/layers.7/self_attn/Gather_4 1 1 /model/layers.7/self_attn/Shape_4_output_0 /model/layers.7/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_255 1 2 /model/layers.7/self_attn/Gather_4_output_0 /model/layers.7/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.7/self_attn/Gather_5 1 1 /model/layers.7/self_attn/Shape_5_output_0 /model/layers.7/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_256 1 2 /model/layers.7/self_attn/Gather_5_output_0 /model/layers.7/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.7/self_attn/Gather_6 1 1 /model/layers.7/self_attn/Shape_6_output_0 /model/layers.7/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_257 1 2 /model/layers.7/self_attn/Gather_6_output_0 /model/layers.7/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.7/self_attn/Gather_7 1 1 /model/layers.7/self_attn/Shape_7_output_0 /model/layers.7/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_258 1 2 /model/layers.7/self_attn/Gather_7_output_0 /model/layers.7/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.7/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.7/self_attn/Unsqueeze_11 2 1 /model/layers.7/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_33_output_0 /model/layers.7/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_15 2 1 /model/layers.7/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_2587 /model/layers.7/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_12 2 1 /model/layers.7/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_34_output_0 /model/layers.7/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.7/self_attn/Mul_5 2 1 /model/layers.7/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.7/self_attn/Constant_40_output_0 /model/layers.7/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.7/self_attn/Unsqueeze_13 2 1 /model/layers.7/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_36_output_0 /model/layers.7/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_17 2 1 /model/layers.7/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_2591 /model/layers.7/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_14 2 1 /model/layers.7/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_37_output_0 /model/layers.7/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_18 2 1 /model/layers.7/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_2593 /model/layers.7/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.7/self_attn/Unsqueeze_16 2 1 /model/layers.7/self_attn/Mul_5_output_0 onnx::Unsqueeze_2589 /model/layers.7/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.7/self_attn/Concat_7 5 1 /model/layers.7/self_attn/Unsqueeze_11_output_0 /model/layers.7/self_attn/Unsqueeze_12_output_0 /model/layers.7/self_attn/Constant_35_output_0 /model/layers.7/self_attn/Unsqueeze_13_output_0 /model/layers.7/self_attn/Unsqueeze_14_output_0 /model/layers.7/self_attn/Concat_7_output_0 0=0 Concat /model/layers.7/self_attn/Concat_8 4 1 /model/layers.7/self_attn/Unsqueeze_15_output_0 /model/layers.7/self_attn/Unsqueeze_16_output_0 /model/layers.7/self_attn/Unsqueeze_17_output_0 /model/layers.7/self_attn/Unsqueeze_18_output_0 /model/layers.7/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.7/self_attn/Reshape_3 1 1 /model/layers.7/self_attn/Concat_7_output_0 /model/layers.7/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_259 1 3 /model/layers.7/self_attn/Reshape_3_output_0 /model/layers.7/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.7/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.7/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.7/self_attn/Shape_8 1 1 /model/layers.7/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.7/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.7/self_attn/ConstantOfShape 1 1 /model/layers.7/self_attn/Shape_8_output_0 /model/layers.7/self_attn/ConstantOfShape_output_0 Split splitncnn_260 1 2 /model/layers.7/self_attn/ConstantOfShape_output_0 /model/layers.7/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.7/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.7/self_attn/Mul_4 2 1 /model/layers.7/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.7/self_attn/Constant_39_output_0 /model/layers.7/self_attn/Mul_4_output_0 0=2 Equal /model/layers.7/self_attn/Equal 2 1 /model/layers.7/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.7/self_attn/Mul_4_output_0 /model/layers.7/self_attn/Equal_output_0 Where /model/layers.7/self_attn/Where 3 1 /model/layers.7/self_attn/Equal_output_0 /model/layers.7/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.7/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.7/self_attn/Where_output_0 Expand /model/layers.7/self_attn/Expand 2 1 /model/layers.7/self_attn/Unsqueeze_10_output_0 /model/layers.7/self_attn/Where_output_0 /model/layers.7/self_attn/Expand_output_0 Reshape /model/layers.7/self_attn/Reshape_4 2 1 /model/layers.7/self_attn/Expand_output_0 /model/layers.7/self_attn/Concat_8_output_0 /model/layers.7/self_attn/Reshape_4_output_0 Split splitncnn_261 1 2 /model/layers.7/self_attn/Reshape_4_output_0 /model/layers.7/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.7/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.7/self_attn/Shape_14 1 1 /model/layers.7/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.7/self_attn/Shape_14_output_0 Permute /model/layers.7/self_attn/Transpose_3 1 1 /model/layers.7/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.7/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.7/self_attn/Gather_12 1 1 /model/layers.7/self_attn/Shape_14_output_0 /model/layers.7/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.7/self_attn/Mul_9 1 1 /model/layers.7/self_attn/Transpose_3_output_0 /model/layers.7/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.7/self_attn/Unsqueeze_28 2 1 /model/layers.7/self_attn/Gather_12_output_0 /model/layers.7/self_attn/Constant_57_output_0 /model/layers.7/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.7/self_attn/MatMul 2 1 /model/layers.7/self_attn/Mul_8_output_0 /model/layers.7/self_attn/Mul_9_output_0 /model/layers.7/self_attn/MatMul_output_0 Crop /model/layers.7/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_16 /model/layers.7/self_attn/Unsqueeze_28_output_0 /model/layers.7/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.7/self_attn/Where_2 3 1 /model/layers.7/self_attn/Slice_4_output_0 /model/layers.7/self_attn/Constant_61_output_0 /model/layers.7/self_attn/Constant_62_output_0 /model/layers.7/self_attn/Where_2_output_0 BinaryOp /model/layers.7/self_attn/Add_2 2 1 /model/layers.7/self_attn/MatMul_output_0 /model/layers.7/self_attn/Where_2_output_0 /model/layers.7/self_attn/Add_2_output_0 0=0 Softmax /model/layers.7/self_attn/Softmax 1 1 /model/layers.7/self_attn/Add_2_output_0 /model/layers.7/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_262 1 2 /model/layers.7/self_attn/Softmax_output_0 /model/layers.7/self_attn/Softmax_output_0_splitncnn_0 /model/layers.7/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.7/self_attn/IsNaN 1 1 /model/layers.7/self_attn/Softmax_output_0_splitncnn_1 /model/layers.7/self_attn/IsNaN_output_0 Where /model/layers.7/self_attn/Where_3 3 1 /model/layers.7/self_attn/IsNaN_output_0 /model/layers.7/self_attn/Constant_63_output_0 /model/layers.7/self_attn/Softmax_output_0_splitncnn_0 /model/layers.7/self_attn/Where_3_output_0 Gemm /model/layers.7/self_attn/MatMul_1 2 1 /model/layers.7/self_attn/Where_3_output_0 /model/layers.7/self_attn/Reshape_6_output_0 /model/layers.7/self_attn/MatMul_1_output_0 Permute /model/layers.7/self_attn/Transpose_4 1 1 /model/layers.7/self_attn/MatMul_1_output_0 /model/layers.7/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.7/self_attn/Reshape_7 2 1 /model/layers.7/self_attn/Transpose_4_output_0 /model/layers.7/self_attn/Concat_11_output_0 /model/layers.7/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.7/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.7/self_attn/Reshape_7_output_0 /model/layers.7/self_attn/Reshape_7_output_0_quantized /model/layers.7/self_attn/Reshape_7_output_0_scale /model/layers.7/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.7/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.7/self_attn/Reshape_7_output_0_scale /model/layers.7/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.752461e-03 MatMulInteger /model/layers.7/self_attn/o_proj/MatMul_quant 4 1 /model/layers.7/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7466_quantized /model/layers.7/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7466_zero_point /model/layers.7/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.7/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.7/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.7/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.7/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.7/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.7/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.7/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.7/Add 2 1 /model/layers.7/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.7/self_attn/o_proj/MatMul_output_0 /model/layers.7/Add_output_0 0=0 Cast /model/layers.7/post_attention_layernorm/Cast 1 1 /model/layers.7/Add_output_0 /model/layers.7/post_attention_layernorm/Cast_output_0 Split splitncnn_263 1 3 /model/layers.7/post_attention_layernorm/Cast_output_0 /model/layers.7/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.7/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.7/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.7/post_attention_layernorm/Pow 1 1 /model/layers.7/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.7/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.7/post_attention_layernorm/ReduceMean 2 1 /model/layers.7/post_attention_layernorm/Pow_output_0 /model/layers.7/post_attention_layernorm/Constant_1_output_0 /model/layers.7/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.7/post_attention_layernorm/Add 1 1 /model/layers.7/post_attention_layernorm/ReduceMean_output_0 /model/layers.7/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.7/post_attention_layernorm/Sqrt 1 1 /model/layers.7/post_attention_layernorm/Add_output_0 /model/layers.7/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.7/post_attention_layernorm/Div 1 1 /model/layers.7/post_attention_layernorm/Sqrt_output_0 /model/layers.7/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.7/post_attention_layernorm/Mul 2 1 /model/layers.7/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.7/post_attention_layernorm/Div_output_0 /model/layers.7/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.7/post_attention_layernorm/Cast_1 1 1 /model/layers.7/post_attention_layernorm/Mul_output_0 /model/layers.7/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.7/post_attention_layernorm/Mul_1 2 1 model.layers.7.post_attention_layernorm.weight /model/layers.7/post_attention_layernorm/Cast_1_output_0 /model/layers.7/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.7/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.7/post_attention_layernorm/Mul_1_output_0 /model/layers.7/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.7/post_attention_layernorm/Mul_1_output_0_scale /model/layers.7/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_264 1 2 /model/layers.7/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.7/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.7/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_265 1 2 /model/layers.7/post_attention_layernorm/Mul_1_output_0_scale /model/layers.7/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.7/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_266 1 2 /model/layers.7/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.7/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.7/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.7/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.7/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.7/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.583661e-03 BinaryOp /model/layers.7/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.7/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.7/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.552166e-03 MatMulInteger /model/layers.7/mlp/gate_proj/MatMul_quant 4 1 /model/layers.7/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7467_quantized /model/layers.7/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7467_zero_point /model/layers.7/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.7/mlp/up_proj/MatMul_quant 4 1 /model/layers.7/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7468_quantized /model/layers.7/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7468_zero_point /model/layers.7/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.7/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.7/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.7/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.7/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.7/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.7/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.7/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.7/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.7/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.7/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.7/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.7/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.7/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.7/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.7/mlp/act_fn/Mul 1 1 /model/layers.7/mlp/gate_proj/MatMul_output_0 /model/layers.7/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.7/mlp/Mul 2 1 /model/layers.7/mlp/act_fn/Mul_output_0 /model/layers.7/mlp/up_proj/MatMul_output_0 /model/layers.7/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.7/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.7/mlp/Mul_output_0 /model/layers.7/mlp/Mul_output_0_quantized /model/layers.7/mlp/Mul_output_0_scale /model/layers.7/mlp/Mul_output_0_zero_point BinaryOp /model/layers.7/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.7/mlp/Mul_output_0_scale /model/layers.7/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.275714e-03 MatMulInteger /model/layers.7/mlp/down_proj/MatMul_quant 4 1 /model/layers.7/mlp/Mul_output_0_quantized onnx::MatMul_7469_quantized /model/layers.7/mlp/Mul_output_0_zero_point onnx::MatMul_7469_zero_point /model/layers.7/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.7/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.7/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.7/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.7/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.7/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.7/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.7/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.7/Add_1 2 1 /model/layers.7/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.7/mlp/down_proj/MatMul_output_0 /model/layers.7/Add_1_output_0 0=0 Cast /model/layers.8/input_layernorm/Cast 1 1 /model/layers.7/Add_1_output_0 /model/layers.8/input_layernorm/Cast_output_0 Split splitncnn_267 1 3 /model/layers.8/input_layernorm/Cast_output_0 /model/layers.8/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.8/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.8/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.8/input_layernorm/Pow 1 1 /model/layers.8/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.8/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.8/input_layernorm/ReduceMean 2 1 /model/layers.8/input_layernorm/Pow_output_0 /model/layers.8/input_layernorm/Constant_1_output_0 /model/layers.8/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.8/input_layernorm/Add 1 1 /model/layers.8/input_layernorm/ReduceMean_output_0 /model/layers.8/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.8/input_layernorm/Sqrt 1 1 /model/layers.8/input_layernorm/Add_output_0 /model/layers.8/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.8/input_layernorm/Div 1 1 /model/layers.8/input_layernorm/Sqrt_output_0 /model/layers.8/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.8/input_layernorm/Mul 2 1 /model/layers.8/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.8/input_layernorm/Div_output_0 /model/layers.8/input_layernorm/Mul_output_0 0=2 Cast /model/layers.8/input_layernorm/Cast_1 1 1 /model/layers.8/input_layernorm/Mul_output_0 /model/layers.8/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.8/input_layernorm/Mul_1 2 1 model.layers.8.input_layernorm.weight /model/layers.8/input_layernorm/Cast_1_output_0 /model/layers.8/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_268 1 3 /model/layers.8/input_layernorm/Mul_1_output_0 /model/layers.8/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.8/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.8/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.8/self_attn/Shape 1 1 /model/layers.8/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.8/self_attn/Shape_output_0 Shape /model/layers.8/self_attn/Shape_1 1 1 /model/layers.8/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.8/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.8/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.8/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.8/input_layernorm/Mul_1_output_0_quantized /model/layers.8/input_layernorm/Mul_1_output_0_scale /model/layers.8/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_269 1 3 /model/layers.8/input_layernorm/Mul_1_output_0_quantized /model/layers.8/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.8/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.8/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_270 1 3 /model/layers.8/input_layernorm/Mul_1_output_0_scale /model/layers.8/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.8/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.8/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_271 1 3 /model/layers.8/input_layernorm/Mul_1_output_0_zero_point /model/layers.8/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.8/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.8/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.8/self_attn/Gather 1 1 /model/layers.8/self_attn/Shape_output_0 /model/layers.8/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_272 1 4 /model/layers.8/self_attn/Gather_output_0 /model/layers.8/self_attn/Gather_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_output_0_splitncnn_1 /model/layers.8/self_attn/Gather_output_0_splitncnn_2 /model/layers.8/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.8/self_attn/Gather_1 1 1 /model/layers.8/self_attn/Shape_1_output_0 /model/layers.8/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_273 1 4 /model/layers.8/self_attn/Gather_1_output_0 /model/layers.8/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.8/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.8/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.8/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.8/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.8/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.328740e-02 BinaryOp /model/layers.8/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.8/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.8/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.791646e-03 BinaryOp /model/layers.8/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.8/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.8/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.153420e-03 MatMulInteger /model/layers.8/self_attn/q_proj/MatMul_quant 4 1 /model/layers.8/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7470_quantized /model/layers.8/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7470_zero_point /model/layers.8/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.8/self_attn/k_proj/MatMul_quant 4 1 /model/layers.8/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7477_quantized /model/layers.8/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7477_zero_point /model/layers.8/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.8/self_attn/v_proj/MatMul_quant 4 1 /model/layers.8/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7478_quantized /model/layers.8/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7478_zero_point /model/layers.8/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.8/self_attn/Unsqueeze 2 1 /model/layers.8/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_2729 /model/layers.8/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_2 2 1 /model/layers.8/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_2738 /model/layers.8/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_4 2 1 /model/layers.8/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_2747 /model/layers.8/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_29 2 1 /model/layers.8/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_2952 /model/layers.8/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_1 2 1 /model/layers.8/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_2731 /model/layers.8/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_3 2 1 /model/layers.8/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_2740 /model/layers.8/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_5 2 1 /model/layers.8/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_2749 /model/layers.8/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_30 2 1 /model/layers.8/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_2954 /model/layers.8/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.8/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.8/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.8/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.8/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.8/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.8/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.8/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.8/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.8/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.8/self_attn/Concat 4 1 /model/layers.8/self_attn/Unsqueeze_output_0 /model/layers.8/self_attn/Unsqueeze_1_output_0 /model/layers.8/self_attn/Constant_2_output_0 /model/layers.8/self_attn/Constant_3_output_0 /model/layers.8/self_attn/Concat_output_0 0=0 Concat /model/layers.8/self_attn/Concat_1 4 1 /model/layers.8/self_attn/Unsqueeze_2_output_0 /model/layers.8/self_attn/Unsqueeze_3_output_0 /model/layers.8/self_attn/Constant_4_output_0 /model/layers.8/self_attn/Constant_5_output_0 /model/layers.8/self_attn/Concat_1_output_0 0=0 Concat /model/layers.8/self_attn/Concat_2 4 1 /model/layers.8/self_attn/Unsqueeze_4_output_0 /model/layers.8/self_attn/Unsqueeze_5_output_0 /model/layers.8/self_attn/Constant_6_output_0 /model/layers.8/self_attn/Constant_7_output_0 /model/layers.8/self_attn/Concat_2_output_0 0=0 Concat /model/layers.8/self_attn/Concat_11 3 1 /model/layers.8/self_attn/Unsqueeze_29_output_0 /model/layers.8/self_attn/Unsqueeze_30_output_0 /model/layers.8/self_attn/Constant_64_output_0 /model/layers.8/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.8/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.8/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.8/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.8/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.8/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.8/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.8/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.8/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.8/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.8/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.8/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.8/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.8/self_attn/q_proj/Add 2 1 model.layers.8.self_attn.q_proj.bias /model/layers.8/self_attn/q_proj/MatMul_output_0 /model/layers.8/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.8/self_attn/k_proj/Add 2 1 model.layers.8.self_attn.k_proj.bias /model/layers.8/self_attn/k_proj/MatMul_output_0 /model/layers.8/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.8/self_attn/v_proj/Add 2 1 model.layers.8.self_attn.v_proj.bias /model/layers.8/self_attn/v_proj/MatMul_output_0 /model/layers.8/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.8/self_attn/Reshape 2 1 /model/layers.8/self_attn/q_proj/Add_output_0 /model/layers.8/self_attn/Concat_output_0 /model/layers.8/self_attn/Reshape_output_0 Reshape /model/layers.8/self_attn/Reshape_1 2 1 /model/layers.8/self_attn/k_proj/Add_output_0 /model/layers.8/self_attn/Concat_1_output_0 /model/layers.8/self_attn/Reshape_1_output_0 Reshape /model/layers.8/self_attn/Reshape_2 2 1 /model/layers.8/self_attn/v_proj/Add_output_0 /model/layers.8/self_attn/Concat_2_output_0 /model/layers.8/self_attn/Reshape_2_output_0 Permute /model/layers.8/self_attn/Transpose 1 1 /model/layers.8/self_attn/Reshape_output_0 /model/layers.8/self_attn/Transpose_output_0 0=2 Split splitncnn_274 1 4 /model/layers.8/self_attn/Transpose_output_0 /model/layers.8/self_attn/Transpose_output_0_splitncnn_0 /model/layers.8/self_attn/Transpose_output_0_splitncnn_1 /model/layers.8/self_attn/Transpose_output_0_splitncnn_2 /model/layers.8/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.8/self_attn/Transpose_1 1 1 /model/layers.8/self_attn/Reshape_1_output_0 /model/layers.8/self_attn/Transpose_1_output_0 0=2 Split splitncnn_275 1 4 /model/layers.8/self_attn/Transpose_1_output_0 /model/layers.8/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.8/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.8/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.8/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.8/self_attn/Transpose_2 1 1 /model/layers.8/self_attn/Reshape_2_output_0 /model/layers.8/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.8/self_attn/Mul 2 1 /model/layers.8/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_31 /model/layers.8/self_attn/Mul_output_0 0=2 Shape /model/layers.8/self_attn/Shape_2 1 1 /model/layers.8/self_attn/Transpose_output_0_splitncnn_2 /model/layers.8/self_attn/Shape_2_output_0 BinaryOp /model/layers.8/self_attn/Mul_2 2 1 /model/layers.8/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_30 /model/layers.8/self_attn/Mul_2_output_0 0=2 Shape /model/layers.8/self_attn/Shape_3 1 1 /model/layers.8/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.8/self_attn/Shape_3_output_0 Concat /model/layers.8/self_attn/Concat_6 2 1 past_key_values.8.value /model/layers.8/self_attn/Transpose_2_output_0 present.8.value 0=-2 Split splitncnn_276 1 5 present.8.value present.8.value_splitncnn_0 present.8.value_splitncnn_1 present.8.value_splitncnn_2 present.8.value_splitncnn_3 present.8.value_splitncnn_4 Crop /model/layers.8/self_attn/Gather_2 1 1 /model/layers.8/self_attn/Shape_2_output_0 /model/layers.8/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.8/self_attn/Gather_3 1 1 /model/layers.8/self_attn/Shape_3_output_0 /model/layers.8/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.8/self_attn/Shape_9 1 1 present.8.value_splitncnn_4 /model/layers.8/self_attn/Shape_9_output_0 Shape /model/layers.8/self_attn/Shape_10 1 1 present.8.value_splitncnn_3 /model/layers.8/self_attn/Shape_10_output_0 Shape /model/layers.8/self_attn/Shape_11 1 1 present.8.value_splitncnn_2 /model/layers.8/self_attn/Shape_11_output_0 Shape /model/layers.8/self_attn/Shape_12 1 1 present.8.value_splitncnn_1 /model/layers.8/self_attn/Shape_12_output_0 ExpandDims /model/layers.8/self_attn/Unsqueeze_19 2 1 present.8.value_splitncnn_0 /model/layers.8/self_attn/Constant_45_output_0 /model/layers.8/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.8/self_attn/Div 2 1 /model/layers.8/self_attn/Gather_2_output_0 /model/layers.8/self_attn/Constant_9_output_0 /model/layers.8/self_attn/Div_output_0 0=3 BinaryOp /model/layers.8/self_attn/Div_1 2 1 /model/layers.8/self_attn/Gather_3_output_0 /model/layers.8/self_attn/Constant_19_output_0 /model/layers.8/self_attn/Div_1_output_0 0=3 Crop /model/layers.8/self_attn/Gather_8 1 1 /model/layers.8/self_attn/Shape_9_output_0 /model/layers.8/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_277 1 2 /model/layers.8/self_attn/Gather_8_output_0 /model/layers.8/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.8/self_attn/Gather_9 1 1 /model/layers.8/self_attn/Shape_10_output_0 /model/layers.8/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_278 1 2 /model/layers.8/self_attn/Gather_9_output_0 /model/layers.8/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.8/self_attn/Gather_10 1 1 /model/layers.8/self_attn/Shape_11_output_0 /model/layers.8/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_279 1 2 /model/layers.8/self_attn/Gather_10_output_0 /model/layers.8/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.8/self_attn/Gather_11 1 1 /model/layers.8/self_attn/Shape_12_output_0 /model/layers.8/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_280 1 2 /model/layers.8/self_attn/Gather_11_output_0 /model/layers.8/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.8/self_attn/Cast 1 1 /model/layers.8/self_attn/Div_output_0 /model/layers.8/self_attn/Cast_output_0 Cast /model/layers.8/self_attn/Cast_2 1 1 /model/layers.8/self_attn/Div_1_output_0 /model/layers.8/self_attn/Cast_2_output_0 ExpandDims /model/layers.8/self_attn/Unsqueeze_20 2 1 /model/layers.8/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_46_output_0 /model/layers.8/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_24 2 1 /model/layers.8/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_2912 /model/layers.8/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_21 2 1 /model/layers.8/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_47_output_0 /model/layers.8/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.8/self_attn/Mul_7 2 1 /model/layers.8/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.8/self_attn/Constant_53_output_0 /model/layers.8/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.8/self_attn/Unsqueeze_22 2 1 /model/layers.8/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_49_output_0 /model/layers.8/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_26 2 1 /model/layers.8/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_2916 /model/layers.8/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_23 2 1 /model/layers.8/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_50_output_0 /model/layers.8/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_27 2 1 /model/layers.8/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_2918 /model/layers.8/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.8/self_attn/Cast_1 1 1 /model/layers.8/self_attn/Cast_output_0 /model/layers.8/self_attn/Cast_1_output_0 Split splitncnn_281 1 2 /model/layers.8/self_attn/Cast_1_output_0 /model/layers.8/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.8/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.8/self_attn/Cast_3 1 1 /model/layers.8/self_attn/Cast_2_output_0 /model/layers.8/self_attn/Cast_3_output_0 Split splitncnn_282 1 2 /model/layers.8/self_attn/Cast_3_output_0 /model/layers.8/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.8/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.8/self_attn/Unsqueeze_25 2 1 /model/layers.8/self_attn/Mul_7_output_0 onnx::Unsqueeze_2914 /model/layers.8/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.8/self_attn/Concat_9 5 1 /model/layers.8/self_attn/Unsqueeze_20_output_0 /model/layers.8/self_attn/Unsqueeze_21_output_0 /model/layers.8/self_attn/Constant_48_output_0 /model/layers.8/self_attn/Unsqueeze_22_output_0 /model/layers.8/self_attn/Unsqueeze_23_output_0 /model/layers.8/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_6 2 1 /model/layers.8/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_12_output_0 /model/layers.8/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_7 2 1 /model/layers.8/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.8/self_attn/Constant_15_output_0 /model/layers.8/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_8 2 1 /model/layers.8/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_22_output_0 /model/layers.8/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_9 2 1 /model/layers.8/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.8/self_attn/Constant_25_output_0 /model/layers.8/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.8/self_attn/Concat_10 4 1 /model/layers.8/self_attn/Unsqueeze_24_output_0 /model/layers.8/self_attn/Unsqueeze_25_output_0 /model/layers.8/self_attn/Unsqueeze_26_output_0 /model/layers.8/self_attn/Unsqueeze_27_output_0 /model/layers.8/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.8/self_attn/Reshape_5 1 1 /model/layers.8/self_attn/Concat_9_output_0 /model/layers.8/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_283 1 3 /model/layers.8/self_attn/Reshape_5_output_0 /model/layers.8/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.8/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.8/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.8/self_attn/Slice 2 1 /model/layers.8/self_attn/Transpose_output_0_splitncnn_1 /model/layers.8/self_attn/Unsqueeze_6_output_0 /model/layers.8/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.8/self_attn/Slice_1 2 1 /model/layers.8/self_attn/Transpose_output_0_splitncnn_0 /model/layers.8/self_attn/Unsqueeze_7_output_0 /model/layers.8/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.8/self_attn/Slice_2 2 1 /model/layers.8/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.8/self_attn/Unsqueeze_8_output_0 /model/layers.8/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.8/self_attn/Slice_3 2 1 /model/layers.8/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.8/self_attn/Unsqueeze_9_output_0 /model/layers.8/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.8/self_attn/Shape_13 1 1 /model/layers.8/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.8/self_attn/Shape_13_output_0 UnaryOp /model/layers.8/self_attn/Neg 1 1 /model/layers.8/self_attn/Slice_1_output_0 /model/layers.8/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.8/self_attn/Neg_1 1 1 /model/layers.8/self_attn/Slice_3_output_0 /model/layers.8/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.8/self_attn/ConstantOfShape_1 1 1 /model/layers.8/self_attn/Shape_13_output_0 /model/layers.8/self_attn/ConstantOfShape_1_output_0 Split splitncnn_284 1 2 /model/layers.8/self_attn/ConstantOfShape_1_output_0 /model/layers.8/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.8/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.8/self_attn/Concat_3 2 1 /model/layers.8/self_attn/Neg_output_0 /model/layers.8/self_attn/Slice_output_0 /model/layers.8/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.8/self_attn/Concat_4 2 1 /model/layers.8/self_attn/Neg_1_output_0 /model/layers.8/self_attn/Slice_2_output_0 /model/layers.8/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.8/self_attn/Mul_6 2 1 /model/layers.8/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_52_output_0 /model/layers.8/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.8/self_attn/Mul_1 2 1 /model/layers.8/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_31 /model/layers.8/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.8/self_attn/Mul_3 2 1 /model/layers.8/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_30 /model/layers.8/self_attn/Mul_3_output_0 0=2 Equal /model/layers.8/self_attn/Equal_1 2 1 /model/layers.8/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.8/self_attn/Mul_6_output_0 /model/layers.8/self_attn/Equal_1_output_0 BinaryOp /model/layers.8/self_attn/Add 2 1 /model/layers.8/self_attn/Mul_output_0 /model/layers.8/self_attn/Mul_1_output_0 /model/layers.8/self_attn/Add_output_0 0=0 BinaryOp /model/layers.8/self_attn/Add_1 2 1 /model/layers.8/self_attn/Mul_2_output_0 /model/layers.8/self_attn/Mul_3_output_0 /model/layers.8/self_attn/Add_1_output_0 0=0 Where /model/layers.8/self_attn/Where_1 3 1 /model/layers.8/self_attn/Equal_1_output_0 /model/layers.8/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.8/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.8/self_attn/Where_1_output_0 BinaryOp /model/layers.8/self_attn/Mul_8 1 1 /model/layers.8/self_attn/Add_output_0 /model/layers.8/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.8/self_attn/Concat_5 2 1 past_key_values.8.key /model/layers.8/self_attn/Add_1_output_0 present.8.key 0=-2 Split splitncnn_285 1 5 present.8.key present.8.key_splitncnn_0 present.8.key_splitncnn_1 present.8.key_splitncnn_2 present.8.key_splitncnn_3 present.8.key_splitncnn_4 Expand /model/layers.8/self_attn/Expand_1 2 1 /model/layers.8/self_attn/Unsqueeze_19_output_0 /model/layers.8/self_attn/Where_1_output_0 /model/layers.8/self_attn/Expand_1_output_0 Shape /model/layers.8/self_attn/Shape_4 1 1 present.8.key_splitncnn_4 /model/layers.8/self_attn/Shape_4_output_0 Shape /model/layers.8/self_attn/Shape_5 1 1 present.8.key_splitncnn_3 /model/layers.8/self_attn/Shape_5_output_0 Shape /model/layers.8/self_attn/Shape_6 1 1 present.8.key_splitncnn_2 /model/layers.8/self_attn/Shape_6_output_0 Shape /model/layers.8/self_attn/Shape_7 1 1 present.8.key_splitncnn_1 /model/layers.8/self_attn/Shape_7_output_0 ExpandDims /model/layers.8/self_attn/Unsqueeze_10 2 1 present.8.key_splitncnn_0 /model/layers.8/self_attn/Constant_32_output_0 /model/layers.8/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.8/self_attn/Reshape_6 2 1 /model/layers.8/self_attn/Expand_1_output_0 /model/layers.8/self_attn/Concat_10_output_0 /model/layers.8/self_attn/Reshape_6_output_0 Crop /model/layers.8/self_attn/Gather_4 1 1 /model/layers.8/self_attn/Shape_4_output_0 /model/layers.8/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_286 1 2 /model/layers.8/self_attn/Gather_4_output_0 /model/layers.8/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.8/self_attn/Gather_5 1 1 /model/layers.8/self_attn/Shape_5_output_0 /model/layers.8/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_287 1 2 /model/layers.8/self_attn/Gather_5_output_0 /model/layers.8/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.8/self_attn/Gather_6 1 1 /model/layers.8/self_attn/Shape_6_output_0 /model/layers.8/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_288 1 2 /model/layers.8/self_attn/Gather_6_output_0 /model/layers.8/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.8/self_attn/Gather_7 1 1 /model/layers.8/self_attn/Shape_7_output_0 /model/layers.8/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_289 1 2 /model/layers.8/self_attn/Gather_7_output_0 /model/layers.8/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.8/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.8/self_attn/Unsqueeze_11 2 1 /model/layers.8/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_33_output_0 /model/layers.8/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_15 2 1 /model/layers.8/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_2866 /model/layers.8/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_12 2 1 /model/layers.8/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_34_output_0 /model/layers.8/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.8/self_attn/Mul_5 2 1 /model/layers.8/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.8/self_attn/Constant_40_output_0 /model/layers.8/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.8/self_attn/Unsqueeze_13 2 1 /model/layers.8/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_36_output_0 /model/layers.8/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_17 2 1 /model/layers.8/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_2870 /model/layers.8/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_14 2 1 /model/layers.8/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_37_output_0 /model/layers.8/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_18 2 1 /model/layers.8/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_2872 /model/layers.8/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.8/self_attn/Unsqueeze_16 2 1 /model/layers.8/self_attn/Mul_5_output_0 onnx::Unsqueeze_2868 /model/layers.8/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.8/self_attn/Concat_7 5 1 /model/layers.8/self_attn/Unsqueeze_11_output_0 /model/layers.8/self_attn/Unsqueeze_12_output_0 /model/layers.8/self_attn/Constant_35_output_0 /model/layers.8/self_attn/Unsqueeze_13_output_0 /model/layers.8/self_attn/Unsqueeze_14_output_0 /model/layers.8/self_attn/Concat_7_output_0 0=0 Concat /model/layers.8/self_attn/Concat_8 4 1 /model/layers.8/self_attn/Unsqueeze_15_output_0 /model/layers.8/self_attn/Unsqueeze_16_output_0 /model/layers.8/self_attn/Unsqueeze_17_output_0 /model/layers.8/self_attn/Unsqueeze_18_output_0 /model/layers.8/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.8/self_attn/Reshape_3 1 1 /model/layers.8/self_attn/Concat_7_output_0 /model/layers.8/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_290 1 3 /model/layers.8/self_attn/Reshape_3_output_0 /model/layers.8/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.8/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.8/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.8/self_attn/Shape_8 1 1 /model/layers.8/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.8/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.8/self_attn/ConstantOfShape 1 1 /model/layers.8/self_attn/Shape_8_output_0 /model/layers.8/self_attn/ConstantOfShape_output_0 Split splitncnn_291 1 2 /model/layers.8/self_attn/ConstantOfShape_output_0 /model/layers.8/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.8/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.8/self_attn/Mul_4 2 1 /model/layers.8/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.8/self_attn/Constant_39_output_0 /model/layers.8/self_attn/Mul_4_output_0 0=2 Equal /model/layers.8/self_attn/Equal 2 1 /model/layers.8/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.8/self_attn/Mul_4_output_0 /model/layers.8/self_attn/Equal_output_0 Where /model/layers.8/self_attn/Where 3 1 /model/layers.8/self_attn/Equal_output_0 /model/layers.8/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.8/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.8/self_attn/Where_output_0 Expand /model/layers.8/self_attn/Expand 2 1 /model/layers.8/self_attn/Unsqueeze_10_output_0 /model/layers.8/self_attn/Where_output_0 /model/layers.8/self_attn/Expand_output_0 Reshape /model/layers.8/self_attn/Reshape_4 2 1 /model/layers.8/self_attn/Expand_output_0 /model/layers.8/self_attn/Concat_8_output_0 /model/layers.8/self_attn/Reshape_4_output_0 Split splitncnn_292 1 2 /model/layers.8/self_attn/Reshape_4_output_0 /model/layers.8/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.8/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.8/self_attn/Shape_14 1 1 /model/layers.8/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.8/self_attn/Shape_14_output_0 Permute /model/layers.8/self_attn/Transpose_3 1 1 /model/layers.8/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.8/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.8/self_attn/Gather_12 1 1 /model/layers.8/self_attn/Shape_14_output_0 /model/layers.8/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.8/self_attn/Mul_9 1 1 /model/layers.8/self_attn/Transpose_3_output_0 /model/layers.8/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.8/self_attn/Unsqueeze_28 2 1 /model/layers.8/self_attn/Gather_12_output_0 /model/layers.8/self_attn/Constant_57_output_0 /model/layers.8/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.8/self_attn/MatMul 2 1 /model/layers.8/self_attn/Mul_8_output_0 /model/layers.8/self_attn/Mul_9_output_0 /model/layers.8/self_attn/MatMul_output_0 Crop /model/layers.8/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_15 /model/layers.8/self_attn/Unsqueeze_28_output_0 /model/layers.8/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.8/self_attn/Where_2 3 1 /model/layers.8/self_attn/Slice_4_output_0 /model/layers.8/self_attn/Constant_61_output_0 /model/layers.8/self_attn/Constant_62_output_0 /model/layers.8/self_attn/Where_2_output_0 BinaryOp /model/layers.8/self_attn/Add_2 2 1 /model/layers.8/self_attn/MatMul_output_0 /model/layers.8/self_attn/Where_2_output_0 /model/layers.8/self_attn/Add_2_output_0 0=0 Softmax /model/layers.8/self_attn/Softmax 1 1 /model/layers.8/self_attn/Add_2_output_0 /model/layers.8/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_293 1 2 /model/layers.8/self_attn/Softmax_output_0 /model/layers.8/self_attn/Softmax_output_0_splitncnn_0 /model/layers.8/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.8/self_attn/IsNaN 1 1 /model/layers.8/self_attn/Softmax_output_0_splitncnn_1 /model/layers.8/self_attn/IsNaN_output_0 Where /model/layers.8/self_attn/Where_3 3 1 /model/layers.8/self_attn/IsNaN_output_0 /model/layers.8/self_attn/Constant_63_output_0 /model/layers.8/self_attn/Softmax_output_0_splitncnn_0 /model/layers.8/self_attn/Where_3_output_0 Gemm /model/layers.8/self_attn/MatMul_1 2 1 /model/layers.8/self_attn/Where_3_output_0 /model/layers.8/self_attn/Reshape_6_output_0 /model/layers.8/self_attn/MatMul_1_output_0 Permute /model/layers.8/self_attn/Transpose_4 1 1 /model/layers.8/self_attn/MatMul_1_output_0 /model/layers.8/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.8/self_attn/Reshape_7 2 1 /model/layers.8/self_attn/Transpose_4_output_0 /model/layers.8/self_attn/Concat_11_output_0 /model/layers.8/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.8/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.8/self_attn/Reshape_7_output_0 /model/layers.8/self_attn/Reshape_7_output_0_quantized /model/layers.8/self_attn/Reshape_7_output_0_scale /model/layers.8/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.8/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.8/self_attn/Reshape_7_output_0_scale /model/layers.8/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.629798e-03 MatMulInteger /model/layers.8/self_attn/o_proj/MatMul_quant 4 1 /model/layers.8/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7501_quantized /model/layers.8/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7501_zero_point /model/layers.8/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.8/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.8/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.8/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.8/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.8/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.8/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.8/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.8/Add 2 1 /model/layers.8/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.8/self_attn/o_proj/MatMul_output_0 /model/layers.8/Add_output_0 0=0 Cast /model/layers.8/post_attention_layernorm/Cast 1 1 /model/layers.8/Add_output_0 /model/layers.8/post_attention_layernorm/Cast_output_0 Split splitncnn_294 1 3 /model/layers.8/post_attention_layernorm/Cast_output_0 /model/layers.8/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.8/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.8/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.8/post_attention_layernorm/Pow 1 1 /model/layers.8/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.8/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.8/post_attention_layernorm/ReduceMean 2 1 /model/layers.8/post_attention_layernorm/Pow_output_0 /model/layers.8/post_attention_layernorm/Constant_1_output_0 /model/layers.8/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.8/post_attention_layernorm/Add 1 1 /model/layers.8/post_attention_layernorm/ReduceMean_output_0 /model/layers.8/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.8/post_attention_layernorm/Sqrt 1 1 /model/layers.8/post_attention_layernorm/Add_output_0 /model/layers.8/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.8/post_attention_layernorm/Div 1 1 /model/layers.8/post_attention_layernorm/Sqrt_output_0 /model/layers.8/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.8/post_attention_layernorm/Mul 2 1 /model/layers.8/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.8/post_attention_layernorm/Div_output_0 /model/layers.8/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.8/post_attention_layernorm/Cast_1 1 1 /model/layers.8/post_attention_layernorm/Mul_output_0 /model/layers.8/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.8/post_attention_layernorm/Mul_1 2 1 model.layers.8.post_attention_layernorm.weight /model/layers.8/post_attention_layernorm/Cast_1_output_0 /model/layers.8/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.8/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.8/post_attention_layernorm/Mul_1_output_0 /model/layers.8/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.8/post_attention_layernorm/Mul_1_output_0_scale /model/layers.8/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_295 1 2 /model/layers.8/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.8/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.8/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_296 1 2 /model/layers.8/post_attention_layernorm/Mul_1_output_0_scale /model/layers.8/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.8/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_297 1 2 /model/layers.8/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.8/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.8/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.8/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.8/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.8/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.537525e-03 BinaryOp /model/layers.8/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.8/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.8/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.029651e-03 MatMulInteger /model/layers.8/mlp/gate_proj/MatMul_quant 4 1 /model/layers.8/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7502_quantized /model/layers.8/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7502_zero_point /model/layers.8/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.8/mlp/up_proj/MatMul_quant 4 1 /model/layers.8/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7503_quantized /model/layers.8/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7503_zero_point /model/layers.8/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.8/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.8/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.8/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.8/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.8/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.8/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.8/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.8/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.8/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.8/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.8/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.8/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.8/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.8/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.8/mlp/act_fn/Mul 1 1 /model/layers.8/mlp/gate_proj/MatMul_output_0 /model/layers.8/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.8/mlp/Mul 2 1 /model/layers.8/mlp/act_fn/Mul_output_0 /model/layers.8/mlp/up_proj/MatMul_output_0 /model/layers.8/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.8/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.8/mlp/Mul_output_0 /model/layers.8/mlp/Mul_output_0_quantized /model/layers.8/mlp/Mul_output_0_scale /model/layers.8/mlp/Mul_output_0_zero_point BinaryOp /model/layers.8/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.8/mlp/Mul_output_0_scale /model/layers.8/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.706693e-03 MatMulInteger /model/layers.8/mlp/down_proj/MatMul_quant 4 1 /model/layers.8/mlp/Mul_output_0_quantized onnx::MatMul_7504_quantized /model/layers.8/mlp/Mul_output_0_zero_point onnx::MatMul_7504_zero_point /model/layers.8/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.8/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.8/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.8/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.8/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.8/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.8/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.8/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.8/Add_1 2 1 /model/layers.8/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.8/mlp/down_proj/MatMul_output_0 /model/layers.8/Add_1_output_0 0=0 Cast /model/layers.9/input_layernorm/Cast 1 1 /model/layers.8/Add_1_output_0 /model/layers.9/input_layernorm/Cast_output_0 Split splitncnn_298 1 3 /model/layers.9/input_layernorm/Cast_output_0 /model/layers.9/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.9/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.9/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.9/input_layernorm/Pow 1 1 /model/layers.9/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.9/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.9/input_layernorm/ReduceMean 2 1 /model/layers.9/input_layernorm/Pow_output_0 /model/layers.9/input_layernorm/Constant_1_output_0 /model/layers.9/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.9/input_layernorm/Add 1 1 /model/layers.9/input_layernorm/ReduceMean_output_0 /model/layers.9/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.9/input_layernorm/Sqrt 1 1 /model/layers.9/input_layernorm/Add_output_0 /model/layers.9/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.9/input_layernorm/Div 1 1 /model/layers.9/input_layernorm/Sqrt_output_0 /model/layers.9/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.9/input_layernorm/Mul 2 1 /model/layers.9/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.9/input_layernorm/Div_output_0 /model/layers.9/input_layernorm/Mul_output_0 0=2 Cast /model/layers.9/input_layernorm/Cast_1 1 1 /model/layers.9/input_layernorm/Mul_output_0 /model/layers.9/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.9/input_layernorm/Mul_1 2 1 model.layers.9.input_layernorm.weight /model/layers.9/input_layernorm/Cast_1_output_0 /model/layers.9/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_299 1 3 /model/layers.9/input_layernorm/Mul_1_output_0 /model/layers.9/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.9/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.9/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.9/self_attn/Shape 1 1 /model/layers.9/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.9/self_attn/Shape_output_0 Shape /model/layers.9/self_attn/Shape_1 1 1 /model/layers.9/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.9/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.9/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.9/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.9/input_layernorm/Mul_1_output_0_quantized /model/layers.9/input_layernorm/Mul_1_output_0_scale /model/layers.9/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_300 1 3 /model/layers.9/input_layernorm/Mul_1_output_0_quantized /model/layers.9/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.9/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.9/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_301 1 3 /model/layers.9/input_layernorm/Mul_1_output_0_scale /model/layers.9/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.9/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.9/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_302 1 3 /model/layers.9/input_layernorm/Mul_1_output_0_zero_point /model/layers.9/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.9/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.9/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.9/self_attn/Gather 1 1 /model/layers.9/self_attn/Shape_output_0 /model/layers.9/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_303 1 4 /model/layers.9/self_attn/Gather_output_0 /model/layers.9/self_attn/Gather_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_output_0_splitncnn_1 /model/layers.9/self_attn/Gather_output_0_splitncnn_2 /model/layers.9/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.9/self_attn/Gather_1 1 1 /model/layers.9/self_attn/Shape_1_output_0 /model/layers.9/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_304 1 4 /model/layers.9/self_attn/Gather_1_output_0 /model/layers.9/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.9/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.9/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.9/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.9/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.9/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.830093e-03 BinaryOp /model/layers.9/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.9/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.9/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.045399e-03 BinaryOp /model/layers.9/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.9/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.9/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=8.612205e-04 MatMulInteger /model/layers.9/self_attn/q_proj/MatMul_quant 4 1 /model/layers.9/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7505_quantized /model/layers.9/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7505_zero_point /model/layers.9/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.9/self_attn/k_proj/MatMul_quant 4 1 /model/layers.9/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7512_quantized /model/layers.9/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7512_zero_point /model/layers.9/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.9/self_attn/v_proj/MatMul_quant 4 1 /model/layers.9/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7513_quantized /model/layers.9/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7513_zero_point /model/layers.9/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.9/self_attn/Unsqueeze 2 1 /model/layers.9/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_3008 /model/layers.9/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_2 2 1 /model/layers.9/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_3017 /model/layers.9/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_4 2 1 /model/layers.9/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_3026 /model/layers.9/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_29 2 1 /model/layers.9/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_3231 /model/layers.9/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_1 2 1 /model/layers.9/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_3010 /model/layers.9/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_3 2 1 /model/layers.9/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_3019 /model/layers.9/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_5 2 1 /model/layers.9/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_3028 /model/layers.9/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_30 2 1 /model/layers.9/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_3233 /model/layers.9/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.9/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.9/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.9/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.9/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.9/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.9/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.9/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.9/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.9/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.9/self_attn/Concat 4 1 /model/layers.9/self_attn/Unsqueeze_output_0 /model/layers.9/self_attn/Unsqueeze_1_output_0 /model/layers.9/self_attn/Constant_2_output_0 /model/layers.9/self_attn/Constant_3_output_0 /model/layers.9/self_attn/Concat_output_0 0=0 Concat /model/layers.9/self_attn/Concat_1 4 1 /model/layers.9/self_attn/Unsqueeze_2_output_0 /model/layers.9/self_attn/Unsqueeze_3_output_0 /model/layers.9/self_attn/Constant_4_output_0 /model/layers.9/self_attn/Constant_5_output_0 /model/layers.9/self_attn/Concat_1_output_0 0=0 Concat /model/layers.9/self_attn/Concat_2 4 1 /model/layers.9/self_attn/Unsqueeze_4_output_0 /model/layers.9/self_attn/Unsqueeze_5_output_0 /model/layers.9/self_attn/Constant_6_output_0 /model/layers.9/self_attn/Constant_7_output_0 /model/layers.9/self_attn/Concat_2_output_0 0=0 Concat /model/layers.9/self_attn/Concat_11 3 1 /model/layers.9/self_attn/Unsqueeze_29_output_0 /model/layers.9/self_attn/Unsqueeze_30_output_0 /model/layers.9/self_attn/Constant_64_output_0 /model/layers.9/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.9/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.9/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.9/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.9/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.9/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.9/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.9/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.9/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.9/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.9/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.9/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.9/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.9/self_attn/q_proj/Add 2 1 model.layers.9.self_attn.q_proj.bias /model/layers.9/self_attn/q_proj/MatMul_output_0 /model/layers.9/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.9/self_attn/k_proj/Add 2 1 model.layers.9.self_attn.k_proj.bias /model/layers.9/self_attn/k_proj/MatMul_output_0 /model/layers.9/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.9/self_attn/v_proj/Add 2 1 model.layers.9.self_attn.v_proj.bias /model/layers.9/self_attn/v_proj/MatMul_output_0 /model/layers.9/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.9/self_attn/Reshape 2 1 /model/layers.9/self_attn/q_proj/Add_output_0 /model/layers.9/self_attn/Concat_output_0 /model/layers.9/self_attn/Reshape_output_0 Reshape /model/layers.9/self_attn/Reshape_1 2 1 /model/layers.9/self_attn/k_proj/Add_output_0 /model/layers.9/self_attn/Concat_1_output_0 /model/layers.9/self_attn/Reshape_1_output_0 Reshape /model/layers.9/self_attn/Reshape_2 2 1 /model/layers.9/self_attn/v_proj/Add_output_0 /model/layers.9/self_attn/Concat_2_output_0 /model/layers.9/self_attn/Reshape_2_output_0 Permute /model/layers.9/self_attn/Transpose 1 1 /model/layers.9/self_attn/Reshape_output_0 /model/layers.9/self_attn/Transpose_output_0 0=2 Split splitncnn_305 1 4 /model/layers.9/self_attn/Transpose_output_0 /model/layers.9/self_attn/Transpose_output_0_splitncnn_0 /model/layers.9/self_attn/Transpose_output_0_splitncnn_1 /model/layers.9/self_attn/Transpose_output_0_splitncnn_2 /model/layers.9/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.9/self_attn/Transpose_1 1 1 /model/layers.9/self_attn/Reshape_1_output_0 /model/layers.9/self_attn/Transpose_1_output_0 0=2 Split splitncnn_306 1 4 /model/layers.9/self_attn/Transpose_1_output_0 /model/layers.9/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.9/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.9/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.9/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.9/self_attn/Transpose_2 1 1 /model/layers.9/self_attn/Reshape_2_output_0 /model/layers.9/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.9/self_attn/Mul 2 1 /model/layers.9/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_29 /model/layers.9/self_attn/Mul_output_0 0=2 Shape /model/layers.9/self_attn/Shape_2 1 1 /model/layers.9/self_attn/Transpose_output_0_splitncnn_2 /model/layers.9/self_attn/Shape_2_output_0 BinaryOp /model/layers.9/self_attn/Mul_2 2 1 /model/layers.9/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_28 /model/layers.9/self_attn/Mul_2_output_0 0=2 Shape /model/layers.9/self_attn/Shape_3 1 1 /model/layers.9/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.9/self_attn/Shape_3_output_0 Concat /model/layers.9/self_attn/Concat_6 2 1 past_key_values.9.value /model/layers.9/self_attn/Transpose_2_output_0 present.9.value 0=-2 Split splitncnn_307 1 5 present.9.value present.9.value_splitncnn_0 present.9.value_splitncnn_1 present.9.value_splitncnn_2 present.9.value_splitncnn_3 present.9.value_splitncnn_4 Crop /model/layers.9/self_attn/Gather_2 1 1 /model/layers.9/self_attn/Shape_2_output_0 /model/layers.9/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.9/self_attn/Gather_3 1 1 /model/layers.9/self_attn/Shape_3_output_0 /model/layers.9/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.9/self_attn/Shape_9 1 1 present.9.value_splitncnn_4 /model/layers.9/self_attn/Shape_9_output_0 Shape /model/layers.9/self_attn/Shape_10 1 1 present.9.value_splitncnn_3 /model/layers.9/self_attn/Shape_10_output_0 Shape /model/layers.9/self_attn/Shape_11 1 1 present.9.value_splitncnn_2 /model/layers.9/self_attn/Shape_11_output_0 Shape /model/layers.9/self_attn/Shape_12 1 1 present.9.value_splitncnn_1 /model/layers.9/self_attn/Shape_12_output_0 ExpandDims /model/layers.9/self_attn/Unsqueeze_19 2 1 present.9.value_splitncnn_0 /model/layers.9/self_attn/Constant_45_output_0 /model/layers.9/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.9/self_attn/Div 2 1 /model/layers.9/self_attn/Gather_2_output_0 /model/layers.9/self_attn/Constant_9_output_0 /model/layers.9/self_attn/Div_output_0 0=3 BinaryOp /model/layers.9/self_attn/Div_1 2 1 /model/layers.9/self_attn/Gather_3_output_0 /model/layers.9/self_attn/Constant_19_output_0 /model/layers.9/self_attn/Div_1_output_0 0=3 Crop /model/layers.9/self_attn/Gather_8 1 1 /model/layers.9/self_attn/Shape_9_output_0 /model/layers.9/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_308 1 2 /model/layers.9/self_attn/Gather_8_output_0 /model/layers.9/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.9/self_attn/Gather_9 1 1 /model/layers.9/self_attn/Shape_10_output_0 /model/layers.9/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_309 1 2 /model/layers.9/self_attn/Gather_9_output_0 /model/layers.9/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.9/self_attn/Gather_10 1 1 /model/layers.9/self_attn/Shape_11_output_0 /model/layers.9/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_310 1 2 /model/layers.9/self_attn/Gather_10_output_0 /model/layers.9/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.9/self_attn/Gather_11 1 1 /model/layers.9/self_attn/Shape_12_output_0 /model/layers.9/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_311 1 2 /model/layers.9/self_attn/Gather_11_output_0 /model/layers.9/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.9/self_attn/Cast 1 1 /model/layers.9/self_attn/Div_output_0 /model/layers.9/self_attn/Cast_output_0 Cast /model/layers.9/self_attn/Cast_2 1 1 /model/layers.9/self_attn/Div_1_output_0 /model/layers.9/self_attn/Cast_2_output_0 ExpandDims /model/layers.9/self_attn/Unsqueeze_20 2 1 /model/layers.9/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_46_output_0 /model/layers.9/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_24 2 1 /model/layers.9/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_3191 /model/layers.9/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_21 2 1 /model/layers.9/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_47_output_0 /model/layers.9/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.9/self_attn/Mul_7 2 1 /model/layers.9/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.9/self_attn/Constant_53_output_0 /model/layers.9/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.9/self_attn/Unsqueeze_22 2 1 /model/layers.9/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_49_output_0 /model/layers.9/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_26 2 1 /model/layers.9/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_3195 /model/layers.9/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_23 2 1 /model/layers.9/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_50_output_0 /model/layers.9/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_27 2 1 /model/layers.9/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_3197 /model/layers.9/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.9/self_attn/Cast_1 1 1 /model/layers.9/self_attn/Cast_output_0 /model/layers.9/self_attn/Cast_1_output_0 Split splitncnn_312 1 2 /model/layers.9/self_attn/Cast_1_output_0 /model/layers.9/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.9/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.9/self_attn/Cast_3 1 1 /model/layers.9/self_attn/Cast_2_output_0 /model/layers.9/self_attn/Cast_3_output_0 Split splitncnn_313 1 2 /model/layers.9/self_attn/Cast_3_output_0 /model/layers.9/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.9/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.9/self_attn/Unsqueeze_25 2 1 /model/layers.9/self_attn/Mul_7_output_0 onnx::Unsqueeze_3193 /model/layers.9/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.9/self_attn/Concat_9 5 1 /model/layers.9/self_attn/Unsqueeze_20_output_0 /model/layers.9/self_attn/Unsqueeze_21_output_0 /model/layers.9/self_attn/Constant_48_output_0 /model/layers.9/self_attn/Unsqueeze_22_output_0 /model/layers.9/self_attn/Unsqueeze_23_output_0 /model/layers.9/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_6 2 1 /model/layers.9/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_12_output_0 /model/layers.9/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_7 2 1 /model/layers.9/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.9/self_attn/Constant_15_output_0 /model/layers.9/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_8 2 1 /model/layers.9/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_22_output_0 /model/layers.9/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_9 2 1 /model/layers.9/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.9/self_attn/Constant_25_output_0 /model/layers.9/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.9/self_attn/Concat_10 4 1 /model/layers.9/self_attn/Unsqueeze_24_output_0 /model/layers.9/self_attn/Unsqueeze_25_output_0 /model/layers.9/self_attn/Unsqueeze_26_output_0 /model/layers.9/self_attn/Unsqueeze_27_output_0 /model/layers.9/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.9/self_attn/Reshape_5 1 1 /model/layers.9/self_attn/Concat_9_output_0 /model/layers.9/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_314 1 3 /model/layers.9/self_attn/Reshape_5_output_0 /model/layers.9/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.9/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.9/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.9/self_attn/Slice 2 1 /model/layers.9/self_attn/Transpose_output_0_splitncnn_1 /model/layers.9/self_attn/Unsqueeze_6_output_0 /model/layers.9/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.9/self_attn/Slice_1 2 1 /model/layers.9/self_attn/Transpose_output_0_splitncnn_0 /model/layers.9/self_attn/Unsqueeze_7_output_0 /model/layers.9/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.9/self_attn/Slice_2 2 1 /model/layers.9/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.9/self_attn/Unsqueeze_8_output_0 /model/layers.9/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.9/self_attn/Slice_3 2 1 /model/layers.9/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.9/self_attn/Unsqueeze_9_output_0 /model/layers.9/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.9/self_attn/Shape_13 1 1 /model/layers.9/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.9/self_attn/Shape_13_output_0 UnaryOp /model/layers.9/self_attn/Neg 1 1 /model/layers.9/self_attn/Slice_1_output_0 /model/layers.9/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.9/self_attn/Neg_1 1 1 /model/layers.9/self_attn/Slice_3_output_0 /model/layers.9/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.9/self_attn/ConstantOfShape_1 1 1 /model/layers.9/self_attn/Shape_13_output_0 /model/layers.9/self_attn/ConstantOfShape_1_output_0 Split splitncnn_315 1 2 /model/layers.9/self_attn/ConstantOfShape_1_output_0 /model/layers.9/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.9/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.9/self_attn/Concat_3 2 1 /model/layers.9/self_attn/Neg_output_0 /model/layers.9/self_attn/Slice_output_0 /model/layers.9/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.9/self_attn/Concat_4 2 1 /model/layers.9/self_attn/Neg_1_output_0 /model/layers.9/self_attn/Slice_2_output_0 /model/layers.9/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.9/self_attn/Mul_6 2 1 /model/layers.9/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_52_output_0 /model/layers.9/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.9/self_attn/Mul_1 2 1 /model/layers.9/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_29 /model/layers.9/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.9/self_attn/Mul_3 2 1 /model/layers.9/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_28 /model/layers.9/self_attn/Mul_3_output_0 0=2 Equal /model/layers.9/self_attn/Equal_1 2 1 /model/layers.9/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.9/self_attn/Mul_6_output_0 /model/layers.9/self_attn/Equal_1_output_0 BinaryOp /model/layers.9/self_attn/Add 2 1 /model/layers.9/self_attn/Mul_output_0 /model/layers.9/self_attn/Mul_1_output_0 /model/layers.9/self_attn/Add_output_0 0=0 BinaryOp /model/layers.9/self_attn/Add_1 2 1 /model/layers.9/self_attn/Mul_2_output_0 /model/layers.9/self_attn/Mul_3_output_0 /model/layers.9/self_attn/Add_1_output_0 0=0 Where /model/layers.9/self_attn/Where_1 3 1 /model/layers.9/self_attn/Equal_1_output_0 /model/layers.9/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.9/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.9/self_attn/Where_1_output_0 BinaryOp /model/layers.9/self_attn/Mul_8 1 1 /model/layers.9/self_attn/Add_output_0 /model/layers.9/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.9/self_attn/Concat_5 2 1 past_key_values.9.key /model/layers.9/self_attn/Add_1_output_0 present.9.key 0=-2 Split splitncnn_316 1 5 present.9.key present.9.key_splitncnn_0 present.9.key_splitncnn_1 present.9.key_splitncnn_2 present.9.key_splitncnn_3 present.9.key_splitncnn_4 Expand /model/layers.9/self_attn/Expand_1 2 1 /model/layers.9/self_attn/Unsqueeze_19_output_0 /model/layers.9/self_attn/Where_1_output_0 /model/layers.9/self_attn/Expand_1_output_0 Shape /model/layers.9/self_attn/Shape_4 1 1 present.9.key_splitncnn_4 /model/layers.9/self_attn/Shape_4_output_0 Shape /model/layers.9/self_attn/Shape_5 1 1 present.9.key_splitncnn_3 /model/layers.9/self_attn/Shape_5_output_0 Shape /model/layers.9/self_attn/Shape_6 1 1 present.9.key_splitncnn_2 /model/layers.9/self_attn/Shape_6_output_0 Shape /model/layers.9/self_attn/Shape_7 1 1 present.9.key_splitncnn_1 /model/layers.9/self_attn/Shape_7_output_0 ExpandDims /model/layers.9/self_attn/Unsqueeze_10 2 1 present.9.key_splitncnn_0 /model/layers.9/self_attn/Constant_32_output_0 /model/layers.9/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.9/self_attn/Reshape_6 2 1 /model/layers.9/self_attn/Expand_1_output_0 /model/layers.9/self_attn/Concat_10_output_0 /model/layers.9/self_attn/Reshape_6_output_0 Crop /model/layers.9/self_attn/Gather_4 1 1 /model/layers.9/self_attn/Shape_4_output_0 /model/layers.9/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_317 1 2 /model/layers.9/self_attn/Gather_4_output_0 /model/layers.9/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.9/self_attn/Gather_5 1 1 /model/layers.9/self_attn/Shape_5_output_0 /model/layers.9/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_318 1 2 /model/layers.9/self_attn/Gather_5_output_0 /model/layers.9/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.9/self_attn/Gather_6 1 1 /model/layers.9/self_attn/Shape_6_output_0 /model/layers.9/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_319 1 2 /model/layers.9/self_attn/Gather_6_output_0 /model/layers.9/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.9/self_attn/Gather_7 1 1 /model/layers.9/self_attn/Shape_7_output_0 /model/layers.9/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_320 1 2 /model/layers.9/self_attn/Gather_7_output_0 /model/layers.9/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.9/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.9/self_attn/Unsqueeze_11 2 1 /model/layers.9/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_33_output_0 /model/layers.9/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_15 2 1 /model/layers.9/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_3145 /model/layers.9/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_12 2 1 /model/layers.9/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_34_output_0 /model/layers.9/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.9/self_attn/Mul_5 2 1 /model/layers.9/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.9/self_attn/Constant_40_output_0 /model/layers.9/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.9/self_attn/Unsqueeze_13 2 1 /model/layers.9/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_36_output_0 /model/layers.9/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_17 2 1 /model/layers.9/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_3149 /model/layers.9/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_14 2 1 /model/layers.9/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_37_output_0 /model/layers.9/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_18 2 1 /model/layers.9/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_3151 /model/layers.9/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.9/self_attn/Unsqueeze_16 2 1 /model/layers.9/self_attn/Mul_5_output_0 onnx::Unsqueeze_3147 /model/layers.9/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.9/self_attn/Concat_7 5 1 /model/layers.9/self_attn/Unsqueeze_11_output_0 /model/layers.9/self_attn/Unsqueeze_12_output_0 /model/layers.9/self_attn/Constant_35_output_0 /model/layers.9/self_attn/Unsqueeze_13_output_0 /model/layers.9/self_attn/Unsqueeze_14_output_0 /model/layers.9/self_attn/Concat_7_output_0 0=0 Concat /model/layers.9/self_attn/Concat_8 4 1 /model/layers.9/self_attn/Unsqueeze_15_output_0 /model/layers.9/self_attn/Unsqueeze_16_output_0 /model/layers.9/self_attn/Unsqueeze_17_output_0 /model/layers.9/self_attn/Unsqueeze_18_output_0 /model/layers.9/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.9/self_attn/Reshape_3 1 1 /model/layers.9/self_attn/Concat_7_output_0 /model/layers.9/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_321 1 3 /model/layers.9/self_attn/Reshape_3_output_0 /model/layers.9/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.9/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.9/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.9/self_attn/Shape_8 1 1 /model/layers.9/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.9/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.9/self_attn/ConstantOfShape 1 1 /model/layers.9/self_attn/Shape_8_output_0 /model/layers.9/self_attn/ConstantOfShape_output_0 Split splitncnn_322 1 2 /model/layers.9/self_attn/ConstantOfShape_output_0 /model/layers.9/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.9/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.9/self_attn/Mul_4 2 1 /model/layers.9/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.9/self_attn/Constant_39_output_0 /model/layers.9/self_attn/Mul_4_output_0 0=2 Equal /model/layers.9/self_attn/Equal 2 1 /model/layers.9/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.9/self_attn/Mul_4_output_0 /model/layers.9/self_attn/Equal_output_0 Where /model/layers.9/self_attn/Where 3 1 /model/layers.9/self_attn/Equal_output_0 /model/layers.9/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.9/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.9/self_attn/Where_output_0 Expand /model/layers.9/self_attn/Expand 2 1 /model/layers.9/self_attn/Unsqueeze_10_output_0 /model/layers.9/self_attn/Where_output_0 /model/layers.9/self_attn/Expand_output_0 Reshape /model/layers.9/self_attn/Reshape_4 2 1 /model/layers.9/self_attn/Expand_output_0 /model/layers.9/self_attn/Concat_8_output_0 /model/layers.9/self_attn/Reshape_4_output_0 Split splitncnn_323 1 2 /model/layers.9/self_attn/Reshape_4_output_0 /model/layers.9/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.9/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.9/self_attn/Shape_14 1 1 /model/layers.9/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.9/self_attn/Shape_14_output_0 Permute /model/layers.9/self_attn/Transpose_3 1 1 /model/layers.9/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.9/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.9/self_attn/Gather_12 1 1 /model/layers.9/self_attn/Shape_14_output_0 /model/layers.9/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.9/self_attn/Mul_9 1 1 /model/layers.9/self_attn/Transpose_3_output_0 /model/layers.9/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.9/self_attn/Unsqueeze_28 2 1 /model/layers.9/self_attn/Gather_12_output_0 /model/layers.9/self_attn/Constant_57_output_0 /model/layers.9/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.9/self_attn/MatMul 2 1 /model/layers.9/self_attn/Mul_8_output_0 /model/layers.9/self_attn/Mul_9_output_0 /model/layers.9/self_attn/MatMul_output_0 Crop /model/layers.9/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_14 /model/layers.9/self_attn/Unsqueeze_28_output_0 /model/layers.9/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.9/self_attn/Where_2 3 1 /model/layers.9/self_attn/Slice_4_output_0 /model/layers.9/self_attn/Constant_61_output_0 /model/layers.9/self_attn/Constant_62_output_0 /model/layers.9/self_attn/Where_2_output_0 BinaryOp /model/layers.9/self_attn/Add_2 2 1 /model/layers.9/self_attn/MatMul_output_0 /model/layers.9/self_attn/Where_2_output_0 /model/layers.9/self_attn/Add_2_output_0 0=0 Softmax /model/layers.9/self_attn/Softmax 1 1 /model/layers.9/self_attn/Add_2_output_0 /model/layers.9/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_324 1 2 /model/layers.9/self_attn/Softmax_output_0 /model/layers.9/self_attn/Softmax_output_0_splitncnn_0 /model/layers.9/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.9/self_attn/IsNaN 1 1 /model/layers.9/self_attn/Softmax_output_0_splitncnn_1 /model/layers.9/self_attn/IsNaN_output_0 Where /model/layers.9/self_attn/Where_3 3 1 /model/layers.9/self_attn/IsNaN_output_0 /model/layers.9/self_attn/Constant_63_output_0 /model/layers.9/self_attn/Softmax_output_0_splitncnn_0 /model/layers.9/self_attn/Where_3_output_0 Gemm /model/layers.9/self_attn/MatMul_1 2 1 /model/layers.9/self_attn/Where_3_output_0 /model/layers.9/self_attn/Reshape_6_output_0 /model/layers.9/self_attn/MatMul_1_output_0 Permute /model/layers.9/self_attn/Transpose_4 1 1 /model/layers.9/self_attn/MatMul_1_output_0 /model/layers.9/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.9/self_attn/Reshape_7 2 1 /model/layers.9/self_attn/Transpose_4_output_0 /model/layers.9/self_attn/Concat_11_output_0 /model/layers.9/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.9/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.9/self_attn/Reshape_7_output_0 /model/layers.9/self_attn/Reshape_7_output_0_quantized /model/layers.9/self_attn/Reshape_7_output_0_scale /model/layers.9/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.9/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.9/self_attn/Reshape_7_output_0_scale /model/layers.9/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.629429e-03 MatMulInteger /model/layers.9/self_attn/o_proj/MatMul_quant 4 1 /model/layers.9/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7536_quantized /model/layers.9/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7536_zero_point /model/layers.9/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.9/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.9/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.9/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.9/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.9/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.9/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.9/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.9/Add 2 1 /model/layers.9/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.9/self_attn/o_proj/MatMul_output_0 /model/layers.9/Add_output_0 0=0 Cast /model/layers.9/post_attention_layernorm/Cast 1 1 /model/layers.9/Add_output_0 /model/layers.9/post_attention_layernorm/Cast_output_0 Split splitncnn_325 1 3 /model/layers.9/post_attention_layernorm/Cast_output_0 /model/layers.9/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.9/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.9/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.9/post_attention_layernorm/Pow 1 1 /model/layers.9/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.9/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.9/post_attention_layernorm/ReduceMean 2 1 /model/layers.9/post_attention_layernorm/Pow_output_0 /model/layers.9/post_attention_layernorm/Constant_1_output_0 /model/layers.9/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.9/post_attention_layernorm/Add 1 1 /model/layers.9/post_attention_layernorm/ReduceMean_output_0 /model/layers.9/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.9/post_attention_layernorm/Sqrt 1 1 /model/layers.9/post_attention_layernorm/Add_output_0 /model/layers.9/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.9/post_attention_layernorm/Div 1 1 /model/layers.9/post_attention_layernorm/Sqrt_output_0 /model/layers.9/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.9/post_attention_layernorm/Mul 2 1 /model/layers.9/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.9/post_attention_layernorm/Div_output_0 /model/layers.9/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.9/post_attention_layernorm/Cast_1 1 1 /model/layers.9/post_attention_layernorm/Mul_output_0 /model/layers.9/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.9/post_attention_layernorm/Mul_1 2 1 model.layers.9.post_attention_layernorm.weight /model/layers.9/post_attention_layernorm/Cast_1_output_0 /model/layers.9/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.9/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.9/post_attention_layernorm/Mul_1_output_0 /model/layers.9/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.9/post_attention_layernorm/Mul_1_output_0_scale /model/layers.9/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_326 1 2 /model/layers.9/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.9/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.9/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_327 1 2 /model/layers.9/post_attention_layernorm/Mul_1_output_0_scale /model/layers.9/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.9/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_328 1 2 /model/layers.9/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.9/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.9/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.9/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.9/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.9/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.521777e-03 BinaryOp /model/layers.9/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.9/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.9/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.321850e-03 MatMulInteger /model/layers.9/mlp/gate_proj/MatMul_quant 4 1 /model/layers.9/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7537_quantized /model/layers.9/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7537_zero_point /model/layers.9/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.9/mlp/up_proj/MatMul_quant 4 1 /model/layers.9/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7538_quantized /model/layers.9/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7538_zero_point /model/layers.9/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.9/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.9/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.9/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.9/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.9/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.9/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.9/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.9/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.9/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.9/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.9/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.9/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.9/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.9/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.9/mlp/act_fn/Mul 1 1 /model/layers.9/mlp/gate_proj/MatMul_output_0 /model/layers.9/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.9/mlp/Mul 2 1 /model/layers.9/mlp/act_fn/Mul_output_0 /model/layers.9/mlp/up_proj/MatMul_output_0 /model/layers.9/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.9/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.9/mlp/Mul_output_0 /model/layers.9/mlp/Mul_output_0_quantized /model/layers.9/mlp/Mul_output_0_scale /model/layers.9/mlp/Mul_output_0_zero_point BinaryOp /model/layers.9/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.9/mlp/Mul_output_0_scale /model/layers.9/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.675197e-03 MatMulInteger /model/layers.9/mlp/down_proj/MatMul_quant 4 1 /model/layers.9/mlp/Mul_output_0_quantized onnx::MatMul_7539_quantized /model/layers.9/mlp/Mul_output_0_zero_point onnx::MatMul_7539_zero_point /model/layers.9/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.9/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.9/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.9/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.9/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.9/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.9/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.9/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.9/Add_1 2 1 /model/layers.9/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.9/mlp/down_proj/MatMul_output_0 /model/layers.9/Add_1_output_0 0=0 Cast /model/layers.10/input_layernorm/Cast 1 1 /model/layers.9/Add_1_output_0 /model/layers.10/input_layernorm/Cast_output_0 Split splitncnn_329 1 3 /model/layers.10/input_layernorm/Cast_output_0 /model/layers.10/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.10/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.10/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.10/input_layernorm/Pow 1 1 /model/layers.10/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.10/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.10/input_layernorm/ReduceMean 2 1 /model/layers.10/input_layernorm/Pow_output_0 /model/layers.10/input_layernorm/Constant_1_output_0 /model/layers.10/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.10/input_layernorm/Add 1 1 /model/layers.10/input_layernorm/ReduceMean_output_0 /model/layers.10/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.10/input_layernorm/Sqrt 1 1 /model/layers.10/input_layernorm/Add_output_0 /model/layers.10/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.10/input_layernorm/Div 1 1 /model/layers.10/input_layernorm/Sqrt_output_0 /model/layers.10/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.10/input_layernorm/Mul 2 1 /model/layers.10/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.10/input_layernorm/Div_output_0 /model/layers.10/input_layernorm/Mul_output_0 0=2 Cast /model/layers.10/input_layernorm/Cast_1 1 1 /model/layers.10/input_layernorm/Mul_output_0 /model/layers.10/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.10/input_layernorm/Mul_1 2 1 model.layers.10.input_layernorm.weight /model/layers.10/input_layernorm/Cast_1_output_0 /model/layers.10/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_330 1 3 /model/layers.10/input_layernorm/Mul_1_output_0 /model/layers.10/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.10/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.10/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.10/self_attn/Shape 1 1 /model/layers.10/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.10/self_attn/Shape_output_0 Shape /model/layers.10/self_attn/Shape_1 1 1 /model/layers.10/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.10/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.10/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.10/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.10/input_layernorm/Mul_1_output_0_quantized /model/layers.10/input_layernorm/Mul_1_output_0_scale /model/layers.10/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_331 1 3 /model/layers.10/input_layernorm/Mul_1_output_0_quantized /model/layers.10/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.10/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.10/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_332 1 3 /model/layers.10/input_layernorm/Mul_1_output_0_scale /model/layers.10/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.10/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.10/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_333 1 3 /model/layers.10/input_layernorm/Mul_1_output_0_zero_point /model/layers.10/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.10/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.10/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.10/self_attn/Gather 1 1 /model/layers.10/self_attn/Shape_output_0 /model/layers.10/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_334 1 4 /model/layers.10/self_attn/Gather_output_0 /model/layers.10/self_attn/Gather_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_output_0_splitncnn_1 /model/layers.10/self_attn/Gather_output_0_splitncnn_2 /model/layers.10/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.10/self_attn/Gather_1 1 1 /model/layers.10/self_attn/Shape_1_output_0 /model/layers.10/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_335 1 4 /model/layers.10/self_attn/Gather_1_output_0 /model/layers.10/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.10/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.10/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.10/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.10/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.10/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.460630e-03 BinaryOp /model/layers.10/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.10/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.10/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.514825e-03 BinaryOp /model/layers.10/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.10/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.10/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.045768e-03 MatMulInteger /model/layers.10/self_attn/q_proj/MatMul_quant 4 1 /model/layers.10/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7540_quantized /model/layers.10/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7540_zero_point /model/layers.10/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.10/self_attn/k_proj/MatMul_quant 4 1 /model/layers.10/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7547_quantized /model/layers.10/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7547_zero_point /model/layers.10/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.10/self_attn/v_proj/MatMul_quant 4 1 /model/layers.10/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7548_quantized /model/layers.10/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7548_zero_point /model/layers.10/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.10/self_attn/Unsqueeze 2 1 /model/layers.10/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_3287 /model/layers.10/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_2 2 1 /model/layers.10/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_3296 /model/layers.10/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_4 2 1 /model/layers.10/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_3305 /model/layers.10/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_29 2 1 /model/layers.10/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_3510 /model/layers.10/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_1 2 1 /model/layers.10/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_3289 /model/layers.10/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_3 2 1 /model/layers.10/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_3298 /model/layers.10/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_5 2 1 /model/layers.10/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_3307 /model/layers.10/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_30 2 1 /model/layers.10/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_3512 /model/layers.10/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.10/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.10/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.10/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.10/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.10/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.10/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.10/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.10/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.10/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.10/self_attn/Concat 4 1 /model/layers.10/self_attn/Unsqueeze_output_0 /model/layers.10/self_attn/Unsqueeze_1_output_0 /model/layers.10/self_attn/Constant_2_output_0 /model/layers.10/self_attn/Constant_3_output_0 /model/layers.10/self_attn/Concat_output_0 0=0 Concat /model/layers.10/self_attn/Concat_1 4 1 /model/layers.10/self_attn/Unsqueeze_2_output_0 /model/layers.10/self_attn/Unsqueeze_3_output_0 /model/layers.10/self_attn/Constant_4_output_0 /model/layers.10/self_attn/Constant_5_output_0 /model/layers.10/self_attn/Concat_1_output_0 0=0 Concat /model/layers.10/self_attn/Concat_2 4 1 /model/layers.10/self_attn/Unsqueeze_4_output_0 /model/layers.10/self_attn/Unsqueeze_5_output_0 /model/layers.10/self_attn/Constant_6_output_0 /model/layers.10/self_attn/Constant_7_output_0 /model/layers.10/self_attn/Concat_2_output_0 0=0 Concat /model/layers.10/self_attn/Concat_11 3 1 /model/layers.10/self_attn/Unsqueeze_29_output_0 /model/layers.10/self_attn/Unsqueeze_30_output_0 /model/layers.10/self_attn/Constant_64_output_0 /model/layers.10/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.10/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.10/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.10/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.10/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.10/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.10/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.10/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.10/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.10/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.10/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.10/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.10/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.10/self_attn/q_proj/Add 2 1 model.layers.10.self_attn.q_proj.bias /model/layers.10/self_attn/q_proj/MatMul_output_0 /model/layers.10/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.10/self_attn/k_proj/Add 2 1 model.layers.10.self_attn.k_proj.bias /model/layers.10/self_attn/k_proj/MatMul_output_0 /model/layers.10/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.10/self_attn/v_proj/Add 2 1 model.layers.10.self_attn.v_proj.bias /model/layers.10/self_attn/v_proj/MatMul_output_0 /model/layers.10/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.10/self_attn/Reshape 2 1 /model/layers.10/self_attn/q_proj/Add_output_0 /model/layers.10/self_attn/Concat_output_0 /model/layers.10/self_attn/Reshape_output_0 Reshape /model/layers.10/self_attn/Reshape_1 2 1 /model/layers.10/self_attn/k_proj/Add_output_0 /model/layers.10/self_attn/Concat_1_output_0 /model/layers.10/self_attn/Reshape_1_output_0 Reshape /model/layers.10/self_attn/Reshape_2 2 1 /model/layers.10/self_attn/v_proj/Add_output_0 /model/layers.10/self_attn/Concat_2_output_0 /model/layers.10/self_attn/Reshape_2_output_0 Permute /model/layers.10/self_attn/Transpose 1 1 /model/layers.10/self_attn/Reshape_output_0 /model/layers.10/self_attn/Transpose_output_0 0=2 Split splitncnn_336 1 4 /model/layers.10/self_attn/Transpose_output_0 /model/layers.10/self_attn/Transpose_output_0_splitncnn_0 /model/layers.10/self_attn/Transpose_output_0_splitncnn_1 /model/layers.10/self_attn/Transpose_output_0_splitncnn_2 /model/layers.10/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.10/self_attn/Transpose_1 1 1 /model/layers.10/self_attn/Reshape_1_output_0 /model/layers.10/self_attn/Transpose_1_output_0 0=2 Split splitncnn_337 1 4 /model/layers.10/self_attn/Transpose_1_output_0 /model/layers.10/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.10/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.10/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.10/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.10/self_attn/Transpose_2 1 1 /model/layers.10/self_attn/Reshape_2_output_0 /model/layers.10/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.10/self_attn/Mul 2 1 /model/layers.10/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_27 /model/layers.10/self_attn/Mul_output_0 0=2 Shape /model/layers.10/self_attn/Shape_2 1 1 /model/layers.10/self_attn/Transpose_output_0_splitncnn_2 /model/layers.10/self_attn/Shape_2_output_0 BinaryOp /model/layers.10/self_attn/Mul_2 2 1 /model/layers.10/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_26 /model/layers.10/self_attn/Mul_2_output_0 0=2 Shape /model/layers.10/self_attn/Shape_3 1 1 /model/layers.10/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.10/self_attn/Shape_3_output_0 Concat /model/layers.10/self_attn/Concat_6 2 1 past_key_values.10.value /model/layers.10/self_attn/Transpose_2_output_0 present.10.value 0=-2 Split splitncnn_338 1 5 present.10.value present.10.value_splitncnn_0 present.10.value_splitncnn_1 present.10.value_splitncnn_2 present.10.value_splitncnn_3 present.10.value_splitncnn_4 Crop /model/layers.10/self_attn/Gather_2 1 1 /model/layers.10/self_attn/Shape_2_output_0 /model/layers.10/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.10/self_attn/Gather_3 1 1 /model/layers.10/self_attn/Shape_3_output_0 /model/layers.10/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.10/self_attn/Shape_9 1 1 present.10.value_splitncnn_4 /model/layers.10/self_attn/Shape_9_output_0 Shape /model/layers.10/self_attn/Shape_10 1 1 present.10.value_splitncnn_3 /model/layers.10/self_attn/Shape_10_output_0 Shape /model/layers.10/self_attn/Shape_11 1 1 present.10.value_splitncnn_2 /model/layers.10/self_attn/Shape_11_output_0 Shape /model/layers.10/self_attn/Shape_12 1 1 present.10.value_splitncnn_1 /model/layers.10/self_attn/Shape_12_output_0 ExpandDims /model/layers.10/self_attn/Unsqueeze_19 2 1 present.10.value_splitncnn_0 /model/layers.10/self_attn/Constant_45_output_0 /model/layers.10/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.10/self_attn/Div 2 1 /model/layers.10/self_attn/Gather_2_output_0 /model/layers.10/self_attn/Constant_9_output_0 /model/layers.10/self_attn/Div_output_0 0=3 BinaryOp /model/layers.10/self_attn/Div_1 2 1 /model/layers.10/self_attn/Gather_3_output_0 /model/layers.10/self_attn/Constant_19_output_0 /model/layers.10/self_attn/Div_1_output_0 0=3 Crop /model/layers.10/self_attn/Gather_8 1 1 /model/layers.10/self_attn/Shape_9_output_0 /model/layers.10/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_339 1 2 /model/layers.10/self_attn/Gather_8_output_0 /model/layers.10/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.10/self_attn/Gather_9 1 1 /model/layers.10/self_attn/Shape_10_output_0 /model/layers.10/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_340 1 2 /model/layers.10/self_attn/Gather_9_output_0 /model/layers.10/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.10/self_attn/Gather_10 1 1 /model/layers.10/self_attn/Shape_11_output_0 /model/layers.10/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_341 1 2 /model/layers.10/self_attn/Gather_10_output_0 /model/layers.10/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.10/self_attn/Gather_11 1 1 /model/layers.10/self_attn/Shape_12_output_0 /model/layers.10/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_342 1 2 /model/layers.10/self_attn/Gather_11_output_0 /model/layers.10/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.10/self_attn/Cast 1 1 /model/layers.10/self_attn/Div_output_0 /model/layers.10/self_attn/Cast_output_0 Cast /model/layers.10/self_attn/Cast_2 1 1 /model/layers.10/self_attn/Div_1_output_0 /model/layers.10/self_attn/Cast_2_output_0 ExpandDims /model/layers.10/self_attn/Unsqueeze_20 2 1 /model/layers.10/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_46_output_0 /model/layers.10/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_24 2 1 /model/layers.10/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_3470 /model/layers.10/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_21 2 1 /model/layers.10/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_47_output_0 /model/layers.10/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.10/self_attn/Mul_7 2 1 /model/layers.10/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.10/self_attn/Constant_53_output_0 /model/layers.10/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.10/self_attn/Unsqueeze_22 2 1 /model/layers.10/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_49_output_0 /model/layers.10/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_26 2 1 /model/layers.10/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_3474 /model/layers.10/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_23 2 1 /model/layers.10/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_50_output_0 /model/layers.10/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_27 2 1 /model/layers.10/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_3476 /model/layers.10/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.10/self_attn/Cast_1 1 1 /model/layers.10/self_attn/Cast_output_0 /model/layers.10/self_attn/Cast_1_output_0 Split splitncnn_343 1 2 /model/layers.10/self_attn/Cast_1_output_0 /model/layers.10/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.10/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.10/self_attn/Cast_3 1 1 /model/layers.10/self_attn/Cast_2_output_0 /model/layers.10/self_attn/Cast_3_output_0 Split splitncnn_344 1 2 /model/layers.10/self_attn/Cast_3_output_0 /model/layers.10/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.10/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.10/self_attn/Unsqueeze_25 2 1 /model/layers.10/self_attn/Mul_7_output_0 onnx::Unsqueeze_3472 /model/layers.10/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.10/self_attn/Concat_9 5 1 /model/layers.10/self_attn/Unsqueeze_20_output_0 /model/layers.10/self_attn/Unsqueeze_21_output_0 /model/layers.10/self_attn/Constant_48_output_0 /model/layers.10/self_attn/Unsqueeze_22_output_0 /model/layers.10/self_attn/Unsqueeze_23_output_0 /model/layers.10/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_6 2 1 /model/layers.10/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_12_output_0 /model/layers.10/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_7 2 1 /model/layers.10/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.10/self_attn/Constant_15_output_0 /model/layers.10/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_8 2 1 /model/layers.10/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_22_output_0 /model/layers.10/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_9 2 1 /model/layers.10/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.10/self_attn/Constant_25_output_0 /model/layers.10/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.10/self_attn/Concat_10 4 1 /model/layers.10/self_attn/Unsqueeze_24_output_0 /model/layers.10/self_attn/Unsqueeze_25_output_0 /model/layers.10/self_attn/Unsqueeze_26_output_0 /model/layers.10/self_attn/Unsqueeze_27_output_0 /model/layers.10/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.10/self_attn/Reshape_5 1 1 /model/layers.10/self_attn/Concat_9_output_0 /model/layers.10/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_345 1 3 /model/layers.10/self_attn/Reshape_5_output_0 /model/layers.10/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.10/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.10/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.10/self_attn/Slice 2 1 /model/layers.10/self_attn/Transpose_output_0_splitncnn_1 /model/layers.10/self_attn/Unsqueeze_6_output_0 /model/layers.10/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.10/self_attn/Slice_1 2 1 /model/layers.10/self_attn/Transpose_output_0_splitncnn_0 /model/layers.10/self_attn/Unsqueeze_7_output_0 /model/layers.10/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.10/self_attn/Slice_2 2 1 /model/layers.10/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.10/self_attn/Unsqueeze_8_output_0 /model/layers.10/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.10/self_attn/Slice_3 2 1 /model/layers.10/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.10/self_attn/Unsqueeze_9_output_0 /model/layers.10/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.10/self_attn/Shape_13 1 1 /model/layers.10/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.10/self_attn/Shape_13_output_0 UnaryOp /model/layers.10/self_attn/Neg 1 1 /model/layers.10/self_attn/Slice_1_output_0 /model/layers.10/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.10/self_attn/Neg_1 1 1 /model/layers.10/self_attn/Slice_3_output_0 /model/layers.10/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.10/self_attn/ConstantOfShape_1 1 1 /model/layers.10/self_attn/Shape_13_output_0 /model/layers.10/self_attn/ConstantOfShape_1_output_0 Split splitncnn_346 1 2 /model/layers.10/self_attn/ConstantOfShape_1_output_0 /model/layers.10/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.10/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.10/self_attn/Concat_3 2 1 /model/layers.10/self_attn/Neg_output_0 /model/layers.10/self_attn/Slice_output_0 /model/layers.10/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.10/self_attn/Concat_4 2 1 /model/layers.10/self_attn/Neg_1_output_0 /model/layers.10/self_attn/Slice_2_output_0 /model/layers.10/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.10/self_attn/Mul_6 2 1 /model/layers.10/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_52_output_0 /model/layers.10/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.10/self_attn/Mul_1 2 1 /model/layers.10/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_27 /model/layers.10/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.10/self_attn/Mul_3 2 1 /model/layers.10/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_26 /model/layers.10/self_attn/Mul_3_output_0 0=2 Equal /model/layers.10/self_attn/Equal_1 2 1 /model/layers.10/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.10/self_attn/Mul_6_output_0 /model/layers.10/self_attn/Equal_1_output_0 BinaryOp /model/layers.10/self_attn/Add 2 1 /model/layers.10/self_attn/Mul_output_0 /model/layers.10/self_attn/Mul_1_output_0 /model/layers.10/self_attn/Add_output_0 0=0 BinaryOp /model/layers.10/self_attn/Add_1 2 1 /model/layers.10/self_attn/Mul_2_output_0 /model/layers.10/self_attn/Mul_3_output_0 /model/layers.10/self_attn/Add_1_output_0 0=0 Where /model/layers.10/self_attn/Where_1 3 1 /model/layers.10/self_attn/Equal_1_output_0 /model/layers.10/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.10/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.10/self_attn/Where_1_output_0 BinaryOp /model/layers.10/self_attn/Mul_8 1 1 /model/layers.10/self_attn/Add_output_0 /model/layers.10/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.10/self_attn/Concat_5 2 1 past_key_values.10.key /model/layers.10/self_attn/Add_1_output_0 present.10.key 0=-2 Split splitncnn_347 1 5 present.10.key present.10.key_splitncnn_0 present.10.key_splitncnn_1 present.10.key_splitncnn_2 present.10.key_splitncnn_3 present.10.key_splitncnn_4 Expand /model/layers.10/self_attn/Expand_1 2 1 /model/layers.10/self_attn/Unsqueeze_19_output_0 /model/layers.10/self_attn/Where_1_output_0 /model/layers.10/self_attn/Expand_1_output_0 Shape /model/layers.10/self_attn/Shape_4 1 1 present.10.key_splitncnn_4 /model/layers.10/self_attn/Shape_4_output_0 Shape /model/layers.10/self_attn/Shape_5 1 1 present.10.key_splitncnn_3 /model/layers.10/self_attn/Shape_5_output_0 Shape /model/layers.10/self_attn/Shape_6 1 1 present.10.key_splitncnn_2 /model/layers.10/self_attn/Shape_6_output_0 Shape /model/layers.10/self_attn/Shape_7 1 1 present.10.key_splitncnn_1 /model/layers.10/self_attn/Shape_7_output_0 ExpandDims /model/layers.10/self_attn/Unsqueeze_10 2 1 present.10.key_splitncnn_0 /model/layers.10/self_attn/Constant_32_output_0 /model/layers.10/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.10/self_attn/Reshape_6 2 1 /model/layers.10/self_attn/Expand_1_output_0 /model/layers.10/self_attn/Concat_10_output_0 /model/layers.10/self_attn/Reshape_6_output_0 Crop /model/layers.10/self_attn/Gather_4 1 1 /model/layers.10/self_attn/Shape_4_output_0 /model/layers.10/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_348 1 2 /model/layers.10/self_attn/Gather_4_output_0 /model/layers.10/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.10/self_attn/Gather_5 1 1 /model/layers.10/self_attn/Shape_5_output_0 /model/layers.10/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_349 1 2 /model/layers.10/self_attn/Gather_5_output_0 /model/layers.10/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.10/self_attn/Gather_6 1 1 /model/layers.10/self_attn/Shape_6_output_0 /model/layers.10/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_350 1 2 /model/layers.10/self_attn/Gather_6_output_0 /model/layers.10/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.10/self_attn/Gather_7 1 1 /model/layers.10/self_attn/Shape_7_output_0 /model/layers.10/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_351 1 2 /model/layers.10/self_attn/Gather_7_output_0 /model/layers.10/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.10/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.10/self_attn/Unsqueeze_11 2 1 /model/layers.10/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_33_output_0 /model/layers.10/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_15 2 1 /model/layers.10/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_3424 /model/layers.10/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_12 2 1 /model/layers.10/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_34_output_0 /model/layers.10/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.10/self_attn/Mul_5 2 1 /model/layers.10/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.10/self_attn/Constant_40_output_0 /model/layers.10/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.10/self_attn/Unsqueeze_13 2 1 /model/layers.10/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_36_output_0 /model/layers.10/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_17 2 1 /model/layers.10/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_3428 /model/layers.10/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_14 2 1 /model/layers.10/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_37_output_0 /model/layers.10/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_18 2 1 /model/layers.10/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_3430 /model/layers.10/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.10/self_attn/Unsqueeze_16 2 1 /model/layers.10/self_attn/Mul_5_output_0 onnx::Unsqueeze_3426 /model/layers.10/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.10/self_attn/Concat_7 5 1 /model/layers.10/self_attn/Unsqueeze_11_output_0 /model/layers.10/self_attn/Unsqueeze_12_output_0 /model/layers.10/self_attn/Constant_35_output_0 /model/layers.10/self_attn/Unsqueeze_13_output_0 /model/layers.10/self_attn/Unsqueeze_14_output_0 /model/layers.10/self_attn/Concat_7_output_0 0=0 Concat /model/layers.10/self_attn/Concat_8 4 1 /model/layers.10/self_attn/Unsqueeze_15_output_0 /model/layers.10/self_attn/Unsqueeze_16_output_0 /model/layers.10/self_attn/Unsqueeze_17_output_0 /model/layers.10/self_attn/Unsqueeze_18_output_0 /model/layers.10/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.10/self_attn/Reshape_3 1 1 /model/layers.10/self_attn/Concat_7_output_0 /model/layers.10/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_352 1 3 /model/layers.10/self_attn/Reshape_3_output_0 /model/layers.10/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.10/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.10/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.10/self_attn/Shape_8 1 1 /model/layers.10/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.10/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.10/self_attn/ConstantOfShape 1 1 /model/layers.10/self_attn/Shape_8_output_0 /model/layers.10/self_attn/ConstantOfShape_output_0 Split splitncnn_353 1 2 /model/layers.10/self_attn/ConstantOfShape_output_0 /model/layers.10/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.10/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.10/self_attn/Mul_4 2 1 /model/layers.10/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.10/self_attn/Constant_39_output_0 /model/layers.10/self_attn/Mul_4_output_0 0=2 Equal /model/layers.10/self_attn/Equal 2 1 /model/layers.10/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.10/self_attn/Mul_4_output_0 /model/layers.10/self_attn/Equal_output_0 Where /model/layers.10/self_attn/Where 3 1 /model/layers.10/self_attn/Equal_output_0 /model/layers.10/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.10/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.10/self_attn/Where_output_0 Expand /model/layers.10/self_attn/Expand 2 1 /model/layers.10/self_attn/Unsqueeze_10_output_0 /model/layers.10/self_attn/Where_output_0 /model/layers.10/self_attn/Expand_output_0 Reshape /model/layers.10/self_attn/Reshape_4 2 1 /model/layers.10/self_attn/Expand_output_0 /model/layers.10/self_attn/Concat_8_output_0 /model/layers.10/self_attn/Reshape_4_output_0 Split splitncnn_354 1 2 /model/layers.10/self_attn/Reshape_4_output_0 /model/layers.10/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.10/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.10/self_attn/Shape_14 1 1 /model/layers.10/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.10/self_attn/Shape_14_output_0 Permute /model/layers.10/self_attn/Transpose_3 1 1 /model/layers.10/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.10/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.10/self_attn/Gather_12 1 1 /model/layers.10/self_attn/Shape_14_output_0 /model/layers.10/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.10/self_attn/Mul_9 1 1 /model/layers.10/self_attn/Transpose_3_output_0 /model/layers.10/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.10/self_attn/Unsqueeze_28 2 1 /model/layers.10/self_attn/Gather_12_output_0 /model/layers.10/self_attn/Constant_57_output_0 /model/layers.10/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.10/self_attn/MatMul 2 1 /model/layers.10/self_attn/Mul_8_output_0 /model/layers.10/self_attn/Mul_9_output_0 /model/layers.10/self_attn/MatMul_output_0 Crop /model/layers.10/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_13 /model/layers.10/self_attn/Unsqueeze_28_output_0 /model/layers.10/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.10/self_attn/Where_2 3 1 /model/layers.10/self_attn/Slice_4_output_0 /model/layers.10/self_attn/Constant_61_output_0 /model/layers.10/self_attn/Constant_62_output_0 /model/layers.10/self_attn/Where_2_output_0 BinaryOp /model/layers.10/self_attn/Add_2 2 1 /model/layers.10/self_attn/MatMul_output_0 /model/layers.10/self_attn/Where_2_output_0 /model/layers.10/self_attn/Add_2_output_0 0=0 Softmax /model/layers.10/self_attn/Softmax 1 1 /model/layers.10/self_attn/Add_2_output_0 /model/layers.10/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_355 1 2 /model/layers.10/self_attn/Softmax_output_0 /model/layers.10/self_attn/Softmax_output_0_splitncnn_0 /model/layers.10/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.10/self_attn/IsNaN 1 1 /model/layers.10/self_attn/Softmax_output_0_splitncnn_1 /model/layers.10/self_attn/IsNaN_output_0 Where /model/layers.10/self_attn/Where_3 3 1 /model/layers.10/self_attn/IsNaN_output_0 /model/layers.10/self_attn/Constant_63_output_0 /model/layers.10/self_attn/Softmax_output_0_splitncnn_0 /model/layers.10/self_attn/Where_3_output_0 Gemm /model/layers.10/self_attn/MatMul_1 2 1 /model/layers.10/self_attn/Where_3_output_0 /model/layers.10/self_attn/Reshape_6_output_0 /model/layers.10/self_attn/MatMul_1_output_0 Permute /model/layers.10/self_attn/Transpose_4 1 1 /model/layers.10/self_attn/MatMul_1_output_0 /model/layers.10/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.10/self_attn/Reshape_7 2 1 /model/layers.10/self_attn/Transpose_4_output_0 /model/layers.10/self_attn/Concat_11_output_0 /model/layers.10/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.10/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.10/self_attn/Reshape_7_output_0 /model/layers.10/self_attn/Reshape_7_output_0_quantized /model/layers.10/self_attn/Reshape_7_output_0_scale /model/layers.10/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.10/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.10/self_attn/Reshape_7_output_0_scale /model/layers.10/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.629798e-03 MatMulInteger /model/layers.10/self_attn/o_proj/MatMul_quant 4 1 /model/layers.10/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7571_quantized /model/layers.10/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7571_zero_point /model/layers.10/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.10/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.10/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.10/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.10/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.10/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.10/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.10/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.10/Add 2 1 /model/layers.10/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.10/self_attn/o_proj/MatMul_output_0 /model/layers.10/Add_output_0 0=0 Cast /model/layers.10/post_attention_layernorm/Cast 1 1 /model/layers.10/Add_output_0 /model/layers.10/post_attention_layernorm/Cast_output_0 Split splitncnn_356 1 3 /model/layers.10/post_attention_layernorm/Cast_output_0 /model/layers.10/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.10/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.10/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.10/post_attention_layernorm/Pow 1 1 /model/layers.10/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.10/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.10/post_attention_layernorm/ReduceMean 2 1 /model/layers.10/post_attention_layernorm/Pow_output_0 /model/layers.10/post_attention_layernorm/Constant_1_output_0 /model/layers.10/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.10/post_attention_layernorm/Add 1 1 /model/layers.10/post_attention_layernorm/ReduceMean_output_0 /model/layers.10/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.10/post_attention_layernorm/Sqrt 1 1 /model/layers.10/post_attention_layernorm/Add_output_0 /model/layers.10/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.10/post_attention_layernorm/Div 1 1 /model/layers.10/post_attention_layernorm/Sqrt_output_0 /model/layers.10/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.10/post_attention_layernorm/Mul 2 1 /model/layers.10/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.10/post_attention_layernorm/Div_output_0 /model/layers.10/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.10/post_attention_layernorm/Cast_1 1 1 /model/layers.10/post_attention_layernorm/Mul_output_0 /model/layers.10/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.10/post_attention_layernorm/Mul_1 2 1 model.layers.10.post_attention_layernorm.weight /model/layers.10/post_attention_layernorm/Cast_1_output_0 /model/layers.10/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.10/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.10/post_attention_layernorm/Mul_1_output_0 /model/layers.10/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.10/post_attention_layernorm/Mul_1_output_0_scale /model/layers.10/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_357 1 2 /model/layers.10/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.10/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.10/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_358 1 2 /model/layers.10/post_attention_layernorm/Mul_1_output_0_scale /model/layers.10/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.10/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_359 1 2 /model/layers.10/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.10/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.10/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.10/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.10/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.10/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.613681e-03 BinaryOp /model/layers.10/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.10/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.10/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.383366e-03 MatMulInteger /model/layers.10/mlp/gate_proj/MatMul_quant 4 1 /model/layers.10/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7572_quantized /model/layers.10/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7572_zero_point /model/layers.10/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.10/mlp/up_proj/MatMul_quant 4 1 /model/layers.10/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7573_quantized /model/layers.10/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7573_zero_point /model/layers.10/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.10/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.10/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.10/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.10/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.10/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.10/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.10/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.10/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.10/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.10/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.10/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.10/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.10/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.10/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.10/mlp/act_fn/Mul 1 1 /model/layers.10/mlp/gate_proj/MatMul_output_0 /model/layers.10/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.10/mlp/Mul 2 1 /model/layers.10/mlp/act_fn/Mul_output_0 /model/layers.10/mlp/up_proj/MatMul_output_0 /model/layers.10/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.10/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.10/mlp/Mul_output_0 /model/layers.10/mlp/Mul_output_0_quantized /model/layers.10/mlp/Mul_output_0_scale /model/layers.10/mlp/Mul_output_0_zero_point BinaryOp /model/layers.10/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.10/mlp/Mul_output_0_scale /model/layers.10/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.583292e-03 MatMulInteger /model/layers.10/mlp/down_proj/MatMul_quant 4 1 /model/layers.10/mlp/Mul_output_0_quantized onnx::MatMul_7574_quantized /model/layers.10/mlp/Mul_output_0_zero_point onnx::MatMul_7574_zero_point /model/layers.10/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.10/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.10/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.10/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.10/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.10/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.10/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.10/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.10/Add_1 2 1 /model/layers.10/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.10/mlp/down_proj/MatMul_output_0 /model/layers.10/Add_1_output_0 0=0 Cast /model/layers.11/input_layernorm/Cast 1 1 /model/layers.10/Add_1_output_0 /model/layers.11/input_layernorm/Cast_output_0 Split splitncnn_360 1 3 /model/layers.11/input_layernorm/Cast_output_0 /model/layers.11/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.11/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.11/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.11/input_layernorm/Pow 1 1 /model/layers.11/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.11/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.11/input_layernorm/ReduceMean 2 1 /model/layers.11/input_layernorm/Pow_output_0 /model/layers.11/input_layernorm/Constant_1_output_0 /model/layers.11/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.11/input_layernorm/Add 1 1 /model/layers.11/input_layernorm/ReduceMean_output_0 /model/layers.11/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.11/input_layernorm/Sqrt 1 1 /model/layers.11/input_layernorm/Add_output_0 /model/layers.11/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.11/input_layernorm/Div 1 1 /model/layers.11/input_layernorm/Sqrt_output_0 /model/layers.11/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.11/input_layernorm/Mul 2 1 /model/layers.11/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.11/input_layernorm/Div_output_0 /model/layers.11/input_layernorm/Mul_output_0 0=2 Cast /model/layers.11/input_layernorm/Cast_1 1 1 /model/layers.11/input_layernorm/Mul_output_0 /model/layers.11/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.11/input_layernorm/Mul_1 2 1 model.layers.11.input_layernorm.weight /model/layers.11/input_layernorm/Cast_1_output_0 /model/layers.11/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_361 1 3 /model/layers.11/input_layernorm/Mul_1_output_0 /model/layers.11/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.11/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.11/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.11/self_attn/Shape 1 1 /model/layers.11/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.11/self_attn/Shape_output_0 Shape /model/layers.11/self_attn/Shape_1 1 1 /model/layers.11/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.11/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.11/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.11/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.11/input_layernorm/Mul_1_output_0_quantized /model/layers.11/input_layernorm/Mul_1_output_0_scale /model/layers.11/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_362 1 3 /model/layers.11/input_layernorm/Mul_1_output_0_quantized /model/layers.11/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.11/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.11/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_363 1 3 /model/layers.11/input_layernorm/Mul_1_output_0_scale /model/layers.11/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.11/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.11/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_364 1 3 /model/layers.11/input_layernorm/Mul_1_output_0_zero_point /model/layers.11/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.11/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.11/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.11/self_attn/Gather 1 1 /model/layers.11/self_attn/Shape_output_0 /model/layers.11/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_365 1 4 /model/layers.11/self_attn/Gather_output_0 /model/layers.11/self_attn/Gather_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_output_0_splitncnn_1 /model/layers.11/self_attn/Gather_output_0_splitncnn_2 /model/layers.11/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.11/self_attn/Gather_1 1 1 /model/layers.11/self_attn/Shape_1_output_0 /model/layers.11/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_366 1 4 /model/layers.11/self_attn/Gather_1_output_0 /model/layers.11/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.11/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.11/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.11/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.11/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.11/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.822404e-03 BinaryOp /model/layers.11/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.11/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.11/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.845472e-03 BinaryOp /model/layers.11/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.11/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.11/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=7.151206e-04 MatMulInteger /model/layers.11/self_attn/q_proj/MatMul_quant 4 1 /model/layers.11/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7575_quantized /model/layers.11/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7575_zero_point /model/layers.11/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.11/self_attn/k_proj/MatMul_quant 4 1 /model/layers.11/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7582_quantized /model/layers.11/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7582_zero_point /model/layers.11/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.11/self_attn/v_proj/MatMul_quant 4 1 /model/layers.11/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7583_quantized /model/layers.11/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7583_zero_point /model/layers.11/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.11/self_attn/Unsqueeze 2 1 /model/layers.11/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_3566 /model/layers.11/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_2 2 1 /model/layers.11/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_3575 /model/layers.11/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_4 2 1 /model/layers.11/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_3584 /model/layers.11/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_29 2 1 /model/layers.11/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_3789 /model/layers.11/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_1 2 1 /model/layers.11/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_3568 /model/layers.11/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_3 2 1 /model/layers.11/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_3577 /model/layers.11/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_5 2 1 /model/layers.11/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_3586 /model/layers.11/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_30 2 1 /model/layers.11/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_3791 /model/layers.11/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.11/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.11/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.11/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.11/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.11/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.11/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.11/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.11/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.11/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.11/self_attn/Concat 4 1 /model/layers.11/self_attn/Unsqueeze_output_0 /model/layers.11/self_attn/Unsqueeze_1_output_0 /model/layers.11/self_attn/Constant_2_output_0 /model/layers.11/self_attn/Constant_3_output_0 /model/layers.11/self_attn/Concat_output_0 0=0 Concat /model/layers.11/self_attn/Concat_1 4 1 /model/layers.11/self_attn/Unsqueeze_2_output_0 /model/layers.11/self_attn/Unsqueeze_3_output_0 /model/layers.11/self_attn/Constant_4_output_0 /model/layers.11/self_attn/Constant_5_output_0 /model/layers.11/self_attn/Concat_1_output_0 0=0 Concat /model/layers.11/self_attn/Concat_2 4 1 /model/layers.11/self_attn/Unsqueeze_4_output_0 /model/layers.11/self_attn/Unsqueeze_5_output_0 /model/layers.11/self_attn/Constant_6_output_0 /model/layers.11/self_attn/Constant_7_output_0 /model/layers.11/self_attn/Concat_2_output_0 0=0 Concat /model/layers.11/self_attn/Concat_11 3 1 /model/layers.11/self_attn/Unsqueeze_29_output_0 /model/layers.11/self_attn/Unsqueeze_30_output_0 /model/layers.11/self_attn/Constant_64_output_0 /model/layers.11/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.11/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.11/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.11/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.11/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.11/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.11/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.11/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.11/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.11/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.11/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.11/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.11/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.11/self_attn/q_proj/Add 2 1 model.layers.11.self_attn.q_proj.bias /model/layers.11/self_attn/q_proj/MatMul_output_0 /model/layers.11/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.11/self_attn/k_proj/Add 2 1 model.layers.11.self_attn.k_proj.bias /model/layers.11/self_attn/k_proj/MatMul_output_0 /model/layers.11/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.11/self_attn/v_proj/Add 2 1 model.layers.11.self_attn.v_proj.bias /model/layers.11/self_attn/v_proj/MatMul_output_0 /model/layers.11/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.11/self_attn/Reshape 2 1 /model/layers.11/self_attn/q_proj/Add_output_0 /model/layers.11/self_attn/Concat_output_0 /model/layers.11/self_attn/Reshape_output_0 Reshape /model/layers.11/self_attn/Reshape_1 2 1 /model/layers.11/self_attn/k_proj/Add_output_0 /model/layers.11/self_attn/Concat_1_output_0 /model/layers.11/self_attn/Reshape_1_output_0 Reshape /model/layers.11/self_attn/Reshape_2 2 1 /model/layers.11/self_attn/v_proj/Add_output_0 /model/layers.11/self_attn/Concat_2_output_0 /model/layers.11/self_attn/Reshape_2_output_0 Permute /model/layers.11/self_attn/Transpose 1 1 /model/layers.11/self_attn/Reshape_output_0 /model/layers.11/self_attn/Transpose_output_0 0=2 Split splitncnn_367 1 4 /model/layers.11/self_attn/Transpose_output_0 /model/layers.11/self_attn/Transpose_output_0_splitncnn_0 /model/layers.11/self_attn/Transpose_output_0_splitncnn_1 /model/layers.11/self_attn/Transpose_output_0_splitncnn_2 /model/layers.11/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.11/self_attn/Transpose_1 1 1 /model/layers.11/self_attn/Reshape_1_output_0 /model/layers.11/self_attn/Transpose_1_output_0 0=2 Split splitncnn_368 1 4 /model/layers.11/self_attn/Transpose_1_output_0 /model/layers.11/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.11/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.11/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.11/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.11/self_attn/Transpose_2 1 1 /model/layers.11/self_attn/Reshape_2_output_0 /model/layers.11/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.11/self_attn/Mul 2 1 /model/layers.11/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_25 /model/layers.11/self_attn/Mul_output_0 0=2 Shape /model/layers.11/self_attn/Shape_2 1 1 /model/layers.11/self_attn/Transpose_output_0_splitncnn_2 /model/layers.11/self_attn/Shape_2_output_0 BinaryOp /model/layers.11/self_attn/Mul_2 2 1 /model/layers.11/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_24 /model/layers.11/self_attn/Mul_2_output_0 0=2 Shape /model/layers.11/self_attn/Shape_3 1 1 /model/layers.11/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.11/self_attn/Shape_3_output_0 Concat /model/layers.11/self_attn/Concat_6 2 1 past_key_values.11.value /model/layers.11/self_attn/Transpose_2_output_0 present.11.value 0=-2 Split splitncnn_369 1 5 present.11.value present.11.value_splitncnn_0 present.11.value_splitncnn_1 present.11.value_splitncnn_2 present.11.value_splitncnn_3 present.11.value_splitncnn_4 Crop /model/layers.11/self_attn/Gather_2 1 1 /model/layers.11/self_attn/Shape_2_output_0 /model/layers.11/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.11/self_attn/Gather_3 1 1 /model/layers.11/self_attn/Shape_3_output_0 /model/layers.11/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.11/self_attn/Shape_9 1 1 present.11.value_splitncnn_4 /model/layers.11/self_attn/Shape_9_output_0 Shape /model/layers.11/self_attn/Shape_10 1 1 present.11.value_splitncnn_3 /model/layers.11/self_attn/Shape_10_output_0 Shape /model/layers.11/self_attn/Shape_11 1 1 present.11.value_splitncnn_2 /model/layers.11/self_attn/Shape_11_output_0 Shape /model/layers.11/self_attn/Shape_12 1 1 present.11.value_splitncnn_1 /model/layers.11/self_attn/Shape_12_output_0 ExpandDims /model/layers.11/self_attn/Unsqueeze_19 2 1 present.11.value_splitncnn_0 /model/layers.11/self_attn/Constant_45_output_0 /model/layers.11/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.11/self_attn/Div 2 1 /model/layers.11/self_attn/Gather_2_output_0 /model/layers.11/self_attn/Constant_9_output_0 /model/layers.11/self_attn/Div_output_0 0=3 BinaryOp /model/layers.11/self_attn/Div_1 2 1 /model/layers.11/self_attn/Gather_3_output_0 /model/layers.11/self_attn/Constant_19_output_0 /model/layers.11/self_attn/Div_1_output_0 0=3 Crop /model/layers.11/self_attn/Gather_8 1 1 /model/layers.11/self_attn/Shape_9_output_0 /model/layers.11/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_370 1 2 /model/layers.11/self_attn/Gather_8_output_0 /model/layers.11/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.11/self_attn/Gather_9 1 1 /model/layers.11/self_attn/Shape_10_output_0 /model/layers.11/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_371 1 2 /model/layers.11/self_attn/Gather_9_output_0 /model/layers.11/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.11/self_attn/Gather_10 1 1 /model/layers.11/self_attn/Shape_11_output_0 /model/layers.11/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_372 1 2 /model/layers.11/self_attn/Gather_10_output_0 /model/layers.11/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.11/self_attn/Gather_11 1 1 /model/layers.11/self_attn/Shape_12_output_0 /model/layers.11/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_373 1 2 /model/layers.11/self_attn/Gather_11_output_0 /model/layers.11/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.11/self_attn/Cast 1 1 /model/layers.11/self_attn/Div_output_0 /model/layers.11/self_attn/Cast_output_0 Cast /model/layers.11/self_attn/Cast_2 1 1 /model/layers.11/self_attn/Div_1_output_0 /model/layers.11/self_attn/Cast_2_output_0 ExpandDims /model/layers.11/self_attn/Unsqueeze_20 2 1 /model/layers.11/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_46_output_0 /model/layers.11/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_24 2 1 /model/layers.11/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_3749 /model/layers.11/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_21 2 1 /model/layers.11/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_47_output_0 /model/layers.11/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.11/self_attn/Mul_7 2 1 /model/layers.11/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.11/self_attn/Constant_53_output_0 /model/layers.11/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.11/self_attn/Unsqueeze_22 2 1 /model/layers.11/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_49_output_0 /model/layers.11/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_26 2 1 /model/layers.11/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_3753 /model/layers.11/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_23 2 1 /model/layers.11/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_50_output_0 /model/layers.11/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_27 2 1 /model/layers.11/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_3755 /model/layers.11/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.11/self_attn/Cast_1 1 1 /model/layers.11/self_attn/Cast_output_0 /model/layers.11/self_attn/Cast_1_output_0 Split splitncnn_374 1 2 /model/layers.11/self_attn/Cast_1_output_0 /model/layers.11/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.11/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.11/self_attn/Cast_3 1 1 /model/layers.11/self_attn/Cast_2_output_0 /model/layers.11/self_attn/Cast_3_output_0 Split splitncnn_375 1 2 /model/layers.11/self_attn/Cast_3_output_0 /model/layers.11/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.11/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.11/self_attn/Unsqueeze_25 2 1 /model/layers.11/self_attn/Mul_7_output_0 onnx::Unsqueeze_3751 /model/layers.11/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.11/self_attn/Concat_9 5 1 /model/layers.11/self_attn/Unsqueeze_20_output_0 /model/layers.11/self_attn/Unsqueeze_21_output_0 /model/layers.11/self_attn/Constant_48_output_0 /model/layers.11/self_attn/Unsqueeze_22_output_0 /model/layers.11/self_attn/Unsqueeze_23_output_0 /model/layers.11/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_6 2 1 /model/layers.11/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_12_output_0 /model/layers.11/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_7 2 1 /model/layers.11/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.11/self_attn/Constant_15_output_0 /model/layers.11/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_8 2 1 /model/layers.11/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_22_output_0 /model/layers.11/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_9 2 1 /model/layers.11/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.11/self_attn/Constant_25_output_0 /model/layers.11/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.11/self_attn/Concat_10 4 1 /model/layers.11/self_attn/Unsqueeze_24_output_0 /model/layers.11/self_attn/Unsqueeze_25_output_0 /model/layers.11/self_attn/Unsqueeze_26_output_0 /model/layers.11/self_attn/Unsqueeze_27_output_0 /model/layers.11/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.11/self_attn/Reshape_5 1 1 /model/layers.11/self_attn/Concat_9_output_0 /model/layers.11/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_376 1 3 /model/layers.11/self_attn/Reshape_5_output_0 /model/layers.11/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.11/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.11/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.11/self_attn/Slice 2 1 /model/layers.11/self_attn/Transpose_output_0_splitncnn_1 /model/layers.11/self_attn/Unsqueeze_6_output_0 /model/layers.11/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.11/self_attn/Slice_1 2 1 /model/layers.11/self_attn/Transpose_output_0_splitncnn_0 /model/layers.11/self_attn/Unsqueeze_7_output_0 /model/layers.11/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.11/self_attn/Slice_2 2 1 /model/layers.11/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.11/self_attn/Unsqueeze_8_output_0 /model/layers.11/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.11/self_attn/Slice_3 2 1 /model/layers.11/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.11/self_attn/Unsqueeze_9_output_0 /model/layers.11/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.11/self_attn/Shape_13 1 1 /model/layers.11/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.11/self_attn/Shape_13_output_0 UnaryOp /model/layers.11/self_attn/Neg 1 1 /model/layers.11/self_attn/Slice_1_output_0 /model/layers.11/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.11/self_attn/Neg_1 1 1 /model/layers.11/self_attn/Slice_3_output_0 /model/layers.11/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.11/self_attn/ConstantOfShape_1 1 1 /model/layers.11/self_attn/Shape_13_output_0 /model/layers.11/self_attn/ConstantOfShape_1_output_0 Split splitncnn_377 1 2 /model/layers.11/self_attn/ConstantOfShape_1_output_0 /model/layers.11/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.11/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.11/self_attn/Concat_3 2 1 /model/layers.11/self_attn/Neg_output_0 /model/layers.11/self_attn/Slice_output_0 /model/layers.11/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.11/self_attn/Concat_4 2 1 /model/layers.11/self_attn/Neg_1_output_0 /model/layers.11/self_attn/Slice_2_output_0 /model/layers.11/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.11/self_attn/Mul_6 2 1 /model/layers.11/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_52_output_0 /model/layers.11/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.11/self_attn/Mul_1 2 1 /model/layers.11/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_25 /model/layers.11/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.11/self_attn/Mul_3 2 1 /model/layers.11/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_24 /model/layers.11/self_attn/Mul_3_output_0 0=2 Equal /model/layers.11/self_attn/Equal_1 2 1 /model/layers.11/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.11/self_attn/Mul_6_output_0 /model/layers.11/self_attn/Equal_1_output_0 BinaryOp /model/layers.11/self_attn/Add 2 1 /model/layers.11/self_attn/Mul_output_0 /model/layers.11/self_attn/Mul_1_output_0 /model/layers.11/self_attn/Add_output_0 0=0 BinaryOp /model/layers.11/self_attn/Add_1 2 1 /model/layers.11/self_attn/Mul_2_output_0 /model/layers.11/self_attn/Mul_3_output_0 /model/layers.11/self_attn/Add_1_output_0 0=0 Where /model/layers.11/self_attn/Where_1 3 1 /model/layers.11/self_attn/Equal_1_output_0 /model/layers.11/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.11/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.11/self_attn/Where_1_output_0 BinaryOp /model/layers.11/self_attn/Mul_8 1 1 /model/layers.11/self_attn/Add_output_0 /model/layers.11/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.11/self_attn/Concat_5 2 1 past_key_values.11.key /model/layers.11/self_attn/Add_1_output_0 present.11.key 0=-2 Split splitncnn_378 1 5 present.11.key present.11.key_splitncnn_0 present.11.key_splitncnn_1 present.11.key_splitncnn_2 present.11.key_splitncnn_3 present.11.key_splitncnn_4 Expand /model/layers.11/self_attn/Expand_1 2 1 /model/layers.11/self_attn/Unsqueeze_19_output_0 /model/layers.11/self_attn/Where_1_output_0 /model/layers.11/self_attn/Expand_1_output_0 Shape /model/layers.11/self_attn/Shape_4 1 1 present.11.key_splitncnn_4 /model/layers.11/self_attn/Shape_4_output_0 Shape /model/layers.11/self_attn/Shape_5 1 1 present.11.key_splitncnn_3 /model/layers.11/self_attn/Shape_5_output_0 Shape /model/layers.11/self_attn/Shape_6 1 1 present.11.key_splitncnn_2 /model/layers.11/self_attn/Shape_6_output_0 Shape /model/layers.11/self_attn/Shape_7 1 1 present.11.key_splitncnn_1 /model/layers.11/self_attn/Shape_7_output_0 ExpandDims /model/layers.11/self_attn/Unsqueeze_10 2 1 present.11.key_splitncnn_0 /model/layers.11/self_attn/Constant_32_output_0 /model/layers.11/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.11/self_attn/Reshape_6 2 1 /model/layers.11/self_attn/Expand_1_output_0 /model/layers.11/self_attn/Concat_10_output_0 /model/layers.11/self_attn/Reshape_6_output_0 Crop /model/layers.11/self_attn/Gather_4 1 1 /model/layers.11/self_attn/Shape_4_output_0 /model/layers.11/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_379 1 2 /model/layers.11/self_attn/Gather_4_output_0 /model/layers.11/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.11/self_attn/Gather_5 1 1 /model/layers.11/self_attn/Shape_5_output_0 /model/layers.11/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_380 1 2 /model/layers.11/self_attn/Gather_5_output_0 /model/layers.11/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.11/self_attn/Gather_6 1 1 /model/layers.11/self_attn/Shape_6_output_0 /model/layers.11/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_381 1 2 /model/layers.11/self_attn/Gather_6_output_0 /model/layers.11/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.11/self_attn/Gather_7 1 1 /model/layers.11/self_attn/Shape_7_output_0 /model/layers.11/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_382 1 2 /model/layers.11/self_attn/Gather_7_output_0 /model/layers.11/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.11/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.11/self_attn/Unsqueeze_11 2 1 /model/layers.11/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_33_output_0 /model/layers.11/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_15 2 1 /model/layers.11/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_3703 /model/layers.11/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_12 2 1 /model/layers.11/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_34_output_0 /model/layers.11/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.11/self_attn/Mul_5 2 1 /model/layers.11/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.11/self_attn/Constant_40_output_0 /model/layers.11/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.11/self_attn/Unsqueeze_13 2 1 /model/layers.11/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_36_output_0 /model/layers.11/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_17 2 1 /model/layers.11/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_3707 /model/layers.11/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_14 2 1 /model/layers.11/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_37_output_0 /model/layers.11/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_18 2 1 /model/layers.11/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_3709 /model/layers.11/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.11/self_attn/Unsqueeze_16 2 1 /model/layers.11/self_attn/Mul_5_output_0 onnx::Unsqueeze_3705 /model/layers.11/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.11/self_attn/Concat_7 5 1 /model/layers.11/self_attn/Unsqueeze_11_output_0 /model/layers.11/self_attn/Unsqueeze_12_output_0 /model/layers.11/self_attn/Constant_35_output_0 /model/layers.11/self_attn/Unsqueeze_13_output_0 /model/layers.11/self_attn/Unsqueeze_14_output_0 /model/layers.11/self_attn/Concat_7_output_0 0=0 Concat /model/layers.11/self_attn/Concat_8 4 1 /model/layers.11/self_attn/Unsqueeze_15_output_0 /model/layers.11/self_attn/Unsqueeze_16_output_0 /model/layers.11/self_attn/Unsqueeze_17_output_0 /model/layers.11/self_attn/Unsqueeze_18_output_0 /model/layers.11/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.11/self_attn/Reshape_3 1 1 /model/layers.11/self_attn/Concat_7_output_0 /model/layers.11/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_383 1 3 /model/layers.11/self_attn/Reshape_3_output_0 /model/layers.11/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.11/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.11/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.11/self_attn/Shape_8 1 1 /model/layers.11/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.11/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.11/self_attn/ConstantOfShape 1 1 /model/layers.11/self_attn/Shape_8_output_0 /model/layers.11/self_attn/ConstantOfShape_output_0 Split splitncnn_384 1 2 /model/layers.11/self_attn/ConstantOfShape_output_0 /model/layers.11/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.11/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.11/self_attn/Mul_4 2 1 /model/layers.11/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.11/self_attn/Constant_39_output_0 /model/layers.11/self_attn/Mul_4_output_0 0=2 Equal /model/layers.11/self_attn/Equal 2 1 /model/layers.11/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.11/self_attn/Mul_4_output_0 /model/layers.11/self_attn/Equal_output_0 Where /model/layers.11/self_attn/Where 3 1 /model/layers.11/self_attn/Equal_output_0 /model/layers.11/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.11/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.11/self_attn/Where_output_0 Expand /model/layers.11/self_attn/Expand 2 1 /model/layers.11/self_attn/Unsqueeze_10_output_0 /model/layers.11/self_attn/Where_output_0 /model/layers.11/self_attn/Expand_output_0 Reshape /model/layers.11/self_attn/Reshape_4 2 1 /model/layers.11/self_attn/Expand_output_0 /model/layers.11/self_attn/Concat_8_output_0 /model/layers.11/self_attn/Reshape_4_output_0 Split splitncnn_385 1 2 /model/layers.11/self_attn/Reshape_4_output_0 /model/layers.11/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.11/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.11/self_attn/Shape_14 1 1 /model/layers.11/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.11/self_attn/Shape_14_output_0 Permute /model/layers.11/self_attn/Transpose_3 1 1 /model/layers.11/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.11/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.11/self_attn/Gather_12 1 1 /model/layers.11/self_attn/Shape_14_output_0 /model/layers.11/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.11/self_attn/Mul_9 1 1 /model/layers.11/self_attn/Transpose_3_output_0 /model/layers.11/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.11/self_attn/Unsqueeze_28 2 1 /model/layers.11/self_attn/Gather_12_output_0 /model/layers.11/self_attn/Constant_57_output_0 /model/layers.11/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.11/self_attn/MatMul 2 1 /model/layers.11/self_attn/Mul_8_output_0 /model/layers.11/self_attn/Mul_9_output_0 /model/layers.11/self_attn/MatMul_output_0 Crop /model/layers.11/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_12 /model/layers.11/self_attn/Unsqueeze_28_output_0 /model/layers.11/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.11/self_attn/Where_2 3 1 /model/layers.11/self_attn/Slice_4_output_0 /model/layers.11/self_attn/Constant_61_output_0 /model/layers.11/self_attn/Constant_62_output_0 /model/layers.11/self_attn/Where_2_output_0 BinaryOp /model/layers.11/self_attn/Add_2 2 1 /model/layers.11/self_attn/MatMul_output_0 /model/layers.11/self_attn/Where_2_output_0 /model/layers.11/self_attn/Add_2_output_0 0=0 Softmax /model/layers.11/self_attn/Softmax 1 1 /model/layers.11/self_attn/Add_2_output_0 /model/layers.11/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_386 1 2 /model/layers.11/self_attn/Softmax_output_0 /model/layers.11/self_attn/Softmax_output_0_splitncnn_0 /model/layers.11/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.11/self_attn/IsNaN 1 1 /model/layers.11/self_attn/Softmax_output_0_splitncnn_1 /model/layers.11/self_attn/IsNaN_output_0 Where /model/layers.11/self_attn/Where_3 3 1 /model/layers.11/self_attn/IsNaN_output_0 /model/layers.11/self_attn/Constant_63_output_0 /model/layers.11/self_attn/Softmax_output_0_splitncnn_0 /model/layers.11/self_attn/Where_3_output_0 Gemm /model/layers.11/self_attn/MatMul_1 2 1 /model/layers.11/self_attn/Where_3_output_0 /model/layers.11/self_attn/Reshape_6_output_0 /model/layers.11/self_attn/MatMul_1_output_0 Permute /model/layers.11/self_attn/Transpose_4 1 1 /model/layers.11/self_attn/MatMul_1_output_0 /model/layers.11/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.11/self_attn/Reshape_7 2 1 /model/layers.11/self_attn/Transpose_4_output_0 /model/layers.11/self_attn/Concat_11_output_0 /model/layers.11/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.11/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.11/self_attn/Reshape_7_output_0 /model/layers.11/self_attn/Reshape_7_output_0_quantized /model/layers.11/self_attn/Reshape_7_output_0_scale /model/layers.11/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.11/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.11/self_attn/Reshape_7_output_0_scale /model/layers.11/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.552166e-03 MatMulInteger /model/layers.11/self_attn/o_proj/MatMul_quant 4 1 /model/layers.11/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7606_quantized /model/layers.11/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7606_zero_point /model/layers.11/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.11/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.11/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.11/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.11/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.11/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.11/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.11/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.11/Add 2 1 /model/layers.11/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.11/self_attn/o_proj/MatMul_output_0 /model/layers.11/Add_output_0 0=0 Cast /model/layers.11/post_attention_layernorm/Cast 1 1 /model/layers.11/Add_output_0 /model/layers.11/post_attention_layernorm/Cast_output_0 Split splitncnn_387 1 3 /model/layers.11/post_attention_layernorm/Cast_output_0 /model/layers.11/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.11/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.11/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.11/post_attention_layernorm/Pow 1 1 /model/layers.11/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.11/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.11/post_attention_layernorm/ReduceMean 2 1 /model/layers.11/post_attention_layernorm/Pow_output_0 /model/layers.11/post_attention_layernorm/Constant_1_output_0 /model/layers.11/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.11/post_attention_layernorm/Add 1 1 /model/layers.11/post_attention_layernorm/ReduceMean_output_0 /model/layers.11/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.11/post_attention_layernorm/Sqrt 1 1 /model/layers.11/post_attention_layernorm/Add_output_0 /model/layers.11/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.11/post_attention_layernorm/Div 1 1 /model/layers.11/post_attention_layernorm/Sqrt_output_0 /model/layers.11/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.11/post_attention_layernorm/Mul 2 1 /model/layers.11/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.11/post_attention_layernorm/Div_output_0 /model/layers.11/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.11/post_attention_layernorm/Cast_1 1 1 /model/layers.11/post_attention_layernorm/Mul_output_0 /model/layers.11/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.11/post_attention_layernorm/Mul_1 2 1 model.layers.11.post_attention_layernorm.weight /model/layers.11/post_attention_layernorm/Cast_1_output_0 /model/layers.11/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.11/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.11/post_attention_layernorm/Mul_1_output_0 /model/layers.11/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.11/post_attention_layernorm/Mul_1_output_0_scale /model/layers.11/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_388 1 2 /model/layers.11/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.11/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.11/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_389 1 2 /model/layers.11/post_attention_layernorm/Mul_1_output_0_scale /model/layers.11/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.11/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_390 1 2 /model/layers.11/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.11/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.11/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.11/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.11/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.11/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.890871e-03 BinaryOp /model/layers.11/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.11/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.11/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=5.351870e-03 MatMulInteger /model/layers.11/mlp/gate_proj/MatMul_quant 4 1 /model/layers.11/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7607_quantized /model/layers.11/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7607_zero_point /model/layers.11/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.11/mlp/up_proj/MatMul_quant 4 1 /model/layers.11/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7608_quantized /model/layers.11/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7608_zero_point /model/layers.11/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.11/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.11/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.11/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.11/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.11/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.11/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.11/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.11/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.11/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.11/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.11/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.11/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.11/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.11/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.11/mlp/act_fn/Mul 1 1 /model/layers.11/mlp/gate_proj/MatMul_output_0 /model/layers.11/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.11/mlp/Mul 2 1 /model/layers.11/mlp/act_fn/Mul_output_0 /model/layers.11/mlp/up_proj/MatMul_output_0 /model/layers.11/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.11/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.11/mlp/Mul_output_0 /model/layers.11/mlp/Mul_output_0_quantized /model/layers.11/mlp/Mul_output_0_scale /model/layers.11/mlp/Mul_output_0_zero_point BinaryOp /model/layers.11/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.11/mlp/Mul_output_0_scale /model/layers.11/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.705955e-03 MatMulInteger /model/layers.11/mlp/down_proj/MatMul_quant 4 1 /model/layers.11/mlp/Mul_output_0_quantized onnx::MatMul_7609_quantized /model/layers.11/mlp/Mul_output_0_zero_point onnx::MatMul_7609_zero_point /model/layers.11/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.11/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.11/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.11/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.11/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.11/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.11/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.11/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.11/Add_1 2 1 /model/layers.11/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.11/mlp/down_proj/MatMul_output_0 /model/layers.11/Add_1_output_0 0=0 Cast /model/layers.12/input_layernorm/Cast 1 1 /model/layers.11/Add_1_output_0 /model/layers.12/input_layernorm/Cast_output_0 Split splitncnn_391 1 3 /model/layers.12/input_layernorm/Cast_output_0 /model/layers.12/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.12/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.12/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.12/input_layernorm/Pow 1 1 /model/layers.12/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.12/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.12/input_layernorm/ReduceMean 2 1 /model/layers.12/input_layernorm/Pow_output_0 /model/layers.12/input_layernorm/Constant_1_output_0 /model/layers.12/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.12/input_layernorm/Add 1 1 /model/layers.12/input_layernorm/ReduceMean_output_0 /model/layers.12/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.12/input_layernorm/Sqrt 1 1 /model/layers.12/input_layernorm/Add_output_0 /model/layers.12/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.12/input_layernorm/Div 1 1 /model/layers.12/input_layernorm/Sqrt_output_0 /model/layers.12/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.12/input_layernorm/Mul 2 1 /model/layers.12/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.12/input_layernorm/Div_output_0 /model/layers.12/input_layernorm/Mul_output_0 0=2 Cast /model/layers.12/input_layernorm/Cast_1 1 1 /model/layers.12/input_layernorm/Mul_output_0 /model/layers.12/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.12/input_layernorm/Mul_1 2 1 model.layers.12.input_layernorm.weight /model/layers.12/input_layernorm/Cast_1_output_0 /model/layers.12/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_392 1 3 /model/layers.12/input_layernorm/Mul_1_output_0 /model/layers.12/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.12/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.12/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.12/self_attn/Shape 1 1 /model/layers.12/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.12/self_attn/Shape_output_0 Shape /model/layers.12/self_attn/Shape_1 1 1 /model/layers.12/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.12/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.12/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.12/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.12/input_layernorm/Mul_1_output_0_quantized /model/layers.12/input_layernorm/Mul_1_output_0_scale /model/layers.12/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_393 1 3 /model/layers.12/input_layernorm/Mul_1_output_0_quantized /model/layers.12/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.12/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.12/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_394 1 3 /model/layers.12/input_layernorm/Mul_1_output_0_scale /model/layers.12/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.12/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.12/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_395 1 3 /model/layers.12/input_layernorm/Mul_1_output_0_zero_point /model/layers.12/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.12/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.12/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.12/self_attn/Gather 1 1 /model/layers.12/self_attn/Shape_output_0 /model/layers.12/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_396 1 4 /model/layers.12/self_attn/Gather_output_0 /model/layers.12/self_attn/Gather_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_output_0_splitncnn_1 /model/layers.12/self_attn/Gather_output_0_splitncnn_2 /model/layers.12/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.12/self_attn/Gather_1 1 1 /model/layers.12/self_attn/Shape_1_output_0 /model/layers.12/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_397 1 4 /model/layers.12/self_attn/Gather_1_output_0 /model/layers.12/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.12/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.12/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.12/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.12/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.12/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.045399e-03 BinaryOp /model/layers.12/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.12/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.12/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.484067e-03 BinaryOp /model/layers.12/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.12/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.12/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.061147e-03 MatMulInteger /model/layers.12/self_attn/q_proj/MatMul_quant 4 1 /model/layers.12/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7610_quantized /model/layers.12/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7610_zero_point /model/layers.12/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.12/self_attn/k_proj/MatMul_quant 4 1 /model/layers.12/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7617_quantized /model/layers.12/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7617_zero_point /model/layers.12/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.12/self_attn/v_proj/MatMul_quant 4 1 /model/layers.12/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7618_quantized /model/layers.12/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7618_zero_point /model/layers.12/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.12/self_attn/Unsqueeze 2 1 /model/layers.12/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_3845 /model/layers.12/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_2 2 1 /model/layers.12/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_3854 /model/layers.12/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_4 2 1 /model/layers.12/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_3863 /model/layers.12/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_29 2 1 /model/layers.12/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_4068 /model/layers.12/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_1 2 1 /model/layers.12/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_3847 /model/layers.12/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_3 2 1 /model/layers.12/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_3856 /model/layers.12/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_5 2 1 /model/layers.12/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_3865 /model/layers.12/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_30 2 1 /model/layers.12/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_4070 /model/layers.12/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.12/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.12/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.12/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.12/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.12/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.12/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.12/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.12/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.12/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.12/self_attn/Concat 4 1 /model/layers.12/self_attn/Unsqueeze_output_0 /model/layers.12/self_attn/Unsqueeze_1_output_0 /model/layers.12/self_attn/Constant_2_output_0 /model/layers.12/self_attn/Constant_3_output_0 /model/layers.12/self_attn/Concat_output_0 0=0 Concat /model/layers.12/self_attn/Concat_1 4 1 /model/layers.12/self_attn/Unsqueeze_2_output_0 /model/layers.12/self_attn/Unsqueeze_3_output_0 /model/layers.12/self_attn/Constant_4_output_0 /model/layers.12/self_attn/Constant_5_output_0 /model/layers.12/self_attn/Concat_1_output_0 0=0 Concat /model/layers.12/self_attn/Concat_2 4 1 /model/layers.12/self_attn/Unsqueeze_4_output_0 /model/layers.12/self_attn/Unsqueeze_5_output_0 /model/layers.12/self_attn/Constant_6_output_0 /model/layers.12/self_attn/Constant_7_output_0 /model/layers.12/self_attn/Concat_2_output_0 0=0 Concat /model/layers.12/self_attn/Concat_11 3 1 /model/layers.12/self_attn/Unsqueeze_29_output_0 /model/layers.12/self_attn/Unsqueeze_30_output_0 /model/layers.12/self_attn/Constant_64_output_0 /model/layers.12/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.12/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.12/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.12/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.12/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.12/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.12/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.12/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.12/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.12/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.12/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.12/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.12/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.12/self_attn/q_proj/Add 2 1 model.layers.12.self_attn.q_proj.bias /model/layers.12/self_attn/q_proj/MatMul_output_0 /model/layers.12/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.12/self_attn/k_proj/Add 2 1 model.layers.12.self_attn.k_proj.bias /model/layers.12/self_attn/k_proj/MatMul_output_0 /model/layers.12/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.12/self_attn/v_proj/Add 2 1 model.layers.12.self_attn.v_proj.bias /model/layers.12/self_attn/v_proj/MatMul_output_0 /model/layers.12/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.12/self_attn/Reshape 2 1 /model/layers.12/self_attn/q_proj/Add_output_0 /model/layers.12/self_attn/Concat_output_0 /model/layers.12/self_attn/Reshape_output_0 Reshape /model/layers.12/self_attn/Reshape_1 2 1 /model/layers.12/self_attn/k_proj/Add_output_0 /model/layers.12/self_attn/Concat_1_output_0 /model/layers.12/self_attn/Reshape_1_output_0 Reshape /model/layers.12/self_attn/Reshape_2 2 1 /model/layers.12/self_attn/v_proj/Add_output_0 /model/layers.12/self_attn/Concat_2_output_0 /model/layers.12/self_attn/Reshape_2_output_0 Permute /model/layers.12/self_attn/Transpose 1 1 /model/layers.12/self_attn/Reshape_output_0 /model/layers.12/self_attn/Transpose_output_0 0=2 Split splitncnn_398 1 4 /model/layers.12/self_attn/Transpose_output_0 /model/layers.12/self_attn/Transpose_output_0_splitncnn_0 /model/layers.12/self_attn/Transpose_output_0_splitncnn_1 /model/layers.12/self_attn/Transpose_output_0_splitncnn_2 /model/layers.12/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.12/self_attn/Transpose_1 1 1 /model/layers.12/self_attn/Reshape_1_output_0 /model/layers.12/self_attn/Transpose_1_output_0 0=2 Split splitncnn_399 1 4 /model/layers.12/self_attn/Transpose_1_output_0 /model/layers.12/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.12/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.12/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.12/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.12/self_attn/Transpose_2 1 1 /model/layers.12/self_attn/Reshape_2_output_0 /model/layers.12/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.12/self_attn/Mul 2 1 /model/layers.12/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_23 /model/layers.12/self_attn/Mul_output_0 0=2 Shape /model/layers.12/self_attn/Shape_2 1 1 /model/layers.12/self_attn/Transpose_output_0_splitncnn_2 /model/layers.12/self_attn/Shape_2_output_0 BinaryOp /model/layers.12/self_attn/Mul_2 2 1 /model/layers.12/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_22 /model/layers.12/self_attn/Mul_2_output_0 0=2 Shape /model/layers.12/self_attn/Shape_3 1 1 /model/layers.12/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.12/self_attn/Shape_3_output_0 Concat /model/layers.12/self_attn/Concat_6 2 1 past_key_values.12.value /model/layers.12/self_attn/Transpose_2_output_0 present.12.value 0=-2 Split splitncnn_400 1 5 present.12.value present.12.value_splitncnn_0 present.12.value_splitncnn_1 present.12.value_splitncnn_2 present.12.value_splitncnn_3 present.12.value_splitncnn_4 Crop /model/layers.12/self_attn/Gather_2 1 1 /model/layers.12/self_attn/Shape_2_output_0 /model/layers.12/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.12/self_attn/Gather_3 1 1 /model/layers.12/self_attn/Shape_3_output_0 /model/layers.12/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.12/self_attn/Shape_9 1 1 present.12.value_splitncnn_4 /model/layers.12/self_attn/Shape_9_output_0 Shape /model/layers.12/self_attn/Shape_10 1 1 present.12.value_splitncnn_3 /model/layers.12/self_attn/Shape_10_output_0 Shape /model/layers.12/self_attn/Shape_11 1 1 present.12.value_splitncnn_2 /model/layers.12/self_attn/Shape_11_output_0 Shape /model/layers.12/self_attn/Shape_12 1 1 present.12.value_splitncnn_1 /model/layers.12/self_attn/Shape_12_output_0 ExpandDims /model/layers.12/self_attn/Unsqueeze_19 2 1 present.12.value_splitncnn_0 /model/layers.12/self_attn/Constant_45_output_0 /model/layers.12/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.12/self_attn/Div 2 1 /model/layers.12/self_attn/Gather_2_output_0 /model/layers.12/self_attn/Constant_9_output_0 /model/layers.12/self_attn/Div_output_0 0=3 BinaryOp /model/layers.12/self_attn/Div_1 2 1 /model/layers.12/self_attn/Gather_3_output_0 /model/layers.12/self_attn/Constant_19_output_0 /model/layers.12/self_attn/Div_1_output_0 0=3 Crop /model/layers.12/self_attn/Gather_8 1 1 /model/layers.12/self_attn/Shape_9_output_0 /model/layers.12/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_401 1 2 /model/layers.12/self_attn/Gather_8_output_0 /model/layers.12/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.12/self_attn/Gather_9 1 1 /model/layers.12/self_attn/Shape_10_output_0 /model/layers.12/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_402 1 2 /model/layers.12/self_attn/Gather_9_output_0 /model/layers.12/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.12/self_attn/Gather_10 1 1 /model/layers.12/self_attn/Shape_11_output_0 /model/layers.12/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_403 1 2 /model/layers.12/self_attn/Gather_10_output_0 /model/layers.12/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.12/self_attn/Gather_11 1 1 /model/layers.12/self_attn/Shape_12_output_0 /model/layers.12/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_404 1 2 /model/layers.12/self_attn/Gather_11_output_0 /model/layers.12/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.12/self_attn/Cast 1 1 /model/layers.12/self_attn/Div_output_0 /model/layers.12/self_attn/Cast_output_0 Cast /model/layers.12/self_attn/Cast_2 1 1 /model/layers.12/self_attn/Div_1_output_0 /model/layers.12/self_attn/Cast_2_output_0 ExpandDims /model/layers.12/self_attn/Unsqueeze_20 2 1 /model/layers.12/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_46_output_0 /model/layers.12/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_24 2 1 /model/layers.12/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_4028 /model/layers.12/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_21 2 1 /model/layers.12/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_47_output_0 /model/layers.12/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.12/self_attn/Mul_7 2 1 /model/layers.12/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.12/self_attn/Constant_53_output_0 /model/layers.12/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.12/self_attn/Unsqueeze_22 2 1 /model/layers.12/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_49_output_0 /model/layers.12/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_26 2 1 /model/layers.12/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_4032 /model/layers.12/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_23 2 1 /model/layers.12/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_50_output_0 /model/layers.12/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_27 2 1 /model/layers.12/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_4034 /model/layers.12/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.12/self_attn/Cast_1 1 1 /model/layers.12/self_attn/Cast_output_0 /model/layers.12/self_attn/Cast_1_output_0 Split splitncnn_405 1 2 /model/layers.12/self_attn/Cast_1_output_0 /model/layers.12/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.12/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.12/self_attn/Cast_3 1 1 /model/layers.12/self_attn/Cast_2_output_0 /model/layers.12/self_attn/Cast_3_output_0 Split splitncnn_406 1 2 /model/layers.12/self_attn/Cast_3_output_0 /model/layers.12/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.12/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.12/self_attn/Unsqueeze_25 2 1 /model/layers.12/self_attn/Mul_7_output_0 onnx::Unsqueeze_4030 /model/layers.12/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.12/self_attn/Concat_9 5 1 /model/layers.12/self_attn/Unsqueeze_20_output_0 /model/layers.12/self_attn/Unsqueeze_21_output_0 /model/layers.12/self_attn/Constant_48_output_0 /model/layers.12/self_attn/Unsqueeze_22_output_0 /model/layers.12/self_attn/Unsqueeze_23_output_0 /model/layers.12/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_6 2 1 /model/layers.12/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_12_output_0 /model/layers.12/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_7 2 1 /model/layers.12/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.12/self_attn/Constant_15_output_0 /model/layers.12/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_8 2 1 /model/layers.12/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_22_output_0 /model/layers.12/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_9 2 1 /model/layers.12/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.12/self_attn/Constant_25_output_0 /model/layers.12/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.12/self_attn/Concat_10 4 1 /model/layers.12/self_attn/Unsqueeze_24_output_0 /model/layers.12/self_attn/Unsqueeze_25_output_0 /model/layers.12/self_attn/Unsqueeze_26_output_0 /model/layers.12/self_attn/Unsqueeze_27_output_0 /model/layers.12/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.12/self_attn/Reshape_5 1 1 /model/layers.12/self_attn/Concat_9_output_0 /model/layers.12/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_407 1 3 /model/layers.12/self_attn/Reshape_5_output_0 /model/layers.12/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.12/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.12/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.12/self_attn/Slice 2 1 /model/layers.12/self_attn/Transpose_output_0_splitncnn_1 /model/layers.12/self_attn/Unsqueeze_6_output_0 /model/layers.12/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.12/self_attn/Slice_1 2 1 /model/layers.12/self_attn/Transpose_output_0_splitncnn_0 /model/layers.12/self_attn/Unsqueeze_7_output_0 /model/layers.12/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.12/self_attn/Slice_2 2 1 /model/layers.12/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.12/self_attn/Unsqueeze_8_output_0 /model/layers.12/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.12/self_attn/Slice_3 2 1 /model/layers.12/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.12/self_attn/Unsqueeze_9_output_0 /model/layers.12/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.12/self_attn/Shape_13 1 1 /model/layers.12/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.12/self_attn/Shape_13_output_0 UnaryOp /model/layers.12/self_attn/Neg 1 1 /model/layers.12/self_attn/Slice_1_output_0 /model/layers.12/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.12/self_attn/Neg_1 1 1 /model/layers.12/self_attn/Slice_3_output_0 /model/layers.12/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.12/self_attn/ConstantOfShape_1 1 1 /model/layers.12/self_attn/Shape_13_output_0 /model/layers.12/self_attn/ConstantOfShape_1_output_0 Split splitncnn_408 1 2 /model/layers.12/self_attn/ConstantOfShape_1_output_0 /model/layers.12/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.12/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.12/self_attn/Concat_3 2 1 /model/layers.12/self_attn/Neg_output_0 /model/layers.12/self_attn/Slice_output_0 /model/layers.12/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.12/self_attn/Concat_4 2 1 /model/layers.12/self_attn/Neg_1_output_0 /model/layers.12/self_attn/Slice_2_output_0 /model/layers.12/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.12/self_attn/Mul_6 2 1 /model/layers.12/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_52_output_0 /model/layers.12/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.12/self_attn/Mul_1 2 1 /model/layers.12/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_23 /model/layers.12/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.12/self_attn/Mul_3 2 1 /model/layers.12/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_22 /model/layers.12/self_attn/Mul_3_output_0 0=2 Equal /model/layers.12/self_attn/Equal_1 2 1 /model/layers.12/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.12/self_attn/Mul_6_output_0 /model/layers.12/self_attn/Equal_1_output_0 BinaryOp /model/layers.12/self_attn/Add 2 1 /model/layers.12/self_attn/Mul_output_0 /model/layers.12/self_attn/Mul_1_output_0 /model/layers.12/self_attn/Add_output_0 0=0 BinaryOp /model/layers.12/self_attn/Add_1 2 1 /model/layers.12/self_attn/Mul_2_output_0 /model/layers.12/self_attn/Mul_3_output_0 /model/layers.12/self_attn/Add_1_output_0 0=0 Where /model/layers.12/self_attn/Where_1 3 1 /model/layers.12/self_attn/Equal_1_output_0 /model/layers.12/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.12/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.12/self_attn/Where_1_output_0 BinaryOp /model/layers.12/self_attn/Mul_8 1 1 /model/layers.12/self_attn/Add_output_0 /model/layers.12/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.12/self_attn/Concat_5 2 1 past_key_values.12.key /model/layers.12/self_attn/Add_1_output_0 present.12.key 0=-2 Split splitncnn_409 1 5 present.12.key present.12.key_splitncnn_0 present.12.key_splitncnn_1 present.12.key_splitncnn_2 present.12.key_splitncnn_3 present.12.key_splitncnn_4 Expand /model/layers.12/self_attn/Expand_1 2 1 /model/layers.12/self_attn/Unsqueeze_19_output_0 /model/layers.12/self_attn/Where_1_output_0 /model/layers.12/self_attn/Expand_1_output_0 Shape /model/layers.12/self_attn/Shape_4 1 1 present.12.key_splitncnn_4 /model/layers.12/self_attn/Shape_4_output_0 Shape /model/layers.12/self_attn/Shape_5 1 1 present.12.key_splitncnn_3 /model/layers.12/self_attn/Shape_5_output_0 Shape /model/layers.12/self_attn/Shape_6 1 1 present.12.key_splitncnn_2 /model/layers.12/self_attn/Shape_6_output_0 Shape /model/layers.12/self_attn/Shape_7 1 1 present.12.key_splitncnn_1 /model/layers.12/self_attn/Shape_7_output_0 ExpandDims /model/layers.12/self_attn/Unsqueeze_10 2 1 present.12.key_splitncnn_0 /model/layers.12/self_attn/Constant_32_output_0 /model/layers.12/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.12/self_attn/Reshape_6 2 1 /model/layers.12/self_attn/Expand_1_output_0 /model/layers.12/self_attn/Concat_10_output_0 /model/layers.12/self_attn/Reshape_6_output_0 Crop /model/layers.12/self_attn/Gather_4 1 1 /model/layers.12/self_attn/Shape_4_output_0 /model/layers.12/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_410 1 2 /model/layers.12/self_attn/Gather_4_output_0 /model/layers.12/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.12/self_attn/Gather_5 1 1 /model/layers.12/self_attn/Shape_5_output_0 /model/layers.12/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_411 1 2 /model/layers.12/self_attn/Gather_5_output_0 /model/layers.12/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.12/self_attn/Gather_6 1 1 /model/layers.12/self_attn/Shape_6_output_0 /model/layers.12/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_412 1 2 /model/layers.12/self_attn/Gather_6_output_0 /model/layers.12/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.12/self_attn/Gather_7 1 1 /model/layers.12/self_attn/Shape_7_output_0 /model/layers.12/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_413 1 2 /model/layers.12/self_attn/Gather_7_output_0 /model/layers.12/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.12/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.12/self_attn/Unsqueeze_11 2 1 /model/layers.12/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_33_output_0 /model/layers.12/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_15 2 1 /model/layers.12/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_3982 /model/layers.12/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_12 2 1 /model/layers.12/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_34_output_0 /model/layers.12/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.12/self_attn/Mul_5 2 1 /model/layers.12/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.12/self_attn/Constant_40_output_0 /model/layers.12/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.12/self_attn/Unsqueeze_13 2 1 /model/layers.12/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_36_output_0 /model/layers.12/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_17 2 1 /model/layers.12/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_3986 /model/layers.12/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_14 2 1 /model/layers.12/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_37_output_0 /model/layers.12/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_18 2 1 /model/layers.12/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_3988 /model/layers.12/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.12/self_attn/Unsqueeze_16 2 1 /model/layers.12/self_attn/Mul_5_output_0 onnx::Unsqueeze_3984 /model/layers.12/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.12/self_attn/Concat_7 5 1 /model/layers.12/self_attn/Unsqueeze_11_output_0 /model/layers.12/self_attn/Unsqueeze_12_output_0 /model/layers.12/self_attn/Constant_35_output_0 /model/layers.12/self_attn/Unsqueeze_13_output_0 /model/layers.12/self_attn/Unsqueeze_14_output_0 /model/layers.12/self_attn/Concat_7_output_0 0=0 Concat /model/layers.12/self_attn/Concat_8 4 1 /model/layers.12/self_attn/Unsqueeze_15_output_0 /model/layers.12/self_attn/Unsqueeze_16_output_0 /model/layers.12/self_attn/Unsqueeze_17_output_0 /model/layers.12/self_attn/Unsqueeze_18_output_0 /model/layers.12/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.12/self_attn/Reshape_3 1 1 /model/layers.12/self_attn/Concat_7_output_0 /model/layers.12/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_414 1 3 /model/layers.12/self_attn/Reshape_3_output_0 /model/layers.12/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.12/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.12/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.12/self_attn/Shape_8 1 1 /model/layers.12/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.12/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.12/self_attn/ConstantOfShape 1 1 /model/layers.12/self_attn/Shape_8_output_0 /model/layers.12/self_attn/ConstantOfShape_output_0 Split splitncnn_415 1 2 /model/layers.12/self_attn/ConstantOfShape_output_0 /model/layers.12/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.12/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.12/self_attn/Mul_4 2 1 /model/layers.12/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.12/self_attn/Constant_39_output_0 /model/layers.12/self_attn/Mul_4_output_0 0=2 Equal /model/layers.12/self_attn/Equal 2 1 /model/layers.12/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.12/self_attn/Mul_4_output_0 /model/layers.12/self_attn/Equal_output_0 Where /model/layers.12/self_attn/Where 3 1 /model/layers.12/self_attn/Equal_output_0 /model/layers.12/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.12/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.12/self_attn/Where_output_0 Expand /model/layers.12/self_attn/Expand 2 1 /model/layers.12/self_attn/Unsqueeze_10_output_0 /model/layers.12/self_attn/Where_output_0 /model/layers.12/self_attn/Expand_output_0 Reshape /model/layers.12/self_attn/Reshape_4 2 1 /model/layers.12/self_attn/Expand_output_0 /model/layers.12/self_attn/Concat_8_output_0 /model/layers.12/self_attn/Reshape_4_output_0 Split splitncnn_416 1 2 /model/layers.12/self_attn/Reshape_4_output_0 /model/layers.12/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.12/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.12/self_attn/Shape_14 1 1 /model/layers.12/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.12/self_attn/Shape_14_output_0 Permute /model/layers.12/self_attn/Transpose_3 1 1 /model/layers.12/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.12/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.12/self_attn/Gather_12 1 1 /model/layers.12/self_attn/Shape_14_output_0 /model/layers.12/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.12/self_attn/Mul_9 1 1 /model/layers.12/self_attn/Transpose_3_output_0 /model/layers.12/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.12/self_attn/Unsqueeze_28 2 1 /model/layers.12/self_attn/Gather_12_output_0 /model/layers.12/self_attn/Constant_57_output_0 /model/layers.12/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.12/self_attn/MatMul 2 1 /model/layers.12/self_attn/Mul_8_output_0 /model/layers.12/self_attn/Mul_9_output_0 /model/layers.12/self_attn/MatMul_output_0 Crop /model/layers.12/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_11 /model/layers.12/self_attn/Unsqueeze_28_output_0 /model/layers.12/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.12/self_attn/Where_2 3 1 /model/layers.12/self_attn/Slice_4_output_0 /model/layers.12/self_attn/Constant_61_output_0 /model/layers.12/self_attn/Constant_62_output_0 /model/layers.12/self_attn/Where_2_output_0 BinaryOp /model/layers.12/self_attn/Add_2 2 1 /model/layers.12/self_attn/MatMul_output_0 /model/layers.12/self_attn/Where_2_output_0 /model/layers.12/self_attn/Add_2_output_0 0=0 Softmax /model/layers.12/self_attn/Softmax 1 1 /model/layers.12/self_attn/Add_2_output_0 /model/layers.12/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_417 1 2 /model/layers.12/self_attn/Softmax_output_0 /model/layers.12/self_attn/Softmax_output_0_splitncnn_0 /model/layers.12/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.12/self_attn/IsNaN 1 1 /model/layers.12/self_attn/Softmax_output_0_splitncnn_1 /model/layers.12/self_attn/IsNaN_output_0 Where /model/layers.12/self_attn/Where_3 3 1 /model/layers.12/self_attn/IsNaN_output_0 /model/layers.12/self_attn/Constant_63_output_0 /model/layers.12/self_attn/Softmax_output_0_splitncnn_0 /model/layers.12/self_attn/Where_3_output_0 Gemm /model/layers.12/self_attn/MatMul_1 2 1 /model/layers.12/self_attn/Where_3_output_0 /model/layers.12/self_attn/Reshape_6_output_0 /model/layers.12/self_attn/MatMul_1_output_0 Permute /model/layers.12/self_attn/Transpose_4 1 1 /model/layers.12/self_attn/MatMul_1_output_0 /model/layers.12/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.12/self_attn/Reshape_7 2 1 /model/layers.12/self_attn/Transpose_4_output_0 /model/layers.12/self_attn/Concat_11_output_0 /model/layers.12/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.12/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.12/self_attn/Reshape_7_output_0 /model/layers.12/self_attn/Reshape_7_output_0_quantized /model/layers.12/self_attn/Reshape_7_output_0_scale /model/layers.12/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.12/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.12/self_attn/Reshape_7_output_0_scale /model/layers.12/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.552903e-03 MatMulInteger /model/layers.12/self_attn/o_proj/MatMul_quant 4 1 /model/layers.12/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7641_quantized /model/layers.12/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7641_zero_point /model/layers.12/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.12/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.12/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.12/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.12/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.12/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.12/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.12/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.12/Add 2 1 /model/layers.12/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.12/self_attn/o_proj/MatMul_output_0 /model/layers.12/Add_output_0 0=0 Cast /model/layers.12/post_attention_layernorm/Cast 1 1 /model/layers.12/Add_output_0 /model/layers.12/post_attention_layernorm/Cast_output_0 Split splitncnn_418 1 3 /model/layers.12/post_attention_layernorm/Cast_output_0 /model/layers.12/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.12/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.12/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.12/post_attention_layernorm/Pow 1 1 /model/layers.12/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.12/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.12/post_attention_layernorm/ReduceMean 2 1 /model/layers.12/post_attention_layernorm/Pow_output_0 /model/layers.12/post_attention_layernorm/Constant_1_output_0 /model/layers.12/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.12/post_attention_layernorm/Add 1 1 /model/layers.12/post_attention_layernorm/ReduceMean_output_0 /model/layers.12/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.12/post_attention_layernorm/Sqrt 1 1 /model/layers.12/post_attention_layernorm/Add_output_0 /model/layers.12/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.12/post_attention_layernorm/Div 1 1 /model/layers.12/post_attention_layernorm/Sqrt_output_0 /model/layers.12/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.12/post_attention_layernorm/Mul 2 1 /model/layers.12/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.12/post_attention_layernorm/Div_output_0 /model/layers.12/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.12/post_attention_layernorm/Cast_1 1 1 /model/layers.12/post_attention_layernorm/Mul_output_0 /model/layers.12/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.12/post_attention_layernorm/Mul_1 2 1 model.layers.12.post_attention_layernorm.weight /model/layers.12/post_attention_layernorm/Cast_1_output_0 /model/layers.12/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.12/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.12/post_attention_layernorm/Mul_1_output_0 /model/layers.12/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.12/post_attention_layernorm/Mul_1_output_0_scale /model/layers.12/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_419 1 2 /model/layers.12/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.12/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.12/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_420 1 2 /model/layers.12/post_attention_layernorm/Mul_1_output_0_scale /model/layers.12/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.12/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_421 1 2 /model/layers.12/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.12/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.12/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.12/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.12/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.12/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.645177e-03 BinaryOp /model/layers.12/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.12/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.12/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.029651e-03 MatMulInteger /model/layers.12/mlp/gate_proj/MatMul_quant 4 1 /model/layers.12/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7642_quantized /model/layers.12/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7642_zero_point /model/layers.12/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.12/mlp/up_proj/MatMul_quant 4 1 /model/layers.12/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7643_quantized /model/layers.12/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7643_zero_point /model/layers.12/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.12/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.12/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.12/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.12/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.12/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.12/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.12/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.12/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.12/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.12/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.12/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.12/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.12/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.12/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.12/mlp/act_fn/Mul 1 1 /model/layers.12/mlp/gate_proj/MatMul_output_0 /model/layers.12/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.12/mlp/Mul 2 1 /model/layers.12/mlp/act_fn/Mul_output_0 /model/layers.12/mlp/up_proj/MatMul_output_0 /model/layers.12/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.12/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.12/mlp/Mul_output_0 /model/layers.12/mlp/Mul_output_0_quantized /model/layers.12/mlp/Mul_output_0_scale /model/layers.12/mlp/Mul_output_0_zero_point BinaryOp /model/layers.12/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.12/mlp/Mul_output_0_scale /model/layers.12/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.460261e-03 MatMulInteger /model/layers.12/mlp/down_proj/MatMul_quant 4 1 /model/layers.12/mlp/Mul_output_0_quantized onnx::MatMul_7644_quantized /model/layers.12/mlp/Mul_output_0_zero_point onnx::MatMul_7644_zero_point /model/layers.12/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.12/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.12/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.12/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.12/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.12/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.12/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.12/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.12/Add_1 2 1 /model/layers.12/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.12/mlp/down_proj/MatMul_output_0 /model/layers.12/Add_1_output_0 0=0 Cast /model/layers.13/input_layernorm/Cast 1 1 /model/layers.12/Add_1_output_0 /model/layers.13/input_layernorm/Cast_output_0 Split splitncnn_422 1 3 /model/layers.13/input_layernorm/Cast_output_0 /model/layers.13/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.13/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.13/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.13/input_layernorm/Pow 1 1 /model/layers.13/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.13/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.13/input_layernorm/ReduceMean 2 1 /model/layers.13/input_layernorm/Pow_output_0 /model/layers.13/input_layernorm/Constant_1_output_0 /model/layers.13/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.13/input_layernorm/Add 1 1 /model/layers.13/input_layernorm/ReduceMean_output_0 /model/layers.13/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.13/input_layernorm/Sqrt 1 1 /model/layers.13/input_layernorm/Add_output_0 /model/layers.13/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.13/input_layernorm/Div 1 1 /model/layers.13/input_layernorm/Sqrt_output_0 /model/layers.13/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.13/input_layernorm/Mul 2 1 /model/layers.13/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.13/input_layernorm/Div_output_0 /model/layers.13/input_layernorm/Mul_output_0 0=2 Cast /model/layers.13/input_layernorm/Cast_1 1 1 /model/layers.13/input_layernorm/Mul_output_0 /model/layers.13/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.13/input_layernorm/Mul_1 2 1 model.layers.13.input_layernorm.weight /model/layers.13/input_layernorm/Cast_1_output_0 /model/layers.13/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_423 1 3 /model/layers.13/input_layernorm/Mul_1_output_0 /model/layers.13/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.13/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.13/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.13/self_attn/Shape 1 1 /model/layers.13/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.13/self_attn/Shape_output_0 Shape /model/layers.13/self_attn/Shape_1 1 1 /model/layers.13/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.13/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.13/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.13/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.13/input_layernorm/Mul_1_output_0_quantized /model/layers.13/input_layernorm/Mul_1_output_0_scale /model/layers.13/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_424 1 3 /model/layers.13/input_layernorm/Mul_1_output_0_quantized /model/layers.13/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.13/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.13/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_425 1 3 /model/layers.13/input_layernorm/Mul_1_output_0_scale /model/layers.13/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.13/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.13/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_426 1 3 /model/layers.13/input_layernorm/Mul_1_output_0_zero_point /model/layers.13/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.13/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.13/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.13/self_attn/Gather 1 1 /model/layers.13/self_attn/Shape_output_0 /model/layers.13/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_427 1 4 /model/layers.13/self_attn/Gather_output_0 /model/layers.13/self_attn/Gather_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_output_0_splitncnn_1 /model/layers.13/self_attn/Gather_output_0_splitncnn_2 /model/layers.13/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.13/self_attn/Gather_1 1 1 /model/layers.13/self_attn/Shape_1_output_0 /model/layers.13/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_428 1 4 /model/layers.13/self_attn/Gather_1_output_0 /model/layers.13/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.13/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.13/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.13/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.13/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.13/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.537155e-03 BinaryOp /model/layers.13/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.13/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.13/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.245325e-03 BinaryOp /model/layers.13/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.13/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.13/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=9.035125e-04 MatMulInteger /model/layers.13/self_attn/q_proj/MatMul_quant 4 1 /model/layers.13/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7645_quantized /model/layers.13/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7645_zero_point /model/layers.13/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.13/self_attn/k_proj/MatMul_quant 4 1 /model/layers.13/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7652_quantized /model/layers.13/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7652_zero_point /model/layers.13/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.13/self_attn/v_proj/MatMul_quant 4 1 /model/layers.13/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7653_quantized /model/layers.13/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7653_zero_point /model/layers.13/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.13/self_attn/Unsqueeze 2 1 /model/layers.13/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_4124 /model/layers.13/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_2 2 1 /model/layers.13/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_4133 /model/layers.13/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_4 2 1 /model/layers.13/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_4142 /model/layers.13/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_29 2 1 /model/layers.13/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_4347 /model/layers.13/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_1 2 1 /model/layers.13/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_4126 /model/layers.13/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_3 2 1 /model/layers.13/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_4135 /model/layers.13/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_5 2 1 /model/layers.13/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_4144 /model/layers.13/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_30 2 1 /model/layers.13/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_4349 /model/layers.13/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.13/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.13/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.13/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.13/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.13/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.13/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.13/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.13/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.13/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.13/self_attn/Concat 4 1 /model/layers.13/self_attn/Unsqueeze_output_0 /model/layers.13/self_attn/Unsqueeze_1_output_0 /model/layers.13/self_attn/Constant_2_output_0 /model/layers.13/self_attn/Constant_3_output_0 /model/layers.13/self_attn/Concat_output_0 0=0 Concat /model/layers.13/self_attn/Concat_1 4 1 /model/layers.13/self_attn/Unsqueeze_2_output_0 /model/layers.13/self_attn/Unsqueeze_3_output_0 /model/layers.13/self_attn/Constant_4_output_0 /model/layers.13/self_attn/Constant_5_output_0 /model/layers.13/self_attn/Concat_1_output_0 0=0 Concat /model/layers.13/self_attn/Concat_2 4 1 /model/layers.13/self_attn/Unsqueeze_4_output_0 /model/layers.13/self_attn/Unsqueeze_5_output_0 /model/layers.13/self_attn/Constant_6_output_0 /model/layers.13/self_attn/Constant_7_output_0 /model/layers.13/self_attn/Concat_2_output_0 0=0 Concat /model/layers.13/self_attn/Concat_11 3 1 /model/layers.13/self_attn/Unsqueeze_29_output_0 /model/layers.13/self_attn/Unsqueeze_30_output_0 /model/layers.13/self_attn/Constant_64_output_0 /model/layers.13/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.13/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.13/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.13/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.13/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.13/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.13/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.13/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.13/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.13/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.13/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.13/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.13/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.13/self_attn/q_proj/Add 2 1 model.layers.13.self_attn.q_proj.bias /model/layers.13/self_attn/q_proj/MatMul_output_0 /model/layers.13/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.13/self_attn/k_proj/Add 2 1 model.layers.13.self_attn.k_proj.bias /model/layers.13/self_attn/k_proj/MatMul_output_0 /model/layers.13/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.13/self_attn/v_proj/Add 2 1 model.layers.13.self_attn.v_proj.bias /model/layers.13/self_attn/v_proj/MatMul_output_0 /model/layers.13/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.13/self_attn/Reshape 2 1 /model/layers.13/self_attn/q_proj/Add_output_0 /model/layers.13/self_attn/Concat_output_0 /model/layers.13/self_attn/Reshape_output_0 Reshape /model/layers.13/self_attn/Reshape_1 2 1 /model/layers.13/self_attn/k_proj/Add_output_0 /model/layers.13/self_attn/Concat_1_output_0 /model/layers.13/self_attn/Reshape_1_output_0 Reshape /model/layers.13/self_attn/Reshape_2 2 1 /model/layers.13/self_attn/v_proj/Add_output_0 /model/layers.13/self_attn/Concat_2_output_0 /model/layers.13/self_attn/Reshape_2_output_0 Permute /model/layers.13/self_attn/Transpose 1 1 /model/layers.13/self_attn/Reshape_output_0 /model/layers.13/self_attn/Transpose_output_0 0=2 Split splitncnn_429 1 4 /model/layers.13/self_attn/Transpose_output_0 /model/layers.13/self_attn/Transpose_output_0_splitncnn_0 /model/layers.13/self_attn/Transpose_output_0_splitncnn_1 /model/layers.13/self_attn/Transpose_output_0_splitncnn_2 /model/layers.13/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.13/self_attn/Transpose_1 1 1 /model/layers.13/self_attn/Reshape_1_output_0 /model/layers.13/self_attn/Transpose_1_output_0 0=2 Split splitncnn_430 1 4 /model/layers.13/self_attn/Transpose_1_output_0 /model/layers.13/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.13/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.13/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.13/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.13/self_attn/Transpose_2 1 1 /model/layers.13/self_attn/Reshape_2_output_0 /model/layers.13/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.13/self_attn/Mul 2 1 /model/layers.13/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_21 /model/layers.13/self_attn/Mul_output_0 0=2 Shape /model/layers.13/self_attn/Shape_2 1 1 /model/layers.13/self_attn/Transpose_output_0_splitncnn_2 /model/layers.13/self_attn/Shape_2_output_0 BinaryOp /model/layers.13/self_attn/Mul_2 2 1 /model/layers.13/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_20 /model/layers.13/self_attn/Mul_2_output_0 0=2 Shape /model/layers.13/self_attn/Shape_3 1 1 /model/layers.13/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.13/self_attn/Shape_3_output_0 Concat /model/layers.13/self_attn/Concat_6 2 1 past_key_values.13.value /model/layers.13/self_attn/Transpose_2_output_0 present.13.value 0=-2 Split splitncnn_431 1 5 present.13.value present.13.value_splitncnn_0 present.13.value_splitncnn_1 present.13.value_splitncnn_2 present.13.value_splitncnn_3 present.13.value_splitncnn_4 Crop /model/layers.13/self_attn/Gather_2 1 1 /model/layers.13/self_attn/Shape_2_output_0 /model/layers.13/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.13/self_attn/Gather_3 1 1 /model/layers.13/self_attn/Shape_3_output_0 /model/layers.13/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.13/self_attn/Shape_9 1 1 present.13.value_splitncnn_4 /model/layers.13/self_attn/Shape_9_output_0 Shape /model/layers.13/self_attn/Shape_10 1 1 present.13.value_splitncnn_3 /model/layers.13/self_attn/Shape_10_output_0 Shape /model/layers.13/self_attn/Shape_11 1 1 present.13.value_splitncnn_2 /model/layers.13/self_attn/Shape_11_output_0 Shape /model/layers.13/self_attn/Shape_12 1 1 present.13.value_splitncnn_1 /model/layers.13/self_attn/Shape_12_output_0 ExpandDims /model/layers.13/self_attn/Unsqueeze_19 2 1 present.13.value_splitncnn_0 /model/layers.13/self_attn/Constant_45_output_0 /model/layers.13/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.13/self_attn/Div 2 1 /model/layers.13/self_attn/Gather_2_output_0 /model/layers.13/self_attn/Constant_9_output_0 /model/layers.13/self_attn/Div_output_0 0=3 BinaryOp /model/layers.13/self_attn/Div_1 2 1 /model/layers.13/self_attn/Gather_3_output_0 /model/layers.13/self_attn/Constant_19_output_0 /model/layers.13/self_attn/Div_1_output_0 0=3 Crop /model/layers.13/self_attn/Gather_8 1 1 /model/layers.13/self_attn/Shape_9_output_0 /model/layers.13/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_432 1 2 /model/layers.13/self_attn/Gather_8_output_0 /model/layers.13/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.13/self_attn/Gather_9 1 1 /model/layers.13/self_attn/Shape_10_output_0 /model/layers.13/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_433 1 2 /model/layers.13/self_attn/Gather_9_output_0 /model/layers.13/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.13/self_attn/Gather_10 1 1 /model/layers.13/self_attn/Shape_11_output_0 /model/layers.13/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_434 1 2 /model/layers.13/self_attn/Gather_10_output_0 /model/layers.13/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.13/self_attn/Gather_11 1 1 /model/layers.13/self_attn/Shape_12_output_0 /model/layers.13/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_435 1 2 /model/layers.13/self_attn/Gather_11_output_0 /model/layers.13/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.13/self_attn/Cast 1 1 /model/layers.13/self_attn/Div_output_0 /model/layers.13/self_attn/Cast_output_0 Cast /model/layers.13/self_attn/Cast_2 1 1 /model/layers.13/self_attn/Div_1_output_0 /model/layers.13/self_attn/Cast_2_output_0 ExpandDims /model/layers.13/self_attn/Unsqueeze_20 2 1 /model/layers.13/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_46_output_0 /model/layers.13/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_24 2 1 /model/layers.13/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_4307 /model/layers.13/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_21 2 1 /model/layers.13/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_47_output_0 /model/layers.13/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.13/self_attn/Mul_7 2 1 /model/layers.13/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.13/self_attn/Constant_53_output_0 /model/layers.13/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.13/self_attn/Unsqueeze_22 2 1 /model/layers.13/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_49_output_0 /model/layers.13/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_26 2 1 /model/layers.13/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_4311 /model/layers.13/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_23 2 1 /model/layers.13/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_50_output_0 /model/layers.13/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_27 2 1 /model/layers.13/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_4313 /model/layers.13/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.13/self_attn/Cast_1 1 1 /model/layers.13/self_attn/Cast_output_0 /model/layers.13/self_attn/Cast_1_output_0 Split splitncnn_436 1 2 /model/layers.13/self_attn/Cast_1_output_0 /model/layers.13/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.13/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.13/self_attn/Cast_3 1 1 /model/layers.13/self_attn/Cast_2_output_0 /model/layers.13/self_attn/Cast_3_output_0 Split splitncnn_437 1 2 /model/layers.13/self_attn/Cast_3_output_0 /model/layers.13/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.13/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.13/self_attn/Unsqueeze_25 2 1 /model/layers.13/self_attn/Mul_7_output_0 onnx::Unsqueeze_4309 /model/layers.13/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.13/self_attn/Concat_9 5 1 /model/layers.13/self_attn/Unsqueeze_20_output_0 /model/layers.13/self_attn/Unsqueeze_21_output_0 /model/layers.13/self_attn/Constant_48_output_0 /model/layers.13/self_attn/Unsqueeze_22_output_0 /model/layers.13/self_attn/Unsqueeze_23_output_0 /model/layers.13/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_6 2 1 /model/layers.13/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_12_output_0 /model/layers.13/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_7 2 1 /model/layers.13/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.13/self_attn/Constant_15_output_0 /model/layers.13/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_8 2 1 /model/layers.13/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_22_output_0 /model/layers.13/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_9 2 1 /model/layers.13/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.13/self_attn/Constant_25_output_0 /model/layers.13/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.13/self_attn/Concat_10 4 1 /model/layers.13/self_attn/Unsqueeze_24_output_0 /model/layers.13/self_attn/Unsqueeze_25_output_0 /model/layers.13/self_attn/Unsqueeze_26_output_0 /model/layers.13/self_attn/Unsqueeze_27_output_0 /model/layers.13/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.13/self_attn/Reshape_5 1 1 /model/layers.13/self_attn/Concat_9_output_0 /model/layers.13/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_438 1 3 /model/layers.13/self_attn/Reshape_5_output_0 /model/layers.13/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.13/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.13/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.13/self_attn/Slice 2 1 /model/layers.13/self_attn/Transpose_output_0_splitncnn_1 /model/layers.13/self_attn/Unsqueeze_6_output_0 /model/layers.13/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.13/self_attn/Slice_1 2 1 /model/layers.13/self_attn/Transpose_output_0_splitncnn_0 /model/layers.13/self_attn/Unsqueeze_7_output_0 /model/layers.13/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.13/self_attn/Slice_2 2 1 /model/layers.13/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.13/self_attn/Unsqueeze_8_output_0 /model/layers.13/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.13/self_attn/Slice_3 2 1 /model/layers.13/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.13/self_attn/Unsqueeze_9_output_0 /model/layers.13/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.13/self_attn/Shape_13 1 1 /model/layers.13/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.13/self_attn/Shape_13_output_0 UnaryOp /model/layers.13/self_attn/Neg 1 1 /model/layers.13/self_attn/Slice_1_output_0 /model/layers.13/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.13/self_attn/Neg_1 1 1 /model/layers.13/self_attn/Slice_3_output_0 /model/layers.13/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.13/self_attn/ConstantOfShape_1 1 1 /model/layers.13/self_attn/Shape_13_output_0 /model/layers.13/self_attn/ConstantOfShape_1_output_0 Split splitncnn_439 1 2 /model/layers.13/self_attn/ConstantOfShape_1_output_0 /model/layers.13/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.13/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.13/self_attn/Concat_3 2 1 /model/layers.13/self_attn/Neg_output_0 /model/layers.13/self_attn/Slice_output_0 /model/layers.13/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.13/self_attn/Concat_4 2 1 /model/layers.13/self_attn/Neg_1_output_0 /model/layers.13/self_attn/Slice_2_output_0 /model/layers.13/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.13/self_attn/Mul_6 2 1 /model/layers.13/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_52_output_0 /model/layers.13/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.13/self_attn/Mul_1 2 1 /model/layers.13/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_21 /model/layers.13/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.13/self_attn/Mul_3 2 1 /model/layers.13/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_20 /model/layers.13/self_attn/Mul_3_output_0 0=2 Equal /model/layers.13/self_attn/Equal_1 2 1 /model/layers.13/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.13/self_attn/Mul_6_output_0 /model/layers.13/self_attn/Equal_1_output_0 BinaryOp /model/layers.13/self_attn/Add 2 1 /model/layers.13/self_attn/Mul_output_0 /model/layers.13/self_attn/Mul_1_output_0 /model/layers.13/self_attn/Add_output_0 0=0 BinaryOp /model/layers.13/self_attn/Add_1 2 1 /model/layers.13/self_attn/Mul_2_output_0 /model/layers.13/self_attn/Mul_3_output_0 /model/layers.13/self_attn/Add_1_output_0 0=0 Where /model/layers.13/self_attn/Where_1 3 1 /model/layers.13/self_attn/Equal_1_output_0 /model/layers.13/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.13/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.13/self_attn/Where_1_output_0 BinaryOp /model/layers.13/self_attn/Mul_8 1 1 /model/layers.13/self_attn/Add_output_0 /model/layers.13/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.13/self_attn/Concat_5 2 1 past_key_values.13.key /model/layers.13/self_attn/Add_1_output_0 present.13.key 0=-2 Split splitncnn_440 1 5 present.13.key present.13.key_splitncnn_0 present.13.key_splitncnn_1 present.13.key_splitncnn_2 present.13.key_splitncnn_3 present.13.key_splitncnn_4 Expand /model/layers.13/self_attn/Expand_1 2 1 /model/layers.13/self_attn/Unsqueeze_19_output_0 /model/layers.13/self_attn/Where_1_output_0 /model/layers.13/self_attn/Expand_1_output_0 Shape /model/layers.13/self_attn/Shape_4 1 1 present.13.key_splitncnn_4 /model/layers.13/self_attn/Shape_4_output_0 Shape /model/layers.13/self_attn/Shape_5 1 1 present.13.key_splitncnn_3 /model/layers.13/self_attn/Shape_5_output_0 Shape /model/layers.13/self_attn/Shape_6 1 1 present.13.key_splitncnn_2 /model/layers.13/self_attn/Shape_6_output_0 Shape /model/layers.13/self_attn/Shape_7 1 1 present.13.key_splitncnn_1 /model/layers.13/self_attn/Shape_7_output_0 ExpandDims /model/layers.13/self_attn/Unsqueeze_10 2 1 present.13.key_splitncnn_0 /model/layers.13/self_attn/Constant_32_output_0 /model/layers.13/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.13/self_attn/Reshape_6 2 1 /model/layers.13/self_attn/Expand_1_output_0 /model/layers.13/self_attn/Concat_10_output_0 /model/layers.13/self_attn/Reshape_6_output_0 Crop /model/layers.13/self_attn/Gather_4 1 1 /model/layers.13/self_attn/Shape_4_output_0 /model/layers.13/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_441 1 2 /model/layers.13/self_attn/Gather_4_output_0 /model/layers.13/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.13/self_attn/Gather_5 1 1 /model/layers.13/self_attn/Shape_5_output_0 /model/layers.13/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_442 1 2 /model/layers.13/self_attn/Gather_5_output_0 /model/layers.13/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.13/self_attn/Gather_6 1 1 /model/layers.13/self_attn/Shape_6_output_0 /model/layers.13/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_443 1 2 /model/layers.13/self_attn/Gather_6_output_0 /model/layers.13/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.13/self_attn/Gather_7 1 1 /model/layers.13/self_attn/Shape_7_output_0 /model/layers.13/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_444 1 2 /model/layers.13/self_attn/Gather_7_output_0 /model/layers.13/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.13/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.13/self_attn/Unsqueeze_11 2 1 /model/layers.13/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_33_output_0 /model/layers.13/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_15 2 1 /model/layers.13/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_4261 /model/layers.13/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_12 2 1 /model/layers.13/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_34_output_0 /model/layers.13/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.13/self_attn/Mul_5 2 1 /model/layers.13/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.13/self_attn/Constant_40_output_0 /model/layers.13/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.13/self_attn/Unsqueeze_13 2 1 /model/layers.13/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_36_output_0 /model/layers.13/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_17 2 1 /model/layers.13/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_4265 /model/layers.13/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_14 2 1 /model/layers.13/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_37_output_0 /model/layers.13/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_18 2 1 /model/layers.13/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_4267 /model/layers.13/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.13/self_attn/Unsqueeze_16 2 1 /model/layers.13/self_attn/Mul_5_output_0 onnx::Unsqueeze_4263 /model/layers.13/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.13/self_attn/Concat_7 5 1 /model/layers.13/self_attn/Unsqueeze_11_output_0 /model/layers.13/self_attn/Unsqueeze_12_output_0 /model/layers.13/self_attn/Constant_35_output_0 /model/layers.13/self_attn/Unsqueeze_13_output_0 /model/layers.13/self_attn/Unsqueeze_14_output_0 /model/layers.13/self_attn/Concat_7_output_0 0=0 Concat /model/layers.13/self_attn/Concat_8 4 1 /model/layers.13/self_attn/Unsqueeze_15_output_0 /model/layers.13/self_attn/Unsqueeze_16_output_0 /model/layers.13/self_attn/Unsqueeze_17_output_0 /model/layers.13/self_attn/Unsqueeze_18_output_0 /model/layers.13/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.13/self_attn/Reshape_3 1 1 /model/layers.13/self_attn/Concat_7_output_0 /model/layers.13/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_445 1 3 /model/layers.13/self_attn/Reshape_3_output_0 /model/layers.13/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.13/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.13/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.13/self_attn/Shape_8 1 1 /model/layers.13/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.13/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.13/self_attn/ConstantOfShape 1 1 /model/layers.13/self_attn/Shape_8_output_0 /model/layers.13/self_attn/ConstantOfShape_output_0 Split splitncnn_446 1 2 /model/layers.13/self_attn/ConstantOfShape_output_0 /model/layers.13/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.13/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.13/self_attn/Mul_4 2 1 /model/layers.13/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.13/self_attn/Constant_39_output_0 /model/layers.13/self_attn/Mul_4_output_0 0=2 Equal /model/layers.13/self_attn/Equal 2 1 /model/layers.13/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.13/self_attn/Mul_4_output_0 /model/layers.13/self_attn/Equal_output_0 Where /model/layers.13/self_attn/Where 3 1 /model/layers.13/self_attn/Equal_output_0 /model/layers.13/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.13/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.13/self_attn/Where_output_0 Expand /model/layers.13/self_attn/Expand 2 1 /model/layers.13/self_attn/Unsqueeze_10_output_0 /model/layers.13/self_attn/Where_output_0 /model/layers.13/self_attn/Expand_output_0 Reshape /model/layers.13/self_attn/Reshape_4 2 1 /model/layers.13/self_attn/Expand_output_0 /model/layers.13/self_attn/Concat_8_output_0 /model/layers.13/self_attn/Reshape_4_output_0 Split splitncnn_447 1 2 /model/layers.13/self_attn/Reshape_4_output_0 /model/layers.13/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.13/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.13/self_attn/Shape_14 1 1 /model/layers.13/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.13/self_attn/Shape_14_output_0 Permute /model/layers.13/self_attn/Transpose_3 1 1 /model/layers.13/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.13/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.13/self_attn/Gather_12 1 1 /model/layers.13/self_attn/Shape_14_output_0 /model/layers.13/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.13/self_attn/Mul_9 1 1 /model/layers.13/self_attn/Transpose_3_output_0 /model/layers.13/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.13/self_attn/Unsqueeze_28 2 1 /model/layers.13/self_attn/Gather_12_output_0 /model/layers.13/self_attn/Constant_57_output_0 /model/layers.13/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.13/self_attn/MatMul 2 1 /model/layers.13/self_attn/Mul_8_output_0 /model/layers.13/self_attn/Mul_9_output_0 /model/layers.13/self_attn/MatMul_output_0 Crop /model/layers.13/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_10 /model/layers.13/self_attn/Unsqueeze_28_output_0 /model/layers.13/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.13/self_attn/Where_2 3 1 /model/layers.13/self_attn/Slice_4_output_0 /model/layers.13/self_attn/Constant_61_output_0 /model/layers.13/self_attn/Constant_62_output_0 /model/layers.13/self_attn/Where_2_output_0 BinaryOp /model/layers.13/self_attn/Add_2 2 1 /model/layers.13/self_attn/MatMul_output_0 /model/layers.13/self_attn/Where_2_output_0 /model/layers.13/self_attn/Add_2_output_0 0=0 Softmax /model/layers.13/self_attn/Softmax 1 1 /model/layers.13/self_attn/Add_2_output_0 /model/layers.13/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_448 1 2 /model/layers.13/self_attn/Softmax_output_0 /model/layers.13/self_attn/Softmax_output_0_splitncnn_0 /model/layers.13/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.13/self_attn/IsNaN 1 1 /model/layers.13/self_attn/Softmax_output_0_splitncnn_1 /model/layers.13/self_attn/IsNaN_output_0 Where /model/layers.13/self_attn/Where_3 3 1 /model/layers.13/self_attn/IsNaN_output_0 /model/layers.13/self_attn/Constant_63_output_0 /model/layers.13/self_attn/Softmax_output_0_splitncnn_0 /model/layers.13/self_attn/Where_3_output_0 Gemm /model/layers.13/self_attn/MatMul_1 2 1 /model/layers.13/self_attn/Where_3_output_0 /model/layers.13/self_attn/Reshape_6_output_0 /model/layers.13/self_attn/MatMul_1_output_0 Permute /model/layers.13/self_attn/Transpose_4 1 1 /model/layers.13/self_attn/MatMul_1_output_0 /model/layers.13/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.13/self_attn/Reshape_7 2 1 /model/layers.13/self_attn/Transpose_4_output_0 /model/layers.13/self_attn/Concat_11_output_0 /model/layers.13/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.13/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.13/self_attn/Reshape_7_output_0 /model/layers.13/self_attn/Reshape_7_output_0_quantized /model/layers.13/self_attn/Reshape_7_output_0_scale /model/layers.13/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.13/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.13/self_attn/Reshape_7_output_0_scale /model/layers.13/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.875861e-03 MatMulInteger /model/layers.13/self_attn/o_proj/MatMul_quant 4 1 /model/layers.13/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7676_quantized /model/layers.13/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7676_zero_point /model/layers.13/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.13/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.13/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.13/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.13/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.13/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.13/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.13/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.13/Add 2 1 /model/layers.13/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.13/self_attn/o_proj/MatMul_output_0 /model/layers.13/Add_output_0 0=0 Cast /model/layers.13/post_attention_layernorm/Cast 1 1 /model/layers.13/Add_output_0 /model/layers.13/post_attention_layernorm/Cast_output_0 Split splitncnn_449 1 3 /model/layers.13/post_attention_layernorm/Cast_output_0 /model/layers.13/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.13/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.13/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.13/post_attention_layernorm/Pow 1 1 /model/layers.13/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.13/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.13/post_attention_layernorm/ReduceMean 2 1 /model/layers.13/post_attention_layernorm/Pow_output_0 /model/layers.13/post_attention_layernorm/Constant_1_output_0 /model/layers.13/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.13/post_attention_layernorm/Add 1 1 /model/layers.13/post_attention_layernorm/ReduceMean_output_0 /model/layers.13/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.13/post_attention_layernorm/Sqrt 1 1 /model/layers.13/post_attention_layernorm/Add_output_0 /model/layers.13/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.13/post_attention_layernorm/Div 1 1 /model/layers.13/post_attention_layernorm/Sqrt_output_0 /model/layers.13/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.13/post_attention_layernorm/Mul 2 1 /model/layers.13/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.13/post_attention_layernorm/Div_output_0 /model/layers.13/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.13/post_attention_layernorm/Cast_1 1 1 /model/layers.13/post_attention_layernorm/Mul_output_0 /model/layers.13/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.13/post_attention_layernorm/Mul_1 2 1 model.layers.13.post_attention_layernorm.weight /model/layers.13/post_attention_layernorm/Cast_1_output_0 /model/layers.13/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.13/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.13/post_attention_layernorm/Mul_1_output_0 /model/layers.13/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.13/post_attention_layernorm/Mul_1_output_0_scale /model/layers.13/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_450 1 2 /model/layers.13/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.13/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.13/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_451 1 2 /model/layers.13/post_attention_layernorm/Mul_1_output_0_scale /model/layers.13/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.13/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_452 1 2 /model/layers.13/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.13/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.13/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.13/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.13/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.13/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.460261e-03 BinaryOp /model/layers.13/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.13/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.13/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.752461e-03 MatMulInteger /model/layers.13/mlp/gate_proj/MatMul_quant 4 1 /model/layers.13/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7677_quantized /model/layers.13/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7677_zero_point /model/layers.13/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.13/mlp/up_proj/MatMul_quant 4 1 /model/layers.13/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7678_quantized /model/layers.13/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7678_zero_point /model/layers.13/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.13/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.13/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.13/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.13/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.13/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.13/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.13/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.13/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.13/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.13/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.13/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.13/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.13/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.13/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.13/mlp/act_fn/Mul 1 1 /model/layers.13/mlp/gate_proj/MatMul_output_0 /model/layers.13/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.13/mlp/Mul 2 1 /model/layers.13/mlp/act_fn/Mul_output_0 /model/layers.13/mlp/up_proj/MatMul_output_0 /model/layers.13/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.13/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.13/mlp/Mul_output_0 /model/layers.13/mlp/Mul_output_0_quantized /model/layers.13/mlp/Mul_output_0_scale /model/layers.13/mlp/Mul_output_0_zero_point BinaryOp /model/layers.13/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.13/mlp/Mul_output_0_scale /model/layers.13/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.060039e-03 MatMulInteger /model/layers.13/mlp/down_proj/MatMul_quant 4 1 /model/layers.13/mlp/Mul_output_0_quantized onnx::MatMul_7679_quantized /model/layers.13/mlp/Mul_output_0_zero_point onnx::MatMul_7679_zero_point /model/layers.13/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.13/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.13/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.13/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.13/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.13/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.13/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.13/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.13/Add_1 2 1 /model/layers.13/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.13/mlp/down_proj/MatMul_output_0 /model/layers.13/Add_1_output_0 0=0 Cast /model/layers.14/input_layernorm/Cast 1 1 /model/layers.13/Add_1_output_0 /model/layers.14/input_layernorm/Cast_output_0 Split splitncnn_453 1 3 /model/layers.14/input_layernorm/Cast_output_0 /model/layers.14/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.14/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.14/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.14/input_layernorm/Pow 1 1 /model/layers.14/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.14/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.14/input_layernorm/ReduceMean 2 1 /model/layers.14/input_layernorm/Pow_output_0 /model/layers.14/input_layernorm/Constant_1_output_0 /model/layers.14/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.14/input_layernorm/Add 1 1 /model/layers.14/input_layernorm/ReduceMean_output_0 /model/layers.14/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.14/input_layernorm/Sqrt 1 1 /model/layers.14/input_layernorm/Add_output_0 /model/layers.14/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.14/input_layernorm/Div 1 1 /model/layers.14/input_layernorm/Sqrt_output_0 /model/layers.14/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.14/input_layernorm/Mul 2 1 /model/layers.14/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.14/input_layernorm/Div_output_0 /model/layers.14/input_layernorm/Mul_output_0 0=2 Cast /model/layers.14/input_layernorm/Cast_1 1 1 /model/layers.14/input_layernorm/Mul_output_0 /model/layers.14/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.14/input_layernorm/Mul_1 2 1 model.layers.14.input_layernorm.weight /model/layers.14/input_layernorm/Cast_1_output_0 /model/layers.14/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_454 1 3 /model/layers.14/input_layernorm/Mul_1_output_0 /model/layers.14/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.14/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.14/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.14/self_attn/Shape 1 1 /model/layers.14/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.14/self_attn/Shape_output_0 Shape /model/layers.14/self_attn/Shape_1 1 1 /model/layers.14/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.14/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.14/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.14/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.14/input_layernorm/Mul_1_output_0_quantized /model/layers.14/input_layernorm/Mul_1_output_0_scale /model/layers.14/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_455 1 3 /model/layers.14/input_layernorm/Mul_1_output_0_quantized /model/layers.14/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.14/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.14/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_456 1 3 /model/layers.14/input_layernorm/Mul_1_output_0_scale /model/layers.14/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.14/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.14/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_457 1 3 /model/layers.14/input_layernorm/Mul_1_output_0_zero_point /model/layers.14/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.14/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.14/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.14/self_attn/Gather 1 1 /model/layers.14/self_attn/Shape_output_0 /model/layers.14/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_458 1 4 /model/layers.14/self_attn/Gather_output_0 /model/layers.14/self_attn/Gather_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_output_0_splitncnn_1 /model/layers.14/self_attn/Gather_output_0_splitncnn_2 /model/layers.14/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.14/self_attn/Gather_1 1 1 /model/layers.14/self_attn/Shape_1_output_0 /model/layers.14/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_459 1 4 /model/layers.14/self_attn/Gather_1_output_0 /model/layers.14/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.14/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.14/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.14/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.14/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.14/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.968504e-03 BinaryOp /model/layers.14/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.14/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.14/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.468688e-03 BinaryOp /model/layers.14/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.14/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.14/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.061147e-03 MatMulInteger /model/layers.14/self_attn/q_proj/MatMul_quant 4 1 /model/layers.14/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7680_quantized /model/layers.14/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7680_zero_point /model/layers.14/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.14/self_attn/k_proj/MatMul_quant 4 1 /model/layers.14/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7687_quantized /model/layers.14/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7687_zero_point /model/layers.14/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.14/self_attn/v_proj/MatMul_quant 4 1 /model/layers.14/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7688_quantized /model/layers.14/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7688_zero_point /model/layers.14/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.14/self_attn/Unsqueeze 2 1 /model/layers.14/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_4403 /model/layers.14/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_2 2 1 /model/layers.14/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_4412 /model/layers.14/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_4 2 1 /model/layers.14/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_4421 /model/layers.14/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_29 2 1 /model/layers.14/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_4626 /model/layers.14/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_1 2 1 /model/layers.14/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_4405 /model/layers.14/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_3 2 1 /model/layers.14/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_4414 /model/layers.14/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_5 2 1 /model/layers.14/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_4423 /model/layers.14/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_30 2 1 /model/layers.14/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_4628 /model/layers.14/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.14/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.14/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.14/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.14/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.14/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.14/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.14/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.14/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.14/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.14/self_attn/Concat 4 1 /model/layers.14/self_attn/Unsqueeze_output_0 /model/layers.14/self_attn/Unsqueeze_1_output_0 /model/layers.14/self_attn/Constant_2_output_0 /model/layers.14/self_attn/Constant_3_output_0 /model/layers.14/self_attn/Concat_output_0 0=0 Concat /model/layers.14/self_attn/Concat_1 4 1 /model/layers.14/self_attn/Unsqueeze_2_output_0 /model/layers.14/self_attn/Unsqueeze_3_output_0 /model/layers.14/self_attn/Constant_4_output_0 /model/layers.14/self_attn/Constant_5_output_0 /model/layers.14/self_attn/Concat_1_output_0 0=0 Concat /model/layers.14/self_attn/Concat_2 4 1 /model/layers.14/self_attn/Unsqueeze_4_output_0 /model/layers.14/self_attn/Unsqueeze_5_output_0 /model/layers.14/self_attn/Constant_6_output_0 /model/layers.14/self_attn/Constant_7_output_0 /model/layers.14/self_attn/Concat_2_output_0 0=0 Concat /model/layers.14/self_attn/Concat_11 3 1 /model/layers.14/self_attn/Unsqueeze_29_output_0 /model/layers.14/self_attn/Unsqueeze_30_output_0 /model/layers.14/self_attn/Constant_64_output_0 /model/layers.14/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.14/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.14/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.14/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.14/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.14/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.14/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.14/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.14/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.14/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.14/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.14/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.14/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.14/self_attn/q_proj/Add 2 1 model.layers.14.self_attn.q_proj.bias /model/layers.14/self_attn/q_proj/MatMul_output_0 /model/layers.14/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.14/self_attn/k_proj/Add 2 1 model.layers.14.self_attn.k_proj.bias /model/layers.14/self_attn/k_proj/MatMul_output_0 /model/layers.14/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.14/self_attn/v_proj/Add 2 1 model.layers.14.self_attn.v_proj.bias /model/layers.14/self_attn/v_proj/MatMul_output_0 /model/layers.14/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.14/self_attn/Reshape 2 1 /model/layers.14/self_attn/q_proj/Add_output_0 /model/layers.14/self_attn/Concat_output_0 /model/layers.14/self_attn/Reshape_output_0 Reshape /model/layers.14/self_attn/Reshape_1 2 1 /model/layers.14/self_attn/k_proj/Add_output_0 /model/layers.14/self_attn/Concat_1_output_0 /model/layers.14/self_attn/Reshape_1_output_0 Reshape /model/layers.14/self_attn/Reshape_2 2 1 /model/layers.14/self_attn/v_proj/Add_output_0 /model/layers.14/self_attn/Concat_2_output_0 /model/layers.14/self_attn/Reshape_2_output_0 Permute /model/layers.14/self_attn/Transpose 1 1 /model/layers.14/self_attn/Reshape_output_0 /model/layers.14/self_attn/Transpose_output_0 0=2 Split splitncnn_460 1 4 /model/layers.14/self_attn/Transpose_output_0 /model/layers.14/self_attn/Transpose_output_0_splitncnn_0 /model/layers.14/self_attn/Transpose_output_0_splitncnn_1 /model/layers.14/self_attn/Transpose_output_0_splitncnn_2 /model/layers.14/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.14/self_attn/Transpose_1 1 1 /model/layers.14/self_attn/Reshape_1_output_0 /model/layers.14/self_attn/Transpose_1_output_0 0=2 Split splitncnn_461 1 4 /model/layers.14/self_attn/Transpose_1_output_0 /model/layers.14/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.14/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.14/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.14/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.14/self_attn/Transpose_2 1 1 /model/layers.14/self_attn/Reshape_2_output_0 /model/layers.14/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.14/self_attn/Mul 2 1 /model/layers.14/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_19 /model/layers.14/self_attn/Mul_output_0 0=2 Shape /model/layers.14/self_attn/Shape_2 1 1 /model/layers.14/self_attn/Transpose_output_0_splitncnn_2 /model/layers.14/self_attn/Shape_2_output_0 BinaryOp /model/layers.14/self_attn/Mul_2 2 1 /model/layers.14/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_18 /model/layers.14/self_attn/Mul_2_output_0 0=2 Shape /model/layers.14/self_attn/Shape_3 1 1 /model/layers.14/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.14/self_attn/Shape_3_output_0 Concat /model/layers.14/self_attn/Concat_6 2 1 past_key_values.14.value /model/layers.14/self_attn/Transpose_2_output_0 present.14.value 0=-2 Split splitncnn_462 1 5 present.14.value present.14.value_splitncnn_0 present.14.value_splitncnn_1 present.14.value_splitncnn_2 present.14.value_splitncnn_3 present.14.value_splitncnn_4 Crop /model/layers.14/self_attn/Gather_2 1 1 /model/layers.14/self_attn/Shape_2_output_0 /model/layers.14/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.14/self_attn/Gather_3 1 1 /model/layers.14/self_attn/Shape_3_output_0 /model/layers.14/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.14/self_attn/Shape_9 1 1 present.14.value_splitncnn_4 /model/layers.14/self_attn/Shape_9_output_0 Shape /model/layers.14/self_attn/Shape_10 1 1 present.14.value_splitncnn_3 /model/layers.14/self_attn/Shape_10_output_0 Shape /model/layers.14/self_attn/Shape_11 1 1 present.14.value_splitncnn_2 /model/layers.14/self_attn/Shape_11_output_0 Shape /model/layers.14/self_attn/Shape_12 1 1 present.14.value_splitncnn_1 /model/layers.14/self_attn/Shape_12_output_0 ExpandDims /model/layers.14/self_attn/Unsqueeze_19 2 1 present.14.value_splitncnn_0 /model/layers.14/self_attn/Constant_45_output_0 /model/layers.14/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.14/self_attn/Div 2 1 /model/layers.14/self_attn/Gather_2_output_0 /model/layers.14/self_attn/Constant_9_output_0 /model/layers.14/self_attn/Div_output_0 0=3 BinaryOp /model/layers.14/self_attn/Div_1 2 1 /model/layers.14/self_attn/Gather_3_output_0 /model/layers.14/self_attn/Constant_19_output_0 /model/layers.14/self_attn/Div_1_output_0 0=3 Crop /model/layers.14/self_attn/Gather_8 1 1 /model/layers.14/self_attn/Shape_9_output_0 /model/layers.14/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_463 1 2 /model/layers.14/self_attn/Gather_8_output_0 /model/layers.14/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.14/self_attn/Gather_9 1 1 /model/layers.14/self_attn/Shape_10_output_0 /model/layers.14/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_464 1 2 /model/layers.14/self_attn/Gather_9_output_0 /model/layers.14/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.14/self_attn/Gather_10 1 1 /model/layers.14/self_attn/Shape_11_output_0 /model/layers.14/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_465 1 2 /model/layers.14/self_attn/Gather_10_output_0 /model/layers.14/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.14/self_attn/Gather_11 1 1 /model/layers.14/self_attn/Shape_12_output_0 /model/layers.14/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_466 1 2 /model/layers.14/self_attn/Gather_11_output_0 /model/layers.14/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.14/self_attn/Cast 1 1 /model/layers.14/self_attn/Div_output_0 /model/layers.14/self_attn/Cast_output_0 Cast /model/layers.14/self_attn/Cast_2 1 1 /model/layers.14/self_attn/Div_1_output_0 /model/layers.14/self_attn/Cast_2_output_0 ExpandDims /model/layers.14/self_attn/Unsqueeze_20 2 1 /model/layers.14/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_46_output_0 /model/layers.14/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_24 2 1 /model/layers.14/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_4586 /model/layers.14/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_21 2 1 /model/layers.14/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_47_output_0 /model/layers.14/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.14/self_attn/Mul_7 2 1 /model/layers.14/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.14/self_attn/Constant_53_output_0 /model/layers.14/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.14/self_attn/Unsqueeze_22 2 1 /model/layers.14/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_49_output_0 /model/layers.14/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_26 2 1 /model/layers.14/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_4590 /model/layers.14/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_23 2 1 /model/layers.14/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_50_output_0 /model/layers.14/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_27 2 1 /model/layers.14/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_4592 /model/layers.14/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.14/self_attn/Cast_1 1 1 /model/layers.14/self_attn/Cast_output_0 /model/layers.14/self_attn/Cast_1_output_0 Split splitncnn_467 1 2 /model/layers.14/self_attn/Cast_1_output_0 /model/layers.14/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.14/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.14/self_attn/Cast_3 1 1 /model/layers.14/self_attn/Cast_2_output_0 /model/layers.14/self_attn/Cast_3_output_0 Split splitncnn_468 1 2 /model/layers.14/self_attn/Cast_3_output_0 /model/layers.14/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.14/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.14/self_attn/Unsqueeze_25 2 1 /model/layers.14/self_attn/Mul_7_output_0 onnx::Unsqueeze_4588 /model/layers.14/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.14/self_attn/Concat_9 5 1 /model/layers.14/self_attn/Unsqueeze_20_output_0 /model/layers.14/self_attn/Unsqueeze_21_output_0 /model/layers.14/self_attn/Constant_48_output_0 /model/layers.14/self_attn/Unsqueeze_22_output_0 /model/layers.14/self_attn/Unsqueeze_23_output_0 /model/layers.14/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_6 2 1 /model/layers.14/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_12_output_0 /model/layers.14/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_7 2 1 /model/layers.14/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.14/self_attn/Constant_15_output_0 /model/layers.14/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_8 2 1 /model/layers.14/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_22_output_0 /model/layers.14/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_9 2 1 /model/layers.14/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.14/self_attn/Constant_25_output_0 /model/layers.14/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.14/self_attn/Concat_10 4 1 /model/layers.14/self_attn/Unsqueeze_24_output_0 /model/layers.14/self_attn/Unsqueeze_25_output_0 /model/layers.14/self_attn/Unsqueeze_26_output_0 /model/layers.14/self_attn/Unsqueeze_27_output_0 /model/layers.14/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.14/self_attn/Reshape_5 1 1 /model/layers.14/self_attn/Concat_9_output_0 /model/layers.14/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_469 1 3 /model/layers.14/self_attn/Reshape_5_output_0 /model/layers.14/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.14/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.14/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.14/self_attn/Slice 2 1 /model/layers.14/self_attn/Transpose_output_0_splitncnn_1 /model/layers.14/self_attn/Unsqueeze_6_output_0 /model/layers.14/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.14/self_attn/Slice_1 2 1 /model/layers.14/self_attn/Transpose_output_0_splitncnn_0 /model/layers.14/self_attn/Unsqueeze_7_output_0 /model/layers.14/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.14/self_attn/Slice_2 2 1 /model/layers.14/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.14/self_attn/Unsqueeze_8_output_0 /model/layers.14/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.14/self_attn/Slice_3 2 1 /model/layers.14/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.14/self_attn/Unsqueeze_9_output_0 /model/layers.14/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.14/self_attn/Shape_13 1 1 /model/layers.14/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.14/self_attn/Shape_13_output_0 UnaryOp /model/layers.14/self_attn/Neg 1 1 /model/layers.14/self_attn/Slice_1_output_0 /model/layers.14/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.14/self_attn/Neg_1 1 1 /model/layers.14/self_attn/Slice_3_output_0 /model/layers.14/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.14/self_attn/ConstantOfShape_1 1 1 /model/layers.14/self_attn/Shape_13_output_0 /model/layers.14/self_attn/ConstantOfShape_1_output_0 Split splitncnn_470 1 2 /model/layers.14/self_attn/ConstantOfShape_1_output_0 /model/layers.14/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.14/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.14/self_attn/Concat_3 2 1 /model/layers.14/self_attn/Neg_output_0 /model/layers.14/self_attn/Slice_output_0 /model/layers.14/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.14/self_attn/Concat_4 2 1 /model/layers.14/self_attn/Neg_1_output_0 /model/layers.14/self_attn/Slice_2_output_0 /model/layers.14/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.14/self_attn/Mul_6 2 1 /model/layers.14/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_52_output_0 /model/layers.14/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.14/self_attn/Mul_1 2 1 /model/layers.14/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_19 /model/layers.14/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.14/self_attn/Mul_3 2 1 /model/layers.14/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_18 /model/layers.14/self_attn/Mul_3_output_0 0=2 Equal /model/layers.14/self_attn/Equal_1 2 1 /model/layers.14/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.14/self_attn/Mul_6_output_0 /model/layers.14/self_attn/Equal_1_output_0 BinaryOp /model/layers.14/self_attn/Add 2 1 /model/layers.14/self_attn/Mul_output_0 /model/layers.14/self_attn/Mul_1_output_0 /model/layers.14/self_attn/Add_output_0 0=0 BinaryOp /model/layers.14/self_attn/Add_1 2 1 /model/layers.14/self_attn/Mul_2_output_0 /model/layers.14/self_attn/Mul_3_output_0 /model/layers.14/self_attn/Add_1_output_0 0=0 Where /model/layers.14/self_attn/Where_1 3 1 /model/layers.14/self_attn/Equal_1_output_0 /model/layers.14/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.14/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.14/self_attn/Where_1_output_0 BinaryOp /model/layers.14/self_attn/Mul_8 1 1 /model/layers.14/self_attn/Add_output_0 /model/layers.14/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.14/self_attn/Concat_5 2 1 past_key_values.14.key /model/layers.14/self_attn/Add_1_output_0 present.14.key 0=-2 Split splitncnn_471 1 5 present.14.key present.14.key_splitncnn_0 present.14.key_splitncnn_1 present.14.key_splitncnn_2 present.14.key_splitncnn_3 present.14.key_splitncnn_4 Expand /model/layers.14/self_attn/Expand_1 2 1 /model/layers.14/self_attn/Unsqueeze_19_output_0 /model/layers.14/self_attn/Where_1_output_0 /model/layers.14/self_attn/Expand_1_output_0 Shape /model/layers.14/self_attn/Shape_4 1 1 present.14.key_splitncnn_4 /model/layers.14/self_attn/Shape_4_output_0 Shape /model/layers.14/self_attn/Shape_5 1 1 present.14.key_splitncnn_3 /model/layers.14/self_attn/Shape_5_output_0 Shape /model/layers.14/self_attn/Shape_6 1 1 present.14.key_splitncnn_2 /model/layers.14/self_attn/Shape_6_output_0 Shape /model/layers.14/self_attn/Shape_7 1 1 present.14.key_splitncnn_1 /model/layers.14/self_attn/Shape_7_output_0 ExpandDims /model/layers.14/self_attn/Unsqueeze_10 2 1 present.14.key_splitncnn_0 /model/layers.14/self_attn/Constant_32_output_0 /model/layers.14/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.14/self_attn/Reshape_6 2 1 /model/layers.14/self_attn/Expand_1_output_0 /model/layers.14/self_attn/Concat_10_output_0 /model/layers.14/self_attn/Reshape_6_output_0 Crop /model/layers.14/self_attn/Gather_4 1 1 /model/layers.14/self_attn/Shape_4_output_0 /model/layers.14/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_472 1 2 /model/layers.14/self_attn/Gather_4_output_0 /model/layers.14/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.14/self_attn/Gather_5 1 1 /model/layers.14/self_attn/Shape_5_output_0 /model/layers.14/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_473 1 2 /model/layers.14/self_attn/Gather_5_output_0 /model/layers.14/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.14/self_attn/Gather_6 1 1 /model/layers.14/self_attn/Shape_6_output_0 /model/layers.14/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_474 1 2 /model/layers.14/self_attn/Gather_6_output_0 /model/layers.14/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.14/self_attn/Gather_7 1 1 /model/layers.14/self_attn/Shape_7_output_0 /model/layers.14/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_475 1 2 /model/layers.14/self_attn/Gather_7_output_0 /model/layers.14/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.14/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.14/self_attn/Unsqueeze_11 2 1 /model/layers.14/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_33_output_0 /model/layers.14/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_15 2 1 /model/layers.14/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_4540 /model/layers.14/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_12 2 1 /model/layers.14/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_34_output_0 /model/layers.14/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.14/self_attn/Mul_5 2 1 /model/layers.14/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.14/self_attn/Constant_40_output_0 /model/layers.14/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.14/self_attn/Unsqueeze_13 2 1 /model/layers.14/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_36_output_0 /model/layers.14/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_17 2 1 /model/layers.14/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_4544 /model/layers.14/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_14 2 1 /model/layers.14/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_37_output_0 /model/layers.14/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_18 2 1 /model/layers.14/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_4546 /model/layers.14/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.14/self_attn/Unsqueeze_16 2 1 /model/layers.14/self_attn/Mul_5_output_0 onnx::Unsqueeze_4542 /model/layers.14/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.14/self_attn/Concat_7 5 1 /model/layers.14/self_attn/Unsqueeze_11_output_0 /model/layers.14/self_attn/Unsqueeze_12_output_0 /model/layers.14/self_attn/Constant_35_output_0 /model/layers.14/self_attn/Unsqueeze_13_output_0 /model/layers.14/self_attn/Unsqueeze_14_output_0 /model/layers.14/self_attn/Concat_7_output_0 0=0 Concat /model/layers.14/self_attn/Concat_8 4 1 /model/layers.14/self_attn/Unsqueeze_15_output_0 /model/layers.14/self_attn/Unsqueeze_16_output_0 /model/layers.14/self_attn/Unsqueeze_17_output_0 /model/layers.14/self_attn/Unsqueeze_18_output_0 /model/layers.14/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.14/self_attn/Reshape_3 1 1 /model/layers.14/self_attn/Concat_7_output_0 /model/layers.14/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_476 1 3 /model/layers.14/self_attn/Reshape_3_output_0 /model/layers.14/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.14/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.14/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.14/self_attn/Shape_8 1 1 /model/layers.14/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.14/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.14/self_attn/ConstantOfShape 1 1 /model/layers.14/self_attn/Shape_8_output_0 /model/layers.14/self_attn/ConstantOfShape_output_0 Split splitncnn_477 1 2 /model/layers.14/self_attn/ConstantOfShape_output_0 /model/layers.14/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.14/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.14/self_attn/Mul_4 2 1 /model/layers.14/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.14/self_attn/Constant_39_output_0 /model/layers.14/self_attn/Mul_4_output_0 0=2 Equal /model/layers.14/self_attn/Equal 2 1 /model/layers.14/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.14/self_attn/Mul_4_output_0 /model/layers.14/self_attn/Equal_output_0 Where /model/layers.14/self_attn/Where 3 1 /model/layers.14/self_attn/Equal_output_0 /model/layers.14/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.14/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.14/self_attn/Where_output_0 Expand /model/layers.14/self_attn/Expand 2 1 /model/layers.14/self_attn/Unsqueeze_10_output_0 /model/layers.14/self_attn/Where_output_0 /model/layers.14/self_attn/Expand_output_0 Reshape /model/layers.14/self_attn/Reshape_4 2 1 /model/layers.14/self_attn/Expand_output_0 /model/layers.14/self_attn/Concat_8_output_0 /model/layers.14/self_attn/Reshape_4_output_0 Split splitncnn_478 1 2 /model/layers.14/self_attn/Reshape_4_output_0 /model/layers.14/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.14/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.14/self_attn/Shape_14 1 1 /model/layers.14/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.14/self_attn/Shape_14_output_0 Permute /model/layers.14/self_attn/Transpose_3 1 1 /model/layers.14/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.14/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.14/self_attn/Gather_12 1 1 /model/layers.14/self_attn/Shape_14_output_0 /model/layers.14/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.14/self_attn/Mul_9 1 1 /model/layers.14/self_attn/Transpose_3_output_0 /model/layers.14/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.14/self_attn/Unsqueeze_28 2 1 /model/layers.14/self_attn/Gather_12_output_0 /model/layers.14/self_attn/Constant_57_output_0 /model/layers.14/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.14/self_attn/MatMul 2 1 /model/layers.14/self_attn/Mul_8_output_0 /model/layers.14/self_attn/Mul_9_output_0 /model/layers.14/self_attn/MatMul_output_0 Crop /model/layers.14/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_9 /model/layers.14/self_attn/Unsqueeze_28_output_0 /model/layers.14/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.14/self_attn/Where_2 3 1 /model/layers.14/self_attn/Slice_4_output_0 /model/layers.14/self_attn/Constant_61_output_0 /model/layers.14/self_attn/Constant_62_output_0 /model/layers.14/self_attn/Where_2_output_0 BinaryOp /model/layers.14/self_attn/Add_2 2 1 /model/layers.14/self_attn/MatMul_output_0 /model/layers.14/self_attn/Where_2_output_0 /model/layers.14/self_attn/Add_2_output_0 0=0 Softmax /model/layers.14/self_attn/Softmax 1 1 /model/layers.14/self_attn/Add_2_output_0 /model/layers.14/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_479 1 2 /model/layers.14/self_attn/Softmax_output_0 /model/layers.14/self_attn/Softmax_output_0_splitncnn_0 /model/layers.14/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.14/self_attn/IsNaN 1 1 /model/layers.14/self_attn/Softmax_output_0_splitncnn_1 /model/layers.14/self_attn/IsNaN_output_0 Where /model/layers.14/self_attn/Where_3 3 1 /model/layers.14/self_attn/IsNaN_output_0 /model/layers.14/self_attn/Constant_63_output_0 /model/layers.14/self_attn/Softmax_output_0_splitncnn_0 /model/layers.14/self_attn/Where_3_output_0 Gemm /model/layers.14/self_attn/MatMul_1 2 1 /model/layers.14/self_attn/Where_3_output_0 /model/layers.14/self_attn/Reshape_6_output_0 /model/layers.14/self_attn/MatMul_1_output_0 Permute /model/layers.14/self_attn/Transpose_4 1 1 /model/layers.14/self_attn/MatMul_1_output_0 /model/layers.14/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.14/self_attn/Reshape_7 2 1 /model/layers.14/self_attn/Transpose_4_output_0 /model/layers.14/self_attn/Concat_11_output_0 /model/layers.14/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.14/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.14/self_attn/Reshape_7_output_0 /model/layers.14/self_attn/Reshape_7_output_0_quantized /model/layers.14/self_attn/Reshape_7_output_0_scale /model/layers.14/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.14/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.14/self_attn/Reshape_7_output_0_scale /model/layers.14/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.983514e-03 MatMulInteger /model/layers.14/self_attn/o_proj/MatMul_quant 4 1 /model/layers.14/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7711_quantized /model/layers.14/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7711_zero_point /model/layers.14/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.14/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.14/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.14/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.14/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.14/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.14/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.14/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.14/Add 2 1 /model/layers.14/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.14/self_attn/o_proj/MatMul_output_0 /model/layers.14/Add_output_0 0=0 Cast /model/layers.14/post_attention_layernorm/Cast 1 1 /model/layers.14/Add_output_0 /model/layers.14/post_attention_layernorm/Cast_output_0 Split splitncnn_480 1 3 /model/layers.14/post_attention_layernorm/Cast_output_0 /model/layers.14/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.14/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.14/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.14/post_attention_layernorm/Pow 1 1 /model/layers.14/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.14/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.14/post_attention_layernorm/ReduceMean 2 1 /model/layers.14/post_attention_layernorm/Pow_output_0 /model/layers.14/post_attention_layernorm/Constant_1_output_0 /model/layers.14/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.14/post_attention_layernorm/Add 1 1 /model/layers.14/post_attention_layernorm/ReduceMean_output_0 /model/layers.14/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.14/post_attention_layernorm/Sqrt 1 1 /model/layers.14/post_attention_layernorm/Add_output_0 /model/layers.14/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.14/post_attention_layernorm/Div 1 1 /model/layers.14/post_attention_layernorm/Sqrt_output_0 /model/layers.14/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.14/post_attention_layernorm/Mul 2 1 /model/layers.14/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.14/post_attention_layernorm/Div_output_0 /model/layers.14/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.14/post_attention_layernorm/Cast_1 1 1 /model/layers.14/post_attention_layernorm/Mul_output_0 /model/layers.14/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.14/post_attention_layernorm/Mul_1 2 1 model.layers.14.post_attention_layernorm.weight /model/layers.14/post_attention_layernorm/Cast_1_output_0 /model/layers.14/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.14/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.14/post_attention_layernorm/Mul_1_output_0 /model/layers.14/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.14/post_attention_layernorm/Mul_1_output_0_scale /model/layers.14/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_481 1 2 /model/layers.14/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.14/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.14/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_482 1 2 /model/layers.14/post_attention_layernorm/Mul_1_output_0_scale /model/layers.14/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.14/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_483 1 2 /model/layers.14/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.14/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.14/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.14/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.14/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.14/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.737082e-03 BinaryOp /model/layers.14/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.14/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.14/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.737082e-03 MatMulInteger /model/layers.14/mlp/gate_proj/MatMul_quant 4 1 /model/layers.14/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7712_quantized /model/layers.14/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7712_zero_point /model/layers.14/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.14/mlp/up_proj/MatMul_quant 4 1 /model/layers.14/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7713_quantized /model/layers.14/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7713_zero_point /model/layers.14/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.14/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.14/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.14/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.14/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.14/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.14/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.14/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.14/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.14/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.14/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.14/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.14/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.14/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.14/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.14/mlp/act_fn/Mul 1 1 /model/layers.14/mlp/gate_proj/MatMul_output_0 /model/layers.14/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.14/mlp/Mul 2 1 /model/layers.14/mlp/act_fn/Mul_output_0 /model/layers.14/mlp/up_proj/MatMul_output_0 /model/layers.14/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.14/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.14/mlp/Mul_output_0 /model/layers.14/mlp/Mul_output_0_quantized /model/layers.14/mlp/Mul_output_0_scale /model/layers.14/mlp/Mul_output_0_zero_point BinaryOp /model/layers.14/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.14/mlp/Mul_output_0_scale /model/layers.14/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.337229e-03 MatMulInteger /model/layers.14/mlp/down_proj/MatMul_quant 4 1 /model/layers.14/mlp/Mul_output_0_quantized onnx::MatMul_7714_quantized /model/layers.14/mlp/Mul_output_0_zero_point onnx::MatMul_7714_zero_point /model/layers.14/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.14/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.14/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.14/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.14/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.14/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.14/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.14/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.14/Add_1 2 1 /model/layers.14/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.14/mlp/down_proj/MatMul_output_0 /model/layers.14/Add_1_output_0 0=0 Cast /model/layers.15/input_layernorm/Cast 1 1 /model/layers.14/Add_1_output_0 /model/layers.15/input_layernorm/Cast_output_0 Split splitncnn_484 1 3 /model/layers.15/input_layernorm/Cast_output_0 /model/layers.15/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.15/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.15/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.15/input_layernorm/Pow 1 1 /model/layers.15/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.15/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.15/input_layernorm/ReduceMean 2 1 /model/layers.15/input_layernorm/Pow_output_0 /model/layers.15/input_layernorm/Constant_1_output_0 /model/layers.15/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.15/input_layernorm/Add 1 1 /model/layers.15/input_layernorm/ReduceMean_output_0 /model/layers.15/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.15/input_layernorm/Sqrt 1 1 /model/layers.15/input_layernorm/Add_output_0 /model/layers.15/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.15/input_layernorm/Div 1 1 /model/layers.15/input_layernorm/Sqrt_output_0 /model/layers.15/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.15/input_layernorm/Mul 2 1 /model/layers.15/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.15/input_layernorm/Div_output_0 /model/layers.15/input_layernorm/Mul_output_0 0=2 Cast /model/layers.15/input_layernorm/Cast_1 1 1 /model/layers.15/input_layernorm/Mul_output_0 /model/layers.15/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.15/input_layernorm/Mul_1 2 1 model.layers.15.input_layernorm.weight /model/layers.15/input_layernorm/Cast_1_output_0 /model/layers.15/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_485 1 3 /model/layers.15/input_layernorm/Mul_1_output_0 /model/layers.15/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.15/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.15/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.15/self_attn/Shape 1 1 /model/layers.15/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.15/self_attn/Shape_output_0 Shape /model/layers.15/self_attn/Shape_1 1 1 /model/layers.15/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.15/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.15/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.15/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.15/input_layernorm/Mul_1_output_0_quantized /model/layers.15/input_layernorm/Mul_1_output_0_scale /model/layers.15/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_486 1 3 /model/layers.15/input_layernorm/Mul_1_output_0_quantized /model/layers.15/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.15/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.15/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_487 1 3 /model/layers.15/input_layernorm/Mul_1_output_0_scale /model/layers.15/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.15/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.15/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_488 1 3 /model/layers.15/input_layernorm/Mul_1_output_0_zero_point /model/layers.15/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.15/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.15/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.15/self_attn/Gather 1 1 /model/layers.15/self_attn/Shape_output_0 /model/layers.15/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_489 1 4 /model/layers.15/self_attn/Gather_output_0 /model/layers.15/self_attn/Gather_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_output_0_splitncnn_1 /model/layers.15/self_attn/Gather_output_0_splitncnn_2 /model/layers.15/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.15/self_attn/Gather_1 1 1 /model/layers.15/self_attn/Shape_1_output_0 /model/layers.15/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_490 1 4 /model/layers.15/self_attn/Gather_1_output_0 /model/layers.15/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.15/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.15/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.15/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.15/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.15/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.660925e-03 BinaryOp /model/layers.15/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.15/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.15/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.491757e-03 BinaryOp /model/layers.15/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.15/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.15/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.022699e-03 MatMulInteger /model/layers.15/self_attn/q_proj/MatMul_quant 4 1 /model/layers.15/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7715_quantized /model/layers.15/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7715_zero_point /model/layers.15/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.15/self_attn/k_proj/MatMul_quant 4 1 /model/layers.15/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7722_quantized /model/layers.15/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7722_zero_point /model/layers.15/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.15/self_attn/v_proj/MatMul_quant 4 1 /model/layers.15/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7723_quantized /model/layers.15/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7723_zero_point /model/layers.15/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.15/self_attn/Unsqueeze 2 1 /model/layers.15/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_4682 /model/layers.15/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_2 2 1 /model/layers.15/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_4691 /model/layers.15/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_4 2 1 /model/layers.15/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_4700 /model/layers.15/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_29 2 1 /model/layers.15/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_4905 /model/layers.15/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_1 2 1 /model/layers.15/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_4684 /model/layers.15/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_3 2 1 /model/layers.15/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_4693 /model/layers.15/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_5 2 1 /model/layers.15/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_4702 /model/layers.15/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_30 2 1 /model/layers.15/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_4907 /model/layers.15/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.15/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.15/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.15/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.15/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.15/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.15/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.15/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.15/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.15/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.15/self_attn/Concat 4 1 /model/layers.15/self_attn/Unsqueeze_output_0 /model/layers.15/self_attn/Unsqueeze_1_output_0 /model/layers.15/self_attn/Constant_2_output_0 /model/layers.15/self_attn/Constant_3_output_0 /model/layers.15/self_attn/Concat_output_0 0=0 Concat /model/layers.15/self_attn/Concat_1 4 1 /model/layers.15/self_attn/Unsqueeze_2_output_0 /model/layers.15/self_attn/Unsqueeze_3_output_0 /model/layers.15/self_attn/Constant_4_output_0 /model/layers.15/self_attn/Constant_5_output_0 /model/layers.15/self_attn/Concat_1_output_0 0=0 Concat /model/layers.15/self_attn/Concat_2 4 1 /model/layers.15/self_attn/Unsqueeze_4_output_0 /model/layers.15/self_attn/Unsqueeze_5_output_0 /model/layers.15/self_attn/Constant_6_output_0 /model/layers.15/self_attn/Constant_7_output_0 /model/layers.15/self_attn/Concat_2_output_0 0=0 Concat /model/layers.15/self_attn/Concat_11 3 1 /model/layers.15/self_attn/Unsqueeze_29_output_0 /model/layers.15/self_attn/Unsqueeze_30_output_0 /model/layers.15/self_attn/Constant_64_output_0 /model/layers.15/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.15/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.15/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.15/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.15/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.15/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.15/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.15/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.15/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.15/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.15/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.15/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.15/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.15/self_attn/q_proj/Add 2 1 model.layers.15.self_attn.q_proj.bias /model/layers.15/self_attn/q_proj/MatMul_output_0 /model/layers.15/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.15/self_attn/k_proj/Add 2 1 model.layers.15.self_attn.k_proj.bias /model/layers.15/self_attn/k_proj/MatMul_output_0 /model/layers.15/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.15/self_attn/v_proj/Add 2 1 model.layers.15.self_attn.v_proj.bias /model/layers.15/self_attn/v_proj/MatMul_output_0 /model/layers.15/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.15/self_attn/Reshape 2 1 /model/layers.15/self_attn/q_proj/Add_output_0 /model/layers.15/self_attn/Concat_output_0 /model/layers.15/self_attn/Reshape_output_0 Reshape /model/layers.15/self_attn/Reshape_1 2 1 /model/layers.15/self_attn/k_proj/Add_output_0 /model/layers.15/self_attn/Concat_1_output_0 /model/layers.15/self_attn/Reshape_1_output_0 Reshape /model/layers.15/self_attn/Reshape_2 2 1 /model/layers.15/self_attn/v_proj/Add_output_0 /model/layers.15/self_attn/Concat_2_output_0 /model/layers.15/self_attn/Reshape_2_output_0 Permute /model/layers.15/self_attn/Transpose 1 1 /model/layers.15/self_attn/Reshape_output_0 /model/layers.15/self_attn/Transpose_output_0 0=2 Split splitncnn_491 1 4 /model/layers.15/self_attn/Transpose_output_0 /model/layers.15/self_attn/Transpose_output_0_splitncnn_0 /model/layers.15/self_attn/Transpose_output_0_splitncnn_1 /model/layers.15/self_attn/Transpose_output_0_splitncnn_2 /model/layers.15/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.15/self_attn/Transpose_1 1 1 /model/layers.15/self_attn/Reshape_1_output_0 /model/layers.15/self_attn/Transpose_1_output_0 0=2 Split splitncnn_492 1 4 /model/layers.15/self_attn/Transpose_1_output_0 /model/layers.15/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.15/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.15/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.15/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.15/self_attn/Transpose_2 1 1 /model/layers.15/self_attn/Reshape_2_output_0 /model/layers.15/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.15/self_attn/Mul 2 1 /model/layers.15/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_17 /model/layers.15/self_attn/Mul_output_0 0=2 Shape /model/layers.15/self_attn/Shape_2 1 1 /model/layers.15/self_attn/Transpose_output_0_splitncnn_2 /model/layers.15/self_attn/Shape_2_output_0 BinaryOp /model/layers.15/self_attn/Mul_2 2 1 /model/layers.15/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_16 /model/layers.15/self_attn/Mul_2_output_0 0=2 Shape /model/layers.15/self_attn/Shape_3 1 1 /model/layers.15/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.15/self_attn/Shape_3_output_0 Concat /model/layers.15/self_attn/Concat_6 2 1 past_key_values.15.value /model/layers.15/self_attn/Transpose_2_output_0 present.15.value 0=-2 Split splitncnn_493 1 5 present.15.value present.15.value_splitncnn_0 present.15.value_splitncnn_1 present.15.value_splitncnn_2 present.15.value_splitncnn_3 present.15.value_splitncnn_4 Crop /model/layers.15/self_attn/Gather_2 1 1 /model/layers.15/self_attn/Shape_2_output_0 /model/layers.15/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.15/self_attn/Gather_3 1 1 /model/layers.15/self_attn/Shape_3_output_0 /model/layers.15/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.15/self_attn/Shape_9 1 1 present.15.value_splitncnn_4 /model/layers.15/self_attn/Shape_9_output_0 Shape /model/layers.15/self_attn/Shape_10 1 1 present.15.value_splitncnn_3 /model/layers.15/self_attn/Shape_10_output_0 Shape /model/layers.15/self_attn/Shape_11 1 1 present.15.value_splitncnn_2 /model/layers.15/self_attn/Shape_11_output_0 Shape /model/layers.15/self_attn/Shape_12 1 1 present.15.value_splitncnn_1 /model/layers.15/self_attn/Shape_12_output_0 ExpandDims /model/layers.15/self_attn/Unsqueeze_19 2 1 present.15.value_splitncnn_0 /model/layers.15/self_attn/Constant_45_output_0 /model/layers.15/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.15/self_attn/Div 2 1 /model/layers.15/self_attn/Gather_2_output_0 /model/layers.15/self_attn/Constant_9_output_0 /model/layers.15/self_attn/Div_output_0 0=3 BinaryOp /model/layers.15/self_attn/Div_1 2 1 /model/layers.15/self_attn/Gather_3_output_0 /model/layers.15/self_attn/Constant_19_output_0 /model/layers.15/self_attn/Div_1_output_0 0=3 Crop /model/layers.15/self_attn/Gather_8 1 1 /model/layers.15/self_attn/Shape_9_output_0 /model/layers.15/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_494 1 2 /model/layers.15/self_attn/Gather_8_output_0 /model/layers.15/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.15/self_attn/Gather_9 1 1 /model/layers.15/self_attn/Shape_10_output_0 /model/layers.15/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_495 1 2 /model/layers.15/self_attn/Gather_9_output_0 /model/layers.15/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.15/self_attn/Gather_10 1 1 /model/layers.15/self_attn/Shape_11_output_0 /model/layers.15/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_496 1 2 /model/layers.15/self_attn/Gather_10_output_0 /model/layers.15/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.15/self_attn/Gather_11 1 1 /model/layers.15/self_attn/Shape_12_output_0 /model/layers.15/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_497 1 2 /model/layers.15/self_attn/Gather_11_output_0 /model/layers.15/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.15/self_attn/Cast 1 1 /model/layers.15/self_attn/Div_output_0 /model/layers.15/self_attn/Cast_output_0 Cast /model/layers.15/self_attn/Cast_2 1 1 /model/layers.15/self_attn/Div_1_output_0 /model/layers.15/self_attn/Cast_2_output_0 ExpandDims /model/layers.15/self_attn/Unsqueeze_20 2 1 /model/layers.15/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_46_output_0 /model/layers.15/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_24 2 1 /model/layers.15/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_4865 /model/layers.15/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_21 2 1 /model/layers.15/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_47_output_0 /model/layers.15/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.15/self_attn/Mul_7 2 1 /model/layers.15/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.15/self_attn/Constant_53_output_0 /model/layers.15/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.15/self_attn/Unsqueeze_22 2 1 /model/layers.15/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_49_output_0 /model/layers.15/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_26 2 1 /model/layers.15/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_4869 /model/layers.15/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_23 2 1 /model/layers.15/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_50_output_0 /model/layers.15/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_27 2 1 /model/layers.15/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_4871 /model/layers.15/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.15/self_attn/Cast_1 1 1 /model/layers.15/self_attn/Cast_output_0 /model/layers.15/self_attn/Cast_1_output_0 Split splitncnn_498 1 2 /model/layers.15/self_attn/Cast_1_output_0 /model/layers.15/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.15/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.15/self_attn/Cast_3 1 1 /model/layers.15/self_attn/Cast_2_output_0 /model/layers.15/self_attn/Cast_3_output_0 Split splitncnn_499 1 2 /model/layers.15/self_attn/Cast_3_output_0 /model/layers.15/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.15/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.15/self_attn/Unsqueeze_25 2 1 /model/layers.15/self_attn/Mul_7_output_0 onnx::Unsqueeze_4867 /model/layers.15/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.15/self_attn/Concat_9 5 1 /model/layers.15/self_attn/Unsqueeze_20_output_0 /model/layers.15/self_attn/Unsqueeze_21_output_0 /model/layers.15/self_attn/Constant_48_output_0 /model/layers.15/self_attn/Unsqueeze_22_output_0 /model/layers.15/self_attn/Unsqueeze_23_output_0 /model/layers.15/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_6 2 1 /model/layers.15/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_12_output_0 /model/layers.15/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_7 2 1 /model/layers.15/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.15/self_attn/Constant_15_output_0 /model/layers.15/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_8 2 1 /model/layers.15/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_22_output_0 /model/layers.15/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_9 2 1 /model/layers.15/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.15/self_attn/Constant_25_output_0 /model/layers.15/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.15/self_attn/Concat_10 4 1 /model/layers.15/self_attn/Unsqueeze_24_output_0 /model/layers.15/self_attn/Unsqueeze_25_output_0 /model/layers.15/self_attn/Unsqueeze_26_output_0 /model/layers.15/self_attn/Unsqueeze_27_output_0 /model/layers.15/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.15/self_attn/Reshape_5 1 1 /model/layers.15/self_attn/Concat_9_output_0 /model/layers.15/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_500 1 3 /model/layers.15/self_attn/Reshape_5_output_0 /model/layers.15/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.15/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.15/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.15/self_attn/Slice 2 1 /model/layers.15/self_attn/Transpose_output_0_splitncnn_1 /model/layers.15/self_attn/Unsqueeze_6_output_0 /model/layers.15/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.15/self_attn/Slice_1 2 1 /model/layers.15/self_attn/Transpose_output_0_splitncnn_0 /model/layers.15/self_attn/Unsqueeze_7_output_0 /model/layers.15/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.15/self_attn/Slice_2 2 1 /model/layers.15/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.15/self_attn/Unsqueeze_8_output_0 /model/layers.15/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.15/self_attn/Slice_3 2 1 /model/layers.15/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.15/self_attn/Unsqueeze_9_output_0 /model/layers.15/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.15/self_attn/Shape_13 1 1 /model/layers.15/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.15/self_attn/Shape_13_output_0 UnaryOp /model/layers.15/self_attn/Neg 1 1 /model/layers.15/self_attn/Slice_1_output_0 /model/layers.15/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.15/self_attn/Neg_1 1 1 /model/layers.15/self_attn/Slice_3_output_0 /model/layers.15/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.15/self_attn/ConstantOfShape_1 1 1 /model/layers.15/self_attn/Shape_13_output_0 /model/layers.15/self_attn/ConstantOfShape_1_output_0 Split splitncnn_501 1 2 /model/layers.15/self_attn/ConstantOfShape_1_output_0 /model/layers.15/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.15/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.15/self_attn/Concat_3 2 1 /model/layers.15/self_attn/Neg_output_0 /model/layers.15/self_attn/Slice_output_0 /model/layers.15/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.15/self_attn/Concat_4 2 1 /model/layers.15/self_attn/Neg_1_output_0 /model/layers.15/self_attn/Slice_2_output_0 /model/layers.15/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.15/self_attn/Mul_6 2 1 /model/layers.15/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_52_output_0 /model/layers.15/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.15/self_attn/Mul_1 2 1 /model/layers.15/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_17 /model/layers.15/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.15/self_attn/Mul_3 2 1 /model/layers.15/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_16 /model/layers.15/self_attn/Mul_3_output_0 0=2 Equal /model/layers.15/self_attn/Equal_1 2 1 /model/layers.15/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.15/self_attn/Mul_6_output_0 /model/layers.15/self_attn/Equal_1_output_0 BinaryOp /model/layers.15/self_attn/Add 2 1 /model/layers.15/self_attn/Mul_output_0 /model/layers.15/self_attn/Mul_1_output_0 /model/layers.15/self_attn/Add_output_0 0=0 BinaryOp /model/layers.15/self_attn/Add_1 2 1 /model/layers.15/self_attn/Mul_2_output_0 /model/layers.15/self_attn/Mul_3_output_0 /model/layers.15/self_attn/Add_1_output_0 0=0 Where /model/layers.15/self_attn/Where_1 3 1 /model/layers.15/self_attn/Equal_1_output_0 /model/layers.15/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.15/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.15/self_attn/Where_1_output_0 BinaryOp /model/layers.15/self_attn/Mul_8 1 1 /model/layers.15/self_attn/Add_output_0 /model/layers.15/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.15/self_attn/Concat_5 2 1 past_key_values.15.key /model/layers.15/self_attn/Add_1_output_0 present.15.key 0=-2 Split splitncnn_502 1 5 present.15.key present.15.key_splitncnn_0 present.15.key_splitncnn_1 present.15.key_splitncnn_2 present.15.key_splitncnn_3 present.15.key_splitncnn_4 Expand /model/layers.15/self_attn/Expand_1 2 1 /model/layers.15/self_attn/Unsqueeze_19_output_0 /model/layers.15/self_attn/Where_1_output_0 /model/layers.15/self_attn/Expand_1_output_0 Shape /model/layers.15/self_attn/Shape_4 1 1 present.15.key_splitncnn_4 /model/layers.15/self_attn/Shape_4_output_0 Shape /model/layers.15/self_attn/Shape_5 1 1 present.15.key_splitncnn_3 /model/layers.15/self_attn/Shape_5_output_0 Shape /model/layers.15/self_attn/Shape_6 1 1 present.15.key_splitncnn_2 /model/layers.15/self_attn/Shape_6_output_0 Shape /model/layers.15/self_attn/Shape_7 1 1 present.15.key_splitncnn_1 /model/layers.15/self_attn/Shape_7_output_0 ExpandDims /model/layers.15/self_attn/Unsqueeze_10 2 1 present.15.key_splitncnn_0 /model/layers.15/self_attn/Constant_32_output_0 /model/layers.15/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.15/self_attn/Reshape_6 2 1 /model/layers.15/self_attn/Expand_1_output_0 /model/layers.15/self_attn/Concat_10_output_0 /model/layers.15/self_attn/Reshape_6_output_0 Crop /model/layers.15/self_attn/Gather_4 1 1 /model/layers.15/self_attn/Shape_4_output_0 /model/layers.15/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_503 1 2 /model/layers.15/self_attn/Gather_4_output_0 /model/layers.15/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.15/self_attn/Gather_5 1 1 /model/layers.15/self_attn/Shape_5_output_0 /model/layers.15/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_504 1 2 /model/layers.15/self_attn/Gather_5_output_0 /model/layers.15/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.15/self_attn/Gather_6 1 1 /model/layers.15/self_attn/Shape_6_output_0 /model/layers.15/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_505 1 2 /model/layers.15/self_attn/Gather_6_output_0 /model/layers.15/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.15/self_attn/Gather_7 1 1 /model/layers.15/self_attn/Shape_7_output_0 /model/layers.15/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_506 1 2 /model/layers.15/self_attn/Gather_7_output_0 /model/layers.15/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.15/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.15/self_attn/Unsqueeze_11 2 1 /model/layers.15/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_33_output_0 /model/layers.15/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_15 2 1 /model/layers.15/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_4819 /model/layers.15/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_12 2 1 /model/layers.15/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_34_output_0 /model/layers.15/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.15/self_attn/Mul_5 2 1 /model/layers.15/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.15/self_attn/Constant_40_output_0 /model/layers.15/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.15/self_attn/Unsqueeze_13 2 1 /model/layers.15/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_36_output_0 /model/layers.15/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_17 2 1 /model/layers.15/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_4823 /model/layers.15/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_14 2 1 /model/layers.15/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_37_output_0 /model/layers.15/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_18 2 1 /model/layers.15/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_4825 /model/layers.15/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.15/self_attn/Unsqueeze_16 2 1 /model/layers.15/self_attn/Mul_5_output_0 onnx::Unsqueeze_4821 /model/layers.15/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.15/self_attn/Concat_7 5 1 /model/layers.15/self_attn/Unsqueeze_11_output_0 /model/layers.15/self_attn/Unsqueeze_12_output_0 /model/layers.15/self_attn/Constant_35_output_0 /model/layers.15/self_attn/Unsqueeze_13_output_0 /model/layers.15/self_attn/Unsqueeze_14_output_0 /model/layers.15/self_attn/Concat_7_output_0 0=0 Concat /model/layers.15/self_attn/Concat_8 4 1 /model/layers.15/self_attn/Unsqueeze_15_output_0 /model/layers.15/self_attn/Unsqueeze_16_output_0 /model/layers.15/self_attn/Unsqueeze_17_output_0 /model/layers.15/self_attn/Unsqueeze_18_output_0 /model/layers.15/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.15/self_attn/Reshape_3 1 1 /model/layers.15/self_attn/Concat_7_output_0 /model/layers.15/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_507 1 3 /model/layers.15/self_attn/Reshape_3_output_0 /model/layers.15/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.15/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.15/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.15/self_attn/Shape_8 1 1 /model/layers.15/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.15/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.15/self_attn/ConstantOfShape 1 1 /model/layers.15/self_attn/Shape_8_output_0 /model/layers.15/self_attn/ConstantOfShape_output_0 Split splitncnn_508 1 2 /model/layers.15/self_attn/ConstantOfShape_output_0 /model/layers.15/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.15/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.15/self_attn/Mul_4 2 1 /model/layers.15/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.15/self_attn/Constant_39_output_0 /model/layers.15/self_attn/Mul_4_output_0 0=2 Equal /model/layers.15/self_attn/Equal 2 1 /model/layers.15/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.15/self_attn/Mul_4_output_0 /model/layers.15/self_attn/Equal_output_0 Where /model/layers.15/self_attn/Where 3 1 /model/layers.15/self_attn/Equal_output_0 /model/layers.15/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.15/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.15/self_attn/Where_output_0 Expand /model/layers.15/self_attn/Expand 2 1 /model/layers.15/self_attn/Unsqueeze_10_output_0 /model/layers.15/self_attn/Where_output_0 /model/layers.15/self_attn/Expand_output_0 Reshape /model/layers.15/self_attn/Reshape_4 2 1 /model/layers.15/self_attn/Expand_output_0 /model/layers.15/self_attn/Concat_8_output_0 /model/layers.15/self_attn/Reshape_4_output_0 Split splitncnn_509 1 2 /model/layers.15/self_attn/Reshape_4_output_0 /model/layers.15/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.15/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.15/self_attn/Shape_14 1 1 /model/layers.15/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.15/self_attn/Shape_14_output_0 Permute /model/layers.15/self_attn/Transpose_3 1 1 /model/layers.15/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.15/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.15/self_attn/Gather_12 1 1 /model/layers.15/self_attn/Shape_14_output_0 /model/layers.15/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.15/self_attn/Mul_9 1 1 /model/layers.15/self_attn/Transpose_3_output_0 /model/layers.15/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.15/self_attn/Unsqueeze_28 2 1 /model/layers.15/self_attn/Gather_12_output_0 /model/layers.15/self_attn/Constant_57_output_0 /model/layers.15/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.15/self_attn/MatMul 2 1 /model/layers.15/self_attn/Mul_8_output_0 /model/layers.15/self_attn/Mul_9_output_0 /model/layers.15/self_attn/MatMul_output_0 Crop /model/layers.15/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_8 /model/layers.15/self_attn/Unsqueeze_28_output_0 /model/layers.15/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.15/self_attn/Where_2 3 1 /model/layers.15/self_attn/Slice_4_output_0 /model/layers.15/self_attn/Constant_61_output_0 /model/layers.15/self_attn/Constant_62_output_0 /model/layers.15/self_attn/Where_2_output_0 BinaryOp /model/layers.15/self_attn/Add_2 2 1 /model/layers.15/self_attn/MatMul_output_0 /model/layers.15/self_attn/Where_2_output_0 /model/layers.15/self_attn/Add_2_output_0 0=0 Softmax /model/layers.15/self_attn/Softmax 1 1 /model/layers.15/self_attn/Add_2_output_0 /model/layers.15/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_510 1 2 /model/layers.15/self_attn/Softmax_output_0 /model/layers.15/self_attn/Softmax_output_0_splitncnn_0 /model/layers.15/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.15/self_attn/IsNaN 1 1 /model/layers.15/self_attn/Softmax_output_0_splitncnn_1 /model/layers.15/self_attn/IsNaN_output_0 Where /model/layers.15/self_attn/Where_3 3 1 /model/layers.15/self_attn/IsNaN_output_0 /model/layers.15/self_attn/Constant_63_output_0 /model/layers.15/self_attn/Softmax_output_0_splitncnn_0 /model/layers.15/self_attn/Where_3_output_0 Gemm /model/layers.15/self_attn/MatMul_1 2 1 /model/layers.15/self_attn/Where_3_output_0 /model/layers.15/self_attn/Reshape_6_output_0 /model/layers.15/self_attn/MatMul_1_output_0 Permute /model/layers.15/self_attn/Transpose_4 1 1 /model/layers.15/self_attn/MatMul_1_output_0 /model/layers.15/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.15/self_attn/Reshape_7 2 1 /model/layers.15/self_attn/Transpose_4_output_0 /model/layers.15/self_attn/Concat_11_output_0 /model/layers.15/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.15/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.15/self_attn/Reshape_7_output_0 /model/layers.15/self_attn/Reshape_7_output_0_quantized /model/layers.15/self_attn/Reshape_7_output_0_scale /model/layers.15/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.15/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.15/self_attn/Reshape_7_output_0_scale /model/layers.15/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.998893e-03 MatMulInteger /model/layers.15/self_attn/o_proj/MatMul_quant 4 1 /model/layers.15/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7746_quantized /model/layers.15/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7746_zero_point /model/layers.15/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.15/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.15/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.15/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.15/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.15/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.15/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.15/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.15/Add 2 1 /model/layers.15/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.15/self_attn/o_proj/MatMul_output_0 /model/layers.15/Add_output_0 0=0 Cast /model/layers.15/post_attention_layernorm/Cast 1 1 /model/layers.15/Add_output_0 /model/layers.15/post_attention_layernorm/Cast_output_0 Split splitncnn_511 1 3 /model/layers.15/post_attention_layernorm/Cast_output_0 /model/layers.15/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.15/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.15/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.15/post_attention_layernorm/Pow 1 1 /model/layers.15/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.15/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.15/post_attention_layernorm/ReduceMean 2 1 /model/layers.15/post_attention_layernorm/Pow_output_0 /model/layers.15/post_attention_layernorm/Constant_1_output_0 /model/layers.15/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.15/post_attention_layernorm/Add 1 1 /model/layers.15/post_attention_layernorm/ReduceMean_output_0 /model/layers.15/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.15/post_attention_layernorm/Sqrt 1 1 /model/layers.15/post_attention_layernorm/Add_output_0 /model/layers.15/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.15/post_attention_layernorm/Div 1 1 /model/layers.15/post_attention_layernorm/Sqrt_output_0 /model/layers.15/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.15/post_attention_layernorm/Mul 2 1 /model/layers.15/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.15/post_attention_layernorm/Div_output_0 /model/layers.15/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.15/post_attention_layernorm/Cast_1 1 1 /model/layers.15/post_attention_layernorm/Mul_output_0 /model/layers.15/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.15/post_attention_layernorm/Mul_1 2 1 model.layers.15.post_attention_layernorm.weight /model/layers.15/post_attention_layernorm/Cast_1_output_0 /model/layers.15/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.15/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.15/post_attention_layernorm/Mul_1_output_0 /model/layers.15/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.15/post_attention_layernorm/Mul_1_output_0_scale /model/layers.15/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_512 1 2 /model/layers.15/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.15/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.15/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_513 1 2 /model/layers.15/post_attention_layernorm/Mul_1_output_0_scale /model/layers.15/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.15/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_514 1 2 /model/layers.15/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.15/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.15/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.15/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.15/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.15/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.783219e-03 BinaryOp /model/layers.15/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.15/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.15/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=5.967028e-03 MatMulInteger /model/layers.15/mlp/gate_proj/MatMul_quant 4 1 /model/layers.15/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7747_quantized /model/layers.15/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7747_zero_point /model/layers.15/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.15/mlp/up_proj/MatMul_quant 4 1 /model/layers.15/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7748_quantized /model/layers.15/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7748_zero_point /model/layers.15/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.15/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.15/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.15/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.15/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.15/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.15/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.15/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.15/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.15/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.15/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.15/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.15/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.15/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.15/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.15/mlp/act_fn/Mul 1 1 /model/layers.15/mlp/gate_proj/MatMul_output_0 /model/layers.15/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.15/mlp/Mul 2 1 /model/layers.15/mlp/act_fn/Mul_output_0 /model/layers.15/mlp/up_proj/MatMul_output_0 /model/layers.15/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.15/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.15/mlp/Mul_output_0 /model/layers.15/mlp/Mul_output_0_quantized /model/layers.15/mlp/Mul_output_0_scale /model/layers.15/mlp/Mul_output_0_zero_point BinaryOp /model/layers.15/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.15/mlp/Mul_output_0_scale /model/layers.15/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.429503e-03 MatMulInteger /model/layers.15/mlp/down_proj/MatMul_quant 4 1 /model/layers.15/mlp/Mul_output_0_quantized onnx::MatMul_7749_quantized /model/layers.15/mlp/Mul_output_0_zero_point onnx::MatMul_7749_zero_point /model/layers.15/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.15/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.15/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.15/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.15/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.15/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.15/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.15/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.15/Add_1 2 1 /model/layers.15/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.15/mlp/down_proj/MatMul_output_0 /model/layers.15/Add_1_output_0 0=0 Cast /model/layers.16/input_layernorm/Cast 1 1 /model/layers.15/Add_1_output_0 /model/layers.16/input_layernorm/Cast_output_0 Split splitncnn_515 1 3 /model/layers.16/input_layernorm/Cast_output_0 /model/layers.16/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.16/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.16/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.16/input_layernorm/Pow 1 1 /model/layers.16/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.16/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.16/input_layernorm/ReduceMean 2 1 /model/layers.16/input_layernorm/Pow_output_0 /model/layers.16/input_layernorm/Constant_1_output_0 /model/layers.16/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.16/input_layernorm/Add 1 1 /model/layers.16/input_layernorm/ReduceMean_output_0 /model/layers.16/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.16/input_layernorm/Sqrt 1 1 /model/layers.16/input_layernorm/Add_output_0 /model/layers.16/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.16/input_layernorm/Div 1 1 /model/layers.16/input_layernorm/Sqrt_output_0 /model/layers.16/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.16/input_layernorm/Mul 2 1 /model/layers.16/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.16/input_layernorm/Div_output_0 /model/layers.16/input_layernorm/Mul_output_0 0=2 Cast /model/layers.16/input_layernorm/Cast_1 1 1 /model/layers.16/input_layernorm/Mul_output_0 /model/layers.16/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.16/input_layernorm/Mul_1 2 1 model.layers.16.input_layernorm.weight /model/layers.16/input_layernorm/Cast_1_output_0 /model/layers.16/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_516 1 3 /model/layers.16/input_layernorm/Mul_1_output_0 /model/layers.16/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.16/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.16/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.16/self_attn/Shape 1 1 /model/layers.16/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.16/self_attn/Shape_output_0 Shape /model/layers.16/self_attn/Shape_1 1 1 /model/layers.16/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.16/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.16/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.16/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.16/input_layernorm/Mul_1_output_0_quantized /model/layers.16/input_layernorm/Mul_1_output_0_scale /model/layers.16/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_517 1 3 /model/layers.16/input_layernorm/Mul_1_output_0_quantized /model/layers.16/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.16/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.16/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_518 1 3 /model/layers.16/input_layernorm/Mul_1_output_0_scale /model/layers.16/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.16/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.16/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_519 1 3 /model/layers.16/input_layernorm/Mul_1_output_0_zero_point /model/layers.16/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.16/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.16/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.16/self_attn/Gather 1 1 /model/layers.16/self_attn/Shape_output_0 /model/layers.16/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_520 1 4 /model/layers.16/self_attn/Gather_output_0 /model/layers.16/self_attn/Gather_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_output_0_splitncnn_1 /model/layers.16/self_attn/Gather_output_0_splitncnn_2 /model/layers.16/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.16/self_attn/Gather_1 1 1 /model/layers.16/self_attn/Shape_1_output_0 /model/layers.16/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_521 1 4 /model/layers.16/self_attn/Gather_1_output_0 /model/layers.16/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.16/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.16/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.16/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.16/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.16/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.121924e-03 BinaryOp /model/layers.16/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.16/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.16/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.699373e-03 BinaryOp /model/layers.16/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.16/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.16/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.507136e-03 MatMulInteger /model/layers.16/self_attn/q_proj/MatMul_quant 4 1 /model/layers.16/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7750_quantized /model/layers.16/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7750_zero_point /model/layers.16/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.16/self_attn/k_proj/MatMul_quant 4 1 /model/layers.16/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7757_quantized /model/layers.16/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7757_zero_point /model/layers.16/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.16/self_attn/v_proj/MatMul_quant 4 1 /model/layers.16/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7758_quantized /model/layers.16/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7758_zero_point /model/layers.16/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.16/self_attn/Unsqueeze 2 1 /model/layers.16/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_4961 /model/layers.16/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_2 2 1 /model/layers.16/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_4970 /model/layers.16/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_4 2 1 /model/layers.16/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_4979 /model/layers.16/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_29 2 1 /model/layers.16/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_5184 /model/layers.16/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_1 2 1 /model/layers.16/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_4963 /model/layers.16/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_3 2 1 /model/layers.16/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_4972 /model/layers.16/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_5 2 1 /model/layers.16/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_4981 /model/layers.16/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_30 2 1 /model/layers.16/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_5186 /model/layers.16/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.16/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.16/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.16/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.16/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.16/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.16/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.16/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.16/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.16/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.16/self_attn/Concat 4 1 /model/layers.16/self_attn/Unsqueeze_output_0 /model/layers.16/self_attn/Unsqueeze_1_output_0 /model/layers.16/self_attn/Constant_2_output_0 /model/layers.16/self_attn/Constant_3_output_0 /model/layers.16/self_attn/Concat_output_0 0=0 Concat /model/layers.16/self_attn/Concat_1 4 1 /model/layers.16/self_attn/Unsqueeze_2_output_0 /model/layers.16/self_attn/Unsqueeze_3_output_0 /model/layers.16/self_attn/Constant_4_output_0 /model/layers.16/self_attn/Constant_5_output_0 /model/layers.16/self_attn/Concat_1_output_0 0=0 Concat /model/layers.16/self_attn/Concat_2 4 1 /model/layers.16/self_attn/Unsqueeze_4_output_0 /model/layers.16/self_attn/Unsqueeze_5_output_0 /model/layers.16/self_attn/Constant_6_output_0 /model/layers.16/self_attn/Constant_7_output_0 /model/layers.16/self_attn/Concat_2_output_0 0=0 Concat /model/layers.16/self_attn/Concat_11 3 1 /model/layers.16/self_attn/Unsqueeze_29_output_0 /model/layers.16/self_attn/Unsqueeze_30_output_0 /model/layers.16/self_attn/Constant_64_output_0 /model/layers.16/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.16/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.16/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.16/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.16/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.16/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.16/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.16/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.16/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.16/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.16/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.16/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.16/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.16/self_attn/q_proj/Add 2 1 model.layers.16.self_attn.q_proj.bias /model/layers.16/self_attn/q_proj/MatMul_output_0 /model/layers.16/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.16/self_attn/k_proj/Add 2 1 model.layers.16.self_attn.k_proj.bias /model/layers.16/self_attn/k_proj/MatMul_output_0 /model/layers.16/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.16/self_attn/v_proj/Add 2 1 model.layers.16.self_attn.v_proj.bias /model/layers.16/self_attn/v_proj/MatMul_output_0 /model/layers.16/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.16/self_attn/Reshape 2 1 /model/layers.16/self_attn/q_proj/Add_output_0 /model/layers.16/self_attn/Concat_output_0 /model/layers.16/self_attn/Reshape_output_0 Reshape /model/layers.16/self_attn/Reshape_1 2 1 /model/layers.16/self_attn/k_proj/Add_output_0 /model/layers.16/self_attn/Concat_1_output_0 /model/layers.16/self_attn/Reshape_1_output_0 Reshape /model/layers.16/self_attn/Reshape_2 2 1 /model/layers.16/self_attn/v_proj/Add_output_0 /model/layers.16/self_attn/Concat_2_output_0 /model/layers.16/self_attn/Reshape_2_output_0 Permute /model/layers.16/self_attn/Transpose 1 1 /model/layers.16/self_attn/Reshape_output_0 /model/layers.16/self_attn/Transpose_output_0 0=2 Split splitncnn_522 1 4 /model/layers.16/self_attn/Transpose_output_0 /model/layers.16/self_attn/Transpose_output_0_splitncnn_0 /model/layers.16/self_attn/Transpose_output_0_splitncnn_1 /model/layers.16/self_attn/Transpose_output_0_splitncnn_2 /model/layers.16/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.16/self_attn/Transpose_1 1 1 /model/layers.16/self_attn/Reshape_1_output_0 /model/layers.16/self_attn/Transpose_1_output_0 0=2 Split splitncnn_523 1 4 /model/layers.16/self_attn/Transpose_1_output_0 /model/layers.16/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.16/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.16/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.16/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.16/self_attn/Transpose_2 1 1 /model/layers.16/self_attn/Reshape_2_output_0 /model/layers.16/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.16/self_attn/Mul 2 1 /model/layers.16/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_15 /model/layers.16/self_attn/Mul_output_0 0=2 Shape /model/layers.16/self_attn/Shape_2 1 1 /model/layers.16/self_attn/Transpose_output_0_splitncnn_2 /model/layers.16/self_attn/Shape_2_output_0 BinaryOp /model/layers.16/self_attn/Mul_2 2 1 /model/layers.16/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_14 /model/layers.16/self_attn/Mul_2_output_0 0=2 Shape /model/layers.16/self_attn/Shape_3 1 1 /model/layers.16/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.16/self_attn/Shape_3_output_0 Concat /model/layers.16/self_attn/Concat_6 2 1 past_key_values.16.value /model/layers.16/self_attn/Transpose_2_output_0 present.16.value 0=-2 Split splitncnn_524 1 5 present.16.value present.16.value_splitncnn_0 present.16.value_splitncnn_1 present.16.value_splitncnn_2 present.16.value_splitncnn_3 present.16.value_splitncnn_4 Crop /model/layers.16/self_attn/Gather_2 1 1 /model/layers.16/self_attn/Shape_2_output_0 /model/layers.16/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.16/self_attn/Gather_3 1 1 /model/layers.16/self_attn/Shape_3_output_0 /model/layers.16/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.16/self_attn/Shape_9 1 1 present.16.value_splitncnn_4 /model/layers.16/self_attn/Shape_9_output_0 Shape /model/layers.16/self_attn/Shape_10 1 1 present.16.value_splitncnn_3 /model/layers.16/self_attn/Shape_10_output_0 Shape /model/layers.16/self_attn/Shape_11 1 1 present.16.value_splitncnn_2 /model/layers.16/self_attn/Shape_11_output_0 Shape /model/layers.16/self_attn/Shape_12 1 1 present.16.value_splitncnn_1 /model/layers.16/self_attn/Shape_12_output_0 ExpandDims /model/layers.16/self_attn/Unsqueeze_19 2 1 present.16.value_splitncnn_0 /model/layers.16/self_attn/Constant_45_output_0 /model/layers.16/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.16/self_attn/Div 2 1 /model/layers.16/self_attn/Gather_2_output_0 /model/layers.16/self_attn/Constant_9_output_0 /model/layers.16/self_attn/Div_output_0 0=3 BinaryOp /model/layers.16/self_attn/Div_1 2 1 /model/layers.16/self_attn/Gather_3_output_0 /model/layers.16/self_attn/Constant_19_output_0 /model/layers.16/self_attn/Div_1_output_0 0=3 Crop /model/layers.16/self_attn/Gather_8 1 1 /model/layers.16/self_attn/Shape_9_output_0 /model/layers.16/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_525 1 2 /model/layers.16/self_attn/Gather_8_output_0 /model/layers.16/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.16/self_attn/Gather_9 1 1 /model/layers.16/self_attn/Shape_10_output_0 /model/layers.16/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_526 1 2 /model/layers.16/self_attn/Gather_9_output_0 /model/layers.16/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.16/self_attn/Gather_10 1 1 /model/layers.16/self_attn/Shape_11_output_0 /model/layers.16/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_527 1 2 /model/layers.16/self_attn/Gather_10_output_0 /model/layers.16/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.16/self_attn/Gather_11 1 1 /model/layers.16/self_attn/Shape_12_output_0 /model/layers.16/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_528 1 2 /model/layers.16/self_attn/Gather_11_output_0 /model/layers.16/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.16/self_attn/Cast 1 1 /model/layers.16/self_attn/Div_output_0 /model/layers.16/self_attn/Cast_output_0 Cast /model/layers.16/self_attn/Cast_2 1 1 /model/layers.16/self_attn/Div_1_output_0 /model/layers.16/self_attn/Cast_2_output_0 ExpandDims /model/layers.16/self_attn/Unsqueeze_20 2 1 /model/layers.16/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_46_output_0 /model/layers.16/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_24 2 1 /model/layers.16/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_5144 /model/layers.16/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_21 2 1 /model/layers.16/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_47_output_0 /model/layers.16/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.16/self_attn/Mul_7 2 1 /model/layers.16/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.16/self_attn/Constant_53_output_0 /model/layers.16/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.16/self_attn/Unsqueeze_22 2 1 /model/layers.16/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_49_output_0 /model/layers.16/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_26 2 1 /model/layers.16/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_5148 /model/layers.16/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_23 2 1 /model/layers.16/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_50_output_0 /model/layers.16/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_27 2 1 /model/layers.16/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_5150 /model/layers.16/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.16/self_attn/Cast_1 1 1 /model/layers.16/self_attn/Cast_output_0 /model/layers.16/self_attn/Cast_1_output_0 Split splitncnn_529 1 2 /model/layers.16/self_attn/Cast_1_output_0 /model/layers.16/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.16/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.16/self_attn/Cast_3 1 1 /model/layers.16/self_attn/Cast_2_output_0 /model/layers.16/self_attn/Cast_3_output_0 Split splitncnn_530 1 2 /model/layers.16/self_attn/Cast_3_output_0 /model/layers.16/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.16/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.16/self_attn/Unsqueeze_25 2 1 /model/layers.16/self_attn/Mul_7_output_0 onnx::Unsqueeze_5146 /model/layers.16/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.16/self_attn/Concat_9 5 1 /model/layers.16/self_attn/Unsqueeze_20_output_0 /model/layers.16/self_attn/Unsqueeze_21_output_0 /model/layers.16/self_attn/Constant_48_output_0 /model/layers.16/self_attn/Unsqueeze_22_output_0 /model/layers.16/self_attn/Unsqueeze_23_output_0 /model/layers.16/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_6 2 1 /model/layers.16/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_12_output_0 /model/layers.16/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_7 2 1 /model/layers.16/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.16/self_attn/Constant_15_output_0 /model/layers.16/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_8 2 1 /model/layers.16/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_22_output_0 /model/layers.16/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_9 2 1 /model/layers.16/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.16/self_attn/Constant_25_output_0 /model/layers.16/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.16/self_attn/Concat_10 4 1 /model/layers.16/self_attn/Unsqueeze_24_output_0 /model/layers.16/self_attn/Unsqueeze_25_output_0 /model/layers.16/self_attn/Unsqueeze_26_output_0 /model/layers.16/self_attn/Unsqueeze_27_output_0 /model/layers.16/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.16/self_attn/Reshape_5 1 1 /model/layers.16/self_attn/Concat_9_output_0 /model/layers.16/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_531 1 3 /model/layers.16/self_attn/Reshape_5_output_0 /model/layers.16/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.16/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.16/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.16/self_attn/Slice 2 1 /model/layers.16/self_attn/Transpose_output_0_splitncnn_1 /model/layers.16/self_attn/Unsqueeze_6_output_0 /model/layers.16/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.16/self_attn/Slice_1 2 1 /model/layers.16/self_attn/Transpose_output_0_splitncnn_0 /model/layers.16/self_attn/Unsqueeze_7_output_0 /model/layers.16/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.16/self_attn/Slice_2 2 1 /model/layers.16/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.16/self_attn/Unsqueeze_8_output_0 /model/layers.16/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.16/self_attn/Slice_3 2 1 /model/layers.16/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.16/self_attn/Unsqueeze_9_output_0 /model/layers.16/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.16/self_attn/Shape_13 1 1 /model/layers.16/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.16/self_attn/Shape_13_output_0 UnaryOp /model/layers.16/self_attn/Neg 1 1 /model/layers.16/self_attn/Slice_1_output_0 /model/layers.16/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.16/self_attn/Neg_1 1 1 /model/layers.16/self_attn/Slice_3_output_0 /model/layers.16/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.16/self_attn/ConstantOfShape_1 1 1 /model/layers.16/self_attn/Shape_13_output_0 /model/layers.16/self_attn/ConstantOfShape_1_output_0 Split splitncnn_532 1 2 /model/layers.16/self_attn/ConstantOfShape_1_output_0 /model/layers.16/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.16/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.16/self_attn/Concat_3 2 1 /model/layers.16/self_attn/Neg_output_0 /model/layers.16/self_attn/Slice_output_0 /model/layers.16/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.16/self_attn/Concat_4 2 1 /model/layers.16/self_attn/Neg_1_output_0 /model/layers.16/self_attn/Slice_2_output_0 /model/layers.16/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.16/self_attn/Mul_6 2 1 /model/layers.16/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_52_output_0 /model/layers.16/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.16/self_attn/Mul_1 2 1 /model/layers.16/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_15 /model/layers.16/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.16/self_attn/Mul_3 2 1 /model/layers.16/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_14 /model/layers.16/self_attn/Mul_3_output_0 0=2 Equal /model/layers.16/self_attn/Equal_1 2 1 /model/layers.16/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.16/self_attn/Mul_6_output_0 /model/layers.16/self_attn/Equal_1_output_0 BinaryOp /model/layers.16/self_attn/Add 2 1 /model/layers.16/self_attn/Mul_output_0 /model/layers.16/self_attn/Mul_1_output_0 /model/layers.16/self_attn/Add_output_0 0=0 BinaryOp /model/layers.16/self_attn/Add_1 2 1 /model/layers.16/self_attn/Mul_2_output_0 /model/layers.16/self_attn/Mul_3_output_0 /model/layers.16/self_attn/Add_1_output_0 0=0 Where /model/layers.16/self_attn/Where_1 3 1 /model/layers.16/self_attn/Equal_1_output_0 /model/layers.16/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.16/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.16/self_attn/Where_1_output_0 BinaryOp /model/layers.16/self_attn/Mul_8 1 1 /model/layers.16/self_attn/Add_output_0 /model/layers.16/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.16/self_attn/Concat_5 2 1 past_key_values.16.key /model/layers.16/self_attn/Add_1_output_0 present.16.key 0=-2 Split splitncnn_533 1 5 present.16.key present.16.key_splitncnn_0 present.16.key_splitncnn_1 present.16.key_splitncnn_2 present.16.key_splitncnn_3 present.16.key_splitncnn_4 Expand /model/layers.16/self_attn/Expand_1 2 1 /model/layers.16/self_attn/Unsqueeze_19_output_0 /model/layers.16/self_attn/Where_1_output_0 /model/layers.16/self_attn/Expand_1_output_0 Shape /model/layers.16/self_attn/Shape_4 1 1 present.16.key_splitncnn_4 /model/layers.16/self_attn/Shape_4_output_0 Shape /model/layers.16/self_attn/Shape_5 1 1 present.16.key_splitncnn_3 /model/layers.16/self_attn/Shape_5_output_0 Shape /model/layers.16/self_attn/Shape_6 1 1 present.16.key_splitncnn_2 /model/layers.16/self_attn/Shape_6_output_0 Shape /model/layers.16/self_attn/Shape_7 1 1 present.16.key_splitncnn_1 /model/layers.16/self_attn/Shape_7_output_0 ExpandDims /model/layers.16/self_attn/Unsqueeze_10 2 1 present.16.key_splitncnn_0 /model/layers.16/self_attn/Constant_32_output_0 /model/layers.16/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.16/self_attn/Reshape_6 2 1 /model/layers.16/self_attn/Expand_1_output_0 /model/layers.16/self_attn/Concat_10_output_0 /model/layers.16/self_attn/Reshape_6_output_0 Crop /model/layers.16/self_attn/Gather_4 1 1 /model/layers.16/self_attn/Shape_4_output_0 /model/layers.16/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_534 1 2 /model/layers.16/self_attn/Gather_4_output_0 /model/layers.16/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.16/self_attn/Gather_5 1 1 /model/layers.16/self_attn/Shape_5_output_0 /model/layers.16/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_535 1 2 /model/layers.16/self_attn/Gather_5_output_0 /model/layers.16/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.16/self_attn/Gather_6 1 1 /model/layers.16/self_attn/Shape_6_output_0 /model/layers.16/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_536 1 2 /model/layers.16/self_attn/Gather_6_output_0 /model/layers.16/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.16/self_attn/Gather_7 1 1 /model/layers.16/self_attn/Shape_7_output_0 /model/layers.16/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_537 1 2 /model/layers.16/self_attn/Gather_7_output_0 /model/layers.16/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.16/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.16/self_attn/Unsqueeze_11 2 1 /model/layers.16/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_33_output_0 /model/layers.16/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_15 2 1 /model/layers.16/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_5098 /model/layers.16/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_12 2 1 /model/layers.16/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_34_output_0 /model/layers.16/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.16/self_attn/Mul_5 2 1 /model/layers.16/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.16/self_attn/Constant_40_output_0 /model/layers.16/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.16/self_attn/Unsqueeze_13 2 1 /model/layers.16/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_36_output_0 /model/layers.16/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_17 2 1 /model/layers.16/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_5102 /model/layers.16/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_14 2 1 /model/layers.16/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_37_output_0 /model/layers.16/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_18 2 1 /model/layers.16/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_5104 /model/layers.16/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.16/self_attn/Unsqueeze_16 2 1 /model/layers.16/self_attn/Mul_5_output_0 onnx::Unsqueeze_5100 /model/layers.16/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.16/self_attn/Concat_7 5 1 /model/layers.16/self_attn/Unsqueeze_11_output_0 /model/layers.16/self_attn/Unsqueeze_12_output_0 /model/layers.16/self_attn/Constant_35_output_0 /model/layers.16/self_attn/Unsqueeze_13_output_0 /model/layers.16/self_attn/Unsqueeze_14_output_0 /model/layers.16/self_attn/Concat_7_output_0 0=0 Concat /model/layers.16/self_attn/Concat_8 4 1 /model/layers.16/self_attn/Unsqueeze_15_output_0 /model/layers.16/self_attn/Unsqueeze_16_output_0 /model/layers.16/self_attn/Unsqueeze_17_output_0 /model/layers.16/self_attn/Unsqueeze_18_output_0 /model/layers.16/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.16/self_attn/Reshape_3 1 1 /model/layers.16/self_attn/Concat_7_output_0 /model/layers.16/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_538 1 3 /model/layers.16/self_attn/Reshape_3_output_0 /model/layers.16/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.16/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.16/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.16/self_attn/Shape_8 1 1 /model/layers.16/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.16/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.16/self_attn/ConstantOfShape 1 1 /model/layers.16/self_attn/Shape_8_output_0 /model/layers.16/self_attn/ConstantOfShape_output_0 Split splitncnn_539 1 2 /model/layers.16/self_attn/ConstantOfShape_output_0 /model/layers.16/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.16/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.16/self_attn/Mul_4 2 1 /model/layers.16/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.16/self_attn/Constant_39_output_0 /model/layers.16/self_attn/Mul_4_output_0 0=2 Equal /model/layers.16/self_attn/Equal 2 1 /model/layers.16/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.16/self_attn/Mul_4_output_0 /model/layers.16/self_attn/Equal_output_0 Where /model/layers.16/self_attn/Where 3 1 /model/layers.16/self_attn/Equal_output_0 /model/layers.16/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.16/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.16/self_attn/Where_output_0 Expand /model/layers.16/self_attn/Expand 2 1 /model/layers.16/self_attn/Unsqueeze_10_output_0 /model/layers.16/self_attn/Where_output_0 /model/layers.16/self_attn/Expand_output_0 Reshape /model/layers.16/self_attn/Reshape_4 2 1 /model/layers.16/self_attn/Expand_output_0 /model/layers.16/self_attn/Concat_8_output_0 /model/layers.16/self_attn/Reshape_4_output_0 Split splitncnn_540 1 2 /model/layers.16/self_attn/Reshape_4_output_0 /model/layers.16/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.16/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.16/self_attn/Shape_14 1 1 /model/layers.16/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.16/self_attn/Shape_14_output_0 Permute /model/layers.16/self_attn/Transpose_3 1 1 /model/layers.16/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.16/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.16/self_attn/Gather_12 1 1 /model/layers.16/self_attn/Shape_14_output_0 /model/layers.16/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.16/self_attn/Mul_9 1 1 /model/layers.16/self_attn/Transpose_3_output_0 /model/layers.16/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.16/self_attn/Unsqueeze_28 2 1 /model/layers.16/self_attn/Gather_12_output_0 /model/layers.16/self_attn/Constant_57_output_0 /model/layers.16/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.16/self_attn/MatMul 2 1 /model/layers.16/self_attn/Mul_8_output_0 /model/layers.16/self_attn/Mul_9_output_0 /model/layers.16/self_attn/MatMul_output_0 Crop /model/layers.16/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_7 /model/layers.16/self_attn/Unsqueeze_28_output_0 /model/layers.16/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.16/self_attn/Where_2 3 1 /model/layers.16/self_attn/Slice_4_output_0 /model/layers.16/self_attn/Constant_61_output_0 /model/layers.16/self_attn/Constant_62_output_0 /model/layers.16/self_attn/Where_2_output_0 BinaryOp /model/layers.16/self_attn/Add_2 2 1 /model/layers.16/self_attn/MatMul_output_0 /model/layers.16/self_attn/Where_2_output_0 /model/layers.16/self_attn/Add_2_output_0 0=0 Softmax /model/layers.16/self_attn/Softmax 1 1 /model/layers.16/self_attn/Add_2_output_0 /model/layers.16/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_541 1 2 /model/layers.16/self_attn/Softmax_output_0 /model/layers.16/self_attn/Softmax_output_0_splitncnn_0 /model/layers.16/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.16/self_attn/IsNaN 1 1 /model/layers.16/self_attn/Softmax_output_0_splitncnn_1 /model/layers.16/self_attn/IsNaN_output_0 Where /model/layers.16/self_attn/Where_3 3 1 /model/layers.16/self_attn/IsNaN_output_0 /model/layers.16/self_attn/Constant_63_output_0 /model/layers.16/self_attn/Softmax_output_0_splitncnn_0 /model/layers.16/self_attn/Where_3_output_0 Gemm /model/layers.16/self_attn/MatMul_1 2 1 /model/layers.16/self_attn/Where_3_output_0 /model/layers.16/self_attn/Reshape_6_output_0 /model/layers.16/self_attn/MatMul_1_output_0 Permute /model/layers.16/self_attn/Transpose_4 1 1 /model/layers.16/self_attn/MatMul_1_output_0 /model/layers.16/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.16/self_attn/Reshape_7 2 1 /model/layers.16/self_attn/Transpose_4_output_0 /model/layers.16/self_attn/Concat_11_output_0 /model/layers.16/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.16/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.16/self_attn/Reshape_7_output_0 /model/layers.16/self_attn/Reshape_7_output_0_quantized /model/layers.16/self_attn/Reshape_7_output_0_scale /model/layers.16/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.16/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.16/self_attn/Reshape_7_output_0_scale /model/layers.16/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.476009e-03 MatMulInteger /model/layers.16/self_attn/o_proj/MatMul_quant 4 1 /model/layers.16/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7781_quantized /model/layers.16/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7781_zero_point /model/layers.16/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.16/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.16/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.16/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.16/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.16/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.16/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.16/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.16/Add 2 1 /model/layers.16/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.16/self_attn/o_proj/MatMul_output_0 /model/layers.16/Add_output_0 0=0 Cast /model/layers.16/post_attention_layernorm/Cast 1 1 /model/layers.16/Add_output_0 /model/layers.16/post_attention_layernorm/Cast_output_0 Split splitncnn_542 1 3 /model/layers.16/post_attention_layernorm/Cast_output_0 /model/layers.16/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.16/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.16/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.16/post_attention_layernorm/Pow 1 1 /model/layers.16/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.16/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.16/post_attention_layernorm/ReduceMean 2 1 /model/layers.16/post_attention_layernorm/Pow_output_0 /model/layers.16/post_attention_layernorm/Constant_1_output_0 /model/layers.16/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.16/post_attention_layernorm/Add 1 1 /model/layers.16/post_attention_layernorm/ReduceMean_output_0 /model/layers.16/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.16/post_attention_layernorm/Sqrt 1 1 /model/layers.16/post_attention_layernorm/Add_output_0 /model/layers.16/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.16/post_attention_layernorm/Div 1 1 /model/layers.16/post_attention_layernorm/Sqrt_output_0 /model/layers.16/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.16/post_attention_layernorm/Mul 2 1 /model/layers.16/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.16/post_attention_layernorm/Div_output_0 /model/layers.16/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.16/post_attention_layernorm/Cast_1 1 1 /model/layers.16/post_attention_layernorm/Mul_output_0 /model/layers.16/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.16/post_attention_layernorm/Mul_1 2 1 model.layers.16.post_attention_layernorm.weight /model/layers.16/post_attention_layernorm/Cast_1_output_0 /model/layers.16/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.16/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.16/post_attention_layernorm/Mul_1_output_0 /model/layers.16/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.16/post_attention_layernorm/Mul_1_output_0_scale /model/layers.16/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_543 1 2 /model/layers.16/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.16/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.16/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_544 1 2 /model/layers.16/post_attention_layernorm/Mul_1_output_0_scale /model/layers.16/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.16/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_545 1 2 /model/layers.16/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.16/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.16/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.16/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.16/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.16/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.291093e-03 BinaryOp /model/layers.16/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.16/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.16/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.014272e-03 MatMulInteger /model/layers.16/mlp/gate_proj/MatMul_quant 4 1 /model/layers.16/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7782_quantized /model/layers.16/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7782_zero_point /model/layers.16/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.16/mlp/up_proj/MatMul_quant 4 1 /model/layers.16/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7783_quantized /model/layers.16/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7783_zero_point /model/layers.16/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.16/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.16/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.16/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.16/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.16/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.16/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.16/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.16/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.16/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.16/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.16/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.16/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.16/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.16/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.16/mlp/act_fn/Mul 1 1 /model/layers.16/mlp/gate_proj/MatMul_output_0 /model/layers.16/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.16/mlp/Mul 2 1 /model/layers.16/mlp/act_fn/Mul_output_0 /model/layers.16/mlp/up_proj/MatMul_output_0 /model/layers.16/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.16/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.16/mlp/Mul_output_0 /model/layers.16/mlp/Mul_output_0_quantized /model/layers.16/mlp/Mul_output_0_scale /model/layers.16/mlp/Mul_output_0_zero_point BinaryOp /model/layers.16/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.16/mlp/Mul_output_0_scale /model/layers.16/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.491019e-03 MatMulInteger /model/layers.16/mlp/down_proj/MatMul_quant 4 1 /model/layers.16/mlp/Mul_output_0_quantized onnx::MatMul_7784_quantized /model/layers.16/mlp/Mul_output_0_zero_point onnx::MatMul_7784_zero_point /model/layers.16/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.16/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.16/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.16/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.16/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.16/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.16/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.16/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.16/Add_1 2 1 /model/layers.16/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.16/mlp/down_proj/MatMul_output_0 /model/layers.16/Add_1_output_0 0=0 Cast /model/layers.17/input_layernorm/Cast 1 1 /model/layers.16/Add_1_output_0 /model/layers.17/input_layernorm/Cast_output_0 Split splitncnn_546 1 3 /model/layers.17/input_layernorm/Cast_output_0 /model/layers.17/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.17/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.17/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.17/input_layernorm/Pow 1 1 /model/layers.17/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.17/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.17/input_layernorm/ReduceMean 2 1 /model/layers.17/input_layernorm/Pow_output_0 /model/layers.17/input_layernorm/Constant_1_output_0 /model/layers.17/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.17/input_layernorm/Add 1 1 /model/layers.17/input_layernorm/ReduceMean_output_0 /model/layers.17/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.17/input_layernorm/Sqrt 1 1 /model/layers.17/input_layernorm/Add_output_0 /model/layers.17/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.17/input_layernorm/Div 1 1 /model/layers.17/input_layernorm/Sqrt_output_0 /model/layers.17/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.17/input_layernorm/Mul 2 1 /model/layers.17/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.17/input_layernorm/Div_output_0 /model/layers.17/input_layernorm/Mul_output_0 0=2 Cast /model/layers.17/input_layernorm/Cast_1 1 1 /model/layers.17/input_layernorm/Mul_output_0 /model/layers.17/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.17/input_layernorm/Mul_1 2 1 model.layers.17.input_layernorm.weight /model/layers.17/input_layernorm/Cast_1_output_0 /model/layers.17/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_547 1 3 /model/layers.17/input_layernorm/Mul_1_output_0 /model/layers.17/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.17/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.17/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.17/self_attn/Shape 1 1 /model/layers.17/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.17/self_attn/Shape_output_0 Shape /model/layers.17/self_attn/Shape_1 1 1 /model/layers.17/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.17/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.17/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.17/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.17/input_layernorm/Mul_1_output_0_quantized /model/layers.17/input_layernorm/Mul_1_output_0_scale /model/layers.17/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_548 1 3 /model/layers.17/input_layernorm/Mul_1_output_0_quantized /model/layers.17/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.17/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.17/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_549 1 3 /model/layers.17/input_layernorm/Mul_1_output_0_scale /model/layers.17/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.17/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.17/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_550 1 3 /model/layers.17/input_layernorm/Mul_1_output_0_zero_point /model/layers.17/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.17/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.17/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.17/self_attn/Gather 1 1 /model/layers.17/self_attn/Shape_output_0 /model/layers.17/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_551 1 4 /model/layers.17/self_attn/Gather_output_0 /model/layers.17/self_attn/Gather_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_output_0_splitncnn_1 /model/layers.17/self_attn/Gather_output_0_splitncnn_2 /model/layers.17/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.17/self_attn/Gather_1 1 1 /model/layers.17/self_attn/Shape_1_output_0 /model/layers.17/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_552 1 4 /model/layers.17/self_attn/Gather_1_output_0 /model/layers.17/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.17/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.17/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.17/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.17/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.17/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.153051e-03 BinaryOp /model/layers.17/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.17/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.17/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.614788e-03 BinaryOp /model/layers.17/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.17/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.17/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.245694e-03 MatMulInteger /model/layers.17/self_attn/q_proj/MatMul_quant 4 1 /model/layers.17/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7785_quantized /model/layers.17/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7785_zero_point /model/layers.17/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.17/self_attn/k_proj/MatMul_quant 4 1 /model/layers.17/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7792_quantized /model/layers.17/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7792_zero_point /model/layers.17/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.17/self_attn/v_proj/MatMul_quant 4 1 /model/layers.17/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7793_quantized /model/layers.17/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7793_zero_point /model/layers.17/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.17/self_attn/Unsqueeze 2 1 /model/layers.17/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_5240 /model/layers.17/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_2 2 1 /model/layers.17/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_5249 /model/layers.17/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_4 2 1 /model/layers.17/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_5258 /model/layers.17/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_29 2 1 /model/layers.17/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_5463 /model/layers.17/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_1 2 1 /model/layers.17/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_5242 /model/layers.17/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_3 2 1 /model/layers.17/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_5251 /model/layers.17/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_5 2 1 /model/layers.17/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_5260 /model/layers.17/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_30 2 1 /model/layers.17/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_5465 /model/layers.17/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.17/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.17/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.17/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.17/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.17/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.17/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.17/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.17/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.17/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.17/self_attn/Concat 4 1 /model/layers.17/self_attn/Unsqueeze_output_0 /model/layers.17/self_attn/Unsqueeze_1_output_0 /model/layers.17/self_attn/Constant_2_output_0 /model/layers.17/self_attn/Constant_3_output_0 /model/layers.17/self_attn/Concat_output_0 0=0 Concat /model/layers.17/self_attn/Concat_1 4 1 /model/layers.17/self_attn/Unsqueeze_2_output_0 /model/layers.17/self_attn/Unsqueeze_3_output_0 /model/layers.17/self_attn/Constant_4_output_0 /model/layers.17/self_attn/Constant_5_output_0 /model/layers.17/self_attn/Concat_1_output_0 0=0 Concat /model/layers.17/self_attn/Concat_2 4 1 /model/layers.17/self_attn/Unsqueeze_4_output_0 /model/layers.17/self_attn/Unsqueeze_5_output_0 /model/layers.17/self_attn/Constant_6_output_0 /model/layers.17/self_attn/Constant_7_output_0 /model/layers.17/self_attn/Concat_2_output_0 0=0 Concat /model/layers.17/self_attn/Concat_11 3 1 /model/layers.17/self_attn/Unsqueeze_29_output_0 /model/layers.17/self_attn/Unsqueeze_30_output_0 /model/layers.17/self_attn/Constant_64_output_0 /model/layers.17/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.17/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.17/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.17/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.17/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.17/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.17/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.17/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.17/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.17/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.17/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.17/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.17/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.17/self_attn/q_proj/Add 2 1 model.layers.17.self_attn.q_proj.bias /model/layers.17/self_attn/q_proj/MatMul_output_0 /model/layers.17/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.17/self_attn/k_proj/Add 2 1 model.layers.17.self_attn.k_proj.bias /model/layers.17/self_attn/k_proj/MatMul_output_0 /model/layers.17/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.17/self_attn/v_proj/Add 2 1 model.layers.17.self_attn.v_proj.bias /model/layers.17/self_attn/v_proj/MatMul_output_0 /model/layers.17/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.17/self_attn/Reshape 2 1 /model/layers.17/self_attn/q_proj/Add_output_0 /model/layers.17/self_attn/Concat_output_0 /model/layers.17/self_attn/Reshape_output_0 Reshape /model/layers.17/self_attn/Reshape_1 2 1 /model/layers.17/self_attn/k_proj/Add_output_0 /model/layers.17/self_attn/Concat_1_output_0 /model/layers.17/self_attn/Reshape_1_output_0 Reshape /model/layers.17/self_attn/Reshape_2 2 1 /model/layers.17/self_attn/v_proj/Add_output_0 /model/layers.17/self_attn/Concat_2_output_0 /model/layers.17/self_attn/Reshape_2_output_0 Permute /model/layers.17/self_attn/Transpose 1 1 /model/layers.17/self_attn/Reshape_output_0 /model/layers.17/self_attn/Transpose_output_0 0=2 Split splitncnn_553 1 4 /model/layers.17/self_attn/Transpose_output_0 /model/layers.17/self_attn/Transpose_output_0_splitncnn_0 /model/layers.17/self_attn/Transpose_output_0_splitncnn_1 /model/layers.17/self_attn/Transpose_output_0_splitncnn_2 /model/layers.17/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.17/self_attn/Transpose_1 1 1 /model/layers.17/self_attn/Reshape_1_output_0 /model/layers.17/self_attn/Transpose_1_output_0 0=2 Split splitncnn_554 1 4 /model/layers.17/self_attn/Transpose_1_output_0 /model/layers.17/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.17/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.17/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.17/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.17/self_attn/Transpose_2 1 1 /model/layers.17/self_attn/Reshape_2_output_0 /model/layers.17/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.17/self_attn/Mul 2 1 /model/layers.17/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_13 /model/layers.17/self_attn/Mul_output_0 0=2 Shape /model/layers.17/self_attn/Shape_2 1 1 /model/layers.17/self_attn/Transpose_output_0_splitncnn_2 /model/layers.17/self_attn/Shape_2_output_0 BinaryOp /model/layers.17/self_attn/Mul_2 2 1 /model/layers.17/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_12 /model/layers.17/self_attn/Mul_2_output_0 0=2 Shape /model/layers.17/self_attn/Shape_3 1 1 /model/layers.17/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.17/self_attn/Shape_3_output_0 Concat /model/layers.17/self_attn/Concat_6 2 1 past_key_values.17.value /model/layers.17/self_attn/Transpose_2_output_0 present.17.value 0=-2 Split splitncnn_555 1 5 present.17.value present.17.value_splitncnn_0 present.17.value_splitncnn_1 present.17.value_splitncnn_2 present.17.value_splitncnn_3 present.17.value_splitncnn_4 Crop /model/layers.17/self_attn/Gather_2 1 1 /model/layers.17/self_attn/Shape_2_output_0 /model/layers.17/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.17/self_attn/Gather_3 1 1 /model/layers.17/self_attn/Shape_3_output_0 /model/layers.17/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.17/self_attn/Shape_9 1 1 present.17.value_splitncnn_4 /model/layers.17/self_attn/Shape_9_output_0 Shape /model/layers.17/self_attn/Shape_10 1 1 present.17.value_splitncnn_3 /model/layers.17/self_attn/Shape_10_output_0 Shape /model/layers.17/self_attn/Shape_11 1 1 present.17.value_splitncnn_2 /model/layers.17/self_attn/Shape_11_output_0 Shape /model/layers.17/self_attn/Shape_12 1 1 present.17.value_splitncnn_1 /model/layers.17/self_attn/Shape_12_output_0 ExpandDims /model/layers.17/self_attn/Unsqueeze_19 2 1 present.17.value_splitncnn_0 /model/layers.17/self_attn/Constant_45_output_0 /model/layers.17/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.17/self_attn/Div 2 1 /model/layers.17/self_attn/Gather_2_output_0 /model/layers.17/self_attn/Constant_9_output_0 /model/layers.17/self_attn/Div_output_0 0=3 BinaryOp /model/layers.17/self_attn/Div_1 2 1 /model/layers.17/self_attn/Gather_3_output_0 /model/layers.17/self_attn/Constant_19_output_0 /model/layers.17/self_attn/Div_1_output_0 0=3 Crop /model/layers.17/self_attn/Gather_8 1 1 /model/layers.17/self_attn/Shape_9_output_0 /model/layers.17/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_556 1 2 /model/layers.17/self_attn/Gather_8_output_0 /model/layers.17/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.17/self_attn/Gather_9 1 1 /model/layers.17/self_attn/Shape_10_output_0 /model/layers.17/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_557 1 2 /model/layers.17/self_attn/Gather_9_output_0 /model/layers.17/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.17/self_attn/Gather_10 1 1 /model/layers.17/self_attn/Shape_11_output_0 /model/layers.17/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_558 1 2 /model/layers.17/self_attn/Gather_10_output_0 /model/layers.17/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.17/self_attn/Gather_11 1 1 /model/layers.17/self_attn/Shape_12_output_0 /model/layers.17/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_559 1 2 /model/layers.17/self_attn/Gather_11_output_0 /model/layers.17/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.17/self_attn/Cast 1 1 /model/layers.17/self_attn/Div_output_0 /model/layers.17/self_attn/Cast_output_0 Cast /model/layers.17/self_attn/Cast_2 1 1 /model/layers.17/self_attn/Div_1_output_0 /model/layers.17/self_attn/Cast_2_output_0 ExpandDims /model/layers.17/self_attn/Unsqueeze_20 2 1 /model/layers.17/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_46_output_0 /model/layers.17/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_24 2 1 /model/layers.17/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_5423 /model/layers.17/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_21 2 1 /model/layers.17/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_47_output_0 /model/layers.17/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.17/self_attn/Mul_7 2 1 /model/layers.17/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.17/self_attn/Constant_53_output_0 /model/layers.17/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.17/self_attn/Unsqueeze_22 2 1 /model/layers.17/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_49_output_0 /model/layers.17/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_26 2 1 /model/layers.17/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_5427 /model/layers.17/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_23 2 1 /model/layers.17/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_50_output_0 /model/layers.17/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_27 2 1 /model/layers.17/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_5429 /model/layers.17/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.17/self_attn/Cast_1 1 1 /model/layers.17/self_attn/Cast_output_0 /model/layers.17/self_attn/Cast_1_output_0 Split splitncnn_560 1 2 /model/layers.17/self_attn/Cast_1_output_0 /model/layers.17/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.17/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.17/self_attn/Cast_3 1 1 /model/layers.17/self_attn/Cast_2_output_0 /model/layers.17/self_attn/Cast_3_output_0 Split splitncnn_561 1 2 /model/layers.17/self_attn/Cast_3_output_0 /model/layers.17/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.17/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.17/self_attn/Unsqueeze_25 2 1 /model/layers.17/self_attn/Mul_7_output_0 onnx::Unsqueeze_5425 /model/layers.17/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.17/self_attn/Concat_9 5 1 /model/layers.17/self_attn/Unsqueeze_20_output_0 /model/layers.17/self_attn/Unsqueeze_21_output_0 /model/layers.17/self_attn/Constant_48_output_0 /model/layers.17/self_attn/Unsqueeze_22_output_0 /model/layers.17/self_attn/Unsqueeze_23_output_0 /model/layers.17/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_6 2 1 /model/layers.17/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_12_output_0 /model/layers.17/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_7 2 1 /model/layers.17/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.17/self_attn/Constant_15_output_0 /model/layers.17/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_8 2 1 /model/layers.17/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_22_output_0 /model/layers.17/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_9 2 1 /model/layers.17/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.17/self_attn/Constant_25_output_0 /model/layers.17/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.17/self_attn/Concat_10 4 1 /model/layers.17/self_attn/Unsqueeze_24_output_0 /model/layers.17/self_attn/Unsqueeze_25_output_0 /model/layers.17/self_attn/Unsqueeze_26_output_0 /model/layers.17/self_attn/Unsqueeze_27_output_0 /model/layers.17/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.17/self_attn/Reshape_5 1 1 /model/layers.17/self_attn/Concat_9_output_0 /model/layers.17/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_562 1 3 /model/layers.17/self_attn/Reshape_5_output_0 /model/layers.17/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.17/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.17/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.17/self_attn/Slice 2 1 /model/layers.17/self_attn/Transpose_output_0_splitncnn_1 /model/layers.17/self_attn/Unsqueeze_6_output_0 /model/layers.17/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.17/self_attn/Slice_1 2 1 /model/layers.17/self_attn/Transpose_output_0_splitncnn_0 /model/layers.17/self_attn/Unsqueeze_7_output_0 /model/layers.17/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.17/self_attn/Slice_2 2 1 /model/layers.17/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.17/self_attn/Unsqueeze_8_output_0 /model/layers.17/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.17/self_attn/Slice_3 2 1 /model/layers.17/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.17/self_attn/Unsqueeze_9_output_0 /model/layers.17/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.17/self_attn/Shape_13 1 1 /model/layers.17/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.17/self_attn/Shape_13_output_0 UnaryOp /model/layers.17/self_attn/Neg 1 1 /model/layers.17/self_attn/Slice_1_output_0 /model/layers.17/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.17/self_attn/Neg_1 1 1 /model/layers.17/self_attn/Slice_3_output_0 /model/layers.17/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.17/self_attn/ConstantOfShape_1 1 1 /model/layers.17/self_attn/Shape_13_output_0 /model/layers.17/self_attn/ConstantOfShape_1_output_0 Split splitncnn_563 1 2 /model/layers.17/self_attn/ConstantOfShape_1_output_0 /model/layers.17/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.17/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.17/self_attn/Concat_3 2 1 /model/layers.17/self_attn/Neg_output_0 /model/layers.17/self_attn/Slice_output_0 /model/layers.17/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.17/self_attn/Concat_4 2 1 /model/layers.17/self_attn/Neg_1_output_0 /model/layers.17/self_attn/Slice_2_output_0 /model/layers.17/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.17/self_attn/Mul_6 2 1 /model/layers.17/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_52_output_0 /model/layers.17/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.17/self_attn/Mul_1 2 1 /model/layers.17/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_13 /model/layers.17/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.17/self_attn/Mul_3 2 1 /model/layers.17/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_12 /model/layers.17/self_attn/Mul_3_output_0 0=2 Equal /model/layers.17/self_attn/Equal_1 2 1 /model/layers.17/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.17/self_attn/Mul_6_output_0 /model/layers.17/self_attn/Equal_1_output_0 BinaryOp /model/layers.17/self_attn/Add 2 1 /model/layers.17/self_attn/Mul_output_0 /model/layers.17/self_attn/Mul_1_output_0 /model/layers.17/self_attn/Add_output_0 0=0 BinaryOp /model/layers.17/self_attn/Add_1 2 1 /model/layers.17/self_attn/Mul_2_output_0 /model/layers.17/self_attn/Mul_3_output_0 /model/layers.17/self_attn/Add_1_output_0 0=0 Where /model/layers.17/self_attn/Where_1 3 1 /model/layers.17/self_attn/Equal_1_output_0 /model/layers.17/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.17/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.17/self_attn/Where_1_output_0 BinaryOp /model/layers.17/self_attn/Mul_8 1 1 /model/layers.17/self_attn/Add_output_0 /model/layers.17/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.17/self_attn/Concat_5 2 1 past_key_values.17.key /model/layers.17/self_attn/Add_1_output_0 present.17.key 0=-2 Split splitncnn_564 1 5 present.17.key present.17.key_splitncnn_0 present.17.key_splitncnn_1 present.17.key_splitncnn_2 present.17.key_splitncnn_3 present.17.key_splitncnn_4 Expand /model/layers.17/self_attn/Expand_1 2 1 /model/layers.17/self_attn/Unsqueeze_19_output_0 /model/layers.17/self_attn/Where_1_output_0 /model/layers.17/self_attn/Expand_1_output_0 Shape /model/layers.17/self_attn/Shape_4 1 1 present.17.key_splitncnn_4 /model/layers.17/self_attn/Shape_4_output_0 Shape /model/layers.17/self_attn/Shape_5 1 1 present.17.key_splitncnn_3 /model/layers.17/self_attn/Shape_5_output_0 Shape /model/layers.17/self_attn/Shape_6 1 1 present.17.key_splitncnn_2 /model/layers.17/self_attn/Shape_6_output_0 Shape /model/layers.17/self_attn/Shape_7 1 1 present.17.key_splitncnn_1 /model/layers.17/self_attn/Shape_7_output_0 ExpandDims /model/layers.17/self_attn/Unsqueeze_10 2 1 present.17.key_splitncnn_0 /model/layers.17/self_attn/Constant_32_output_0 /model/layers.17/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.17/self_attn/Reshape_6 2 1 /model/layers.17/self_attn/Expand_1_output_0 /model/layers.17/self_attn/Concat_10_output_0 /model/layers.17/self_attn/Reshape_6_output_0 Crop /model/layers.17/self_attn/Gather_4 1 1 /model/layers.17/self_attn/Shape_4_output_0 /model/layers.17/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_565 1 2 /model/layers.17/self_attn/Gather_4_output_0 /model/layers.17/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.17/self_attn/Gather_5 1 1 /model/layers.17/self_attn/Shape_5_output_0 /model/layers.17/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_566 1 2 /model/layers.17/self_attn/Gather_5_output_0 /model/layers.17/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.17/self_attn/Gather_6 1 1 /model/layers.17/self_attn/Shape_6_output_0 /model/layers.17/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_567 1 2 /model/layers.17/self_attn/Gather_6_output_0 /model/layers.17/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.17/self_attn/Gather_7 1 1 /model/layers.17/self_attn/Shape_7_output_0 /model/layers.17/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_568 1 2 /model/layers.17/self_attn/Gather_7_output_0 /model/layers.17/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.17/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.17/self_attn/Unsqueeze_11 2 1 /model/layers.17/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_33_output_0 /model/layers.17/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_15 2 1 /model/layers.17/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_5377 /model/layers.17/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_12 2 1 /model/layers.17/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_34_output_0 /model/layers.17/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.17/self_attn/Mul_5 2 1 /model/layers.17/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.17/self_attn/Constant_40_output_0 /model/layers.17/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.17/self_attn/Unsqueeze_13 2 1 /model/layers.17/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_36_output_0 /model/layers.17/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_17 2 1 /model/layers.17/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_5381 /model/layers.17/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_14 2 1 /model/layers.17/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_37_output_0 /model/layers.17/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_18 2 1 /model/layers.17/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_5383 /model/layers.17/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.17/self_attn/Unsqueeze_16 2 1 /model/layers.17/self_attn/Mul_5_output_0 onnx::Unsqueeze_5379 /model/layers.17/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.17/self_attn/Concat_7 5 1 /model/layers.17/self_attn/Unsqueeze_11_output_0 /model/layers.17/self_attn/Unsqueeze_12_output_0 /model/layers.17/self_attn/Constant_35_output_0 /model/layers.17/self_attn/Unsqueeze_13_output_0 /model/layers.17/self_attn/Unsqueeze_14_output_0 /model/layers.17/self_attn/Concat_7_output_0 0=0 Concat /model/layers.17/self_attn/Concat_8 4 1 /model/layers.17/self_attn/Unsqueeze_15_output_0 /model/layers.17/self_attn/Unsqueeze_16_output_0 /model/layers.17/self_attn/Unsqueeze_17_output_0 /model/layers.17/self_attn/Unsqueeze_18_output_0 /model/layers.17/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.17/self_attn/Reshape_3 1 1 /model/layers.17/self_attn/Concat_7_output_0 /model/layers.17/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_569 1 3 /model/layers.17/self_attn/Reshape_3_output_0 /model/layers.17/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.17/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.17/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.17/self_attn/Shape_8 1 1 /model/layers.17/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.17/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.17/self_attn/ConstantOfShape 1 1 /model/layers.17/self_attn/Shape_8_output_0 /model/layers.17/self_attn/ConstantOfShape_output_0 Split splitncnn_570 1 2 /model/layers.17/self_attn/ConstantOfShape_output_0 /model/layers.17/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.17/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.17/self_attn/Mul_4 2 1 /model/layers.17/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.17/self_attn/Constant_39_output_0 /model/layers.17/self_attn/Mul_4_output_0 0=2 Equal /model/layers.17/self_attn/Equal 2 1 /model/layers.17/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.17/self_attn/Mul_4_output_0 /model/layers.17/self_attn/Equal_output_0 Where /model/layers.17/self_attn/Where 3 1 /model/layers.17/self_attn/Equal_output_0 /model/layers.17/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.17/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.17/self_attn/Where_output_0 Expand /model/layers.17/self_attn/Expand 2 1 /model/layers.17/self_attn/Unsqueeze_10_output_0 /model/layers.17/self_attn/Where_output_0 /model/layers.17/self_attn/Expand_output_0 Reshape /model/layers.17/self_attn/Reshape_4 2 1 /model/layers.17/self_attn/Expand_output_0 /model/layers.17/self_attn/Concat_8_output_0 /model/layers.17/self_attn/Reshape_4_output_0 Split splitncnn_571 1 2 /model/layers.17/self_attn/Reshape_4_output_0 /model/layers.17/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.17/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.17/self_attn/Shape_14 1 1 /model/layers.17/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.17/self_attn/Shape_14_output_0 Permute /model/layers.17/self_attn/Transpose_3 1 1 /model/layers.17/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.17/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.17/self_attn/Gather_12 1 1 /model/layers.17/self_attn/Shape_14_output_0 /model/layers.17/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.17/self_attn/Mul_9 1 1 /model/layers.17/self_attn/Transpose_3_output_0 /model/layers.17/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.17/self_attn/Unsqueeze_28 2 1 /model/layers.17/self_attn/Gather_12_output_0 /model/layers.17/self_attn/Constant_57_output_0 /model/layers.17/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.17/self_attn/MatMul 2 1 /model/layers.17/self_attn/Mul_8_output_0 /model/layers.17/self_attn/Mul_9_output_0 /model/layers.17/self_attn/MatMul_output_0 Crop /model/layers.17/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_6 /model/layers.17/self_attn/Unsqueeze_28_output_0 /model/layers.17/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.17/self_attn/Where_2 3 1 /model/layers.17/self_attn/Slice_4_output_0 /model/layers.17/self_attn/Constant_61_output_0 /model/layers.17/self_attn/Constant_62_output_0 /model/layers.17/self_attn/Where_2_output_0 BinaryOp /model/layers.17/self_attn/Add_2 2 1 /model/layers.17/self_attn/MatMul_output_0 /model/layers.17/self_attn/Where_2_output_0 /model/layers.17/self_attn/Add_2_output_0 0=0 Softmax /model/layers.17/self_attn/Softmax 1 1 /model/layers.17/self_attn/Add_2_output_0 /model/layers.17/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_572 1 2 /model/layers.17/self_attn/Softmax_output_0 /model/layers.17/self_attn/Softmax_output_0_splitncnn_0 /model/layers.17/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.17/self_attn/IsNaN 1 1 /model/layers.17/self_attn/Softmax_output_0_splitncnn_1 /model/layers.17/self_attn/IsNaN_output_0 Where /model/layers.17/self_attn/Where_3 3 1 /model/layers.17/self_attn/IsNaN_output_0 /model/layers.17/self_attn/Constant_63_output_0 /model/layers.17/self_attn/Softmax_output_0_splitncnn_0 /model/layers.17/self_attn/Where_3_output_0 Gemm /model/layers.17/self_attn/MatMul_1 2 1 /model/layers.17/self_attn/Where_3_output_0 /model/layers.17/self_attn/Reshape_6_output_0 /model/layers.17/self_attn/MatMul_1_output_0 Permute /model/layers.17/self_attn/Transpose_4 1 1 /model/layers.17/self_attn/MatMul_1_output_0 /model/layers.17/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.17/self_attn/Reshape_7 2 1 /model/layers.17/self_attn/Transpose_4_output_0 /model/layers.17/self_attn/Concat_11_output_0 /model/layers.17/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.17/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.17/self_attn/Reshape_7_output_0 /model/layers.17/self_attn/Reshape_7_output_0_quantized /model/layers.17/self_attn/Reshape_7_output_0_scale /model/layers.17/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.17/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.17/self_attn/Reshape_7_output_0_scale /model/layers.17/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.168430e-03 MatMulInteger /model/layers.17/self_attn/o_proj/MatMul_quant 4 1 /model/layers.17/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7816_quantized /model/layers.17/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7816_zero_point /model/layers.17/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.17/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.17/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.17/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.17/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.17/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.17/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.17/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.17/Add 2 1 /model/layers.17/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.17/self_attn/o_proj/MatMul_output_0 /model/layers.17/Add_output_0 0=0 Cast /model/layers.17/post_attention_layernorm/Cast 1 1 /model/layers.17/Add_output_0 /model/layers.17/post_attention_layernorm/Cast_output_0 Split splitncnn_573 1 3 /model/layers.17/post_attention_layernorm/Cast_output_0 /model/layers.17/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.17/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.17/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.17/post_attention_layernorm/Pow 1 1 /model/layers.17/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.17/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.17/post_attention_layernorm/ReduceMean 2 1 /model/layers.17/post_attention_layernorm/Pow_output_0 /model/layers.17/post_attention_layernorm/Constant_1_output_0 /model/layers.17/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.17/post_attention_layernorm/Add 1 1 /model/layers.17/post_attention_layernorm/ReduceMean_output_0 /model/layers.17/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.17/post_attention_layernorm/Sqrt 1 1 /model/layers.17/post_attention_layernorm/Add_output_0 /model/layers.17/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.17/post_attention_layernorm/Div 1 1 /model/layers.17/post_attention_layernorm/Sqrt_output_0 /model/layers.17/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.17/post_attention_layernorm/Mul 2 1 /model/layers.17/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.17/post_attention_layernorm/Div_output_0 /model/layers.17/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.17/post_attention_layernorm/Cast_1 1 1 /model/layers.17/post_attention_layernorm/Mul_output_0 /model/layers.17/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.17/post_attention_layernorm/Mul_1 2 1 model.layers.17.post_attention_layernorm.weight /model/layers.17/post_attention_layernorm/Cast_1_output_0 /model/layers.17/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.17/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.17/post_attention_layernorm/Mul_1_output_0 /model/layers.17/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.17/post_attention_layernorm/Mul_1_output_0_scale /model/layers.17/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_574 1 2 /model/layers.17/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.17/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.17/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_575 1 2 /model/layers.17/post_attention_layernorm/Mul_1_output_0_scale /model/layers.17/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.17/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_576 1 2 /model/layers.17/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.17/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.17/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.17/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.17/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.17/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.783219e-03 BinaryOp /model/layers.17/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.17/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.17/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.799336e-03 MatMulInteger /model/layers.17/mlp/gate_proj/MatMul_quant 4 1 /model/layers.17/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7817_quantized /model/layers.17/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7817_zero_point /model/layers.17/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.17/mlp/up_proj/MatMul_quant 4 1 /model/layers.17/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7818_quantized /model/layers.17/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7818_zero_point /model/layers.17/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.17/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.17/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.17/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.17/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.17/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.17/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.17/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.17/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.17/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.17/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.17/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.17/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.17/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.17/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.17/mlp/act_fn/Mul 1 1 /model/layers.17/mlp/gate_proj/MatMul_output_0 /model/layers.17/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.17/mlp/Mul 2 1 /model/layers.17/mlp/act_fn/Mul_output_0 /model/layers.17/mlp/up_proj/MatMul_output_0 /model/layers.17/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.17/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.17/mlp/Mul_output_0 /model/layers.17/mlp/Mul_output_0_quantized /model/layers.17/mlp/Mul_output_0_scale /model/layers.17/mlp/Mul_output_0_zero_point BinaryOp /model/layers.17/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.17/mlp/Mul_output_0_scale /model/layers.17/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.829724e-03 MatMulInteger /model/layers.17/mlp/down_proj/MatMul_quant 4 1 /model/layers.17/mlp/Mul_output_0_quantized onnx::MatMul_7819_quantized /model/layers.17/mlp/Mul_output_0_zero_point onnx::MatMul_7819_zero_point /model/layers.17/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.17/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.17/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.17/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.17/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.17/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.17/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.17/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.17/Add_1 2 1 /model/layers.17/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.17/mlp/down_proj/MatMul_output_0 /model/layers.17/Add_1_output_0 0=0 Cast /model/layers.18/input_layernorm/Cast 1 1 /model/layers.17/Add_1_output_0 /model/layers.18/input_layernorm/Cast_output_0 Split splitncnn_577 1 3 /model/layers.18/input_layernorm/Cast_output_0 /model/layers.18/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.18/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.18/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.18/input_layernorm/Pow 1 1 /model/layers.18/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.18/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.18/input_layernorm/ReduceMean 2 1 /model/layers.18/input_layernorm/Pow_output_0 /model/layers.18/input_layernorm/Constant_1_output_0 /model/layers.18/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.18/input_layernorm/Add 1 1 /model/layers.18/input_layernorm/ReduceMean_output_0 /model/layers.18/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.18/input_layernorm/Sqrt 1 1 /model/layers.18/input_layernorm/Add_output_0 /model/layers.18/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.18/input_layernorm/Div 1 1 /model/layers.18/input_layernorm/Sqrt_output_0 /model/layers.18/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.18/input_layernorm/Mul 2 1 /model/layers.18/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.18/input_layernorm/Div_output_0 /model/layers.18/input_layernorm/Mul_output_0 0=2 Cast /model/layers.18/input_layernorm/Cast_1 1 1 /model/layers.18/input_layernorm/Mul_output_0 /model/layers.18/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.18/input_layernorm/Mul_1 2 1 model.layers.18.input_layernorm.weight /model/layers.18/input_layernorm/Cast_1_output_0 /model/layers.18/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_578 1 3 /model/layers.18/input_layernorm/Mul_1_output_0 /model/layers.18/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.18/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.18/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.18/self_attn/Shape 1 1 /model/layers.18/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.18/self_attn/Shape_output_0 Shape /model/layers.18/self_attn/Shape_1 1 1 /model/layers.18/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.18/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.18/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.18/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.18/input_layernorm/Mul_1_output_0_quantized /model/layers.18/input_layernorm/Mul_1_output_0_scale /model/layers.18/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_579 1 3 /model/layers.18/input_layernorm/Mul_1_output_0_quantized /model/layers.18/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.18/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.18/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_580 1 3 /model/layers.18/input_layernorm/Mul_1_output_0_scale /model/layers.18/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.18/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.18/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_581 1 3 /model/layers.18/input_layernorm/Mul_1_output_0_zero_point /model/layers.18/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.18/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.18/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.18/self_attn/Gather 1 1 /model/layers.18/self_attn/Shape_output_0 /model/layers.18/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_582 1 4 /model/layers.18/self_attn/Gather_output_0 /model/layers.18/self_attn/Gather_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_output_0_splitncnn_1 /model/layers.18/self_attn/Gather_output_0_splitncnn_2 /model/layers.18/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.18/self_attn/Gather_1 1 1 /model/layers.18/self_attn/Shape_1_output_0 /model/layers.18/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_583 1 4 /model/layers.18/self_attn/Gather_1_output_0 /model/layers.18/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.18/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.18/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.18/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.18/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.18/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.045029e-03 BinaryOp /model/layers.18/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.18/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.18/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.899299e-03 BinaryOp /model/layers.18/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.18/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.18/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=9.496494e-04 MatMulInteger /model/layers.18/self_attn/q_proj/MatMul_quant 4 1 /model/layers.18/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7820_quantized /model/layers.18/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7820_zero_point /model/layers.18/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.18/self_attn/k_proj/MatMul_quant 4 1 /model/layers.18/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7827_quantized /model/layers.18/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7827_zero_point /model/layers.18/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.18/self_attn/v_proj/MatMul_quant 4 1 /model/layers.18/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7828_quantized /model/layers.18/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7828_zero_point /model/layers.18/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.18/self_attn/Unsqueeze 2 1 /model/layers.18/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_5519 /model/layers.18/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_2 2 1 /model/layers.18/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_5528 /model/layers.18/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_4 2 1 /model/layers.18/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_5537 /model/layers.18/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_29 2 1 /model/layers.18/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_5742 /model/layers.18/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_1 2 1 /model/layers.18/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_5521 /model/layers.18/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_3 2 1 /model/layers.18/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_5530 /model/layers.18/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_5 2 1 /model/layers.18/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_5539 /model/layers.18/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_30 2 1 /model/layers.18/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_5744 /model/layers.18/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.18/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.18/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.18/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.18/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.18/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.18/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.18/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.18/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.18/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.18/self_attn/Concat 4 1 /model/layers.18/self_attn/Unsqueeze_output_0 /model/layers.18/self_attn/Unsqueeze_1_output_0 /model/layers.18/self_attn/Constant_2_output_0 /model/layers.18/self_attn/Constant_3_output_0 /model/layers.18/self_attn/Concat_output_0 0=0 Concat /model/layers.18/self_attn/Concat_1 4 1 /model/layers.18/self_attn/Unsqueeze_2_output_0 /model/layers.18/self_attn/Unsqueeze_3_output_0 /model/layers.18/self_attn/Constant_4_output_0 /model/layers.18/self_attn/Constant_5_output_0 /model/layers.18/self_attn/Concat_1_output_0 0=0 Concat /model/layers.18/self_attn/Concat_2 4 1 /model/layers.18/self_attn/Unsqueeze_4_output_0 /model/layers.18/self_attn/Unsqueeze_5_output_0 /model/layers.18/self_attn/Constant_6_output_0 /model/layers.18/self_attn/Constant_7_output_0 /model/layers.18/self_attn/Concat_2_output_0 0=0 Concat /model/layers.18/self_attn/Concat_11 3 1 /model/layers.18/self_attn/Unsqueeze_29_output_0 /model/layers.18/self_attn/Unsqueeze_30_output_0 /model/layers.18/self_attn/Constant_64_output_0 /model/layers.18/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.18/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.18/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.18/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.18/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.18/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.18/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.18/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.18/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.18/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.18/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.18/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.18/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.18/self_attn/q_proj/Add 2 1 model.layers.18.self_attn.q_proj.bias /model/layers.18/self_attn/q_proj/MatMul_output_0 /model/layers.18/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.18/self_attn/k_proj/Add 2 1 model.layers.18.self_attn.k_proj.bias /model/layers.18/self_attn/k_proj/MatMul_output_0 /model/layers.18/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.18/self_attn/v_proj/Add 2 1 model.layers.18.self_attn.v_proj.bias /model/layers.18/self_attn/v_proj/MatMul_output_0 /model/layers.18/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.18/self_attn/Reshape 2 1 /model/layers.18/self_attn/q_proj/Add_output_0 /model/layers.18/self_attn/Concat_output_0 /model/layers.18/self_attn/Reshape_output_0 Reshape /model/layers.18/self_attn/Reshape_1 2 1 /model/layers.18/self_attn/k_proj/Add_output_0 /model/layers.18/self_attn/Concat_1_output_0 /model/layers.18/self_attn/Reshape_1_output_0 Reshape /model/layers.18/self_attn/Reshape_2 2 1 /model/layers.18/self_attn/v_proj/Add_output_0 /model/layers.18/self_attn/Concat_2_output_0 /model/layers.18/self_attn/Reshape_2_output_0 Permute /model/layers.18/self_attn/Transpose 1 1 /model/layers.18/self_attn/Reshape_output_0 /model/layers.18/self_attn/Transpose_output_0 0=2 Split splitncnn_584 1 4 /model/layers.18/self_attn/Transpose_output_0 /model/layers.18/self_attn/Transpose_output_0_splitncnn_0 /model/layers.18/self_attn/Transpose_output_0_splitncnn_1 /model/layers.18/self_attn/Transpose_output_0_splitncnn_2 /model/layers.18/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.18/self_attn/Transpose_1 1 1 /model/layers.18/self_attn/Reshape_1_output_0 /model/layers.18/self_attn/Transpose_1_output_0 0=2 Split splitncnn_585 1 4 /model/layers.18/self_attn/Transpose_1_output_0 /model/layers.18/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.18/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.18/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.18/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.18/self_attn/Transpose_2 1 1 /model/layers.18/self_attn/Reshape_2_output_0 /model/layers.18/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.18/self_attn/Mul 2 1 /model/layers.18/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_11 /model/layers.18/self_attn/Mul_output_0 0=2 Shape /model/layers.18/self_attn/Shape_2 1 1 /model/layers.18/self_attn/Transpose_output_0_splitncnn_2 /model/layers.18/self_attn/Shape_2_output_0 BinaryOp /model/layers.18/self_attn/Mul_2 2 1 /model/layers.18/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_10 /model/layers.18/self_attn/Mul_2_output_0 0=2 Shape /model/layers.18/self_attn/Shape_3 1 1 /model/layers.18/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.18/self_attn/Shape_3_output_0 Concat /model/layers.18/self_attn/Concat_6 2 1 past_key_values.18.value /model/layers.18/self_attn/Transpose_2_output_0 present.18.value 0=-2 Split splitncnn_586 1 5 present.18.value present.18.value_splitncnn_0 present.18.value_splitncnn_1 present.18.value_splitncnn_2 present.18.value_splitncnn_3 present.18.value_splitncnn_4 Crop /model/layers.18/self_attn/Gather_2 1 1 /model/layers.18/self_attn/Shape_2_output_0 /model/layers.18/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.18/self_attn/Gather_3 1 1 /model/layers.18/self_attn/Shape_3_output_0 /model/layers.18/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.18/self_attn/Shape_9 1 1 present.18.value_splitncnn_4 /model/layers.18/self_attn/Shape_9_output_0 Shape /model/layers.18/self_attn/Shape_10 1 1 present.18.value_splitncnn_3 /model/layers.18/self_attn/Shape_10_output_0 Shape /model/layers.18/self_attn/Shape_11 1 1 present.18.value_splitncnn_2 /model/layers.18/self_attn/Shape_11_output_0 Shape /model/layers.18/self_attn/Shape_12 1 1 present.18.value_splitncnn_1 /model/layers.18/self_attn/Shape_12_output_0 ExpandDims /model/layers.18/self_attn/Unsqueeze_19 2 1 present.18.value_splitncnn_0 /model/layers.18/self_attn/Constant_45_output_0 /model/layers.18/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.18/self_attn/Div 2 1 /model/layers.18/self_attn/Gather_2_output_0 /model/layers.18/self_attn/Constant_9_output_0 /model/layers.18/self_attn/Div_output_0 0=3 BinaryOp /model/layers.18/self_attn/Div_1 2 1 /model/layers.18/self_attn/Gather_3_output_0 /model/layers.18/self_attn/Constant_19_output_0 /model/layers.18/self_attn/Div_1_output_0 0=3 Crop /model/layers.18/self_attn/Gather_8 1 1 /model/layers.18/self_attn/Shape_9_output_0 /model/layers.18/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_587 1 2 /model/layers.18/self_attn/Gather_8_output_0 /model/layers.18/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.18/self_attn/Gather_9 1 1 /model/layers.18/self_attn/Shape_10_output_0 /model/layers.18/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_588 1 2 /model/layers.18/self_attn/Gather_9_output_0 /model/layers.18/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.18/self_attn/Gather_10 1 1 /model/layers.18/self_attn/Shape_11_output_0 /model/layers.18/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_589 1 2 /model/layers.18/self_attn/Gather_10_output_0 /model/layers.18/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.18/self_attn/Gather_11 1 1 /model/layers.18/self_attn/Shape_12_output_0 /model/layers.18/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_590 1 2 /model/layers.18/self_attn/Gather_11_output_0 /model/layers.18/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.18/self_attn/Cast 1 1 /model/layers.18/self_attn/Div_output_0 /model/layers.18/self_attn/Cast_output_0 Cast /model/layers.18/self_attn/Cast_2 1 1 /model/layers.18/self_attn/Div_1_output_0 /model/layers.18/self_attn/Cast_2_output_0 ExpandDims /model/layers.18/self_attn/Unsqueeze_20 2 1 /model/layers.18/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_46_output_0 /model/layers.18/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_24 2 1 /model/layers.18/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_5702 /model/layers.18/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_21 2 1 /model/layers.18/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_47_output_0 /model/layers.18/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.18/self_attn/Mul_7 2 1 /model/layers.18/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.18/self_attn/Constant_53_output_0 /model/layers.18/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.18/self_attn/Unsqueeze_22 2 1 /model/layers.18/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_49_output_0 /model/layers.18/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_26 2 1 /model/layers.18/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_5706 /model/layers.18/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_23 2 1 /model/layers.18/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_50_output_0 /model/layers.18/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_27 2 1 /model/layers.18/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_5708 /model/layers.18/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.18/self_attn/Cast_1 1 1 /model/layers.18/self_attn/Cast_output_0 /model/layers.18/self_attn/Cast_1_output_0 Split splitncnn_591 1 2 /model/layers.18/self_attn/Cast_1_output_0 /model/layers.18/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.18/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.18/self_attn/Cast_3 1 1 /model/layers.18/self_attn/Cast_2_output_0 /model/layers.18/self_attn/Cast_3_output_0 Split splitncnn_592 1 2 /model/layers.18/self_attn/Cast_3_output_0 /model/layers.18/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.18/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.18/self_attn/Unsqueeze_25 2 1 /model/layers.18/self_attn/Mul_7_output_0 onnx::Unsqueeze_5704 /model/layers.18/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.18/self_attn/Concat_9 5 1 /model/layers.18/self_attn/Unsqueeze_20_output_0 /model/layers.18/self_attn/Unsqueeze_21_output_0 /model/layers.18/self_attn/Constant_48_output_0 /model/layers.18/self_attn/Unsqueeze_22_output_0 /model/layers.18/self_attn/Unsqueeze_23_output_0 /model/layers.18/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_6 2 1 /model/layers.18/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_12_output_0 /model/layers.18/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_7 2 1 /model/layers.18/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.18/self_attn/Constant_15_output_0 /model/layers.18/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_8 2 1 /model/layers.18/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_22_output_0 /model/layers.18/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_9 2 1 /model/layers.18/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.18/self_attn/Constant_25_output_0 /model/layers.18/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.18/self_attn/Concat_10 4 1 /model/layers.18/self_attn/Unsqueeze_24_output_0 /model/layers.18/self_attn/Unsqueeze_25_output_0 /model/layers.18/self_attn/Unsqueeze_26_output_0 /model/layers.18/self_attn/Unsqueeze_27_output_0 /model/layers.18/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.18/self_attn/Reshape_5 1 1 /model/layers.18/self_attn/Concat_9_output_0 /model/layers.18/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_593 1 3 /model/layers.18/self_attn/Reshape_5_output_0 /model/layers.18/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.18/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.18/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.18/self_attn/Slice 2 1 /model/layers.18/self_attn/Transpose_output_0_splitncnn_1 /model/layers.18/self_attn/Unsqueeze_6_output_0 /model/layers.18/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.18/self_attn/Slice_1 2 1 /model/layers.18/self_attn/Transpose_output_0_splitncnn_0 /model/layers.18/self_attn/Unsqueeze_7_output_0 /model/layers.18/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.18/self_attn/Slice_2 2 1 /model/layers.18/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.18/self_attn/Unsqueeze_8_output_0 /model/layers.18/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.18/self_attn/Slice_3 2 1 /model/layers.18/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.18/self_attn/Unsqueeze_9_output_0 /model/layers.18/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.18/self_attn/Shape_13 1 1 /model/layers.18/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.18/self_attn/Shape_13_output_0 UnaryOp /model/layers.18/self_attn/Neg 1 1 /model/layers.18/self_attn/Slice_1_output_0 /model/layers.18/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.18/self_attn/Neg_1 1 1 /model/layers.18/self_attn/Slice_3_output_0 /model/layers.18/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.18/self_attn/ConstantOfShape_1 1 1 /model/layers.18/self_attn/Shape_13_output_0 /model/layers.18/self_attn/ConstantOfShape_1_output_0 Split splitncnn_594 1 2 /model/layers.18/self_attn/ConstantOfShape_1_output_0 /model/layers.18/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.18/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.18/self_attn/Concat_3 2 1 /model/layers.18/self_attn/Neg_output_0 /model/layers.18/self_attn/Slice_output_0 /model/layers.18/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.18/self_attn/Concat_4 2 1 /model/layers.18/self_attn/Neg_1_output_0 /model/layers.18/self_attn/Slice_2_output_0 /model/layers.18/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.18/self_attn/Mul_6 2 1 /model/layers.18/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_52_output_0 /model/layers.18/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.18/self_attn/Mul_1 2 1 /model/layers.18/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_11 /model/layers.18/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.18/self_attn/Mul_3 2 1 /model/layers.18/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_10 /model/layers.18/self_attn/Mul_3_output_0 0=2 Equal /model/layers.18/self_attn/Equal_1 2 1 /model/layers.18/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.18/self_attn/Mul_6_output_0 /model/layers.18/self_attn/Equal_1_output_0 BinaryOp /model/layers.18/self_attn/Add 2 1 /model/layers.18/self_attn/Mul_output_0 /model/layers.18/self_attn/Mul_1_output_0 /model/layers.18/self_attn/Add_output_0 0=0 BinaryOp /model/layers.18/self_attn/Add_1 2 1 /model/layers.18/self_attn/Mul_2_output_0 /model/layers.18/self_attn/Mul_3_output_0 /model/layers.18/self_attn/Add_1_output_0 0=0 Where /model/layers.18/self_attn/Where_1 3 1 /model/layers.18/self_attn/Equal_1_output_0 /model/layers.18/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.18/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.18/self_attn/Where_1_output_0 BinaryOp /model/layers.18/self_attn/Mul_8 1 1 /model/layers.18/self_attn/Add_output_0 /model/layers.18/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.18/self_attn/Concat_5 2 1 past_key_values.18.key /model/layers.18/self_attn/Add_1_output_0 present.18.key 0=-2 Split splitncnn_595 1 5 present.18.key present.18.key_splitncnn_0 present.18.key_splitncnn_1 present.18.key_splitncnn_2 present.18.key_splitncnn_3 present.18.key_splitncnn_4 Expand /model/layers.18/self_attn/Expand_1 2 1 /model/layers.18/self_attn/Unsqueeze_19_output_0 /model/layers.18/self_attn/Where_1_output_0 /model/layers.18/self_attn/Expand_1_output_0 Shape /model/layers.18/self_attn/Shape_4 1 1 present.18.key_splitncnn_4 /model/layers.18/self_attn/Shape_4_output_0 Shape /model/layers.18/self_attn/Shape_5 1 1 present.18.key_splitncnn_3 /model/layers.18/self_attn/Shape_5_output_0 Shape /model/layers.18/self_attn/Shape_6 1 1 present.18.key_splitncnn_2 /model/layers.18/self_attn/Shape_6_output_0 Shape /model/layers.18/self_attn/Shape_7 1 1 present.18.key_splitncnn_1 /model/layers.18/self_attn/Shape_7_output_0 ExpandDims /model/layers.18/self_attn/Unsqueeze_10 2 1 present.18.key_splitncnn_0 /model/layers.18/self_attn/Constant_32_output_0 /model/layers.18/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.18/self_attn/Reshape_6 2 1 /model/layers.18/self_attn/Expand_1_output_0 /model/layers.18/self_attn/Concat_10_output_0 /model/layers.18/self_attn/Reshape_6_output_0 Crop /model/layers.18/self_attn/Gather_4 1 1 /model/layers.18/self_attn/Shape_4_output_0 /model/layers.18/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_596 1 2 /model/layers.18/self_attn/Gather_4_output_0 /model/layers.18/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.18/self_attn/Gather_5 1 1 /model/layers.18/self_attn/Shape_5_output_0 /model/layers.18/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_597 1 2 /model/layers.18/self_attn/Gather_5_output_0 /model/layers.18/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.18/self_attn/Gather_6 1 1 /model/layers.18/self_attn/Shape_6_output_0 /model/layers.18/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_598 1 2 /model/layers.18/self_attn/Gather_6_output_0 /model/layers.18/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.18/self_attn/Gather_7 1 1 /model/layers.18/self_attn/Shape_7_output_0 /model/layers.18/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_599 1 2 /model/layers.18/self_attn/Gather_7_output_0 /model/layers.18/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.18/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.18/self_attn/Unsqueeze_11 2 1 /model/layers.18/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_33_output_0 /model/layers.18/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_15 2 1 /model/layers.18/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_5656 /model/layers.18/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_12 2 1 /model/layers.18/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_34_output_0 /model/layers.18/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.18/self_attn/Mul_5 2 1 /model/layers.18/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.18/self_attn/Constant_40_output_0 /model/layers.18/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.18/self_attn/Unsqueeze_13 2 1 /model/layers.18/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_36_output_0 /model/layers.18/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_17 2 1 /model/layers.18/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_5660 /model/layers.18/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_14 2 1 /model/layers.18/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_37_output_0 /model/layers.18/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_18 2 1 /model/layers.18/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_5662 /model/layers.18/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.18/self_attn/Unsqueeze_16 2 1 /model/layers.18/self_attn/Mul_5_output_0 onnx::Unsqueeze_5658 /model/layers.18/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.18/self_attn/Concat_7 5 1 /model/layers.18/self_attn/Unsqueeze_11_output_0 /model/layers.18/self_attn/Unsqueeze_12_output_0 /model/layers.18/self_attn/Constant_35_output_0 /model/layers.18/self_attn/Unsqueeze_13_output_0 /model/layers.18/self_attn/Unsqueeze_14_output_0 /model/layers.18/self_attn/Concat_7_output_0 0=0 Concat /model/layers.18/self_attn/Concat_8 4 1 /model/layers.18/self_attn/Unsqueeze_15_output_0 /model/layers.18/self_attn/Unsqueeze_16_output_0 /model/layers.18/self_attn/Unsqueeze_17_output_0 /model/layers.18/self_attn/Unsqueeze_18_output_0 /model/layers.18/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.18/self_attn/Reshape_3 1 1 /model/layers.18/self_attn/Concat_7_output_0 /model/layers.18/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_600 1 3 /model/layers.18/self_attn/Reshape_3_output_0 /model/layers.18/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.18/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.18/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.18/self_attn/Shape_8 1 1 /model/layers.18/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.18/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.18/self_attn/ConstantOfShape 1 1 /model/layers.18/self_attn/Shape_8_output_0 /model/layers.18/self_attn/ConstantOfShape_output_0 Split splitncnn_601 1 2 /model/layers.18/self_attn/ConstantOfShape_output_0 /model/layers.18/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.18/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.18/self_attn/Mul_4 2 1 /model/layers.18/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.18/self_attn/Constant_39_output_0 /model/layers.18/self_attn/Mul_4_output_0 0=2 Equal /model/layers.18/self_attn/Equal 2 1 /model/layers.18/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.18/self_attn/Mul_4_output_0 /model/layers.18/self_attn/Equal_output_0 Where /model/layers.18/self_attn/Where 3 1 /model/layers.18/self_attn/Equal_output_0 /model/layers.18/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.18/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.18/self_attn/Where_output_0 Expand /model/layers.18/self_attn/Expand 2 1 /model/layers.18/self_attn/Unsqueeze_10_output_0 /model/layers.18/self_attn/Where_output_0 /model/layers.18/self_attn/Expand_output_0 Reshape /model/layers.18/self_attn/Reshape_4 2 1 /model/layers.18/self_attn/Expand_output_0 /model/layers.18/self_attn/Concat_8_output_0 /model/layers.18/self_attn/Reshape_4_output_0 Split splitncnn_602 1 2 /model/layers.18/self_attn/Reshape_4_output_0 /model/layers.18/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.18/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.18/self_attn/Shape_14 1 1 /model/layers.18/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.18/self_attn/Shape_14_output_0 Permute /model/layers.18/self_attn/Transpose_3 1 1 /model/layers.18/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.18/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.18/self_attn/Gather_12 1 1 /model/layers.18/self_attn/Shape_14_output_0 /model/layers.18/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.18/self_attn/Mul_9 1 1 /model/layers.18/self_attn/Transpose_3_output_0 /model/layers.18/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.18/self_attn/Unsqueeze_28 2 1 /model/layers.18/self_attn/Gather_12_output_0 /model/layers.18/self_attn/Constant_57_output_0 /model/layers.18/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.18/self_attn/MatMul 2 1 /model/layers.18/self_attn/Mul_8_output_0 /model/layers.18/self_attn/Mul_9_output_0 /model/layers.18/self_attn/MatMul_output_0 Crop /model/layers.18/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_5 /model/layers.18/self_attn/Unsqueeze_28_output_0 /model/layers.18/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.18/self_attn/Where_2 3 1 /model/layers.18/self_attn/Slice_4_output_0 /model/layers.18/self_attn/Constant_61_output_0 /model/layers.18/self_attn/Constant_62_output_0 /model/layers.18/self_attn/Where_2_output_0 BinaryOp /model/layers.18/self_attn/Add_2 2 1 /model/layers.18/self_attn/MatMul_output_0 /model/layers.18/self_attn/Where_2_output_0 /model/layers.18/self_attn/Add_2_output_0 0=0 Softmax /model/layers.18/self_attn/Softmax 1 1 /model/layers.18/self_attn/Add_2_output_0 /model/layers.18/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_603 1 2 /model/layers.18/self_attn/Softmax_output_0 /model/layers.18/self_attn/Softmax_output_0_splitncnn_0 /model/layers.18/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.18/self_attn/IsNaN 1 1 /model/layers.18/self_attn/Softmax_output_0_splitncnn_1 /model/layers.18/self_attn/IsNaN_output_0 Where /model/layers.18/self_attn/Where_3 3 1 /model/layers.18/self_attn/IsNaN_output_0 /model/layers.18/self_attn/Constant_63_output_0 /model/layers.18/self_attn/Softmax_output_0_splitncnn_0 /model/layers.18/self_attn/Where_3_output_0 Gemm /model/layers.18/self_attn/MatMul_1 2 1 /model/layers.18/self_attn/Where_3_output_0 /model/layers.18/self_attn/Reshape_6_output_0 /model/layers.18/self_attn/MatMul_1_output_0 Permute /model/layers.18/self_attn/Transpose_4 1 1 /model/layers.18/self_attn/MatMul_1_output_0 /model/layers.18/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.18/self_attn/Reshape_7 2 1 /model/layers.18/self_attn/Transpose_4_output_0 /model/layers.18/self_attn/Concat_11_output_0 /model/layers.18/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.18/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.18/self_attn/Reshape_7_output_0 /model/layers.18/self_attn/Reshape_7_output_0_quantized /model/layers.18/self_attn/Reshape_7_output_0_scale /model/layers.18/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.18/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.18/self_attn/Reshape_7_output_0_scale /model/layers.18/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.960814e-03 MatMulInteger /model/layers.18/self_attn/o_proj/MatMul_quant 4 1 /model/layers.18/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7851_quantized /model/layers.18/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7851_zero_point /model/layers.18/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.18/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.18/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.18/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.18/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.18/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.18/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.18/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.18/Add 2 1 /model/layers.18/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.18/self_attn/o_proj/MatMul_output_0 /model/layers.18/Add_output_0 0=0 Cast /model/layers.18/post_attention_layernorm/Cast 1 1 /model/layers.18/Add_output_0 /model/layers.18/post_attention_layernorm/Cast_output_0 Split splitncnn_604 1 3 /model/layers.18/post_attention_layernorm/Cast_output_0 /model/layers.18/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.18/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.18/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.18/post_attention_layernorm/Pow 1 1 /model/layers.18/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.18/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.18/post_attention_layernorm/ReduceMean 2 1 /model/layers.18/post_attention_layernorm/Pow_output_0 /model/layers.18/post_attention_layernorm/Constant_1_output_0 /model/layers.18/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.18/post_attention_layernorm/Add 1 1 /model/layers.18/post_attention_layernorm/ReduceMean_output_0 /model/layers.18/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.18/post_attention_layernorm/Sqrt 1 1 /model/layers.18/post_attention_layernorm/Add_output_0 /model/layers.18/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.18/post_attention_layernorm/Div 1 1 /model/layers.18/post_attention_layernorm/Sqrt_output_0 /model/layers.18/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.18/post_attention_layernorm/Mul 2 1 /model/layers.18/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.18/post_attention_layernorm/Div_output_0 /model/layers.18/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.18/post_attention_layernorm/Cast_1 1 1 /model/layers.18/post_attention_layernorm/Mul_output_0 /model/layers.18/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.18/post_attention_layernorm/Mul_1 2 1 model.layers.18.post_attention_layernorm.weight /model/layers.18/post_attention_layernorm/Cast_1_output_0 /model/layers.18/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.18/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.18/post_attention_layernorm/Mul_1_output_0 /model/layers.18/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.18/post_attention_layernorm/Mul_1_output_0_scale /model/layers.18/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_605 1 2 /model/layers.18/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.18/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.18/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_606 1 2 /model/layers.18/post_attention_layernorm/Mul_1_output_0_scale /model/layers.18/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.18/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_607 1 2 /model/layers.18/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.18/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.18/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.18/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.18/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.18/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.152682e-03 BinaryOp /model/layers.18/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.18/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.18/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.629798e-03 MatMulInteger /model/layers.18/mlp/gate_proj/MatMul_quant 4 1 /model/layers.18/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7852_quantized /model/layers.18/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7852_zero_point /model/layers.18/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.18/mlp/up_proj/MatMul_quant 4 1 /model/layers.18/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7853_quantized /model/layers.18/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7853_zero_point /model/layers.18/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.18/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.18/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.18/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.18/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.18/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.18/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.18/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.18/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.18/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.18/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.18/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.18/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.18/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.18/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.18/mlp/act_fn/Mul 1 1 /model/layers.18/mlp/gate_proj/MatMul_output_0 /model/layers.18/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.18/mlp/Mul 2 1 /model/layers.18/mlp/act_fn/Mul_output_0 /model/layers.18/mlp/up_proj/MatMul_output_0 /model/layers.18/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.18/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.18/mlp/Mul_output_0 /model/layers.18/mlp/Mul_output_0_quantized /model/layers.18/mlp/Mul_output_0_scale /model/layers.18/mlp/Mul_output_0_zero_point BinaryOp /model/layers.18/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.18/mlp/Mul_output_0_scale /model/layers.18/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.552534e-03 MatMulInteger /model/layers.18/mlp/down_proj/MatMul_quant 4 1 /model/layers.18/mlp/Mul_output_0_quantized onnx::MatMul_7854_quantized /model/layers.18/mlp/Mul_output_0_zero_point onnx::MatMul_7854_zero_point /model/layers.18/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.18/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.18/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.18/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.18/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.18/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.18/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.18/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.18/Add_1 2 1 /model/layers.18/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.18/mlp/down_proj/MatMul_output_0 /model/layers.18/Add_1_output_0 0=0 Cast /model/layers.19/input_layernorm/Cast 1 1 /model/layers.18/Add_1_output_0 /model/layers.19/input_layernorm/Cast_output_0 Split splitncnn_608 1 3 /model/layers.19/input_layernorm/Cast_output_0 /model/layers.19/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.19/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.19/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.19/input_layernorm/Pow 1 1 /model/layers.19/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.19/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.19/input_layernorm/ReduceMean 2 1 /model/layers.19/input_layernorm/Pow_output_0 /model/layers.19/input_layernorm/Constant_1_output_0 /model/layers.19/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.19/input_layernorm/Add 1 1 /model/layers.19/input_layernorm/ReduceMean_output_0 /model/layers.19/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.19/input_layernorm/Sqrt 1 1 /model/layers.19/input_layernorm/Add_output_0 /model/layers.19/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.19/input_layernorm/Div 1 1 /model/layers.19/input_layernorm/Sqrt_output_0 /model/layers.19/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.19/input_layernorm/Mul 2 1 /model/layers.19/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.19/input_layernorm/Div_output_0 /model/layers.19/input_layernorm/Mul_output_0 0=2 Cast /model/layers.19/input_layernorm/Cast_1 1 1 /model/layers.19/input_layernorm/Mul_output_0 /model/layers.19/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.19/input_layernorm/Mul_1 2 1 model.layers.19.input_layernorm.weight /model/layers.19/input_layernorm/Cast_1_output_0 /model/layers.19/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_609 1 3 /model/layers.19/input_layernorm/Mul_1_output_0 /model/layers.19/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.19/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.19/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.19/self_attn/Shape 1 1 /model/layers.19/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.19/self_attn/Shape_output_0 Shape /model/layers.19/self_attn/Shape_1 1 1 /model/layers.19/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.19/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.19/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.19/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.19/input_layernorm/Mul_1_output_0_quantized /model/layers.19/input_layernorm/Mul_1_output_0_scale /model/layers.19/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_610 1 3 /model/layers.19/input_layernorm/Mul_1_output_0_quantized /model/layers.19/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.19/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.19/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_611 1 3 /model/layers.19/input_layernorm/Mul_1_output_0_scale /model/layers.19/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.19/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.19/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_612 1 3 /model/layers.19/input_layernorm/Mul_1_output_0_zero_point /model/layers.19/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.19/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.19/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.19/self_attn/Gather 1 1 /model/layers.19/self_attn/Shape_output_0 /model/layers.19/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_613 1 4 /model/layers.19/self_attn/Gather_output_0 /model/layers.19/self_attn/Gather_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_output_0_splitncnn_1 /model/layers.19/self_attn/Gather_output_0_splitncnn_2 /model/layers.19/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.19/self_attn/Gather_1 1 1 /model/layers.19/self_attn/Shape_1_output_0 /model/layers.19/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_614 1 4 /model/layers.19/self_attn/Gather_1_output_0 /model/layers.19/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.19/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.19/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.19/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.19/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.19/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.807025e-03 BinaryOp /model/layers.19/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.19/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.19/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.445620e-03 BinaryOp /model/layers.19/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.19/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.19/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.114973e-03 MatMulInteger /model/layers.19/self_attn/q_proj/MatMul_quant 4 1 /model/layers.19/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7855_quantized /model/layers.19/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7855_zero_point /model/layers.19/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.19/self_attn/k_proj/MatMul_quant 4 1 /model/layers.19/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7862_quantized /model/layers.19/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7862_zero_point /model/layers.19/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.19/self_attn/v_proj/MatMul_quant 4 1 /model/layers.19/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7863_quantized /model/layers.19/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7863_zero_point /model/layers.19/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.19/self_attn/Unsqueeze 2 1 /model/layers.19/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_5798 /model/layers.19/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_2 2 1 /model/layers.19/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_5807 /model/layers.19/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_4 2 1 /model/layers.19/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_5816 /model/layers.19/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_29 2 1 /model/layers.19/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_6021 /model/layers.19/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_1 2 1 /model/layers.19/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_5800 /model/layers.19/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_3 2 1 /model/layers.19/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_5809 /model/layers.19/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_5 2 1 /model/layers.19/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_5818 /model/layers.19/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_30 2 1 /model/layers.19/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_6023 /model/layers.19/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.19/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.19/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.19/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.19/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.19/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.19/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.19/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.19/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.19/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.19/self_attn/Concat 4 1 /model/layers.19/self_attn/Unsqueeze_output_0 /model/layers.19/self_attn/Unsqueeze_1_output_0 /model/layers.19/self_attn/Constant_2_output_0 /model/layers.19/self_attn/Constant_3_output_0 /model/layers.19/self_attn/Concat_output_0 0=0 Concat /model/layers.19/self_attn/Concat_1 4 1 /model/layers.19/self_attn/Unsqueeze_2_output_0 /model/layers.19/self_attn/Unsqueeze_3_output_0 /model/layers.19/self_attn/Constant_4_output_0 /model/layers.19/self_attn/Constant_5_output_0 /model/layers.19/self_attn/Concat_1_output_0 0=0 Concat /model/layers.19/self_attn/Concat_2 4 1 /model/layers.19/self_attn/Unsqueeze_4_output_0 /model/layers.19/self_attn/Unsqueeze_5_output_0 /model/layers.19/self_attn/Constant_6_output_0 /model/layers.19/self_attn/Constant_7_output_0 /model/layers.19/self_attn/Concat_2_output_0 0=0 Concat /model/layers.19/self_attn/Concat_11 3 1 /model/layers.19/self_attn/Unsqueeze_29_output_0 /model/layers.19/self_attn/Unsqueeze_30_output_0 /model/layers.19/self_attn/Constant_64_output_0 /model/layers.19/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.19/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.19/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.19/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.19/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.19/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.19/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.19/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.19/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.19/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.19/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.19/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.19/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.19/self_attn/q_proj/Add 2 1 model.layers.19.self_attn.q_proj.bias /model/layers.19/self_attn/q_proj/MatMul_output_0 /model/layers.19/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.19/self_attn/k_proj/Add 2 1 model.layers.19.self_attn.k_proj.bias /model/layers.19/self_attn/k_proj/MatMul_output_0 /model/layers.19/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.19/self_attn/v_proj/Add 2 1 model.layers.19.self_attn.v_proj.bias /model/layers.19/self_attn/v_proj/MatMul_output_0 /model/layers.19/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.19/self_attn/Reshape 2 1 /model/layers.19/self_attn/q_proj/Add_output_0 /model/layers.19/self_attn/Concat_output_0 /model/layers.19/self_attn/Reshape_output_0 Reshape /model/layers.19/self_attn/Reshape_1 2 1 /model/layers.19/self_attn/k_proj/Add_output_0 /model/layers.19/self_attn/Concat_1_output_0 /model/layers.19/self_attn/Reshape_1_output_0 Reshape /model/layers.19/self_attn/Reshape_2 2 1 /model/layers.19/self_attn/v_proj/Add_output_0 /model/layers.19/self_attn/Concat_2_output_0 /model/layers.19/self_attn/Reshape_2_output_0 Permute /model/layers.19/self_attn/Transpose 1 1 /model/layers.19/self_attn/Reshape_output_0 /model/layers.19/self_attn/Transpose_output_0 0=2 Split splitncnn_615 1 4 /model/layers.19/self_attn/Transpose_output_0 /model/layers.19/self_attn/Transpose_output_0_splitncnn_0 /model/layers.19/self_attn/Transpose_output_0_splitncnn_1 /model/layers.19/self_attn/Transpose_output_0_splitncnn_2 /model/layers.19/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.19/self_attn/Transpose_1 1 1 /model/layers.19/self_attn/Reshape_1_output_0 /model/layers.19/self_attn/Transpose_1_output_0 0=2 Split splitncnn_616 1 4 /model/layers.19/self_attn/Transpose_1_output_0 /model/layers.19/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.19/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.19/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.19/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.19/self_attn/Transpose_2 1 1 /model/layers.19/self_attn/Reshape_2_output_0 /model/layers.19/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.19/self_attn/Mul 2 1 /model/layers.19/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_9 /model/layers.19/self_attn/Mul_output_0 0=2 Shape /model/layers.19/self_attn/Shape_2 1 1 /model/layers.19/self_attn/Transpose_output_0_splitncnn_2 /model/layers.19/self_attn/Shape_2_output_0 BinaryOp /model/layers.19/self_attn/Mul_2 2 1 /model/layers.19/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_8 /model/layers.19/self_attn/Mul_2_output_0 0=2 Shape /model/layers.19/self_attn/Shape_3 1 1 /model/layers.19/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.19/self_attn/Shape_3_output_0 Concat /model/layers.19/self_attn/Concat_6 2 1 past_key_values.19.value /model/layers.19/self_attn/Transpose_2_output_0 present.19.value 0=-2 Split splitncnn_617 1 5 present.19.value present.19.value_splitncnn_0 present.19.value_splitncnn_1 present.19.value_splitncnn_2 present.19.value_splitncnn_3 present.19.value_splitncnn_4 Crop /model/layers.19/self_attn/Gather_2 1 1 /model/layers.19/self_attn/Shape_2_output_0 /model/layers.19/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.19/self_attn/Gather_3 1 1 /model/layers.19/self_attn/Shape_3_output_0 /model/layers.19/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.19/self_attn/Shape_9 1 1 present.19.value_splitncnn_4 /model/layers.19/self_attn/Shape_9_output_0 Shape /model/layers.19/self_attn/Shape_10 1 1 present.19.value_splitncnn_3 /model/layers.19/self_attn/Shape_10_output_0 Shape /model/layers.19/self_attn/Shape_11 1 1 present.19.value_splitncnn_2 /model/layers.19/self_attn/Shape_11_output_0 Shape /model/layers.19/self_attn/Shape_12 1 1 present.19.value_splitncnn_1 /model/layers.19/self_attn/Shape_12_output_0 ExpandDims /model/layers.19/self_attn/Unsqueeze_19 2 1 present.19.value_splitncnn_0 /model/layers.19/self_attn/Constant_45_output_0 /model/layers.19/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.19/self_attn/Div 2 1 /model/layers.19/self_attn/Gather_2_output_0 /model/layers.19/self_attn/Constant_9_output_0 /model/layers.19/self_attn/Div_output_0 0=3 BinaryOp /model/layers.19/self_attn/Div_1 2 1 /model/layers.19/self_attn/Gather_3_output_0 /model/layers.19/self_attn/Constant_19_output_0 /model/layers.19/self_attn/Div_1_output_0 0=3 Crop /model/layers.19/self_attn/Gather_8 1 1 /model/layers.19/self_attn/Shape_9_output_0 /model/layers.19/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_618 1 2 /model/layers.19/self_attn/Gather_8_output_0 /model/layers.19/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.19/self_attn/Gather_9 1 1 /model/layers.19/self_attn/Shape_10_output_0 /model/layers.19/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_619 1 2 /model/layers.19/self_attn/Gather_9_output_0 /model/layers.19/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.19/self_attn/Gather_10 1 1 /model/layers.19/self_attn/Shape_11_output_0 /model/layers.19/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_620 1 2 /model/layers.19/self_attn/Gather_10_output_0 /model/layers.19/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.19/self_attn/Gather_11 1 1 /model/layers.19/self_attn/Shape_12_output_0 /model/layers.19/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_621 1 2 /model/layers.19/self_attn/Gather_11_output_0 /model/layers.19/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.19/self_attn/Cast 1 1 /model/layers.19/self_attn/Div_output_0 /model/layers.19/self_attn/Cast_output_0 Cast /model/layers.19/self_attn/Cast_2 1 1 /model/layers.19/self_attn/Div_1_output_0 /model/layers.19/self_attn/Cast_2_output_0 ExpandDims /model/layers.19/self_attn/Unsqueeze_20 2 1 /model/layers.19/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_46_output_0 /model/layers.19/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_24 2 1 /model/layers.19/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_5981 /model/layers.19/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_21 2 1 /model/layers.19/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_47_output_0 /model/layers.19/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.19/self_attn/Mul_7 2 1 /model/layers.19/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.19/self_attn/Constant_53_output_0 /model/layers.19/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.19/self_attn/Unsqueeze_22 2 1 /model/layers.19/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_49_output_0 /model/layers.19/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_26 2 1 /model/layers.19/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_5985 /model/layers.19/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_23 2 1 /model/layers.19/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_50_output_0 /model/layers.19/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_27 2 1 /model/layers.19/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_5987 /model/layers.19/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.19/self_attn/Cast_1 1 1 /model/layers.19/self_attn/Cast_output_0 /model/layers.19/self_attn/Cast_1_output_0 Split splitncnn_622 1 2 /model/layers.19/self_attn/Cast_1_output_0 /model/layers.19/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.19/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.19/self_attn/Cast_3 1 1 /model/layers.19/self_attn/Cast_2_output_0 /model/layers.19/self_attn/Cast_3_output_0 Split splitncnn_623 1 2 /model/layers.19/self_attn/Cast_3_output_0 /model/layers.19/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.19/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.19/self_attn/Unsqueeze_25 2 1 /model/layers.19/self_attn/Mul_7_output_0 onnx::Unsqueeze_5983 /model/layers.19/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.19/self_attn/Concat_9 5 1 /model/layers.19/self_attn/Unsqueeze_20_output_0 /model/layers.19/self_attn/Unsqueeze_21_output_0 /model/layers.19/self_attn/Constant_48_output_0 /model/layers.19/self_attn/Unsqueeze_22_output_0 /model/layers.19/self_attn/Unsqueeze_23_output_0 /model/layers.19/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_6 2 1 /model/layers.19/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_12_output_0 /model/layers.19/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_7 2 1 /model/layers.19/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.19/self_attn/Constant_15_output_0 /model/layers.19/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_8 2 1 /model/layers.19/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_22_output_0 /model/layers.19/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_9 2 1 /model/layers.19/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.19/self_attn/Constant_25_output_0 /model/layers.19/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.19/self_attn/Concat_10 4 1 /model/layers.19/self_attn/Unsqueeze_24_output_0 /model/layers.19/self_attn/Unsqueeze_25_output_0 /model/layers.19/self_attn/Unsqueeze_26_output_0 /model/layers.19/self_attn/Unsqueeze_27_output_0 /model/layers.19/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.19/self_attn/Reshape_5 1 1 /model/layers.19/self_attn/Concat_9_output_0 /model/layers.19/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_624 1 3 /model/layers.19/self_attn/Reshape_5_output_0 /model/layers.19/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.19/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.19/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.19/self_attn/Slice 2 1 /model/layers.19/self_attn/Transpose_output_0_splitncnn_1 /model/layers.19/self_attn/Unsqueeze_6_output_0 /model/layers.19/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.19/self_attn/Slice_1 2 1 /model/layers.19/self_attn/Transpose_output_0_splitncnn_0 /model/layers.19/self_attn/Unsqueeze_7_output_0 /model/layers.19/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.19/self_attn/Slice_2 2 1 /model/layers.19/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.19/self_attn/Unsqueeze_8_output_0 /model/layers.19/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.19/self_attn/Slice_3 2 1 /model/layers.19/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.19/self_attn/Unsqueeze_9_output_0 /model/layers.19/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.19/self_attn/Shape_13 1 1 /model/layers.19/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.19/self_attn/Shape_13_output_0 UnaryOp /model/layers.19/self_attn/Neg 1 1 /model/layers.19/self_attn/Slice_1_output_0 /model/layers.19/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.19/self_attn/Neg_1 1 1 /model/layers.19/self_attn/Slice_3_output_0 /model/layers.19/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.19/self_attn/ConstantOfShape_1 1 1 /model/layers.19/self_attn/Shape_13_output_0 /model/layers.19/self_attn/ConstantOfShape_1_output_0 Split splitncnn_625 1 2 /model/layers.19/self_attn/ConstantOfShape_1_output_0 /model/layers.19/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.19/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.19/self_attn/Concat_3 2 1 /model/layers.19/self_attn/Neg_output_0 /model/layers.19/self_attn/Slice_output_0 /model/layers.19/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.19/self_attn/Concat_4 2 1 /model/layers.19/self_attn/Neg_1_output_0 /model/layers.19/self_attn/Slice_2_output_0 /model/layers.19/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.19/self_attn/Mul_6 2 1 /model/layers.19/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_52_output_0 /model/layers.19/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.19/self_attn/Mul_1 2 1 /model/layers.19/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_9 /model/layers.19/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.19/self_attn/Mul_3 2 1 /model/layers.19/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_8 /model/layers.19/self_attn/Mul_3_output_0 0=2 Equal /model/layers.19/self_attn/Equal_1 2 1 /model/layers.19/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.19/self_attn/Mul_6_output_0 /model/layers.19/self_attn/Equal_1_output_0 BinaryOp /model/layers.19/self_attn/Add 2 1 /model/layers.19/self_attn/Mul_output_0 /model/layers.19/self_attn/Mul_1_output_0 /model/layers.19/self_attn/Add_output_0 0=0 BinaryOp /model/layers.19/self_attn/Add_1 2 1 /model/layers.19/self_attn/Mul_2_output_0 /model/layers.19/self_attn/Mul_3_output_0 /model/layers.19/self_attn/Add_1_output_0 0=0 Where /model/layers.19/self_attn/Where_1 3 1 /model/layers.19/self_attn/Equal_1_output_0 /model/layers.19/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.19/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.19/self_attn/Where_1_output_0 BinaryOp /model/layers.19/self_attn/Mul_8 1 1 /model/layers.19/self_attn/Add_output_0 /model/layers.19/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.19/self_attn/Concat_5 2 1 past_key_values.19.key /model/layers.19/self_attn/Add_1_output_0 present.19.key 0=-2 Split splitncnn_626 1 5 present.19.key present.19.key_splitncnn_0 present.19.key_splitncnn_1 present.19.key_splitncnn_2 present.19.key_splitncnn_3 present.19.key_splitncnn_4 Expand /model/layers.19/self_attn/Expand_1 2 1 /model/layers.19/self_attn/Unsqueeze_19_output_0 /model/layers.19/self_attn/Where_1_output_0 /model/layers.19/self_attn/Expand_1_output_0 Shape /model/layers.19/self_attn/Shape_4 1 1 present.19.key_splitncnn_4 /model/layers.19/self_attn/Shape_4_output_0 Shape /model/layers.19/self_attn/Shape_5 1 1 present.19.key_splitncnn_3 /model/layers.19/self_attn/Shape_5_output_0 Shape /model/layers.19/self_attn/Shape_6 1 1 present.19.key_splitncnn_2 /model/layers.19/self_attn/Shape_6_output_0 Shape /model/layers.19/self_attn/Shape_7 1 1 present.19.key_splitncnn_1 /model/layers.19/self_attn/Shape_7_output_0 ExpandDims /model/layers.19/self_attn/Unsqueeze_10 2 1 present.19.key_splitncnn_0 /model/layers.19/self_attn/Constant_32_output_0 /model/layers.19/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.19/self_attn/Reshape_6 2 1 /model/layers.19/self_attn/Expand_1_output_0 /model/layers.19/self_attn/Concat_10_output_0 /model/layers.19/self_attn/Reshape_6_output_0 Crop /model/layers.19/self_attn/Gather_4 1 1 /model/layers.19/self_attn/Shape_4_output_0 /model/layers.19/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_627 1 2 /model/layers.19/self_attn/Gather_4_output_0 /model/layers.19/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.19/self_attn/Gather_5 1 1 /model/layers.19/self_attn/Shape_5_output_0 /model/layers.19/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_628 1 2 /model/layers.19/self_attn/Gather_5_output_0 /model/layers.19/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.19/self_attn/Gather_6 1 1 /model/layers.19/self_attn/Shape_6_output_0 /model/layers.19/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_629 1 2 /model/layers.19/self_attn/Gather_6_output_0 /model/layers.19/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.19/self_attn/Gather_7 1 1 /model/layers.19/self_attn/Shape_7_output_0 /model/layers.19/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_630 1 2 /model/layers.19/self_attn/Gather_7_output_0 /model/layers.19/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.19/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.19/self_attn/Unsqueeze_11 2 1 /model/layers.19/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_33_output_0 /model/layers.19/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_15 2 1 /model/layers.19/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_5935 /model/layers.19/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_12 2 1 /model/layers.19/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_34_output_0 /model/layers.19/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.19/self_attn/Mul_5 2 1 /model/layers.19/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.19/self_attn/Constant_40_output_0 /model/layers.19/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.19/self_attn/Unsqueeze_13 2 1 /model/layers.19/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_36_output_0 /model/layers.19/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_17 2 1 /model/layers.19/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_5939 /model/layers.19/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_14 2 1 /model/layers.19/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_37_output_0 /model/layers.19/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_18 2 1 /model/layers.19/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_5941 /model/layers.19/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.19/self_attn/Unsqueeze_16 2 1 /model/layers.19/self_attn/Mul_5_output_0 onnx::Unsqueeze_5937 /model/layers.19/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.19/self_attn/Concat_7 5 1 /model/layers.19/self_attn/Unsqueeze_11_output_0 /model/layers.19/self_attn/Unsqueeze_12_output_0 /model/layers.19/self_attn/Constant_35_output_0 /model/layers.19/self_attn/Unsqueeze_13_output_0 /model/layers.19/self_attn/Unsqueeze_14_output_0 /model/layers.19/self_attn/Concat_7_output_0 0=0 Concat /model/layers.19/self_attn/Concat_8 4 1 /model/layers.19/self_attn/Unsqueeze_15_output_0 /model/layers.19/self_attn/Unsqueeze_16_output_0 /model/layers.19/self_attn/Unsqueeze_17_output_0 /model/layers.19/self_attn/Unsqueeze_18_output_0 /model/layers.19/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.19/self_attn/Reshape_3 1 1 /model/layers.19/self_attn/Concat_7_output_0 /model/layers.19/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_631 1 3 /model/layers.19/self_attn/Reshape_3_output_0 /model/layers.19/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.19/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.19/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.19/self_attn/Shape_8 1 1 /model/layers.19/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.19/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.19/self_attn/ConstantOfShape 1 1 /model/layers.19/self_attn/Shape_8_output_0 /model/layers.19/self_attn/ConstantOfShape_output_0 Split splitncnn_632 1 2 /model/layers.19/self_attn/ConstantOfShape_output_0 /model/layers.19/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.19/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.19/self_attn/Mul_4 2 1 /model/layers.19/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.19/self_attn/Constant_39_output_0 /model/layers.19/self_attn/Mul_4_output_0 0=2 Equal /model/layers.19/self_attn/Equal 2 1 /model/layers.19/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.19/self_attn/Mul_4_output_0 /model/layers.19/self_attn/Equal_output_0 Where /model/layers.19/self_attn/Where 3 1 /model/layers.19/self_attn/Equal_output_0 /model/layers.19/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.19/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.19/self_attn/Where_output_0 Expand /model/layers.19/self_attn/Expand 2 1 /model/layers.19/self_attn/Unsqueeze_10_output_0 /model/layers.19/self_attn/Where_output_0 /model/layers.19/self_attn/Expand_output_0 Reshape /model/layers.19/self_attn/Reshape_4 2 1 /model/layers.19/self_attn/Expand_output_0 /model/layers.19/self_attn/Concat_8_output_0 /model/layers.19/self_attn/Reshape_4_output_0 Split splitncnn_633 1 2 /model/layers.19/self_attn/Reshape_4_output_0 /model/layers.19/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.19/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.19/self_attn/Shape_14 1 1 /model/layers.19/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.19/self_attn/Shape_14_output_0 Permute /model/layers.19/self_attn/Transpose_3 1 1 /model/layers.19/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.19/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.19/self_attn/Gather_12 1 1 /model/layers.19/self_attn/Shape_14_output_0 /model/layers.19/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.19/self_attn/Mul_9 1 1 /model/layers.19/self_attn/Transpose_3_output_0 /model/layers.19/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.19/self_attn/Unsqueeze_28 2 1 /model/layers.19/self_attn/Gather_12_output_0 /model/layers.19/self_attn/Constant_57_output_0 /model/layers.19/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.19/self_attn/MatMul 2 1 /model/layers.19/self_attn/Mul_8_output_0 /model/layers.19/self_attn/Mul_9_output_0 /model/layers.19/self_attn/MatMul_output_0 Crop /model/layers.19/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_4 /model/layers.19/self_attn/Unsqueeze_28_output_0 /model/layers.19/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.19/self_attn/Where_2 3 1 /model/layers.19/self_attn/Slice_4_output_0 /model/layers.19/self_attn/Constant_61_output_0 /model/layers.19/self_attn/Constant_62_output_0 /model/layers.19/self_attn/Where_2_output_0 BinaryOp /model/layers.19/self_attn/Add_2 2 1 /model/layers.19/self_attn/MatMul_output_0 /model/layers.19/self_attn/Where_2_output_0 /model/layers.19/self_attn/Add_2_output_0 0=0 Softmax /model/layers.19/self_attn/Softmax 1 1 /model/layers.19/self_attn/Add_2_output_0 /model/layers.19/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_634 1 2 /model/layers.19/self_attn/Softmax_output_0 /model/layers.19/self_attn/Softmax_output_0_splitncnn_0 /model/layers.19/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.19/self_attn/IsNaN 1 1 /model/layers.19/self_attn/Softmax_output_0_splitncnn_1 /model/layers.19/self_attn/IsNaN_output_0 Where /model/layers.19/self_attn/Where_3 3 1 /model/layers.19/self_attn/IsNaN_output_0 /model/layers.19/self_attn/Constant_63_output_0 /model/layers.19/self_attn/Softmax_output_0_splitncnn_0 /model/layers.19/self_attn/Where_3_output_0 Gemm /model/layers.19/self_attn/MatMul_1 2 1 /model/layers.19/self_attn/Where_3_output_0 /model/layers.19/self_attn/Reshape_6_output_0 /model/layers.19/self_attn/MatMul_1_output_0 Permute /model/layers.19/self_attn/Transpose_4 1 1 /model/layers.19/self_attn/MatMul_1_output_0 /model/layers.19/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.19/self_attn/Reshape_7 2 1 /model/layers.19/self_attn/Transpose_4_output_0 /model/layers.19/self_attn/Concat_11_output_0 /model/layers.19/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.19/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.19/self_attn/Reshape_7_output_0 /model/layers.19/self_attn/Reshape_7_output_0_quantized /model/layers.19/self_attn/Reshape_7_output_0_scale /model/layers.19/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.19/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.19/self_attn/Reshape_7_output_0_scale /model/layers.19/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.214567e-03 MatMulInteger /model/layers.19/self_attn/o_proj/MatMul_quant 4 1 /model/layers.19/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7886_quantized /model/layers.19/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7886_zero_point /model/layers.19/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.19/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.19/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.19/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.19/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.19/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.19/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.19/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.19/Add 2 1 /model/layers.19/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.19/self_attn/o_proj/MatMul_output_0 /model/layers.19/Add_output_0 0=0 Cast /model/layers.19/post_attention_layernorm/Cast 1 1 /model/layers.19/Add_output_0 /model/layers.19/post_attention_layernorm/Cast_output_0 Split splitncnn_635 1 3 /model/layers.19/post_attention_layernorm/Cast_output_0 /model/layers.19/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.19/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.19/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.19/post_attention_layernorm/Pow 1 1 /model/layers.19/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.19/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.19/post_attention_layernorm/ReduceMean 2 1 /model/layers.19/post_attention_layernorm/Pow_output_0 /model/layers.19/post_attention_layernorm/Constant_1_output_0 /model/layers.19/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.19/post_attention_layernorm/Add 1 1 /model/layers.19/post_attention_layernorm/ReduceMean_output_0 /model/layers.19/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.19/post_attention_layernorm/Sqrt 1 1 /model/layers.19/post_attention_layernorm/Add_output_0 /model/layers.19/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.19/post_attention_layernorm/Div 1 1 /model/layers.19/post_attention_layernorm/Sqrt_output_0 /model/layers.19/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.19/post_attention_layernorm/Mul 2 1 /model/layers.19/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.19/post_attention_layernorm/Div_output_0 /model/layers.19/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.19/post_attention_layernorm/Cast_1 1 1 /model/layers.19/post_attention_layernorm/Mul_output_0 /model/layers.19/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.19/post_attention_layernorm/Mul_1 2 1 model.layers.19.post_attention_layernorm.weight /model/layers.19/post_attention_layernorm/Cast_1_output_0 /model/layers.19/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.19/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.19/post_attention_layernorm/Mul_1_output_0 /model/layers.19/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.19/post_attention_layernorm/Mul_1_output_0_scale /model/layers.19/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_636 1 2 /model/layers.19/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.19/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.19/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_637 1 2 /model/layers.19/post_attention_layernorm/Mul_1_output_0_scale /model/layers.19/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.19/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_638 1 2 /model/layers.19/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.19/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.19/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.19/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.19/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.19/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.582923e-03 BinaryOp /model/layers.19/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.19/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.19/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.337229e-03 MatMulInteger /model/layers.19/mlp/gate_proj/MatMul_quant 4 1 /model/layers.19/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7887_quantized /model/layers.19/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7887_zero_point /model/layers.19/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.19/mlp/up_proj/MatMul_quant 4 1 /model/layers.19/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7888_quantized /model/layers.19/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7888_zero_point /model/layers.19/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.19/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.19/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.19/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.19/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.19/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.19/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.19/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.19/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.19/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.19/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.19/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.19/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.19/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.19/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.19/mlp/act_fn/Mul 1 1 /model/layers.19/mlp/gate_proj/MatMul_output_0 /model/layers.19/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.19/mlp/Mul 2 1 /model/layers.19/mlp/act_fn/Mul_output_0 /model/layers.19/mlp/up_proj/MatMul_output_0 /model/layers.19/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.19/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.19/mlp/Mul_output_0 /model/layers.19/mlp/Mul_output_0_quantized /model/layers.19/mlp/Mul_output_0_scale /model/layers.19/mlp/Mul_output_0_zero_point BinaryOp /model/layers.19/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.19/mlp/Mul_output_0_scale /model/layers.19/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.491388e-03 MatMulInteger /model/layers.19/mlp/down_proj/MatMul_quant 4 1 /model/layers.19/mlp/Mul_output_0_quantized onnx::MatMul_7889_quantized /model/layers.19/mlp/Mul_output_0_zero_point onnx::MatMul_7889_zero_point /model/layers.19/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.19/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.19/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.19/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.19/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.19/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.19/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.19/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.19/Add_1 2 1 /model/layers.19/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.19/mlp/down_proj/MatMul_output_0 /model/layers.19/Add_1_output_0 0=0 Cast /model/layers.20/input_layernorm/Cast 1 1 /model/layers.19/Add_1_output_0 /model/layers.20/input_layernorm/Cast_output_0 Split splitncnn_639 1 3 /model/layers.20/input_layernorm/Cast_output_0 /model/layers.20/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.20/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.20/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.20/input_layernorm/Pow 1 1 /model/layers.20/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.20/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.20/input_layernorm/ReduceMean 2 1 /model/layers.20/input_layernorm/Pow_output_0 /model/layers.20/input_layernorm/Constant_1_output_0 /model/layers.20/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.20/input_layernorm/Add 1 1 /model/layers.20/input_layernorm/ReduceMean_output_0 /model/layers.20/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.20/input_layernorm/Sqrt 1 1 /model/layers.20/input_layernorm/Add_output_0 /model/layers.20/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.20/input_layernorm/Div 1 1 /model/layers.20/input_layernorm/Sqrt_output_0 /model/layers.20/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.20/input_layernorm/Mul 2 1 /model/layers.20/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.20/input_layernorm/Div_output_0 /model/layers.20/input_layernorm/Mul_output_0 0=2 Cast /model/layers.20/input_layernorm/Cast_1 1 1 /model/layers.20/input_layernorm/Mul_output_0 /model/layers.20/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.20/input_layernorm/Mul_1 2 1 model.layers.20.input_layernorm.weight /model/layers.20/input_layernorm/Cast_1_output_0 /model/layers.20/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_640 1 3 /model/layers.20/input_layernorm/Mul_1_output_0 /model/layers.20/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.20/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.20/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.20/self_attn/Shape 1 1 /model/layers.20/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.20/self_attn/Shape_output_0 Shape /model/layers.20/self_attn/Shape_1 1 1 /model/layers.20/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.20/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.20/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.20/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.20/input_layernorm/Mul_1_output_0_quantized /model/layers.20/input_layernorm/Mul_1_output_0_scale /model/layers.20/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_641 1 3 /model/layers.20/input_layernorm/Mul_1_output_0_quantized /model/layers.20/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.20/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.20/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_642 1 3 /model/layers.20/input_layernorm/Mul_1_output_0_scale /model/layers.20/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.20/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.20/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_643 1 3 /model/layers.20/input_layernorm/Mul_1_output_0_zero_point /model/layers.20/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.20/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.20/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.20/self_attn/Gather 1 1 /model/layers.20/self_attn/Shape_output_0 /model/layers.20/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_644 1 4 /model/layers.20/self_attn/Gather_output_0 /model/layers.20/self_attn/Gather_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_output_0_splitncnn_1 /model/layers.20/self_attn/Gather_output_0_splitncnn_2 /model/layers.20/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.20/self_attn/Gather_1 1 1 /model/layers.20/self_attn/Shape_1_output_0 /model/layers.20/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_645 1 4 /model/layers.20/self_attn/Gather_1_output_0 /model/layers.20/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.20/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.20/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.20/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.20/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.20/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.737820e-03 BinaryOp /model/layers.20/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.20/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.20/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.491757e-03 BinaryOp /model/layers.20/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.20/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.20/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.822404e-03 MatMulInteger /model/layers.20/self_attn/q_proj/MatMul_quant 4 1 /model/layers.20/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7890_quantized /model/layers.20/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7890_zero_point /model/layers.20/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.20/self_attn/k_proj/MatMul_quant 4 1 /model/layers.20/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7897_quantized /model/layers.20/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7897_zero_point /model/layers.20/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.20/self_attn/v_proj/MatMul_quant 4 1 /model/layers.20/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7898_quantized /model/layers.20/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7898_zero_point /model/layers.20/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.20/self_attn/Unsqueeze 2 1 /model/layers.20/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_6077 /model/layers.20/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_2 2 1 /model/layers.20/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_6086 /model/layers.20/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_4 2 1 /model/layers.20/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_6095 /model/layers.20/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_29 2 1 /model/layers.20/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_6300 /model/layers.20/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_1 2 1 /model/layers.20/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_6079 /model/layers.20/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_3 2 1 /model/layers.20/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_6088 /model/layers.20/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_5 2 1 /model/layers.20/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_6097 /model/layers.20/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_30 2 1 /model/layers.20/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_6302 /model/layers.20/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.20/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.20/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.20/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.20/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.20/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.20/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.20/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.20/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.20/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.20/self_attn/Concat 4 1 /model/layers.20/self_attn/Unsqueeze_output_0 /model/layers.20/self_attn/Unsqueeze_1_output_0 /model/layers.20/self_attn/Constant_2_output_0 /model/layers.20/self_attn/Constant_3_output_0 /model/layers.20/self_attn/Concat_output_0 0=0 Concat /model/layers.20/self_attn/Concat_1 4 1 /model/layers.20/self_attn/Unsqueeze_2_output_0 /model/layers.20/self_attn/Unsqueeze_3_output_0 /model/layers.20/self_attn/Constant_4_output_0 /model/layers.20/self_attn/Constant_5_output_0 /model/layers.20/self_attn/Concat_1_output_0 0=0 Concat /model/layers.20/self_attn/Concat_2 4 1 /model/layers.20/self_attn/Unsqueeze_4_output_0 /model/layers.20/self_attn/Unsqueeze_5_output_0 /model/layers.20/self_attn/Constant_6_output_0 /model/layers.20/self_attn/Constant_7_output_0 /model/layers.20/self_attn/Concat_2_output_0 0=0 Concat /model/layers.20/self_attn/Concat_11 3 1 /model/layers.20/self_attn/Unsqueeze_29_output_0 /model/layers.20/self_attn/Unsqueeze_30_output_0 /model/layers.20/self_attn/Constant_64_output_0 /model/layers.20/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.20/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.20/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.20/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.20/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.20/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.20/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.20/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.20/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.20/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.20/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.20/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.20/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.20/self_attn/q_proj/Add 2 1 model.layers.20.self_attn.q_proj.bias /model/layers.20/self_attn/q_proj/MatMul_output_0 /model/layers.20/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.20/self_attn/k_proj/Add 2 1 model.layers.20.self_attn.k_proj.bias /model/layers.20/self_attn/k_proj/MatMul_output_0 /model/layers.20/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.20/self_attn/v_proj/Add 2 1 model.layers.20.self_attn.v_proj.bias /model/layers.20/self_attn/v_proj/MatMul_output_0 /model/layers.20/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.20/self_attn/Reshape 2 1 /model/layers.20/self_attn/q_proj/Add_output_0 /model/layers.20/self_attn/Concat_output_0 /model/layers.20/self_attn/Reshape_output_0 Reshape /model/layers.20/self_attn/Reshape_1 2 1 /model/layers.20/self_attn/k_proj/Add_output_0 /model/layers.20/self_attn/Concat_1_output_0 /model/layers.20/self_attn/Reshape_1_output_0 Reshape /model/layers.20/self_attn/Reshape_2 2 1 /model/layers.20/self_attn/v_proj/Add_output_0 /model/layers.20/self_attn/Concat_2_output_0 /model/layers.20/self_attn/Reshape_2_output_0 Permute /model/layers.20/self_attn/Transpose 1 1 /model/layers.20/self_attn/Reshape_output_0 /model/layers.20/self_attn/Transpose_output_0 0=2 Split splitncnn_646 1 4 /model/layers.20/self_attn/Transpose_output_0 /model/layers.20/self_attn/Transpose_output_0_splitncnn_0 /model/layers.20/self_attn/Transpose_output_0_splitncnn_1 /model/layers.20/self_attn/Transpose_output_0_splitncnn_2 /model/layers.20/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.20/self_attn/Transpose_1 1 1 /model/layers.20/self_attn/Reshape_1_output_0 /model/layers.20/self_attn/Transpose_1_output_0 0=2 Split splitncnn_647 1 4 /model/layers.20/self_attn/Transpose_1_output_0 /model/layers.20/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.20/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.20/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.20/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.20/self_attn/Transpose_2 1 1 /model/layers.20/self_attn/Reshape_2_output_0 /model/layers.20/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.20/self_attn/Mul 2 1 /model/layers.20/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_7 /model/layers.20/self_attn/Mul_output_0 0=2 Shape /model/layers.20/self_attn/Shape_2 1 1 /model/layers.20/self_attn/Transpose_output_0_splitncnn_2 /model/layers.20/self_attn/Shape_2_output_0 BinaryOp /model/layers.20/self_attn/Mul_2 2 1 /model/layers.20/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_6 /model/layers.20/self_attn/Mul_2_output_0 0=2 Shape /model/layers.20/self_attn/Shape_3 1 1 /model/layers.20/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.20/self_attn/Shape_3_output_0 Concat /model/layers.20/self_attn/Concat_6 2 1 past_key_values.20.value /model/layers.20/self_attn/Transpose_2_output_0 present.20.value 0=-2 Split splitncnn_648 1 5 present.20.value present.20.value_splitncnn_0 present.20.value_splitncnn_1 present.20.value_splitncnn_2 present.20.value_splitncnn_3 present.20.value_splitncnn_4 Crop /model/layers.20/self_attn/Gather_2 1 1 /model/layers.20/self_attn/Shape_2_output_0 /model/layers.20/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.20/self_attn/Gather_3 1 1 /model/layers.20/self_attn/Shape_3_output_0 /model/layers.20/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.20/self_attn/Shape_9 1 1 present.20.value_splitncnn_4 /model/layers.20/self_attn/Shape_9_output_0 Shape /model/layers.20/self_attn/Shape_10 1 1 present.20.value_splitncnn_3 /model/layers.20/self_attn/Shape_10_output_0 Shape /model/layers.20/self_attn/Shape_11 1 1 present.20.value_splitncnn_2 /model/layers.20/self_attn/Shape_11_output_0 Shape /model/layers.20/self_attn/Shape_12 1 1 present.20.value_splitncnn_1 /model/layers.20/self_attn/Shape_12_output_0 ExpandDims /model/layers.20/self_attn/Unsqueeze_19 2 1 present.20.value_splitncnn_0 /model/layers.20/self_attn/Constant_45_output_0 /model/layers.20/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.20/self_attn/Div 2 1 /model/layers.20/self_attn/Gather_2_output_0 /model/layers.20/self_attn/Constant_9_output_0 /model/layers.20/self_attn/Div_output_0 0=3 BinaryOp /model/layers.20/self_attn/Div_1 2 1 /model/layers.20/self_attn/Gather_3_output_0 /model/layers.20/self_attn/Constant_19_output_0 /model/layers.20/self_attn/Div_1_output_0 0=3 Crop /model/layers.20/self_attn/Gather_8 1 1 /model/layers.20/self_attn/Shape_9_output_0 /model/layers.20/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_649 1 2 /model/layers.20/self_attn/Gather_8_output_0 /model/layers.20/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.20/self_attn/Gather_9 1 1 /model/layers.20/self_attn/Shape_10_output_0 /model/layers.20/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_650 1 2 /model/layers.20/self_attn/Gather_9_output_0 /model/layers.20/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.20/self_attn/Gather_10 1 1 /model/layers.20/self_attn/Shape_11_output_0 /model/layers.20/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_651 1 2 /model/layers.20/self_attn/Gather_10_output_0 /model/layers.20/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.20/self_attn/Gather_11 1 1 /model/layers.20/self_attn/Shape_12_output_0 /model/layers.20/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_652 1 2 /model/layers.20/self_attn/Gather_11_output_0 /model/layers.20/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.20/self_attn/Cast 1 1 /model/layers.20/self_attn/Div_output_0 /model/layers.20/self_attn/Cast_output_0 Cast /model/layers.20/self_attn/Cast_2 1 1 /model/layers.20/self_attn/Div_1_output_0 /model/layers.20/self_attn/Cast_2_output_0 ExpandDims /model/layers.20/self_attn/Unsqueeze_20 2 1 /model/layers.20/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_46_output_0 /model/layers.20/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_24 2 1 /model/layers.20/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_6260 /model/layers.20/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_21 2 1 /model/layers.20/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_47_output_0 /model/layers.20/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.20/self_attn/Mul_7 2 1 /model/layers.20/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.20/self_attn/Constant_53_output_0 /model/layers.20/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.20/self_attn/Unsqueeze_22 2 1 /model/layers.20/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_49_output_0 /model/layers.20/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_26 2 1 /model/layers.20/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_6264 /model/layers.20/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_23 2 1 /model/layers.20/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_50_output_0 /model/layers.20/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_27 2 1 /model/layers.20/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_6266 /model/layers.20/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.20/self_attn/Cast_1 1 1 /model/layers.20/self_attn/Cast_output_0 /model/layers.20/self_attn/Cast_1_output_0 Split splitncnn_653 1 2 /model/layers.20/self_attn/Cast_1_output_0 /model/layers.20/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.20/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.20/self_attn/Cast_3 1 1 /model/layers.20/self_attn/Cast_2_output_0 /model/layers.20/self_attn/Cast_3_output_0 Split splitncnn_654 1 2 /model/layers.20/self_attn/Cast_3_output_0 /model/layers.20/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.20/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.20/self_attn/Unsqueeze_25 2 1 /model/layers.20/self_attn/Mul_7_output_0 onnx::Unsqueeze_6262 /model/layers.20/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.20/self_attn/Concat_9 5 1 /model/layers.20/self_attn/Unsqueeze_20_output_0 /model/layers.20/self_attn/Unsqueeze_21_output_0 /model/layers.20/self_attn/Constant_48_output_0 /model/layers.20/self_attn/Unsqueeze_22_output_0 /model/layers.20/self_attn/Unsqueeze_23_output_0 /model/layers.20/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_6 2 1 /model/layers.20/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_12_output_0 /model/layers.20/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_7 2 1 /model/layers.20/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.20/self_attn/Constant_15_output_0 /model/layers.20/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_8 2 1 /model/layers.20/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_22_output_0 /model/layers.20/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_9 2 1 /model/layers.20/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.20/self_attn/Constant_25_output_0 /model/layers.20/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.20/self_attn/Concat_10 4 1 /model/layers.20/self_attn/Unsqueeze_24_output_0 /model/layers.20/self_attn/Unsqueeze_25_output_0 /model/layers.20/self_attn/Unsqueeze_26_output_0 /model/layers.20/self_attn/Unsqueeze_27_output_0 /model/layers.20/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.20/self_attn/Reshape_5 1 1 /model/layers.20/self_attn/Concat_9_output_0 /model/layers.20/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_655 1 3 /model/layers.20/self_attn/Reshape_5_output_0 /model/layers.20/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.20/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.20/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.20/self_attn/Slice 2 1 /model/layers.20/self_attn/Transpose_output_0_splitncnn_1 /model/layers.20/self_attn/Unsqueeze_6_output_0 /model/layers.20/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.20/self_attn/Slice_1 2 1 /model/layers.20/self_attn/Transpose_output_0_splitncnn_0 /model/layers.20/self_attn/Unsqueeze_7_output_0 /model/layers.20/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.20/self_attn/Slice_2 2 1 /model/layers.20/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.20/self_attn/Unsqueeze_8_output_0 /model/layers.20/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.20/self_attn/Slice_3 2 1 /model/layers.20/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.20/self_attn/Unsqueeze_9_output_0 /model/layers.20/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.20/self_attn/Shape_13 1 1 /model/layers.20/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.20/self_attn/Shape_13_output_0 UnaryOp /model/layers.20/self_attn/Neg 1 1 /model/layers.20/self_attn/Slice_1_output_0 /model/layers.20/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.20/self_attn/Neg_1 1 1 /model/layers.20/self_attn/Slice_3_output_0 /model/layers.20/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.20/self_attn/ConstantOfShape_1 1 1 /model/layers.20/self_attn/Shape_13_output_0 /model/layers.20/self_attn/ConstantOfShape_1_output_0 Split splitncnn_656 1 2 /model/layers.20/self_attn/ConstantOfShape_1_output_0 /model/layers.20/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.20/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.20/self_attn/Concat_3 2 1 /model/layers.20/self_attn/Neg_output_0 /model/layers.20/self_attn/Slice_output_0 /model/layers.20/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.20/self_attn/Concat_4 2 1 /model/layers.20/self_attn/Neg_1_output_0 /model/layers.20/self_attn/Slice_2_output_0 /model/layers.20/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.20/self_attn/Mul_6 2 1 /model/layers.20/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_52_output_0 /model/layers.20/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.20/self_attn/Mul_1 2 1 /model/layers.20/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_7 /model/layers.20/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.20/self_attn/Mul_3 2 1 /model/layers.20/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_6 /model/layers.20/self_attn/Mul_3_output_0 0=2 Equal /model/layers.20/self_attn/Equal_1 2 1 /model/layers.20/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.20/self_attn/Mul_6_output_0 /model/layers.20/self_attn/Equal_1_output_0 BinaryOp /model/layers.20/self_attn/Add 2 1 /model/layers.20/self_attn/Mul_output_0 /model/layers.20/self_attn/Mul_1_output_0 /model/layers.20/self_attn/Add_output_0 0=0 BinaryOp /model/layers.20/self_attn/Add_1 2 1 /model/layers.20/self_attn/Mul_2_output_0 /model/layers.20/self_attn/Mul_3_output_0 /model/layers.20/self_attn/Add_1_output_0 0=0 Where /model/layers.20/self_attn/Where_1 3 1 /model/layers.20/self_attn/Equal_1_output_0 /model/layers.20/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.20/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.20/self_attn/Where_1_output_0 BinaryOp /model/layers.20/self_attn/Mul_8 1 1 /model/layers.20/self_attn/Add_output_0 /model/layers.20/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.20/self_attn/Concat_5 2 1 past_key_values.20.key /model/layers.20/self_attn/Add_1_output_0 present.20.key 0=-2 Split splitncnn_657 1 5 present.20.key present.20.key_splitncnn_0 present.20.key_splitncnn_1 present.20.key_splitncnn_2 present.20.key_splitncnn_3 present.20.key_splitncnn_4 Expand /model/layers.20/self_attn/Expand_1 2 1 /model/layers.20/self_attn/Unsqueeze_19_output_0 /model/layers.20/self_attn/Where_1_output_0 /model/layers.20/self_attn/Expand_1_output_0 Shape /model/layers.20/self_attn/Shape_4 1 1 present.20.key_splitncnn_4 /model/layers.20/self_attn/Shape_4_output_0 Shape /model/layers.20/self_attn/Shape_5 1 1 present.20.key_splitncnn_3 /model/layers.20/self_attn/Shape_5_output_0 Shape /model/layers.20/self_attn/Shape_6 1 1 present.20.key_splitncnn_2 /model/layers.20/self_attn/Shape_6_output_0 Shape /model/layers.20/self_attn/Shape_7 1 1 present.20.key_splitncnn_1 /model/layers.20/self_attn/Shape_7_output_0 ExpandDims /model/layers.20/self_attn/Unsqueeze_10 2 1 present.20.key_splitncnn_0 /model/layers.20/self_attn/Constant_32_output_0 /model/layers.20/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.20/self_attn/Reshape_6 2 1 /model/layers.20/self_attn/Expand_1_output_0 /model/layers.20/self_attn/Concat_10_output_0 /model/layers.20/self_attn/Reshape_6_output_0 Crop /model/layers.20/self_attn/Gather_4 1 1 /model/layers.20/self_attn/Shape_4_output_0 /model/layers.20/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_658 1 2 /model/layers.20/self_attn/Gather_4_output_0 /model/layers.20/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.20/self_attn/Gather_5 1 1 /model/layers.20/self_attn/Shape_5_output_0 /model/layers.20/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_659 1 2 /model/layers.20/self_attn/Gather_5_output_0 /model/layers.20/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.20/self_attn/Gather_6 1 1 /model/layers.20/self_attn/Shape_6_output_0 /model/layers.20/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_660 1 2 /model/layers.20/self_attn/Gather_6_output_0 /model/layers.20/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.20/self_attn/Gather_7 1 1 /model/layers.20/self_attn/Shape_7_output_0 /model/layers.20/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_661 1 2 /model/layers.20/self_attn/Gather_7_output_0 /model/layers.20/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.20/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.20/self_attn/Unsqueeze_11 2 1 /model/layers.20/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_33_output_0 /model/layers.20/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_15 2 1 /model/layers.20/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_6214 /model/layers.20/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_12 2 1 /model/layers.20/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_34_output_0 /model/layers.20/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.20/self_attn/Mul_5 2 1 /model/layers.20/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.20/self_attn/Constant_40_output_0 /model/layers.20/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.20/self_attn/Unsqueeze_13 2 1 /model/layers.20/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_36_output_0 /model/layers.20/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_17 2 1 /model/layers.20/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_6218 /model/layers.20/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_14 2 1 /model/layers.20/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_37_output_0 /model/layers.20/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_18 2 1 /model/layers.20/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_6220 /model/layers.20/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.20/self_attn/Unsqueeze_16 2 1 /model/layers.20/self_attn/Mul_5_output_0 onnx::Unsqueeze_6216 /model/layers.20/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.20/self_attn/Concat_7 5 1 /model/layers.20/self_attn/Unsqueeze_11_output_0 /model/layers.20/self_attn/Unsqueeze_12_output_0 /model/layers.20/self_attn/Constant_35_output_0 /model/layers.20/self_attn/Unsqueeze_13_output_0 /model/layers.20/self_attn/Unsqueeze_14_output_0 /model/layers.20/self_attn/Concat_7_output_0 0=0 Concat /model/layers.20/self_attn/Concat_8 4 1 /model/layers.20/self_attn/Unsqueeze_15_output_0 /model/layers.20/self_attn/Unsqueeze_16_output_0 /model/layers.20/self_attn/Unsqueeze_17_output_0 /model/layers.20/self_attn/Unsqueeze_18_output_0 /model/layers.20/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.20/self_attn/Reshape_3 1 1 /model/layers.20/self_attn/Concat_7_output_0 /model/layers.20/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_662 1 3 /model/layers.20/self_attn/Reshape_3_output_0 /model/layers.20/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.20/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.20/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.20/self_attn/Shape_8 1 1 /model/layers.20/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.20/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.20/self_attn/ConstantOfShape 1 1 /model/layers.20/self_attn/Shape_8_output_0 /model/layers.20/self_attn/ConstantOfShape_output_0 Split splitncnn_663 1 2 /model/layers.20/self_attn/ConstantOfShape_output_0 /model/layers.20/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.20/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.20/self_attn/Mul_4 2 1 /model/layers.20/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.20/self_attn/Constant_39_output_0 /model/layers.20/self_attn/Mul_4_output_0 0=2 Equal /model/layers.20/self_attn/Equal 2 1 /model/layers.20/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.20/self_attn/Mul_4_output_0 /model/layers.20/self_attn/Equal_output_0 Where /model/layers.20/self_attn/Where 3 1 /model/layers.20/self_attn/Equal_output_0 /model/layers.20/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.20/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.20/self_attn/Where_output_0 Expand /model/layers.20/self_attn/Expand 2 1 /model/layers.20/self_attn/Unsqueeze_10_output_0 /model/layers.20/self_attn/Where_output_0 /model/layers.20/self_attn/Expand_output_0 Reshape /model/layers.20/self_attn/Reshape_4 2 1 /model/layers.20/self_attn/Expand_output_0 /model/layers.20/self_attn/Concat_8_output_0 /model/layers.20/self_attn/Reshape_4_output_0 Split splitncnn_664 1 2 /model/layers.20/self_attn/Reshape_4_output_0 /model/layers.20/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.20/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.20/self_attn/Shape_14 1 1 /model/layers.20/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.20/self_attn/Shape_14_output_0 Permute /model/layers.20/self_attn/Transpose_3 1 1 /model/layers.20/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.20/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.20/self_attn/Gather_12 1 1 /model/layers.20/self_attn/Shape_14_output_0 /model/layers.20/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.20/self_attn/Mul_9 1 1 /model/layers.20/self_attn/Transpose_3_output_0 /model/layers.20/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.20/self_attn/Unsqueeze_28 2 1 /model/layers.20/self_attn/Gather_12_output_0 /model/layers.20/self_attn/Constant_57_output_0 /model/layers.20/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.20/self_attn/MatMul 2 1 /model/layers.20/self_attn/Mul_8_output_0 /model/layers.20/self_attn/Mul_9_output_0 /model/layers.20/self_attn/MatMul_output_0 Crop /model/layers.20/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_3 /model/layers.20/self_attn/Unsqueeze_28_output_0 /model/layers.20/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.20/self_attn/Where_2 3 1 /model/layers.20/self_attn/Slice_4_output_0 /model/layers.20/self_attn/Constant_61_output_0 /model/layers.20/self_attn/Constant_62_output_0 /model/layers.20/self_attn/Where_2_output_0 BinaryOp /model/layers.20/self_attn/Add_2 2 1 /model/layers.20/self_attn/MatMul_output_0 /model/layers.20/self_attn/Where_2_output_0 /model/layers.20/self_attn/Add_2_output_0 0=0 Softmax /model/layers.20/self_attn/Softmax 1 1 /model/layers.20/self_attn/Add_2_output_0 /model/layers.20/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_665 1 2 /model/layers.20/self_attn/Softmax_output_0 /model/layers.20/self_attn/Softmax_output_0_splitncnn_0 /model/layers.20/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.20/self_attn/IsNaN 1 1 /model/layers.20/self_attn/Softmax_output_0_splitncnn_1 /model/layers.20/self_attn/IsNaN_output_0 Where /model/layers.20/self_attn/Where_3 3 1 /model/layers.20/self_attn/IsNaN_output_0 /model/layers.20/self_attn/Constant_63_output_0 /model/layers.20/self_attn/Softmax_output_0_splitncnn_0 /model/layers.20/self_attn/Where_3_output_0 Gemm /model/layers.20/self_attn/MatMul_1 2 1 /model/layers.20/self_attn/Where_3_output_0 /model/layers.20/self_attn/Reshape_6_output_0 /model/layers.20/self_attn/MatMul_1_output_0 Permute /model/layers.20/self_attn/Transpose_4 1 1 /model/layers.20/self_attn/MatMul_1_output_0 /model/layers.20/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.20/self_attn/Reshape_7 2 1 /model/layers.20/self_attn/Transpose_4_output_0 /model/layers.20/self_attn/Concat_11_output_0 /model/layers.20/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.20/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.20/self_attn/Reshape_7_output_0 /model/layers.20/self_attn/Reshape_7_output_0_quantized /model/layers.20/self_attn/Reshape_7_output_0_scale /model/layers.20/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.20/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.20/self_attn/Reshape_7_output_0_scale /model/layers.20/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.014272e-03 MatMulInteger /model/layers.20/self_attn/o_proj/MatMul_quant 4 1 /model/layers.20/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7921_quantized /model/layers.20/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7921_zero_point /model/layers.20/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.20/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.20/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.20/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.20/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.20/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.20/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.20/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.20/Add 2 1 /model/layers.20/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.20/self_attn/o_proj/MatMul_output_0 /model/layers.20/Add_output_0 0=0 Cast /model/layers.20/post_attention_layernorm/Cast 1 1 /model/layers.20/Add_output_0 /model/layers.20/post_attention_layernorm/Cast_output_0 Split splitncnn_666 1 3 /model/layers.20/post_attention_layernorm/Cast_output_0 /model/layers.20/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.20/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.20/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.20/post_attention_layernorm/Pow 1 1 /model/layers.20/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.20/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.20/post_attention_layernorm/ReduceMean 2 1 /model/layers.20/post_attention_layernorm/Pow_output_0 /model/layers.20/post_attention_layernorm/Constant_1_output_0 /model/layers.20/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.20/post_attention_layernorm/Add 1 1 /model/layers.20/post_attention_layernorm/ReduceMean_output_0 /model/layers.20/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.20/post_attention_layernorm/Sqrt 1 1 /model/layers.20/post_attention_layernorm/Add_output_0 /model/layers.20/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.20/post_attention_layernorm/Div 1 1 /model/layers.20/post_attention_layernorm/Sqrt_output_0 /model/layers.20/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.20/post_attention_layernorm/Mul 2 1 /model/layers.20/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.20/post_attention_layernorm/Div_output_0 /model/layers.20/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.20/post_attention_layernorm/Cast_1 1 1 /model/layers.20/post_attention_layernorm/Mul_output_0 /model/layers.20/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.20/post_attention_layernorm/Mul_1 2 1 model.layers.20.post_attention_layernorm.weight /model/layers.20/post_attention_layernorm/Cast_1_output_0 /model/layers.20/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.20/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.20/post_attention_layernorm/Mul_1_output_0 /model/layers.20/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.20/post_attention_layernorm/Mul_1_output_0_scale /model/layers.20/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_667 1 2 /model/layers.20/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.20/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.20/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_668 1 2 /model/layers.20/post_attention_layernorm/Mul_1_output_0_scale /model/layers.20/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.20/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_669 1 2 /model/layers.20/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.20/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.20/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.20/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.20/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.20/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.244586e-03 BinaryOp /model/layers.20/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.20/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.20/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=3.337229e-03 MatMulInteger /model/layers.20/mlp/gate_proj/MatMul_quant 4 1 /model/layers.20/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7922_quantized /model/layers.20/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7922_zero_point /model/layers.20/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.20/mlp/up_proj/MatMul_quant 4 1 /model/layers.20/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7923_quantized /model/layers.20/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7923_zero_point /model/layers.20/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.20/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.20/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.20/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.20/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.20/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.20/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.20/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.20/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.20/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.20/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.20/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.20/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.20/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.20/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.20/mlp/act_fn/Mul 1 1 /model/layers.20/mlp/gate_proj/MatMul_output_0 /model/layers.20/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.20/mlp/Mul 2 1 /model/layers.20/mlp/act_fn/Mul_output_0 /model/layers.20/mlp/up_proj/MatMul_output_0 /model/layers.20/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.20/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.20/mlp/Mul_output_0 /model/layers.20/mlp/Mul_output_0_quantized /model/layers.20/mlp/Mul_output_0_scale /model/layers.20/mlp/Mul_output_0_zero_point BinaryOp /model/layers.20/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.20/mlp/Mul_output_0_scale /model/layers.20/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.845103e-03 MatMulInteger /model/layers.20/mlp/down_proj/MatMul_quant 4 1 /model/layers.20/mlp/Mul_output_0_quantized onnx::MatMul_7924_quantized /model/layers.20/mlp/Mul_output_0_zero_point onnx::MatMul_7924_zero_point /model/layers.20/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.20/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.20/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.20/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.20/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.20/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.20/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.20/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.20/Add_1 2 1 /model/layers.20/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.20/mlp/down_proj/MatMul_output_0 /model/layers.20/Add_1_output_0 0=0 Cast /model/layers.21/input_layernorm/Cast 1 1 /model/layers.20/Add_1_output_0 /model/layers.21/input_layernorm/Cast_output_0 Split splitncnn_670 1 3 /model/layers.21/input_layernorm/Cast_output_0 /model/layers.21/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.21/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.21/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.21/input_layernorm/Pow 1 1 /model/layers.21/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.21/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.21/input_layernorm/ReduceMean 2 1 /model/layers.21/input_layernorm/Pow_output_0 /model/layers.21/input_layernorm/Constant_1_output_0 /model/layers.21/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.21/input_layernorm/Add 1 1 /model/layers.21/input_layernorm/ReduceMean_output_0 /model/layers.21/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.21/input_layernorm/Sqrt 1 1 /model/layers.21/input_layernorm/Add_output_0 /model/layers.21/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.21/input_layernorm/Div 1 1 /model/layers.21/input_layernorm/Sqrt_output_0 /model/layers.21/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.21/input_layernorm/Mul 2 1 /model/layers.21/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.21/input_layernorm/Div_output_0 /model/layers.21/input_layernorm/Mul_output_0 0=2 Cast /model/layers.21/input_layernorm/Cast_1 1 1 /model/layers.21/input_layernorm/Mul_output_0 /model/layers.21/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.21/input_layernorm/Mul_1 2 1 model.layers.21.input_layernorm.weight /model/layers.21/input_layernorm/Cast_1_output_0 /model/layers.21/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_671 1 3 /model/layers.21/input_layernorm/Mul_1_output_0 /model/layers.21/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.21/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.21/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.21/self_attn/Shape 1 1 /model/layers.21/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.21/self_attn/Shape_output_0 Shape /model/layers.21/self_attn/Shape_1 1 1 /model/layers.21/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.21/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.21/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.21/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.21/input_layernorm/Mul_1_output_0_quantized /model/layers.21/input_layernorm/Mul_1_output_0_scale /model/layers.21/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_672 1 3 /model/layers.21/input_layernorm/Mul_1_output_0_quantized /model/layers.21/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.21/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.21/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_673 1 3 /model/layers.21/input_layernorm/Mul_1_output_0_scale /model/layers.21/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.21/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.21/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_674 1 3 /model/layers.21/input_layernorm/Mul_1_output_0_zero_point /model/layers.21/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.21/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.21/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.21/self_attn/Gather 1 1 /model/layers.21/self_attn/Shape_output_0 /model/layers.21/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_675 1 4 /model/layers.21/self_attn/Gather_output_0 /model/layers.21/self_attn/Gather_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_output_0_splitncnn_1 /model/layers.21/self_attn/Gather_output_0_splitncnn_2 /model/layers.21/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.21/self_attn/Gather_1 1 1 /model/layers.21/self_attn/Shape_1_output_0 /model/layers.21/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_676 1 4 /model/layers.21/self_attn/Gather_1_output_0 /model/layers.21/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.21/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.21/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.21/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.21/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.21/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.807025e-03 BinaryOp /model/layers.21/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.21/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.21/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.460999e-03 BinaryOp /model/layers.21/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.21/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.21/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.352977e-03 MatMulInteger /model/layers.21/self_attn/q_proj/MatMul_quant 4 1 /model/layers.21/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7925_quantized /model/layers.21/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7925_zero_point /model/layers.21/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.21/self_attn/k_proj/MatMul_quant 4 1 /model/layers.21/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7932_quantized /model/layers.21/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7932_zero_point /model/layers.21/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.21/self_attn/v_proj/MatMul_quant 4 1 /model/layers.21/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7933_quantized /model/layers.21/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7933_zero_point /model/layers.21/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.21/self_attn/Unsqueeze 2 1 /model/layers.21/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_6356 /model/layers.21/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_2 2 1 /model/layers.21/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_6365 /model/layers.21/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_4 2 1 /model/layers.21/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_6374 /model/layers.21/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_29 2 1 /model/layers.21/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_6579 /model/layers.21/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_1 2 1 /model/layers.21/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_6358 /model/layers.21/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_3 2 1 /model/layers.21/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_6367 /model/layers.21/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_5 2 1 /model/layers.21/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_6376 /model/layers.21/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_30 2 1 /model/layers.21/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_6581 /model/layers.21/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.21/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.21/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.21/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.21/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.21/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.21/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.21/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.21/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.21/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.21/self_attn/Concat 4 1 /model/layers.21/self_attn/Unsqueeze_output_0 /model/layers.21/self_attn/Unsqueeze_1_output_0 /model/layers.21/self_attn/Constant_2_output_0 /model/layers.21/self_attn/Constant_3_output_0 /model/layers.21/self_attn/Concat_output_0 0=0 Concat /model/layers.21/self_attn/Concat_1 4 1 /model/layers.21/self_attn/Unsqueeze_2_output_0 /model/layers.21/self_attn/Unsqueeze_3_output_0 /model/layers.21/self_attn/Constant_4_output_0 /model/layers.21/self_attn/Constant_5_output_0 /model/layers.21/self_attn/Concat_1_output_0 0=0 Concat /model/layers.21/self_attn/Concat_2 4 1 /model/layers.21/self_attn/Unsqueeze_4_output_0 /model/layers.21/self_attn/Unsqueeze_5_output_0 /model/layers.21/self_attn/Constant_6_output_0 /model/layers.21/self_attn/Constant_7_output_0 /model/layers.21/self_attn/Concat_2_output_0 0=0 Concat /model/layers.21/self_attn/Concat_11 3 1 /model/layers.21/self_attn/Unsqueeze_29_output_0 /model/layers.21/self_attn/Unsqueeze_30_output_0 /model/layers.21/self_attn/Constant_64_output_0 /model/layers.21/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.21/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.21/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.21/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.21/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.21/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.21/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.21/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.21/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.21/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.21/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.21/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.21/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.21/self_attn/q_proj/Add 2 1 model.layers.21.self_attn.q_proj.bias /model/layers.21/self_attn/q_proj/MatMul_output_0 /model/layers.21/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.21/self_attn/k_proj/Add 2 1 model.layers.21.self_attn.k_proj.bias /model/layers.21/self_attn/k_proj/MatMul_output_0 /model/layers.21/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.21/self_attn/v_proj/Add 2 1 model.layers.21.self_attn.v_proj.bias /model/layers.21/self_attn/v_proj/MatMul_output_0 /model/layers.21/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.21/self_attn/Reshape 2 1 /model/layers.21/self_attn/q_proj/Add_output_0 /model/layers.21/self_attn/Concat_output_0 /model/layers.21/self_attn/Reshape_output_0 Reshape /model/layers.21/self_attn/Reshape_1 2 1 /model/layers.21/self_attn/k_proj/Add_output_0 /model/layers.21/self_attn/Concat_1_output_0 /model/layers.21/self_attn/Reshape_1_output_0 Reshape /model/layers.21/self_attn/Reshape_2 2 1 /model/layers.21/self_attn/v_proj/Add_output_0 /model/layers.21/self_attn/Concat_2_output_0 /model/layers.21/self_attn/Reshape_2_output_0 Permute /model/layers.21/self_attn/Transpose 1 1 /model/layers.21/self_attn/Reshape_output_0 /model/layers.21/self_attn/Transpose_output_0 0=2 Split splitncnn_677 1 4 /model/layers.21/self_attn/Transpose_output_0 /model/layers.21/self_attn/Transpose_output_0_splitncnn_0 /model/layers.21/self_attn/Transpose_output_0_splitncnn_1 /model/layers.21/self_attn/Transpose_output_0_splitncnn_2 /model/layers.21/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.21/self_attn/Transpose_1 1 1 /model/layers.21/self_attn/Reshape_1_output_0 /model/layers.21/self_attn/Transpose_1_output_0 0=2 Split splitncnn_678 1 4 /model/layers.21/self_attn/Transpose_1_output_0 /model/layers.21/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.21/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.21/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.21/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.21/self_attn/Transpose_2 1 1 /model/layers.21/self_attn/Reshape_2_output_0 /model/layers.21/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.21/self_attn/Mul 2 1 /model/layers.21/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_5 /model/layers.21/self_attn/Mul_output_0 0=2 Shape /model/layers.21/self_attn/Shape_2 1 1 /model/layers.21/self_attn/Transpose_output_0_splitncnn_2 /model/layers.21/self_attn/Shape_2_output_0 BinaryOp /model/layers.21/self_attn/Mul_2 2 1 /model/layers.21/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_4 /model/layers.21/self_attn/Mul_2_output_0 0=2 Shape /model/layers.21/self_attn/Shape_3 1 1 /model/layers.21/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.21/self_attn/Shape_3_output_0 Concat /model/layers.21/self_attn/Concat_6 2 1 past_key_values.21.value /model/layers.21/self_attn/Transpose_2_output_0 present.21.value 0=-2 Split splitncnn_679 1 5 present.21.value present.21.value_splitncnn_0 present.21.value_splitncnn_1 present.21.value_splitncnn_2 present.21.value_splitncnn_3 present.21.value_splitncnn_4 Crop /model/layers.21/self_attn/Gather_2 1 1 /model/layers.21/self_attn/Shape_2_output_0 /model/layers.21/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.21/self_attn/Gather_3 1 1 /model/layers.21/self_attn/Shape_3_output_0 /model/layers.21/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.21/self_attn/Shape_9 1 1 present.21.value_splitncnn_4 /model/layers.21/self_attn/Shape_9_output_0 Shape /model/layers.21/self_attn/Shape_10 1 1 present.21.value_splitncnn_3 /model/layers.21/self_attn/Shape_10_output_0 Shape /model/layers.21/self_attn/Shape_11 1 1 present.21.value_splitncnn_2 /model/layers.21/self_attn/Shape_11_output_0 Shape /model/layers.21/self_attn/Shape_12 1 1 present.21.value_splitncnn_1 /model/layers.21/self_attn/Shape_12_output_0 ExpandDims /model/layers.21/self_attn/Unsqueeze_19 2 1 present.21.value_splitncnn_0 /model/layers.21/self_attn/Constant_45_output_0 /model/layers.21/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.21/self_attn/Div 2 1 /model/layers.21/self_attn/Gather_2_output_0 /model/layers.21/self_attn/Constant_9_output_0 /model/layers.21/self_attn/Div_output_0 0=3 BinaryOp /model/layers.21/self_attn/Div_1 2 1 /model/layers.21/self_attn/Gather_3_output_0 /model/layers.21/self_attn/Constant_19_output_0 /model/layers.21/self_attn/Div_1_output_0 0=3 Crop /model/layers.21/self_attn/Gather_8 1 1 /model/layers.21/self_attn/Shape_9_output_0 /model/layers.21/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_680 1 2 /model/layers.21/self_attn/Gather_8_output_0 /model/layers.21/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.21/self_attn/Gather_9 1 1 /model/layers.21/self_attn/Shape_10_output_0 /model/layers.21/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_681 1 2 /model/layers.21/self_attn/Gather_9_output_0 /model/layers.21/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.21/self_attn/Gather_10 1 1 /model/layers.21/self_attn/Shape_11_output_0 /model/layers.21/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_682 1 2 /model/layers.21/self_attn/Gather_10_output_0 /model/layers.21/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.21/self_attn/Gather_11 1 1 /model/layers.21/self_attn/Shape_12_output_0 /model/layers.21/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_683 1 2 /model/layers.21/self_attn/Gather_11_output_0 /model/layers.21/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.21/self_attn/Cast 1 1 /model/layers.21/self_attn/Div_output_0 /model/layers.21/self_attn/Cast_output_0 Cast /model/layers.21/self_attn/Cast_2 1 1 /model/layers.21/self_attn/Div_1_output_0 /model/layers.21/self_attn/Cast_2_output_0 ExpandDims /model/layers.21/self_attn/Unsqueeze_20 2 1 /model/layers.21/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_46_output_0 /model/layers.21/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_24 2 1 /model/layers.21/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_6539 /model/layers.21/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_21 2 1 /model/layers.21/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_47_output_0 /model/layers.21/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.21/self_attn/Mul_7 2 1 /model/layers.21/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.21/self_attn/Constant_53_output_0 /model/layers.21/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.21/self_attn/Unsqueeze_22 2 1 /model/layers.21/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_49_output_0 /model/layers.21/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_26 2 1 /model/layers.21/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_6543 /model/layers.21/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_23 2 1 /model/layers.21/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_50_output_0 /model/layers.21/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_27 2 1 /model/layers.21/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_6545 /model/layers.21/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.21/self_attn/Cast_1 1 1 /model/layers.21/self_attn/Cast_output_0 /model/layers.21/self_attn/Cast_1_output_0 Split splitncnn_684 1 2 /model/layers.21/self_attn/Cast_1_output_0 /model/layers.21/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.21/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.21/self_attn/Cast_3 1 1 /model/layers.21/self_attn/Cast_2_output_0 /model/layers.21/self_attn/Cast_3_output_0 Split splitncnn_685 1 2 /model/layers.21/self_attn/Cast_3_output_0 /model/layers.21/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.21/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.21/self_attn/Unsqueeze_25 2 1 /model/layers.21/self_attn/Mul_7_output_0 onnx::Unsqueeze_6541 /model/layers.21/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.21/self_attn/Concat_9 5 1 /model/layers.21/self_attn/Unsqueeze_20_output_0 /model/layers.21/self_attn/Unsqueeze_21_output_0 /model/layers.21/self_attn/Constant_48_output_0 /model/layers.21/self_attn/Unsqueeze_22_output_0 /model/layers.21/self_attn/Unsqueeze_23_output_0 /model/layers.21/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_6 2 1 /model/layers.21/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_12_output_0 /model/layers.21/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_7 2 1 /model/layers.21/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.21/self_attn/Constant_15_output_0 /model/layers.21/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_8 2 1 /model/layers.21/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_22_output_0 /model/layers.21/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_9 2 1 /model/layers.21/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.21/self_attn/Constant_25_output_0 /model/layers.21/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.21/self_attn/Concat_10 4 1 /model/layers.21/self_attn/Unsqueeze_24_output_0 /model/layers.21/self_attn/Unsqueeze_25_output_0 /model/layers.21/self_attn/Unsqueeze_26_output_0 /model/layers.21/self_attn/Unsqueeze_27_output_0 /model/layers.21/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.21/self_attn/Reshape_5 1 1 /model/layers.21/self_attn/Concat_9_output_0 /model/layers.21/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_686 1 3 /model/layers.21/self_attn/Reshape_5_output_0 /model/layers.21/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.21/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.21/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.21/self_attn/Slice 2 1 /model/layers.21/self_attn/Transpose_output_0_splitncnn_1 /model/layers.21/self_attn/Unsqueeze_6_output_0 /model/layers.21/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.21/self_attn/Slice_1 2 1 /model/layers.21/self_attn/Transpose_output_0_splitncnn_0 /model/layers.21/self_attn/Unsqueeze_7_output_0 /model/layers.21/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.21/self_attn/Slice_2 2 1 /model/layers.21/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.21/self_attn/Unsqueeze_8_output_0 /model/layers.21/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.21/self_attn/Slice_3 2 1 /model/layers.21/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.21/self_attn/Unsqueeze_9_output_0 /model/layers.21/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.21/self_attn/Shape_13 1 1 /model/layers.21/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.21/self_attn/Shape_13_output_0 UnaryOp /model/layers.21/self_attn/Neg 1 1 /model/layers.21/self_attn/Slice_1_output_0 /model/layers.21/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.21/self_attn/Neg_1 1 1 /model/layers.21/self_attn/Slice_3_output_0 /model/layers.21/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.21/self_attn/ConstantOfShape_1 1 1 /model/layers.21/self_attn/Shape_13_output_0 /model/layers.21/self_attn/ConstantOfShape_1_output_0 Split splitncnn_687 1 2 /model/layers.21/self_attn/ConstantOfShape_1_output_0 /model/layers.21/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.21/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.21/self_attn/Concat_3 2 1 /model/layers.21/self_attn/Neg_output_0 /model/layers.21/self_attn/Slice_output_0 /model/layers.21/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.21/self_attn/Concat_4 2 1 /model/layers.21/self_attn/Neg_1_output_0 /model/layers.21/self_attn/Slice_2_output_0 /model/layers.21/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.21/self_attn/Mul_6 2 1 /model/layers.21/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_52_output_0 /model/layers.21/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.21/self_attn/Mul_1 2 1 /model/layers.21/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_5 /model/layers.21/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.21/self_attn/Mul_3 2 1 /model/layers.21/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_4 /model/layers.21/self_attn/Mul_3_output_0 0=2 Equal /model/layers.21/self_attn/Equal_1 2 1 /model/layers.21/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.21/self_attn/Mul_6_output_0 /model/layers.21/self_attn/Equal_1_output_0 BinaryOp /model/layers.21/self_attn/Add 2 1 /model/layers.21/self_attn/Mul_output_0 /model/layers.21/self_attn/Mul_1_output_0 /model/layers.21/self_attn/Add_output_0 0=0 BinaryOp /model/layers.21/self_attn/Add_1 2 1 /model/layers.21/self_attn/Mul_2_output_0 /model/layers.21/self_attn/Mul_3_output_0 /model/layers.21/self_attn/Add_1_output_0 0=0 Where /model/layers.21/self_attn/Where_1 3 1 /model/layers.21/self_attn/Equal_1_output_0 /model/layers.21/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.21/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.21/self_attn/Where_1_output_0 BinaryOp /model/layers.21/self_attn/Mul_8 1 1 /model/layers.21/self_attn/Add_output_0 /model/layers.21/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.21/self_attn/Concat_5 2 1 past_key_values.21.key /model/layers.21/self_attn/Add_1_output_0 present.21.key 0=-2 Split splitncnn_688 1 5 present.21.key present.21.key_splitncnn_0 present.21.key_splitncnn_1 present.21.key_splitncnn_2 present.21.key_splitncnn_3 present.21.key_splitncnn_4 Expand /model/layers.21/self_attn/Expand_1 2 1 /model/layers.21/self_attn/Unsqueeze_19_output_0 /model/layers.21/self_attn/Where_1_output_0 /model/layers.21/self_attn/Expand_1_output_0 Shape /model/layers.21/self_attn/Shape_4 1 1 present.21.key_splitncnn_4 /model/layers.21/self_attn/Shape_4_output_0 Shape /model/layers.21/self_attn/Shape_5 1 1 present.21.key_splitncnn_3 /model/layers.21/self_attn/Shape_5_output_0 Shape /model/layers.21/self_attn/Shape_6 1 1 present.21.key_splitncnn_2 /model/layers.21/self_attn/Shape_6_output_0 Shape /model/layers.21/self_attn/Shape_7 1 1 present.21.key_splitncnn_1 /model/layers.21/self_attn/Shape_7_output_0 ExpandDims /model/layers.21/self_attn/Unsqueeze_10 2 1 present.21.key_splitncnn_0 /model/layers.21/self_attn/Constant_32_output_0 /model/layers.21/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.21/self_attn/Reshape_6 2 1 /model/layers.21/self_attn/Expand_1_output_0 /model/layers.21/self_attn/Concat_10_output_0 /model/layers.21/self_attn/Reshape_6_output_0 Crop /model/layers.21/self_attn/Gather_4 1 1 /model/layers.21/self_attn/Shape_4_output_0 /model/layers.21/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_689 1 2 /model/layers.21/self_attn/Gather_4_output_0 /model/layers.21/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.21/self_attn/Gather_5 1 1 /model/layers.21/self_attn/Shape_5_output_0 /model/layers.21/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_690 1 2 /model/layers.21/self_attn/Gather_5_output_0 /model/layers.21/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.21/self_attn/Gather_6 1 1 /model/layers.21/self_attn/Shape_6_output_0 /model/layers.21/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_691 1 2 /model/layers.21/self_attn/Gather_6_output_0 /model/layers.21/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.21/self_attn/Gather_7 1 1 /model/layers.21/self_attn/Shape_7_output_0 /model/layers.21/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_692 1 2 /model/layers.21/self_attn/Gather_7_output_0 /model/layers.21/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.21/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.21/self_attn/Unsqueeze_11 2 1 /model/layers.21/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_33_output_0 /model/layers.21/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_15 2 1 /model/layers.21/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_6493 /model/layers.21/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_12 2 1 /model/layers.21/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_34_output_0 /model/layers.21/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.21/self_attn/Mul_5 2 1 /model/layers.21/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.21/self_attn/Constant_40_output_0 /model/layers.21/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.21/self_attn/Unsqueeze_13 2 1 /model/layers.21/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_36_output_0 /model/layers.21/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_17 2 1 /model/layers.21/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_6497 /model/layers.21/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_14 2 1 /model/layers.21/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_37_output_0 /model/layers.21/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_18 2 1 /model/layers.21/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_6499 /model/layers.21/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.21/self_attn/Unsqueeze_16 2 1 /model/layers.21/self_attn/Mul_5_output_0 onnx::Unsqueeze_6495 /model/layers.21/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.21/self_attn/Concat_7 5 1 /model/layers.21/self_attn/Unsqueeze_11_output_0 /model/layers.21/self_attn/Unsqueeze_12_output_0 /model/layers.21/self_attn/Constant_35_output_0 /model/layers.21/self_attn/Unsqueeze_13_output_0 /model/layers.21/self_attn/Unsqueeze_14_output_0 /model/layers.21/self_attn/Concat_7_output_0 0=0 Concat /model/layers.21/self_attn/Concat_8 4 1 /model/layers.21/self_attn/Unsqueeze_15_output_0 /model/layers.21/self_attn/Unsqueeze_16_output_0 /model/layers.21/self_attn/Unsqueeze_17_output_0 /model/layers.21/self_attn/Unsqueeze_18_output_0 /model/layers.21/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.21/self_attn/Reshape_3 1 1 /model/layers.21/self_attn/Concat_7_output_0 /model/layers.21/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_693 1 3 /model/layers.21/self_attn/Reshape_3_output_0 /model/layers.21/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.21/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.21/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.21/self_attn/Shape_8 1 1 /model/layers.21/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.21/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.21/self_attn/ConstantOfShape 1 1 /model/layers.21/self_attn/Shape_8_output_0 /model/layers.21/self_attn/ConstantOfShape_output_0 Split splitncnn_694 1 2 /model/layers.21/self_attn/ConstantOfShape_output_0 /model/layers.21/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.21/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.21/self_attn/Mul_4 2 1 /model/layers.21/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.21/self_attn/Constant_39_output_0 /model/layers.21/self_attn/Mul_4_output_0 0=2 Equal /model/layers.21/self_attn/Equal 2 1 /model/layers.21/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.21/self_attn/Mul_4_output_0 /model/layers.21/self_attn/Equal_output_0 Where /model/layers.21/self_attn/Where 3 1 /model/layers.21/self_attn/Equal_output_0 /model/layers.21/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.21/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.21/self_attn/Where_output_0 Expand /model/layers.21/self_attn/Expand 2 1 /model/layers.21/self_attn/Unsqueeze_10_output_0 /model/layers.21/self_attn/Where_output_0 /model/layers.21/self_attn/Expand_output_0 Reshape /model/layers.21/self_attn/Reshape_4 2 1 /model/layers.21/self_attn/Expand_output_0 /model/layers.21/self_attn/Concat_8_output_0 /model/layers.21/self_attn/Reshape_4_output_0 Split splitncnn_695 1 2 /model/layers.21/self_attn/Reshape_4_output_0 /model/layers.21/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.21/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.21/self_attn/Shape_14 1 1 /model/layers.21/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.21/self_attn/Shape_14_output_0 Permute /model/layers.21/self_attn/Transpose_3 1 1 /model/layers.21/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.21/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.21/self_attn/Gather_12 1 1 /model/layers.21/self_attn/Shape_14_output_0 /model/layers.21/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.21/self_attn/Mul_9 1 1 /model/layers.21/self_attn/Transpose_3_output_0 /model/layers.21/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.21/self_attn/Unsqueeze_28 2 1 /model/layers.21/self_attn/Gather_12_output_0 /model/layers.21/self_attn/Constant_57_output_0 /model/layers.21/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.21/self_attn/MatMul 2 1 /model/layers.21/self_attn/Mul_8_output_0 /model/layers.21/self_attn/Mul_9_output_0 /model/layers.21/self_attn/MatMul_output_0 Crop /model/layers.21/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_2 /model/layers.21/self_attn/Unsqueeze_28_output_0 /model/layers.21/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.21/self_attn/Where_2 3 1 /model/layers.21/self_attn/Slice_4_output_0 /model/layers.21/self_attn/Constant_61_output_0 /model/layers.21/self_attn/Constant_62_output_0 /model/layers.21/self_attn/Where_2_output_0 BinaryOp /model/layers.21/self_attn/Add_2 2 1 /model/layers.21/self_attn/MatMul_output_0 /model/layers.21/self_attn/Where_2_output_0 /model/layers.21/self_attn/Add_2_output_0 0=0 Softmax /model/layers.21/self_attn/Softmax 1 1 /model/layers.21/self_attn/Add_2_output_0 /model/layers.21/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_696 1 2 /model/layers.21/self_attn/Softmax_output_0 /model/layers.21/self_attn/Softmax_output_0_splitncnn_0 /model/layers.21/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.21/self_attn/IsNaN 1 1 /model/layers.21/self_attn/Softmax_output_0_splitncnn_1 /model/layers.21/self_attn/IsNaN_output_0 Where /model/layers.21/self_attn/Where_3 3 1 /model/layers.21/self_attn/IsNaN_output_0 /model/layers.21/self_attn/Constant_63_output_0 /model/layers.21/self_attn/Softmax_output_0_splitncnn_0 /model/layers.21/self_attn/Where_3_output_0 Gemm /model/layers.21/self_attn/MatMul_1 2 1 /model/layers.21/self_attn/Where_3_output_0 /model/layers.21/self_attn/Reshape_6_output_0 /model/layers.21/self_attn/MatMul_1_output_0 Permute /model/layers.21/self_attn/Transpose_4 1 1 /model/layers.21/self_attn/MatMul_1_output_0 /model/layers.21/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.21/self_attn/Reshape_7 2 1 /model/layers.21/self_attn/Transpose_4_output_0 /model/layers.21/self_attn/Concat_11_output_0 /model/layers.21/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.21/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.21/self_attn/Reshape_7_output_0 /model/layers.21/self_attn/Reshape_7_output_0_quantized /model/layers.21/self_attn/Reshape_7_output_0_scale /model/layers.21/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.21/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.21/self_attn/Reshape_7_output_0_scale /model/layers.21/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.645177e-03 MatMulInteger /model/layers.21/self_attn/o_proj/MatMul_quant 4 1 /model/layers.21/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7956_quantized /model/layers.21/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7956_zero_point /model/layers.21/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.21/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.21/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.21/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.21/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.21/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.21/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.21/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.21/Add 2 1 /model/layers.21/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.21/self_attn/o_proj/MatMul_output_0 /model/layers.21/Add_output_0 0=0 Cast /model/layers.21/post_attention_layernorm/Cast 1 1 /model/layers.21/Add_output_0 /model/layers.21/post_attention_layernorm/Cast_output_0 Split splitncnn_697 1 3 /model/layers.21/post_attention_layernorm/Cast_output_0 /model/layers.21/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.21/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.21/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.21/post_attention_layernorm/Pow 1 1 /model/layers.21/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.21/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.21/post_attention_layernorm/ReduceMean 2 1 /model/layers.21/post_attention_layernorm/Pow_output_0 /model/layers.21/post_attention_layernorm/Constant_1_output_0 /model/layers.21/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.21/post_attention_layernorm/Add 1 1 /model/layers.21/post_attention_layernorm/ReduceMean_output_0 /model/layers.21/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.21/post_attention_layernorm/Sqrt 1 1 /model/layers.21/post_attention_layernorm/Add_output_0 /model/layers.21/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.21/post_attention_layernorm/Div 1 1 /model/layers.21/post_attention_layernorm/Sqrt_output_0 /model/layers.21/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.21/post_attention_layernorm/Mul 2 1 /model/layers.21/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.21/post_attention_layernorm/Div_output_0 /model/layers.21/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.21/post_attention_layernorm/Cast_1 1 1 /model/layers.21/post_attention_layernorm/Mul_output_0 /model/layers.21/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.21/post_attention_layernorm/Mul_1 2 1 model.layers.21.post_attention_layernorm.weight /model/layers.21/post_attention_layernorm/Cast_1_output_0 /model/layers.21/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.21/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.21/post_attention_layernorm/Mul_1_output_0 /model/layers.21/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.21/post_attention_layernorm/Mul_1_output_0_scale /model/layers.21/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_698 1 2 /model/layers.21/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.21/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.21/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_699 1 2 /model/layers.21/post_attention_layernorm/Mul_1_output_0_scale /model/layers.21/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.21/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_700 1 2 /model/layers.21/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.21/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.21/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.21/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.21/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.21/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.582923e-03 BinaryOp /model/layers.21/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.21/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.21/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=5.967028e-03 MatMulInteger /model/layers.21/mlp/gate_proj/MatMul_quant 4 1 /model/layers.21/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7957_quantized /model/layers.21/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7957_zero_point /model/layers.21/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.21/mlp/up_proj/MatMul_quant 4 1 /model/layers.21/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7958_quantized /model/layers.21/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7958_zero_point /model/layers.21/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.21/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.21/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.21/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.21/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.21/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.21/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.21/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.21/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.21/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.21/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.21/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.21/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.21/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.21/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.21/mlp/act_fn/Mul 1 1 /model/layers.21/mlp/gate_proj/MatMul_output_0 /model/layers.21/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.21/mlp/Mul 2 1 /model/layers.21/mlp/act_fn/Mul_output_0 /model/layers.21/mlp/up_proj/MatMul_output_0 /model/layers.21/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.21/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.21/mlp/Mul_output_0 /model/layers.21/mlp/Mul_output_0_quantized /model/layers.21/mlp/Mul_output_0_scale /model/layers.21/mlp/Mul_output_0_zero_point BinaryOp /model/layers.21/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.21/mlp/Mul_output_0_scale /model/layers.21/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.921260e-03 MatMulInteger /model/layers.21/mlp/down_proj/MatMul_quant 4 1 /model/layers.21/mlp/Mul_output_0_quantized onnx::MatMul_7959_quantized /model/layers.21/mlp/Mul_output_0_zero_point onnx::MatMul_7959_zero_point /model/layers.21/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.21/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.21/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.21/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.21/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.21/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.21/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.21/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.21/Add_1 2 1 /model/layers.21/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.21/mlp/down_proj/MatMul_output_0 /model/layers.21/Add_1_output_0 0=0 Cast /model/layers.22/input_layernorm/Cast 1 1 /model/layers.21/Add_1_output_0 /model/layers.22/input_layernorm/Cast_output_0 Split splitncnn_701 1 3 /model/layers.22/input_layernorm/Cast_output_0 /model/layers.22/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.22/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.22/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.22/input_layernorm/Pow 1 1 /model/layers.22/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.22/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.22/input_layernorm/ReduceMean 2 1 /model/layers.22/input_layernorm/Pow_output_0 /model/layers.22/input_layernorm/Constant_1_output_0 /model/layers.22/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.22/input_layernorm/Add 1 1 /model/layers.22/input_layernorm/ReduceMean_output_0 /model/layers.22/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.22/input_layernorm/Sqrt 1 1 /model/layers.22/input_layernorm/Add_output_0 /model/layers.22/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.22/input_layernorm/Div 1 1 /model/layers.22/input_layernorm/Sqrt_output_0 /model/layers.22/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.22/input_layernorm/Mul 2 1 /model/layers.22/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.22/input_layernorm/Div_output_0 /model/layers.22/input_layernorm/Mul_output_0 0=2 Cast /model/layers.22/input_layernorm/Cast_1 1 1 /model/layers.22/input_layernorm/Mul_output_0 /model/layers.22/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.22/input_layernorm/Mul_1 2 1 model.layers.22.input_layernorm.weight /model/layers.22/input_layernorm/Cast_1_output_0 /model/layers.22/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_702 1 3 /model/layers.22/input_layernorm/Mul_1_output_0 /model/layers.22/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.22/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.22/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.22/self_attn/Shape 1 1 /model/layers.22/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.22/self_attn/Shape_output_0 Shape /model/layers.22/self_attn/Shape_1 1 1 /model/layers.22/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.22/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.22/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.22/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.22/input_layernorm/Mul_1_output_0_quantized /model/layers.22/input_layernorm/Mul_1_output_0_scale /model/layers.22/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_703 1 3 /model/layers.22/input_layernorm/Mul_1_output_0_quantized /model/layers.22/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.22/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.22/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_704 1 3 /model/layers.22/input_layernorm/Mul_1_output_0_scale /model/layers.22/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.22/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.22/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_705 1 3 /model/layers.22/input_layernorm/Mul_1_output_0_zero_point /model/layers.22/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.22/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.22/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.22/self_attn/Gather 1 1 /model/layers.22/self_attn/Shape_output_0 /model/layers.22/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_706 1 4 /model/layers.22/self_attn/Gather_output_0 /model/layers.22/self_attn/Gather_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_output_0_splitncnn_1 /model/layers.22/self_attn/Gather_output_0_splitncnn_2 /model/layers.22/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.22/self_attn/Gather_1 1 1 /model/layers.22/self_attn/Shape_1_output_0 /model/layers.22/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_707 1 4 /model/layers.22/self_attn/Gather_1_output_0 /model/layers.22/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.22/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.22/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.22/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.22/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.22/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.030020e-03 BinaryOp /model/layers.22/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.22/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.22/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.291831e-03 BinaryOp /model/layers.22/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.22/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.22/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.668615e-03 MatMulInteger /model/layers.22/self_attn/q_proj/MatMul_quant 4 1 /model/layers.22/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7960_quantized /model/layers.22/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7960_zero_point /model/layers.22/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.22/self_attn/k_proj/MatMul_quant 4 1 /model/layers.22/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7967_quantized /model/layers.22/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7967_zero_point /model/layers.22/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.22/self_attn/v_proj/MatMul_quant 4 1 /model/layers.22/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7968_quantized /model/layers.22/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7968_zero_point /model/layers.22/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.22/self_attn/Unsqueeze 2 1 /model/layers.22/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_6635 /model/layers.22/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_2 2 1 /model/layers.22/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_6644 /model/layers.22/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_4 2 1 /model/layers.22/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_6653 /model/layers.22/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_29 2 1 /model/layers.22/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_6858 /model/layers.22/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_1 2 1 /model/layers.22/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_6637 /model/layers.22/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_3 2 1 /model/layers.22/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_6646 /model/layers.22/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_5 2 1 /model/layers.22/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_6655 /model/layers.22/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_30 2 1 /model/layers.22/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_6860 /model/layers.22/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.22/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.22/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.22/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.22/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.22/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.22/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.22/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.22/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.22/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.22/self_attn/Concat 4 1 /model/layers.22/self_attn/Unsqueeze_output_0 /model/layers.22/self_attn/Unsqueeze_1_output_0 /model/layers.22/self_attn/Constant_2_output_0 /model/layers.22/self_attn/Constant_3_output_0 /model/layers.22/self_attn/Concat_output_0 0=0 Concat /model/layers.22/self_attn/Concat_1 4 1 /model/layers.22/self_attn/Unsqueeze_2_output_0 /model/layers.22/self_attn/Unsqueeze_3_output_0 /model/layers.22/self_attn/Constant_4_output_0 /model/layers.22/self_attn/Constant_5_output_0 /model/layers.22/self_attn/Concat_1_output_0 0=0 Concat /model/layers.22/self_attn/Concat_2 4 1 /model/layers.22/self_attn/Unsqueeze_4_output_0 /model/layers.22/self_attn/Unsqueeze_5_output_0 /model/layers.22/self_attn/Constant_6_output_0 /model/layers.22/self_attn/Constant_7_output_0 /model/layers.22/self_attn/Concat_2_output_0 0=0 Concat /model/layers.22/self_attn/Concat_11 3 1 /model/layers.22/self_attn/Unsqueeze_29_output_0 /model/layers.22/self_attn/Unsqueeze_30_output_0 /model/layers.22/self_attn/Constant_64_output_0 /model/layers.22/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.22/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.22/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.22/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.22/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.22/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.22/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.22/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.22/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.22/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.22/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.22/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.22/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.22/self_attn/q_proj/Add 2 1 model.layers.22.self_attn.q_proj.bias /model/layers.22/self_attn/q_proj/MatMul_output_0 /model/layers.22/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.22/self_attn/k_proj/Add 2 1 model.layers.22.self_attn.k_proj.bias /model/layers.22/self_attn/k_proj/MatMul_output_0 /model/layers.22/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.22/self_attn/v_proj/Add 2 1 model.layers.22.self_attn.v_proj.bias /model/layers.22/self_attn/v_proj/MatMul_output_0 /model/layers.22/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.22/self_attn/Reshape 2 1 /model/layers.22/self_attn/q_proj/Add_output_0 /model/layers.22/self_attn/Concat_output_0 /model/layers.22/self_attn/Reshape_output_0 Reshape /model/layers.22/self_attn/Reshape_1 2 1 /model/layers.22/self_attn/k_proj/Add_output_0 /model/layers.22/self_attn/Concat_1_output_0 /model/layers.22/self_attn/Reshape_1_output_0 Reshape /model/layers.22/self_attn/Reshape_2 2 1 /model/layers.22/self_attn/v_proj/Add_output_0 /model/layers.22/self_attn/Concat_2_output_0 /model/layers.22/self_attn/Reshape_2_output_0 Permute /model/layers.22/self_attn/Transpose 1 1 /model/layers.22/self_attn/Reshape_output_0 /model/layers.22/self_attn/Transpose_output_0 0=2 Split splitncnn_708 1 4 /model/layers.22/self_attn/Transpose_output_0 /model/layers.22/self_attn/Transpose_output_0_splitncnn_0 /model/layers.22/self_attn/Transpose_output_0_splitncnn_1 /model/layers.22/self_attn/Transpose_output_0_splitncnn_2 /model/layers.22/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.22/self_attn/Transpose_1 1 1 /model/layers.22/self_attn/Reshape_1_output_0 /model/layers.22/self_attn/Transpose_1_output_0 0=2 Split splitncnn_709 1 4 /model/layers.22/self_attn/Transpose_1_output_0 /model/layers.22/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.22/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.22/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.22/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.22/self_attn/Transpose_2 1 1 /model/layers.22/self_attn/Reshape_2_output_0 /model/layers.22/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.22/self_attn/Mul 2 1 /model/layers.22/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_3 /model/layers.22/self_attn/Mul_output_0 0=2 Shape /model/layers.22/self_attn/Shape_2 1 1 /model/layers.22/self_attn/Transpose_output_0_splitncnn_2 /model/layers.22/self_attn/Shape_2_output_0 BinaryOp /model/layers.22/self_attn/Mul_2 2 1 /model/layers.22/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_2 /model/layers.22/self_attn/Mul_2_output_0 0=2 Shape /model/layers.22/self_attn/Shape_3 1 1 /model/layers.22/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.22/self_attn/Shape_3_output_0 Concat /model/layers.22/self_attn/Concat_6 2 1 past_key_values.22.value /model/layers.22/self_attn/Transpose_2_output_0 present.22.value 0=-2 Split splitncnn_710 1 5 present.22.value present.22.value_splitncnn_0 present.22.value_splitncnn_1 present.22.value_splitncnn_2 present.22.value_splitncnn_3 present.22.value_splitncnn_4 Crop /model/layers.22/self_attn/Gather_2 1 1 /model/layers.22/self_attn/Shape_2_output_0 /model/layers.22/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.22/self_attn/Gather_3 1 1 /model/layers.22/self_attn/Shape_3_output_0 /model/layers.22/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.22/self_attn/Shape_9 1 1 present.22.value_splitncnn_4 /model/layers.22/self_attn/Shape_9_output_0 Shape /model/layers.22/self_attn/Shape_10 1 1 present.22.value_splitncnn_3 /model/layers.22/self_attn/Shape_10_output_0 Shape /model/layers.22/self_attn/Shape_11 1 1 present.22.value_splitncnn_2 /model/layers.22/self_attn/Shape_11_output_0 Shape /model/layers.22/self_attn/Shape_12 1 1 present.22.value_splitncnn_1 /model/layers.22/self_attn/Shape_12_output_0 ExpandDims /model/layers.22/self_attn/Unsqueeze_19 2 1 present.22.value_splitncnn_0 /model/layers.22/self_attn/Constant_45_output_0 /model/layers.22/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.22/self_attn/Div 2 1 /model/layers.22/self_attn/Gather_2_output_0 /model/layers.22/self_attn/Constant_9_output_0 /model/layers.22/self_attn/Div_output_0 0=3 BinaryOp /model/layers.22/self_attn/Div_1 2 1 /model/layers.22/self_attn/Gather_3_output_0 /model/layers.22/self_attn/Constant_19_output_0 /model/layers.22/self_attn/Div_1_output_0 0=3 Crop /model/layers.22/self_attn/Gather_8 1 1 /model/layers.22/self_attn/Shape_9_output_0 /model/layers.22/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_711 1 2 /model/layers.22/self_attn/Gather_8_output_0 /model/layers.22/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.22/self_attn/Gather_9 1 1 /model/layers.22/self_attn/Shape_10_output_0 /model/layers.22/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_712 1 2 /model/layers.22/self_attn/Gather_9_output_0 /model/layers.22/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.22/self_attn/Gather_10 1 1 /model/layers.22/self_attn/Shape_11_output_0 /model/layers.22/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_713 1 2 /model/layers.22/self_attn/Gather_10_output_0 /model/layers.22/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.22/self_attn/Gather_11 1 1 /model/layers.22/self_attn/Shape_12_output_0 /model/layers.22/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_714 1 2 /model/layers.22/self_attn/Gather_11_output_0 /model/layers.22/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.22/self_attn/Cast 1 1 /model/layers.22/self_attn/Div_output_0 /model/layers.22/self_attn/Cast_output_0 Cast /model/layers.22/self_attn/Cast_2 1 1 /model/layers.22/self_attn/Div_1_output_0 /model/layers.22/self_attn/Cast_2_output_0 ExpandDims /model/layers.22/self_attn/Unsqueeze_20 2 1 /model/layers.22/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_46_output_0 /model/layers.22/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_24 2 1 /model/layers.22/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_6818 /model/layers.22/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_21 2 1 /model/layers.22/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_47_output_0 /model/layers.22/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.22/self_attn/Mul_7 2 1 /model/layers.22/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.22/self_attn/Constant_53_output_0 /model/layers.22/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.22/self_attn/Unsqueeze_22 2 1 /model/layers.22/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_49_output_0 /model/layers.22/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_26 2 1 /model/layers.22/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_6822 /model/layers.22/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_23 2 1 /model/layers.22/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_50_output_0 /model/layers.22/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_27 2 1 /model/layers.22/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_6824 /model/layers.22/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.22/self_attn/Cast_1 1 1 /model/layers.22/self_attn/Cast_output_0 /model/layers.22/self_attn/Cast_1_output_0 Split splitncnn_715 1 2 /model/layers.22/self_attn/Cast_1_output_0 /model/layers.22/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.22/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.22/self_attn/Cast_3 1 1 /model/layers.22/self_attn/Cast_2_output_0 /model/layers.22/self_attn/Cast_3_output_0 Split splitncnn_716 1 2 /model/layers.22/self_attn/Cast_3_output_0 /model/layers.22/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.22/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.22/self_attn/Unsqueeze_25 2 1 /model/layers.22/self_attn/Mul_7_output_0 onnx::Unsqueeze_6820 /model/layers.22/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.22/self_attn/Concat_9 5 1 /model/layers.22/self_attn/Unsqueeze_20_output_0 /model/layers.22/self_attn/Unsqueeze_21_output_0 /model/layers.22/self_attn/Constant_48_output_0 /model/layers.22/self_attn/Unsqueeze_22_output_0 /model/layers.22/self_attn/Unsqueeze_23_output_0 /model/layers.22/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_6 2 1 /model/layers.22/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_12_output_0 /model/layers.22/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_7 2 1 /model/layers.22/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.22/self_attn/Constant_15_output_0 /model/layers.22/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_8 2 1 /model/layers.22/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_22_output_0 /model/layers.22/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_9 2 1 /model/layers.22/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.22/self_attn/Constant_25_output_0 /model/layers.22/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.22/self_attn/Concat_10 4 1 /model/layers.22/self_attn/Unsqueeze_24_output_0 /model/layers.22/self_attn/Unsqueeze_25_output_0 /model/layers.22/self_attn/Unsqueeze_26_output_0 /model/layers.22/self_attn/Unsqueeze_27_output_0 /model/layers.22/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.22/self_attn/Reshape_5 1 1 /model/layers.22/self_attn/Concat_9_output_0 /model/layers.22/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_717 1 3 /model/layers.22/self_attn/Reshape_5_output_0 /model/layers.22/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.22/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.22/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.22/self_attn/Slice 2 1 /model/layers.22/self_attn/Transpose_output_0_splitncnn_1 /model/layers.22/self_attn/Unsqueeze_6_output_0 /model/layers.22/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.22/self_attn/Slice_1 2 1 /model/layers.22/self_attn/Transpose_output_0_splitncnn_0 /model/layers.22/self_attn/Unsqueeze_7_output_0 /model/layers.22/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.22/self_attn/Slice_2 2 1 /model/layers.22/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.22/self_attn/Unsqueeze_8_output_0 /model/layers.22/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.22/self_attn/Slice_3 2 1 /model/layers.22/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.22/self_attn/Unsqueeze_9_output_0 /model/layers.22/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.22/self_attn/Shape_13 1 1 /model/layers.22/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.22/self_attn/Shape_13_output_0 UnaryOp /model/layers.22/self_attn/Neg 1 1 /model/layers.22/self_attn/Slice_1_output_0 /model/layers.22/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.22/self_attn/Neg_1 1 1 /model/layers.22/self_attn/Slice_3_output_0 /model/layers.22/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.22/self_attn/ConstantOfShape_1 1 1 /model/layers.22/self_attn/Shape_13_output_0 /model/layers.22/self_attn/ConstantOfShape_1_output_0 Split splitncnn_718 1 2 /model/layers.22/self_attn/ConstantOfShape_1_output_0 /model/layers.22/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.22/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.22/self_attn/Concat_3 2 1 /model/layers.22/self_attn/Neg_output_0 /model/layers.22/self_attn/Slice_output_0 /model/layers.22/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.22/self_attn/Concat_4 2 1 /model/layers.22/self_attn/Neg_1_output_0 /model/layers.22/self_attn/Slice_2_output_0 /model/layers.22/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.22/self_attn/Mul_6 2 1 /model/layers.22/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_52_output_0 /model/layers.22/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.22/self_attn/Mul_1 2 1 /model/layers.22/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_3 /model/layers.22/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.22/self_attn/Mul_3 2 1 /model/layers.22/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_2 /model/layers.22/self_attn/Mul_3_output_0 0=2 Equal /model/layers.22/self_attn/Equal_1 2 1 /model/layers.22/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.22/self_attn/Mul_6_output_0 /model/layers.22/self_attn/Equal_1_output_0 BinaryOp /model/layers.22/self_attn/Add 2 1 /model/layers.22/self_attn/Mul_output_0 /model/layers.22/self_attn/Mul_1_output_0 /model/layers.22/self_attn/Add_output_0 0=0 BinaryOp /model/layers.22/self_attn/Add_1 2 1 /model/layers.22/self_attn/Mul_2_output_0 /model/layers.22/self_attn/Mul_3_output_0 /model/layers.22/self_attn/Add_1_output_0 0=0 Where /model/layers.22/self_attn/Where_1 3 1 /model/layers.22/self_attn/Equal_1_output_0 /model/layers.22/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.22/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.22/self_attn/Where_1_output_0 BinaryOp /model/layers.22/self_attn/Mul_8 1 1 /model/layers.22/self_attn/Add_output_0 /model/layers.22/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.22/self_attn/Concat_5 2 1 past_key_values.22.key /model/layers.22/self_attn/Add_1_output_0 present.22.key 0=-2 Split splitncnn_719 1 5 present.22.key present.22.key_splitncnn_0 present.22.key_splitncnn_1 present.22.key_splitncnn_2 present.22.key_splitncnn_3 present.22.key_splitncnn_4 Expand /model/layers.22/self_attn/Expand_1 2 1 /model/layers.22/self_attn/Unsqueeze_19_output_0 /model/layers.22/self_attn/Where_1_output_0 /model/layers.22/self_attn/Expand_1_output_0 Shape /model/layers.22/self_attn/Shape_4 1 1 present.22.key_splitncnn_4 /model/layers.22/self_attn/Shape_4_output_0 Shape /model/layers.22/self_attn/Shape_5 1 1 present.22.key_splitncnn_3 /model/layers.22/self_attn/Shape_5_output_0 Shape /model/layers.22/self_attn/Shape_6 1 1 present.22.key_splitncnn_2 /model/layers.22/self_attn/Shape_6_output_0 Shape /model/layers.22/self_attn/Shape_7 1 1 present.22.key_splitncnn_1 /model/layers.22/self_attn/Shape_7_output_0 ExpandDims /model/layers.22/self_attn/Unsqueeze_10 2 1 present.22.key_splitncnn_0 /model/layers.22/self_attn/Constant_32_output_0 /model/layers.22/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.22/self_attn/Reshape_6 2 1 /model/layers.22/self_attn/Expand_1_output_0 /model/layers.22/self_attn/Concat_10_output_0 /model/layers.22/self_attn/Reshape_6_output_0 Crop /model/layers.22/self_attn/Gather_4 1 1 /model/layers.22/self_attn/Shape_4_output_0 /model/layers.22/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_720 1 2 /model/layers.22/self_attn/Gather_4_output_0 /model/layers.22/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.22/self_attn/Gather_5 1 1 /model/layers.22/self_attn/Shape_5_output_0 /model/layers.22/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_721 1 2 /model/layers.22/self_attn/Gather_5_output_0 /model/layers.22/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.22/self_attn/Gather_6 1 1 /model/layers.22/self_attn/Shape_6_output_0 /model/layers.22/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_722 1 2 /model/layers.22/self_attn/Gather_6_output_0 /model/layers.22/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.22/self_attn/Gather_7 1 1 /model/layers.22/self_attn/Shape_7_output_0 /model/layers.22/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_723 1 2 /model/layers.22/self_attn/Gather_7_output_0 /model/layers.22/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.22/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.22/self_attn/Unsqueeze_11 2 1 /model/layers.22/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_33_output_0 /model/layers.22/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_15 2 1 /model/layers.22/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_6772 /model/layers.22/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_12 2 1 /model/layers.22/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_34_output_0 /model/layers.22/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.22/self_attn/Mul_5 2 1 /model/layers.22/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.22/self_attn/Constant_40_output_0 /model/layers.22/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.22/self_attn/Unsqueeze_13 2 1 /model/layers.22/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_36_output_0 /model/layers.22/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_17 2 1 /model/layers.22/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_6776 /model/layers.22/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_14 2 1 /model/layers.22/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_37_output_0 /model/layers.22/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_18 2 1 /model/layers.22/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_6778 /model/layers.22/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.22/self_attn/Unsqueeze_16 2 1 /model/layers.22/self_attn/Mul_5_output_0 onnx::Unsqueeze_6774 /model/layers.22/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.22/self_attn/Concat_7 5 1 /model/layers.22/self_attn/Unsqueeze_11_output_0 /model/layers.22/self_attn/Unsqueeze_12_output_0 /model/layers.22/self_attn/Constant_35_output_0 /model/layers.22/self_attn/Unsqueeze_13_output_0 /model/layers.22/self_attn/Unsqueeze_14_output_0 /model/layers.22/self_attn/Concat_7_output_0 0=0 Concat /model/layers.22/self_attn/Concat_8 4 1 /model/layers.22/self_attn/Unsqueeze_15_output_0 /model/layers.22/self_attn/Unsqueeze_16_output_0 /model/layers.22/self_attn/Unsqueeze_17_output_0 /model/layers.22/self_attn/Unsqueeze_18_output_0 /model/layers.22/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.22/self_attn/Reshape_3 1 1 /model/layers.22/self_attn/Concat_7_output_0 /model/layers.22/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_724 1 3 /model/layers.22/self_attn/Reshape_3_output_0 /model/layers.22/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.22/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.22/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.22/self_attn/Shape_8 1 1 /model/layers.22/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.22/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.22/self_attn/ConstantOfShape 1 1 /model/layers.22/self_attn/Shape_8_output_0 /model/layers.22/self_attn/ConstantOfShape_output_0 Split splitncnn_725 1 2 /model/layers.22/self_attn/ConstantOfShape_output_0 /model/layers.22/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.22/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.22/self_attn/Mul_4 2 1 /model/layers.22/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.22/self_attn/Constant_39_output_0 /model/layers.22/self_attn/Mul_4_output_0 0=2 Equal /model/layers.22/self_attn/Equal 2 1 /model/layers.22/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.22/self_attn/Mul_4_output_0 /model/layers.22/self_attn/Equal_output_0 Where /model/layers.22/self_attn/Where 3 1 /model/layers.22/self_attn/Equal_output_0 /model/layers.22/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.22/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.22/self_attn/Where_output_0 Expand /model/layers.22/self_attn/Expand 2 1 /model/layers.22/self_attn/Unsqueeze_10_output_0 /model/layers.22/self_attn/Where_output_0 /model/layers.22/self_attn/Expand_output_0 Reshape /model/layers.22/self_attn/Reshape_4 2 1 /model/layers.22/self_attn/Expand_output_0 /model/layers.22/self_attn/Concat_8_output_0 /model/layers.22/self_attn/Reshape_4_output_0 Split splitncnn_726 1 2 /model/layers.22/self_attn/Reshape_4_output_0 /model/layers.22/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.22/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.22/self_attn/Shape_14 1 1 /model/layers.22/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.22/self_attn/Shape_14_output_0 Permute /model/layers.22/self_attn/Transpose_3 1 1 /model/layers.22/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.22/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.22/self_attn/Gather_12 1 1 /model/layers.22/self_attn/Shape_14_output_0 /model/layers.22/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.22/self_attn/Mul_9 1 1 /model/layers.22/self_attn/Transpose_3_output_0 /model/layers.22/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.22/self_attn/Unsqueeze_28 2 1 /model/layers.22/self_attn/Gather_12_output_0 /model/layers.22/self_attn/Constant_57_output_0 /model/layers.22/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.22/self_attn/MatMul 2 1 /model/layers.22/self_attn/Mul_8_output_0 /model/layers.22/self_attn/Mul_9_output_0 /model/layers.22/self_attn/MatMul_output_0 Crop /model/layers.22/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_1 /model/layers.22/self_attn/Unsqueeze_28_output_0 /model/layers.22/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.22/self_attn/Where_2 3 1 /model/layers.22/self_attn/Slice_4_output_0 /model/layers.22/self_attn/Constant_61_output_0 /model/layers.22/self_attn/Constant_62_output_0 /model/layers.22/self_attn/Where_2_output_0 BinaryOp /model/layers.22/self_attn/Add_2 2 1 /model/layers.22/self_attn/MatMul_output_0 /model/layers.22/self_attn/Where_2_output_0 /model/layers.22/self_attn/Add_2_output_0 0=0 Softmax /model/layers.22/self_attn/Softmax 1 1 /model/layers.22/self_attn/Add_2_output_0 /model/layers.22/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_727 1 2 /model/layers.22/self_attn/Softmax_output_0 /model/layers.22/self_attn/Softmax_output_0_splitncnn_0 /model/layers.22/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.22/self_attn/IsNaN 1 1 /model/layers.22/self_attn/Softmax_output_0_splitncnn_1 /model/layers.22/self_attn/IsNaN_output_0 Where /model/layers.22/self_attn/Where_3 3 1 /model/layers.22/self_attn/IsNaN_output_0 /model/layers.22/self_attn/Constant_63_output_0 /model/layers.22/self_attn/Softmax_output_0_splitncnn_0 /model/layers.22/self_attn/Where_3_output_0 Gemm /model/layers.22/self_attn/MatMul_1 2 1 /model/layers.22/self_attn/Where_3_output_0 /model/layers.22/self_attn/Reshape_6_output_0 /model/layers.22/self_attn/MatMul_1_output_0 Permute /model/layers.22/self_attn/Transpose_4 1 1 /model/layers.22/self_attn/MatMul_1_output_0 /model/layers.22/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.22/self_attn/Reshape_7 2 1 /model/layers.22/self_attn/Transpose_4_output_0 /model/layers.22/self_attn/Concat_11_output_0 /model/layers.22/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.22/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.22/self_attn/Reshape_7_output_0 /model/layers.22/self_attn/Reshape_7_output_0_quantized /model/layers.22/self_attn/Reshape_7_output_0_scale /model/layers.22/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.22/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.22/self_attn/Reshape_7_output_0_scale /model/layers.22/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.245325e-03 MatMulInteger /model/layers.22/self_attn/o_proj/MatMul_quant 4 1 /model/layers.22/self_attn/Reshape_7_output_0_quantized onnx::MatMul_7991_quantized /model/layers.22/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_7991_zero_point /model/layers.22/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.22/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.22/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.22/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.22/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.22/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.22/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.22/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.22/Add 2 1 /model/layers.22/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.22/self_attn/o_proj/MatMul_output_0 /model/layers.22/Add_output_0 0=0 Cast /model/layers.22/post_attention_layernorm/Cast 1 1 /model/layers.22/Add_output_0 /model/layers.22/post_attention_layernorm/Cast_output_0 Split splitncnn_728 1 3 /model/layers.22/post_attention_layernorm/Cast_output_0 /model/layers.22/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.22/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.22/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.22/post_attention_layernorm/Pow 1 1 /model/layers.22/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.22/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.22/post_attention_layernorm/ReduceMean 2 1 /model/layers.22/post_attention_layernorm/Pow_output_0 /model/layers.22/post_attention_layernorm/Constant_1_output_0 /model/layers.22/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.22/post_attention_layernorm/Add 1 1 /model/layers.22/post_attention_layernorm/ReduceMean_output_0 /model/layers.22/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.22/post_attention_layernorm/Sqrt 1 1 /model/layers.22/post_attention_layernorm/Add_output_0 /model/layers.22/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.22/post_attention_layernorm/Div 1 1 /model/layers.22/post_attention_layernorm/Sqrt_output_0 /model/layers.22/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.22/post_attention_layernorm/Mul 2 1 /model/layers.22/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.22/post_attention_layernorm/Div_output_0 /model/layers.22/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.22/post_attention_layernorm/Cast_1 1 1 /model/layers.22/post_attention_layernorm/Mul_output_0 /model/layers.22/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.22/post_attention_layernorm/Mul_1 2 1 model.layers.22.post_attention_layernorm.weight /model/layers.22/post_attention_layernorm/Cast_1_output_0 /model/layers.22/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.22/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.22/post_attention_layernorm/Mul_1_output_0 /model/layers.22/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.22/post_attention_layernorm/Mul_1_output_0_scale /model/layers.22/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_729 1 2 /model/layers.22/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.22/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.22/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_730 1 2 /model/layers.22/post_attention_layernorm/Mul_1_output_0_scale /model/layers.22/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.22/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_731 1 2 /model/layers.22/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.22/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.22/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.22/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.22/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.22/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=5.997785e-03 BinaryOp /model/layers.22/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.22/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.22/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.921998e-03 MatMulInteger /model/layers.22/mlp/gate_proj/MatMul_quant 4 1 /model/layers.22/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_7992_quantized /model/layers.22/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_7992_zero_point /model/layers.22/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.22/mlp/up_proj/MatMul_quant 4 1 /model/layers.22/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_7993_quantized /model/layers.22/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_7993_zero_point /model/layers.22/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.22/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.22/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.22/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.22/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.22/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.22/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.22/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.22/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.22/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.22/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.22/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.22/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.22/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.22/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.22/mlp/act_fn/Mul 1 1 /model/layers.22/mlp/gate_proj/MatMul_output_0 /model/layers.22/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.22/mlp/Mul 2 1 /model/layers.22/mlp/act_fn/Mul_output_0 /model/layers.22/mlp/up_proj/MatMul_output_0 /model/layers.22/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.22/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.22/mlp/Mul_output_0 /model/layers.22/mlp/Mul_output_0_quantized /model/layers.22/mlp/Mul_output_0_scale /model/layers.22/mlp/Mul_output_0_zero_point BinaryOp /model/layers.22/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.22/mlp/Mul_output_0_scale /model/layers.22/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.629798e-03 MatMulInteger /model/layers.22/mlp/down_proj/MatMul_quant 4 1 /model/layers.22/mlp/Mul_output_0_quantized onnx::MatMul_7994_quantized /model/layers.22/mlp/Mul_output_0_zero_point onnx::MatMul_7994_zero_point /model/layers.22/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.22/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.22/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.22/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.22/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.22/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.22/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.22/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.22/Add_1 2 1 /model/layers.22/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.22/mlp/down_proj/MatMul_output_0 /model/layers.22/Add_1_output_0 0=0 Cast /model/layers.23/input_layernorm/Cast 1 1 /model/layers.22/Add_1_output_0 /model/layers.23/input_layernorm/Cast_output_0 Split splitncnn_732 1 3 /model/layers.23/input_layernorm/Cast_output_0 /model/layers.23/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.23/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.23/input_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.23/input_layernorm/Pow 1 1 /model/layers.23/input_layernorm/Cast_output_0_splitncnn_2 /model/layers.23/input_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.23/input_layernorm/ReduceMean 2 1 /model/layers.23/input_layernorm/Pow_output_0 /model/layers.23/input_layernorm/Constant_1_output_0 /model/layers.23/input_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.23/input_layernorm/Add 1 1 /model/layers.23/input_layernorm/ReduceMean_output_0 /model/layers.23/input_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.23/input_layernorm/Sqrt 1 1 /model/layers.23/input_layernorm/Add_output_0 /model/layers.23/input_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.23/input_layernorm/Div 1 1 /model/layers.23/input_layernorm/Sqrt_output_0 /model/layers.23/input_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.23/input_layernorm/Mul 2 1 /model/layers.23/input_layernorm/Cast_output_0_splitncnn_1 /model/layers.23/input_layernorm/Div_output_0 /model/layers.23/input_layernorm/Mul_output_0 0=2 Cast /model/layers.23/input_layernorm/Cast_1 1 1 /model/layers.23/input_layernorm/Mul_output_0 /model/layers.23/input_layernorm/Cast_1_output_0 BinaryOp /model/layers.23/input_layernorm/Mul_1 2 1 model.layers.23.input_layernorm.weight /model/layers.23/input_layernorm/Cast_1_output_0 /model/layers.23/input_layernorm/Mul_1_output_0 0=2 Split splitncnn_733 1 3 /model/layers.23/input_layernorm/Mul_1_output_0 /model/layers.23/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.23/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.23/input_layernorm/Mul_1_output_0_splitncnn_2 Shape /model/layers.23/self_attn/Shape 1 1 /model/layers.23/input_layernorm/Mul_1_output_0_splitncnn_2 /model/layers.23/self_attn/Shape_output_0 Shape /model/layers.23/self_attn/Shape_1 1 1 /model/layers.23/input_layernorm/Mul_1_output_0_splitncnn_1 /model/layers.23/self_attn/Shape_1_output_0 DynamicQuantizeLinear /model/layers.23/input_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.23/input_layernorm/Mul_1_output_0_splitncnn_0 /model/layers.23/input_layernorm/Mul_1_output_0_quantized /model/layers.23/input_layernorm/Mul_1_output_0_scale /model/layers.23/input_layernorm/Mul_1_output_0_zero_point Split splitncnn_734 1 3 /model/layers.23/input_layernorm/Mul_1_output_0_quantized /model/layers.23/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.23/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 /model/layers.23/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 Split splitncnn_735 1 3 /model/layers.23/input_layernorm/Mul_1_output_0_scale /model/layers.23/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.23/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.23/input_layernorm/Mul_1_output_0_scale_splitncnn_2 Split splitncnn_736 1 3 /model/layers.23/input_layernorm/Mul_1_output_0_zero_point /model/layers.23/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.23/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 /model/layers.23/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 Crop /model/layers.23/self_attn/Gather 1 1 /model/layers.23/self_attn/Shape_output_0 /model/layers.23/self_attn/Gather_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_737 1 4 /model/layers.23/self_attn/Gather_output_0 /model/layers.23/self_attn/Gather_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_output_0_splitncnn_1 /model/layers.23/self_attn/Gather_output_0_splitncnn_2 /model/layers.23/self_attn/Gather_output_0_splitncnn_3 Crop /model/layers.23/self_attn/Gather_1 1 1 /model/layers.23/self_attn/Shape_1_output_0 /model/layers.23/self_attn/Gather_1_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_738 1 4 /model/layers.23/self_attn/Gather_1_output_0 /model/layers.23/self_attn/Gather_1_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_1_output_0_splitncnn_1 /model/layers.23/self_attn/Gather_1_output_0_splitncnn_2 /model/layers.23/self_attn/Gather_1_output_0_splitncnn_3 BinaryOp /model/layers.23/self_attn/q_proj/MatMul_quant_scales_mul 1 1 /model/layers.23/input_layernorm/Mul_1_output_0_scale_splitncnn_2 /model/layers.23/self_attn/q_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.906988e-03 BinaryOp /model/layers.23/self_attn/k_proj/MatMul_quant_scales_mul 1 1 /model/layers.23/input_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.23/self_attn/k_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.630167e-03 BinaryOp /model/layers.23/self_attn/v_proj/MatMul_quant_scales_mul 1 1 /model/layers.23/input_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.23/self_attn/v_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=1.922367e-03 MatMulInteger /model/layers.23/self_attn/q_proj/MatMul_quant 4 1 /model/layers.23/input_layernorm/Mul_1_output_0_quantized_splitncnn_2 onnx::MatMul_7995_quantized /model/layers.23/input_layernorm/Mul_1_output_0_zero_point_splitncnn_2 onnx::MatMul_7995_zero_point /model/layers.23/self_attn/q_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.23/self_attn/k_proj/MatMul_quant 4 1 /model/layers.23/input_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_8002_quantized /model/layers.23/input_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_8002_zero_point /model/layers.23/self_attn/k_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.23/self_attn/v_proj/MatMul_quant 4 1 /model/layers.23/input_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_8003_quantized /model/layers.23/input_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_8003_zero_point /model/layers.23/self_attn/v_proj/MatMul_output_0_output_quantized ExpandDims /model/layers.23/self_attn/Unsqueeze 2 1 /model/layers.23/self_attn/Gather_output_0_splitncnn_3 onnx::Unsqueeze_6914 /model/layers.23/self_attn/Unsqueeze_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_2 2 1 /model/layers.23/self_attn/Gather_output_0_splitncnn_2 onnx::Unsqueeze_6923 /model/layers.23/self_attn/Unsqueeze_2_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_4 2 1 /model/layers.23/self_attn/Gather_output_0_splitncnn_1 onnx::Unsqueeze_6932 /model/layers.23/self_attn/Unsqueeze_4_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_29 2 1 /model/layers.23/self_attn/Gather_output_0_splitncnn_0 onnx::Unsqueeze_7137 /model/layers.23/self_attn/Unsqueeze_29_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_1 2 1 /model/layers.23/self_attn/Gather_1_output_0_splitncnn_3 onnx::Unsqueeze_6916 /model/layers.23/self_attn/Unsqueeze_1_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_3 2 1 /model/layers.23/self_attn/Gather_1_output_0_splitncnn_2 onnx::Unsqueeze_6925 /model/layers.23/self_attn/Unsqueeze_3_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_5 2 1 /model/layers.23/self_attn/Gather_1_output_0_splitncnn_1 onnx::Unsqueeze_6934 /model/layers.23/self_attn/Unsqueeze_5_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_30 2 1 /model/layers.23/self_attn/Gather_1_output_0_splitncnn_0 onnx::Unsqueeze_7139 /model/layers.23/self_attn/Unsqueeze_30_output_0 -23303=0 Cast /model/layers.23/self_attn/q_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.23/self_attn/q_proj/MatMul_output_0_output_quantized /model/layers.23/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.23/self_attn/k_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.23/self_attn/k_proj/MatMul_output_0_output_quantized /model/layers.23/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.23/self_attn/v_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.23/self_attn/v_proj/MatMul_output_0_output_quantized /model/layers.23/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output Concat /model/layers.23/self_attn/Concat 4 1 /model/layers.23/self_attn/Unsqueeze_output_0 /model/layers.23/self_attn/Unsqueeze_1_output_0 /model/layers.23/self_attn/Constant_2_output_0 /model/layers.23/self_attn/Constant_3_output_0 /model/layers.23/self_attn/Concat_output_0 0=0 Concat /model/layers.23/self_attn/Concat_1 4 1 /model/layers.23/self_attn/Unsqueeze_2_output_0 /model/layers.23/self_attn/Unsqueeze_3_output_0 /model/layers.23/self_attn/Constant_4_output_0 /model/layers.23/self_attn/Constant_5_output_0 /model/layers.23/self_attn/Concat_1_output_0 0=0 Concat /model/layers.23/self_attn/Concat_2 4 1 /model/layers.23/self_attn/Unsqueeze_4_output_0 /model/layers.23/self_attn/Unsqueeze_5_output_0 /model/layers.23/self_attn/Constant_6_output_0 /model/layers.23/self_attn/Constant_7_output_0 /model/layers.23/self_attn/Concat_2_output_0 0=0 Concat /model/layers.23/self_attn/Concat_11 3 1 /model/layers.23/self_attn/Unsqueeze_29_output_0 /model/layers.23/self_attn/Unsqueeze_30_output_0 /model/layers.23/self_attn/Constant_64_output_0 /model/layers.23/self_attn/Concat_11_output_0 0=0 BinaryOp /model/layers.23/self_attn/q_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.23/self_attn/q_proj/MatMul_output_0_output_quantized_cast_output /model/layers.23/self_attn/q_proj/MatMul_quant_scales_mul:0 /model/layers.23/self_attn/q_proj/MatMul_output_0 0=2 BinaryOp /model/layers.23/self_attn/k_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.23/self_attn/k_proj/MatMul_output_0_output_quantized_cast_output /model/layers.23/self_attn/k_proj/MatMul_quant_scales_mul:0 /model/layers.23/self_attn/k_proj/MatMul_output_0 0=2 BinaryOp /model/layers.23/self_attn/v_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.23/self_attn/v_proj/MatMul_output_0_output_quantized_cast_output /model/layers.23/self_attn/v_proj/MatMul_quant_scales_mul:0 /model/layers.23/self_attn/v_proj/MatMul_output_0 0=2 BinaryOp /model/layers.23/self_attn/q_proj/Add 2 1 model.layers.23.self_attn.q_proj.bias /model/layers.23/self_attn/q_proj/MatMul_output_0 /model/layers.23/self_attn/q_proj/Add_output_0 0=0 BinaryOp /model/layers.23/self_attn/k_proj/Add 2 1 model.layers.23.self_attn.k_proj.bias /model/layers.23/self_attn/k_proj/MatMul_output_0 /model/layers.23/self_attn/k_proj/Add_output_0 0=0 BinaryOp /model/layers.23/self_attn/v_proj/Add 2 1 model.layers.23.self_attn.v_proj.bias /model/layers.23/self_attn/v_proj/MatMul_output_0 /model/layers.23/self_attn/v_proj/Add_output_0 0=0 Reshape /model/layers.23/self_attn/Reshape 2 1 /model/layers.23/self_attn/q_proj/Add_output_0 /model/layers.23/self_attn/Concat_output_0 /model/layers.23/self_attn/Reshape_output_0 Reshape /model/layers.23/self_attn/Reshape_1 2 1 /model/layers.23/self_attn/k_proj/Add_output_0 /model/layers.23/self_attn/Concat_1_output_0 /model/layers.23/self_attn/Reshape_1_output_0 Reshape /model/layers.23/self_attn/Reshape_2 2 1 /model/layers.23/self_attn/v_proj/Add_output_0 /model/layers.23/self_attn/Concat_2_output_0 /model/layers.23/self_attn/Reshape_2_output_0 Permute /model/layers.23/self_attn/Transpose 1 1 /model/layers.23/self_attn/Reshape_output_0 /model/layers.23/self_attn/Transpose_output_0 0=2 Split splitncnn_739 1 4 /model/layers.23/self_attn/Transpose_output_0 /model/layers.23/self_attn/Transpose_output_0_splitncnn_0 /model/layers.23/self_attn/Transpose_output_0_splitncnn_1 /model/layers.23/self_attn/Transpose_output_0_splitncnn_2 /model/layers.23/self_attn/Transpose_output_0_splitncnn_3 Permute /model/layers.23/self_attn/Transpose_1 1 1 /model/layers.23/self_attn/Reshape_1_output_0 /model/layers.23/self_attn/Transpose_1_output_0 0=2 Split splitncnn_740 1 4 /model/layers.23/self_attn/Transpose_1_output_0 /model/layers.23/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.23/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.23/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.23/self_attn/Transpose_1_output_0_splitncnn_3 Permute /model/layers.23/self_attn/Transpose_2 1 1 /model/layers.23/self_attn/Reshape_2_output_0 /model/layers.23/self_attn/Transpose_2_output_0 0=2 BinaryOp /model/layers.23/self_attn/Mul 2 1 /model/layers.23/self_attn/Transpose_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_1 /model/layers.23/self_attn/Mul_output_0 0=2 Shape /model/layers.23/self_attn/Shape_2 1 1 /model/layers.23/self_attn/Transpose_output_0_splitncnn_2 /model/layers.23/self_attn/Shape_2_output_0 BinaryOp /model/layers.23/self_attn/Mul_2 2 1 /model/layers.23/self_attn/Transpose_1_output_0_splitncnn_3 /model/layers.0/self_attn/Unsqueeze_6_output_0_splitncnn_0 /model/layers.23/self_attn/Mul_2_output_0 0=2 Shape /model/layers.23/self_attn/Shape_3 1 1 /model/layers.23/self_attn/Transpose_1_output_0_splitncnn_2 /model/layers.23/self_attn/Shape_3_output_0 Concat /model/layers.23/self_attn/Concat_6 2 1 past_key_values.23.value /model/layers.23/self_attn/Transpose_2_output_0 present.23.value 0=-2 Split splitncnn_741 1 5 present.23.value present.23.value_splitncnn_0 present.23.value_splitncnn_1 present.23.value_splitncnn_2 present.23.value_splitncnn_3 present.23.value_splitncnn_4 Crop /model/layers.23/self_attn/Gather_2 1 1 /model/layers.23/self_attn/Shape_2_output_0 /model/layers.23/self_attn/Gather_2_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Crop /model/layers.23/self_attn/Gather_3 1 1 /model/layers.23/self_attn/Shape_3_output_0 /model/layers.23/self_attn/Gather_3_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Shape /model/layers.23/self_attn/Shape_9 1 1 present.23.value_splitncnn_4 /model/layers.23/self_attn/Shape_9_output_0 Shape /model/layers.23/self_attn/Shape_10 1 1 present.23.value_splitncnn_3 /model/layers.23/self_attn/Shape_10_output_0 Shape /model/layers.23/self_attn/Shape_11 1 1 present.23.value_splitncnn_2 /model/layers.23/self_attn/Shape_11_output_0 Shape /model/layers.23/self_attn/Shape_12 1 1 present.23.value_splitncnn_1 /model/layers.23/self_attn/Shape_12_output_0 ExpandDims /model/layers.23/self_attn/Unsqueeze_19 2 1 present.23.value_splitncnn_0 /model/layers.23/self_attn/Constant_45_output_0 /model/layers.23/self_attn/Unsqueeze_19_output_0 -23303=0 BinaryOp /model/layers.23/self_attn/Div 2 1 /model/layers.23/self_attn/Gather_2_output_0 /model/layers.23/self_attn/Constant_9_output_0 /model/layers.23/self_attn/Div_output_0 0=3 BinaryOp /model/layers.23/self_attn/Div_1 2 1 /model/layers.23/self_attn/Gather_3_output_0 /model/layers.23/self_attn/Constant_19_output_0 /model/layers.23/self_attn/Div_1_output_0 0=3 Crop /model/layers.23/self_attn/Gather_8 1 1 /model/layers.23/self_attn/Shape_9_output_0 /model/layers.23/self_attn/Gather_8_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_742 1 2 /model/layers.23/self_attn/Gather_8_output_0 /model/layers.23/self_attn/Gather_8_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_8_output_0_splitncnn_1 Crop /model/layers.23/self_attn/Gather_9 1 1 /model/layers.23/self_attn/Shape_10_output_0 /model/layers.23/self_attn/Gather_9_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_743 1 2 /model/layers.23/self_attn/Gather_9_output_0 /model/layers.23/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_9_output_0_splitncnn_1 Crop /model/layers.23/self_attn/Gather_10 1 1 /model/layers.23/self_attn/Shape_11_output_0 /model/layers.23/self_attn/Gather_10_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_744 1 2 /model/layers.23/self_attn/Gather_10_output_0 /model/layers.23/self_attn/Gather_10_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_10_output_0_splitncnn_1 Crop /model/layers.23/self_attn/Gather_11 1 1 /model/layers.23/self_attn/Shape_12_output_0 /model/layers.23/self_attn/Gather_11_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_745 1 2 /model/layers.23/self_attn/Gather_11_output_0 /model/layers.23/self_attn/Gather_11_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_11_output_0_splitncnn_1 Cast /model/layers.23/self_attn/Cast 1 1 /model/layers.23/self_attn/Div_output_0 /model/layers.23/self_attn/Cast_output_0 Cast /model/layers.23/self_attn/Cast_2 1 1 /model/layers.23/self_attn/Div_1_output_0 /model/layers.23/self_attn/Cast_2_output_0 ExpandDims /model/layers.23/self_attn/Unsqueeze_20 2 1 /model/layers.23/self_attn/Gather_8_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_46_output_0 /model/layers.23/self_attn/Unsqueeze_20_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_24 2 1 /model/layers.23/self_attn/Gather_8_output_0_splitncnn_0 onnx::Unsqueeze_7097 /model/layers.23/self_attn/Unsqueeze_24_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_21 2 1 /model/layers.23/self_attn/Gather_9_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_47_output_0 /model/layers.23/self_attn/Unsqueeze_21_output_0 -23303=0 BinaryOp /model/layers.23/self_attn/Mul_7 2 1 /model/layers.23/self_attn/Gather_9_output_0_splitncnn_0 /model/layers.23/self_attn/Constant_53_output_0 /model/layers.23/self_attn/Mul_7_output_0 0=2 ExpandDims /model/layers.23/self_attn/Unsqueeze_22 2 1 /model/layers.23/self_attn/Gather_10_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_49_output_0 /model/layers.23/self_attn/Unsqueeze_22_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_26 2 1 /model/layers.23/self_attn/Gather_10_output_0_splitncnn_0 onnx::Unsqueeze_7101 /model/layers.23/self_attn/Unsqueeze_26_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_23 2 1 /model/layers.23/self_attn/Gather_11_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_50_output_0 /model/layers.23/self_attn/Unsqueeze_23_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_27 2 1 /model/layers.23/self_attn/Gather_11_output_0_splitncnn_0 onnx::Unsqueeze_7103 /model/layers.23/self_attn/Unsqueeze_27_output_0 -23303=0 Cast /model/layers.23/self_attn/Cast_1 1 1 /model/layers.23/self_attn/Cast_output_0 /model/layers.23/self_attn/Cast_1_output_0 Split splitncnn_746 1 2 /model/layers.23/self_attn/Cast_1_output_0 /model/layers.23/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.23/self_attn/Cast_1_output_0_splitncnn_1 Cast /model/layers.23/self_attn/Cast_3 1 1 /model/layers.23/self_attn/Cast_2_output_0 /model/layers.23/self_attn/Cast_3_output_0 Split splitncnn_747 1 2 /model/layers.23/self_attn/Cast_3_output_0 /model/layers.23/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.23/self_attn/Cast_3_output_0_splitncnn_1 ExpandDims /model/layers.23/self_attn/Unsqueeze_25 2 1 /model/layers.23/self_attn/Mul_7_output_0 onnx::Unsqueeze_7099 /model/layers.23/self_attn/Unsqueeze_25_output_0 -23303=0 Concat /model/layers.23/self_attn/Concat_9 5 1 /model/layers.23/self_attn/Unsqueeze_20_output_0 /model/layers.23/self_attn/Unsqueeze_21_output_0 /model/layers.23/self_attn/Constant_48_output_0 /model/layers.23/self_attn/Unsqueeze_22_output_0 /model/layers.23/self_attn/Unsqueeze_23_output_0 /model/layers.23/self_attn/Concat_9_output_0 0=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_6 2 1 /model/layers.23/self_attn/Cast_1_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_12_output_0 /model/layers.23/self_attn/Unsqueeze_6_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_7 2 1 /model/layers.23/self_attn/Cast_1_output_0_splitncnn_0 /model/layers.23/self_attn/Constant_15_output_0 /model/layers.23/self_attn/Unsqueeze_7_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_8 2 1 /model/layers.23/self_attn/Cast_3_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_22_output_0 /model/layers.23/self_attn/Unsqueeze_8_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_9 2 1 /model/layers.23/self_attn/Cast_3_output_0_splitncnn_0 /model/layers.23/self_attn/Constant_25_output_0 /model/layers.23/self_attn/Unsqueeze_9_output_0 -23303=0 Concat /model/layers.23/self_attn/Concat_10 4 1 /model/layers.23/self_attn/Unsqueeze_24_output_0 /model/layers.23/self_attn/Unsqueeze_25_output_0 /model/layers.23/self_attn/Unsqueeze_26_output_0 /model/layers.23/self_attn/Unsqueeze_27_output_0 /model/layers.23/self_attn/Concat_10_output_0 0=0 Reshape /model/layers.23/self_attn/Reshape_5 1 1 /model/layers.23/self_attn/Concat_9_output_0 /model/layers.23/self_attn/Reshape_5_output_0 0=-1 Split splitncnn_748 1 3 /model/layers.23/self_attn/Reshape_5_output_0 /model/layers.23/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.23/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.23/self_attn/Reshape_5_output_0_splitncnn_2 Crop /model/layers.23/self_attn/Slice 2 1 /model/layers.23/self_attn/Transpose_output_0_splitncnn_1 /model/layers.23/self_attn/Unsqueeze_6_output_0 /model/layers.23/self_attn/Slice_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.23/self_attn/Slice_1 2 1 /model/layers.23/self_attn/Transpose_output_0_splitncnn_0 /model/layers.23/self_attn/Unsqueeze_7_output_0 /model/layers.23/self_attn/Slice_1_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Crop /model/layers.23/self_attn/Slice_2 2 1 /model/layers.23/self_attn/Transpose_1_output_0_splitncnn_1 /model/layers.23/self_attn/Unsqueeze_8_output_0 /model/layers.23/self_attn/Slice_2_output_0 -23309=1,0 -23310=0 -23311=1,2 Crop /model/layers.23/self_attn/Slice_3 2 1 /model/layers.23/self_attn/Transpose_1_output_0_splitncnn_0 /model/layers.23/self_attn/Unsqueeze_9_output_0 /model/layers.23/self_attn/Slice_3_output_0 -23309=0 -23310=1,2147483647 -23311=1,2 Shape /model/layers.23/self_attn/Shape_13 1 1 /model/layers.23/self_attn/Reshape_5_output_0_splitncnn_2 /model/layers.23/self_attn/Shape_13_output_0 UnaryOp /model/layers.23/self_attn/Neg 1 1 /model/layers.23/self_attn/Slice_1_output_0 /model/layers.23/self_attn/Neg_output_0 0=1 UnaryOp /model/layers.23/self_attn/Neg_1 1 1 /model/layers.23/self_attn/Slice_3_output_0 /model/layers.23/self_attn/Neg_1_output_0 0=1 ConstantOfShape /model/layers.23/self_attn/ConstantOfShape_1 1 1 /model/layers.23/self_attn/Shape_13_output_0 /model/layers.23/self_attn/ConstantOfShape_1_output_0 Split splitncnn_749 1 2 /model/layers.23/self_attn/ConstantOfShape_1_output_0 /model/layers.23/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.23/self_attn/ConstantOfShape_1_output_0_splitncnn_1 Concat /model/layers.23/self_attn/Concat_3 2 1 /model/layers.23/self_attn/Neg_output_0 /model/layers.23/self_attn/Slice_output_0 /model/layers.23/self_attn/Concat_3_output_0 0=-1 Concat /model/layers.23/self_attn/Concat_4 2 1 /model/layers.23/self_attn/Neg_1_output_0 /model/layers.23/self_attn/Slice_2_output_0 /model/layers.23/self_attn/Concat_4_output_0 0=-1 BinaryOp /model/layers.23/self_attn/Mul_6 2 1 /model/layers.23/self_attn/ConstantOfShape_1_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_52_output_0 /model/layers.23/self_attn/Mul_6_output_0 0=2 BinaryOp /model/layers.23/self_attn/Mul_1 2 1 /model/layers.23/self_attn/Concat_3_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_1 /model/layers.23/self_attn/Mul_1_output_0 0=2 BinaryOp /model/layers.23/self_attn/Mul_3 2 1 /model/layers.23/self_attn/Concat_4_output_0 /model/layers.0/self_attn/Unsqueeze_7_output_0_splitncnn_0 /model/layers.23/self_attn/Mul_3_output_0 0=2 Equal /model/layers.23/self_attn/Equal_1 2 1 /model/layers.23/self_attn/Reshape_5_output_0_splitncnn_1 /model/layers.23/self_attn/Mul_6_output_0 /model/layers.23/self_attn/Equal_1_output_0 BinaryOp /model/layers.23/self_attn/Add 2 1 /model/layers.23/self_attn/Mul_output_0 /model/layers.23/self_attn/Mul_1_output_0 /model/layers.23/self_attn/Add_output_0 0=0 BinaryOp /model/layers.23/self_attn/Add_1 2 1 /model/layers.23/self_attn/Mul_2_output_0 /model/layers.23/self_attn/Mul_3_output_0 /model/layers.23/self_attn/Add_1_output_0 0=0 Where /model/layers.23/self_attn/Where_1 3 1 /model/layers.23/self_attn/Equal_1_output_0 /model/layers.23/self_attn/ConstantOfShape_1_output_0_splitncnn_0 /model/layers.23/self_attn/Reshape_5_output_0_splitncnn_0 /model/layers.23/self_attn/Where_1_output_0 BinaryOp /model/layers.23/self_attn/Mul_8 1 1 /model/layers.23/self_attn/Add_output_0 /model/layers.23/self_attn/Mul_8_output_0 0=2 1=1 2=3.535534e-01 Concat /model/layers.23/self_attn/Concat_5 2 1 past_key_values.23.key /model/layers.23/self_attn/Add_1_output_0 present.23.key 0=-2 Split splitncnn_750 1 5 present.23.key present.23.key_splitncnn_0 present.23.key_splitncnn_1 present.23.key_splitncnn_2 present.23.key_splitncnn_3 present.23.key_splitncnn_4 Expand /model/layers.23/self_attn/Expand_1 2 1 /model/layers.23/self_attn/Unsqueeze_19_output_0 /model/layers.23/self_attn/Where_1_output_0 /model/layers.23/self_attn/Expand_1_output_0 Shape /model/layers.23/self_attn/Shape_4 1 1 present.23.key_splitncnn_4 /model/layers.23/self_attn/Shape_4_output_0 Shape /model/layers.23/self_attn/Shape_5 1 1 present.23.key_splitncnn_3 /model/layers.23/self_attn/Shape_5_output_0 Shape /model/layers.23/self_attn/Shape_6 1 1 present.23.key_splitncnn_2 /model/layers.23/self_attn/Shape_6_output_0 Shape /model/layers.23/self_attn/Shape_7 1 1 present.23.key_splitncnn_1 /model/layers.23/self_attn/Shape_7_output_0 ExpandDims /model/layers.23/self_attn/Unsqueeze_10 2 1 present.23.key_splitncnn_0 /model/layers.23/self_attn/Constant_32_output_0 /model/layers.23/self_attn/Unsqueeze_10_output_0 -23303=0 Reshape /model/layers.23/self_attn/Reshape_6 2 1 /model/layers.23/self_attn/Expand_1_output_0 /model/layers.23/self_attn/Concat_10_output_0 /model/layers.23/self_attn/Reshape_6_output_0 Crop /model/layers.23/self_attn/Gather_4 1 1 /model/layers.23/self_attn/Shape_4_output_0 /model/layers.23/self_attn/Gather_4_output_0 -23309=1,0 -23310=1,1 -23311=1,-1 Split splitncnn_751 1 2 /model/layers.23/self_attn/Gather_4_output_0 /model/layers.23/self_attn/Gather_4_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_4_output_0_splitncnn_1 Crop /model/layers.23/self_attn/Gather_5 1 1 /model/layers.23/self_attn/Shape_5_output_0 /model/layers.23/self_attn/Gather_5_output_0 -23309=1,1 -23310=1,2 -23311=1,-1 Split splitncnn_752 1 2 /model/layers.23/self_attn/Gather_5_output_0 /model/layers.23/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_5_output_0_splitncnn_1 Crop /model/layers.23/self_attn/Gather_6 1 1 /model/layers.23/self_attn/Shape_6_output_0 /model/layers.23/self_attn/Gather_6_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 Split splitncnn_753 1 2 /model/layers.23/self_attn/Gather_6_output_0 /model/layers.23/self_attn/Gather_6_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_6_output_0_splitncnn_1 Crop /model/layers.23/self_attn/Gather_7 1 1 /model/layers.23/self_attn/Shape_7_output_0 /model/layers.23/self_attn/Gather_7_output_0 -23309=1,3 -23310=1,4 -23311=1,-1 Split splitncnn_754 1 2 /model/layers.23/self_attn/Gather_7_output_0 /model/layers.23/self_attn/Gather_7_output_0_splitncnn_0 /model/layers.23/self_attn/Gather_7_output_0_splitncnn_1 ExpandDims /model/layers.23/self_attn/Unsqueeze_11 2 1 /model/layers.23/self_attn/Gather_4_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_33_output_0 /model/layers.23/self_attn/Unsqueeze_11_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_15 2 1 /model/layers.23/self_attn/Gather_4_output_0_splitncnn_0 onnx::Unsqueeze_7051 /model/layers.23/self_attn/Unsqueeze_15_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_12 2 1 /model/layers.23/self_attn/Gather_5_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_34_output_0 /model/layers.23/self_attn/Unsqueeze_12_output_0 -23303=0 BinaryOp /model/layers.23/self_attn/Mul_5 2 1 /model/layers.23/self_attn/Gather_5_output_0_splitncnn_0 /model/layers.23/self_attn/Constant_40_output_0 /model/layers.23/self_attn/Mul_5_output_0 0=2 ExpandDims /model/layers.23/self_attn/Unsqueeze_13 2 1 /model/layers.23/self_attn/Gather_6_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_36_output_0 /model/layers.23/self_attn/Unsqueeze_13_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_17 2 1 /model/layers.23/self_attn/Gather_6_output_0_splitncnn_0 onnx::Unsqueeze_7055 /model/layers.23/self_attn/Unsqueeze_17_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_14 2 1 /model/layers.23/self_attn/Gather_7_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_37_output_0 /model/layers.23/self_attn/Unsqueeze_14_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_18 2 1 /model/layers.23/self_attn/Gather_7_output_0_splitncnn_0 onnx::Unsqueeze_7057 /model/layers.23/self_attn/Unsqueeze_18_output_0 -23303=0 ExpandDims /model/layers.23/self_attn/Unsqueeze_16 2 1 /model/layers.23/self_attn/Mul_5_output_0 onnx::Unsqueeze_7053 /model/layers.23/self_attn/Unsqueeze_16_output_0 -23303=0 Concat /model/layers.23/self_attn/Concat_7 5 1 /model/layers.23/self_attn/Unsqueeze_11_output_0 /model/layers.23/self_attn/Unsqueeze_12_output_0 /model/layers.23/self_attn/Constant_35_output_0 /model/layers.23/self_attn/Unsqueeze_13_output_0 /model/layers.23/self_attn/Unsqueeze_14_output_0 /model/layers.23/self_attn/Concat_7_output_0 0=0 Concat /model/layers.23/self_attn/Concat_8 4 1 /model/layers.23/self_attn/Unsqueeze_15_output_0 /model/layers.23/self_attn/Unsqueeze_16_output_0 /model/layers.23/self_attn/Unsqueeze_17_output_0 /model/layers.23/self_attn/Unsqueeze_18_output_0 /model/layers.23/self_attn/Concat_8_output_0 0=0 Reshape /model/layers.23/self_attn/Reshape_3 1 1 /model/layers.23/self_attn/Concat_7_output_0 /model/layers.23/self_attn/Reshape_3_output_0 0=-1 Split splitncnn_755 1 3 /model/layers.23/self_attn/Reshape_3_output_0 /model/layers.23/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.23/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.23/self_attn/Reshape_3_output_0_splitncnn_2 Shape /model/layers.23/self_attn/Shape_8 1 1 /model/layers.23/self_attn/Reshape_3_output_0_splitncnn_2 /model/layers.23/self_attn/Shape_8_output_0 ConstantOfShape /model/layers.23/self_attn/ConstantOfShape 1 1 /model/layers.23/self_attn/Shape_8_output_0 /model/layers.23/self_attn/ConstantOfShape_output_0 Split splitncnn_756 1 2 /model/layers.23/self_attn/ConstantOfShape_output_0 /model/layers.23/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.23/self_attn/ConstantOfShape_output_0_splitncnn_1 BinaryOp /model/layers.23/self_attn/Mul_4 2 1 /model/layers.23/self_attn/ConstantOfShape_output_0_splitncnn_1 /model/layers.23/self_attn/Constant_39_output_0 /model/layers.23/self_attn/Mul_4_output_0 0=2 Equal /model/layers.23/self_attn/Equal 2 1 /model/layers.23/self_attn/Reshape_3_output_0_splitncnn_1 /model/layers.23/self_attn/Mul_4_output_0 /model/layers.23/self_attn/Equal_output_0 Where /model/layers.23/self_attn/Where 3 1 /model/layers.23/self_attn/Equal_output_0 /model/layers.23/self_attn/ConstantOfShape_output_0_splitncnn_0 /model/layers.23/self_attn/Reshape_3_output_0_splitncnn_0 /model/layers.23/self_attn/Where_output_0 Expand /model/layers.23/self_attn/Expand 2 1 /model/layers.23/self_attn/Unsqueeze_10_output_0 /model/layers.23/self_attn/Where_output_0 /model/layers.23/self_attn/Expand_output_0 Reshape /model/layers.23/self_attn/Reshape_4 2 1 /model/layers.23/self_attn/Expand_output_0 /model/layers.23/self_attn/Concat_8_output_0 /model/layers.23/self_attn/Reshape_4_output_0 Split splitncnn_757 1 2 /model/layers.23/self_attn/Reshape_4_output_0 /model/layers.23/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.23/self_attn/Reshape_4_output_0_splitncnn_1 Shape /model/layers.23/self_attn/Shape_14 1 1 /model/layers.23/self_attn/Reshape_4_output_0_splitncnn_1 /model/layers.23/self_attn/Shape_14_output_0 Permute /model/layers.23/self_attn/Transpose_3 1 1 /model/layers.23/self_attn/Reshape_4_output_0_splitncnn_0 /model/layers.23/self_attn/Transpose_3_output_0 0=1 Crop /model/layers.23/self_attn/Gather_12 1 1 /model/layers.23/self_attn/Shape_14_output_0 /model/layers.23/self_attn/Gather_12_output_0 -23309=1,2 -23310=1,3 -23311=1,-1 BinaryOp /model/layers.23/self_attn/Mul_9 1 1 /model/layers.23/self_attn/Transpose_3_output_0 /model/layers.23/self_attn/Mul_9_output_0 0=2 1=1 2=3.535534e-01 ExpandDims /model/layers.23/self_attn/Unsqueeze_28 2 1 /model/layers.23/self_attn/Gather_12_output_0 /model/layers.23/self_attn/Constant_57_output_0 /model/layers.23/self_attn/Unsqueeze_28_output_0 -23303=0 Gemm /model/layers.23/self_attn/MatMul 2 1 /model/layers.23/self_attn/Mul_8_output_0 /model/layers.23/self_attn/Mul_9_output_0 /model/layers.23/self_attn/MatMul_output_0 Crop /model/layers.23/self_attn/Slice_4 2 1 /model/Expand_output_0_splitncnn_0 /model/layers.23/self_attn/Unsqueeze_28_output_0 /model/layers.23/self_attn/Slice_4_output_0 -23309=1,0 -23310=0 -23311=1,2 Where /model/layers.23/self_attn/Where_2 3 1 /model/layers.23/self_attn/Slice_4_output_0 /model/layers.23/self_attn/Constant_61_output_0 /model/layers.23/self_attn/Constant_62_output_0 /model/layers.23/self_attn/Where_2_output_0 BinaryOp /model/layers.23/self_attn/Add_2 2 1 /model/layers.23/self_attn/MatMul_output_0 /model/layers.23/self_attn/Where_2_output_0 /model/layers.23/self_attn/Add_2_output_0 0=0 Softmax /model/layers.23/self_attn/Softmax 1 1 /model/layers.23/self_attn/Add_2_output_0 /model/layers.23/self_attn/Softmax_output_0 0=-2 1=1 Split splitncnn_758 1 2 /model/layers.23/self_attn/Softmax_output_0 /model/layers.23/self_attn/Softmax_output_0_splitncnn_0 /model/layers.23/self_attn/Softmax_output_0_splitncnn_1 IsNaN /model/layers.23/self_attn/IsNaN 1 1 /model/layers.23/self_attn/Softmax_output_0_splitncnn_1 /model/layers.23/self_attn/IsNaN_output_0 Where /model/layers.23/self_attn/Where_3 3 1 /model/layers.23/self_attn/IsNaN_output_0 /model/layers.23/self_attn/Constant_63_output_0 /model/layers.23/self_attn/Softmax_output_0_splitncnn_0 /model/layers.23/self_attn/Where_3_output_0 Gemm /model/layers.23/self_attn/MatMul_1 2 1 /model/layers.23/self_attn/Where_3_output_0 /model/layers.23/self_attn/Reshape_6_output_0 /model/layers.23/self_attn/MatMul_1_output_0 Permute /model/layers.23/self_attn/Transpose_4 1 1 /model/layers.23/self_attn/MatMul_1_output_0 /model/layers.23/self_attn/Transpose_4_output_0 0=2 Reshape /model/layers.23/self_attn/Reshape_7 2 1 /model/layers.23/self_attn/Transpose_4_output_0 /model/layers.23/self_attn/Concat_11_output_0 /model/layers.23/self_attn/Reshape_7_output_0 DynamicQuantizeLinear /model/layers.23/self_attn/Reshape_7_output_0_QuantizeLinear 1 3 /model/layers.23/self_attn/Reshape_7_output_0 /model/layers.23/self_attn/Reshape_7_output_0_quantized /model/layers.23/self_attn/Reshape_7_output_0_scale /model/layers.23/self_attn/Reshape_7_output_0_zero_point BinaryOp /model/layers.23/self_attn/o_proj/MatMul_quant_scales_mul 1 1 /model/layers.23/self_attn/Reshape_7_output_0_scale /model/layers.23/self_attn/o_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=7.258858e-03 MatMulInteger /model/layers.23/self_attn/o_proj/MatMul_quant 4 1 /model/layers.23/self_attn/Reshape_7_output_0_quantized onnx::MatMul_8026_quantized /model/layers.23/self_attn/Reshape_7_output_0_zero_point onnx::MatMul_8026_zero_point /model/layers.23/self_attn/o_proj/MatMul_output_0_output_quantized Cast /model/layers.23/self_attn/o_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.23/self_attn/o_proj/MatMul_output_0_output_quantized /model/layers.23/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.23/self_attn/o_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.23/self_attn/o_proj/MatMul_output_0_output_quantized_cast_output /model/layers.23/self_attn/o_proj/MatMul_quant_scales_mul:0 /model/layers.23/self_attn/o_proj/MatMul_output_0 0=2 BinaryOp /model/layers.23/Add 2 1 /model/layers.23/input_layernorm/Cast_output_0_splitncnn_0 /model/layers.23/self_attn/o_proj/MatMul_output_0 /model/layers.23/Add_output_0 0=0 Cast /model/layers.23/post_attention_layernorm/Cast 1 1 /model/layers.23/Add_output_0 /model/layers.23/post_attention_layernorm/Cast_output_0 Split splitncnn_759 1 3 /model/layers.23/post_attention_layernorm/Cast_output_0 /model/layers.23/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.23/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.23/post_attention_layernorm/Cast_output_0_splitncnn_2 BinaryOp /model/layers.23/post_attention_layernorm/Pow 1 1 /model/layers.23/post_attention_layernorm/Cast_output_0_splitncnn_2 /model/layers.23/post_attention_layernorm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/layers.23/post_attention_layernorm/ReduceMean 2 1 /model/layers.23/post_attention_layernorm/Pow_output_0 /model/layers.23/post_attention_layernorm/Constant_1_output_0 /model/layers.23/post_attention_layernorm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/layers.23/post_attention_layernorm/Add 1 1 /model/layers.23/post_attention_layernorm/ReduceMean_output_0 /model/layers.23/post_attention_layernorm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/layers.23/post_attention_layernorm/Sqrt 1 1 /model/layers.23/post_attention_layernorm/Add_output_0 /model/layers.23/post_attention_layernorm/Sqrt_output_0 0=5 BinaryOp /model/layers.23/post_attention_layernorm/Div 1 1 /model/layers.23/post_attention_layernorm/Sqrt_output_0 /model/layers.23/post_attention_layernorm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/layers.23/post_attention_layernorm/Mul 2 1 /model/layers.23/post_attention_layernorm/Cast_output_0_splitncnn_1 /model/layers.23/post_attention_layernorm/Div_output_0 /model/layers.23/post_attention_layernorm/Mul_output_0 0=2 Cast /model/layers.23/post_attention_layernorm/Cast_1 1 1 /model/layers.23/post_attention_layernorm/Mul_output_0 /model/layers.23/post_attention_layernorm/Cast_1_output_0 BinaryOp /model/layers.23/post_attention_layernorm/Mul_1 2 1 model.layers.23.post_attention_layernorm.weight /model/layers.23/post_attention_layernorm/Cast_1_output_0 /model/layers.23/post_attention_layernorm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/layers.23/post_attention_layernorm/Mul_1_output_0_QuantizeLinear 1 3 /model/layers.23/post_attention_layernorm/Mul_1_output_0 /model/layers.23/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.23/post_attention_layernorm/Mul_1_output_0_scale /model/layers.23/post_attention_layernorm/Mul_1_output_0_zero_point Split splitncnn_760 1 2 /model/layers.23/post_attention_layernorm/Mul_1_output_0_quantized /model/layers.23/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 /model/layers.23/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 Split splitncnn_761 1 2 /model/layers.23/post_attention_layernorm/Mul_1_output_0_scale /model/layers.23/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.23/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 Split splitncnn_762 1 2 /model/layers.23/post_attention_layernorm/Mul_1_output_0_zero_point /model/layers.23/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 /model/layers.23/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 BinaryOp /model/layers.23/mlp/gate_proj/MatMul_quant_scales_mul 1 1 /model/layers.23/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_1 /model/layers.23/mlp/gate_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.814346e-03 BinaryOp /model/layers.23/mlp/up_proj/MatMul_quant_scales_mul 1 1 /model/layers.23/post_attention_layernorm/Mul_1_output_0_scale_splitncnn_0 /model/layers.23/mlp/up_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=2.337598e-03 MatMulInteger /model/layers.23/mlp/gate_proj/MatMul_quant 4 1 /model/layers.23/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_1 onnx::MatMul_8027_quantized /model/layers.23/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_1 onnx::MatMul_8027_zero_point /model/layers.23/mlp/gate_proj/MatMul_output_0_output_quantized MatMulInteger /model/layers.23/mlp/up_proj/MatMul_quant 4 1 /model/layers.23/post_attention_layernorm/Mul_1_output_0_quantized_splitncnn_0 onnx::MatMul_8028_quantized /model/layers.23/post_attention_layernorm/Mul_1_output_0_zero_point_splitncnn_0 onnx::MatMul_8028_zero_point /model/layers.23/mlp/up_proj/MatMul_output_0_output_quantized Cast /model/layers.23/mlp/gate_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.23/mlp/gate_proj/MatMul_output_0_output_quantized /model/layers.23/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output Cast /model/layers.23/mlp/up_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.23/mlp/up_proj/MatMul_output_0_output_quantized /model/layers.23/mlp/up_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.23/mlp/gate_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.23/mlp/gate_proj/MatMul_output_0_output_quantized_cast_output /model/layers.23/mlp/gate_proj/MatMul_quant_scales_mul:0 /model/layers.23/mlp/gate_proj/MatMul_output_0 0=2 BinaryOp /model/layers.23/mlp/up_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.23/mlp/up_proj/MatMul_output_0_output_quantized_cast_output /model/layers.23/mlp/up_proj/MatMul_quant_scales_mul:0 /model/layers.23/mlp/up_proj/MatMul_output_0 0=2 Swish /model/layers.23/mlp/act_fn/Mul 1 1 /model/layers.23/mlp/gate_proj/MatMul_output_0 /model/layers.23/mlp/act_fn/Mul_output_0 BinaryOp /model/layers.23/mlp/Mul 2 1 /model/layers.23/mlp/act_fn/Mul_output_0 /model/layers.23/mlp/up_proj/MatMul_output_0 /model/layers.23/mlp/Mul_output_0 0=2 DynamicQuantizeLinear /model/layers.23/mlp/Mul_output_0_QuantizeLinear 1 3 /model/layers.23/mlp/Mul_output_0 /model/layers.23/mlp/Mul_output_0_quantized /model/layers.23/mlp/Mul_output_0_scale /model/layers.23/mlp/Mul_output_0_zero_point BinaryOp /model/layers.23/mlp/down_proj/MatMul_quant_scales_mul 1 1 /model/layers.23/mlp/Mul_output_0_scale /model/layers.23/mlp/down_proj/MatMul_quant_scales_mul:0 0=2 1=1 2=4.152313e-03 MatMulInteger /model/layers.23/mlp/down_proj/MatMul_quant 4 1 /model/layers.23/mlp/Mul_output_0_quantized onnx::MatMul_8029_quantized /model/layers.23/mlp/Mul_output_0_zero_point onnx::MatMul_8029_zero_point /model/layers.23/mlp/down_proj/MatMul_output_0_output_quantized Cast /model/layers.23/mlp/down_proj/MatMul_output_0_output_quantized_cast 1 1 /model/layers.23/mlp/down_proj/MatMul_output_0_output_quantized /model/layers.23/mlp/down_proj/MatMul_output_0_output_quantized_cast_output BinaryOp /model/layers.23/mlp/down_proj/MatMul_quant_output_scale_mul 2 1 /model/layers.23/mlp/down_proj/MatMul_output_0_output_quantized_cast_output /model/layers.23/mlp/down_proj/MatMul_quant_scales_mul:0 /model/layers.23/mlp/down_proj/MatMul_output_0 0=2 BinaryOp /model/layers.23/Add_1 2 1 /model/layers.23/post_attention_layernorm/Cast_output_0_splitncnn_0 /model/layers.23/mlp/down_proj/MatMul_output_0 /model/layers.23/Add_1_output_0 0=0 Cast /model/norm/Cast 1 1 /model/layers.23/Add_1_output_0 /model/norm/Cast_output_0 Split splitncnn_763 1 2 /model/norm/Cast_output_0 /model/norm/Cast_output_0_splitncnn_0 /model/norm/Cast_output_0_splitncnn_1 BinaryOp /model/norm/Pow 1 1 /model/norm/Cast_output_0_splitncnn_1 /model/norm/Pow_output_0 0=6 1=1 2=2.000000e+00 Reduction /model/norm/ReduceMean 2 1 /model/norm/Pow_output_0 /model/norm/Constant_1_output_0 /model/norm/ReduceMean_output_0 0=3 1=1 4=1 5=1 BinaryOp /model/norm/Add 1 1 /model/norm/ReduceMean_output_0 /model/norm/Add_output_0 0=0 1=1 2=1.000000e-06 UnaryOp /model/norm/Sqrt 1 1 /model/norm/Add_output_0 /model/norm/Sqrt_output_0 0=5 BinaryOp /model/norm/Div 1 1 /model/norm/Sqrt_output_0 /model/norm/Div_output_0 0=8 1=1 2=1.000000e+00 BinaryOp /model/norm/Mul 2 1 /model/norm/Cast_output_0_splitncnn_0 /model/norm/Div_output_0 /model/norm/Mul_output_0 0=2 Cast /model/norm/Cast_1 1 1 /model/norm/Mul_output_0 /model/norm/Cast_1_output_0 BinaryOp /model/norm/Mul_1 2 1 model.norm.weight /model/norm/Cast_1_output_0 /model/norm/Mul_1_output_0 0=2 DynamicQuantizeLinear /model/norm/Mul_1_output_0_QuantizeLinear 1 3 /model/norm/Mul_1_output_0 /model/norm/Mul_1_output_0_quantized /model/norm/Mul_1_output_0_scale /model/norm/Mul_1_output_0_zero_point BinaryOp /lm_head/MatMul_quant_scales_mul 1 1 /model/norm/Mul_1_output_0_scale /lm_head/MatMul_quant_scales_mul:0 0=2 1=1 2=1.584030e-03 MatMulInteger /lm_head/MatMul_quant 4 1 /model/norm/Mul_1_output_0_quantized onnx::MatMul_8030_quantized /model/norm/Mul_1_output_0_zero_point onnx::MatMul_8030_zero_point logits_output_quantized Cast logits_output_quantized_cast 1 1 logits_output_quantized logits_output_quantized_cast_output BinaryOp /lm_head/MatMul_quant_output_scale_mul 2 1 logits_output_quantized_cast_output /lm_head/MatMul_quant_scales_mul:0 logits 0=2