/model/model/Shape_1:Shape*input:input_5F_ids(1,0)*output:_2F_model_2F_model_2F_Shape_5F_1_5F_output_5F_0(2) /model/model/Gather_1:Gather*input:_2F_model_2F_model_2F_Shape_5F_1_5F_output_5F_0(2);_2F_model_2F_model_2F_Constant_5F_1_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_Gather_5F_1_5F_output_5F_0()*axis:0 /model/model/Shape_2:Shape*input:pkv0(1,4,0,64)*output:_2F_model_2F_model_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/Gather_2:Gather*input:_2F_model_2F_model_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_Gather_5F_1_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_Unsqueeze_5F_output_5F_0(1) /model/model/Concat:Concat*input:_2F_model_2F_model_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Unsqueeze_5F_output_5F_0(1)*output:_2F_model_2F_model_2F_Concat_5F_output_5F_0(2)*axis:0 /model/model/Reshape:Reshape*input:position_5F_ids(1,0);_2F_model_2F_model_2F_Concat_5F_output_5F_0(2)*output:_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*allowzero:0 /model/model/embed_tokens/Gather:Gather*input:model_2E_model_2E_embed_5F_tokens_2E_weight.bin(float16:32003,2048);input_5F_ids(1,0)*output:_2F_model_2F_model_2F_embed_5F_tokens_2F_Gather_5F_output_5F_0(1,0,2048) /model/model/Concat_1:Concat*input:_2F_model_2F_model_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_Unsqueeze_5F_output_5F_0(1)*output:_2F_model_2F_model_2F_Concat_5F_1_5F_output_5F_0(2)*axis:0 /model/model/ConstantOfShape:ConstantOfShape*input:_2F_model_2F_model_2F_Concat_5F_1_5F_output_5F_0(2)*output:_2F_model_2F_model_2F_ConstantOfShape_5F_output_5F_0(0,0)*value:-3.4028234663852886e+38 /model/model/Shape_3:Shape*input:_2F_model_2F_model_2F_ConstantOfShape_5F_output_5F_0(0,0)*output:_2F_model_2F_model_2F_Shape_5F_3_5F_output_5F_0(2) /model/model/Slice:Slice*input:_2F_model_2F_model_2F_Shape_5F_3_5F_output_5F_0(2);_2F_model_2F_model_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_Slice_5F_output_5F_0(1) /model/model/Squeeze:Squeeze*input:_2F_model_2F_model_2F_Slice_5F_output_5F_0(1);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_Squeeze_5F_output_5F_0() /model/model/Range:Range*input:_2F_model_2F_model_2F_Constant_5F_8_5F_output_5F_0.bin(int64:);_2F_model_2F_model_2F_Squeeze_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_1_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_Range_5F_output_5F_0(0) /model/model/Add:Add*input:_2F_model_2F_model_2F_Range_5F_output_5F_0(0);_2F_model_2F_model_2F_Constant_5F_1_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_Add_5F_output_5F_0(0) /model/model/Reshape_1:Reshape*input:_2F_model_2F_model_2F_Add_5F_output_5F_0(0);3681.bin(int64:2)*output:_2F_model_2F_model_2F_Reshape_5F_1_5F_output_5F_0(0,1)*allowzero:0 /model/model/Less:Less*input:_2F_model_2F_model_2F_Range_5F_output_5F_0(0);_2F_model_2F_model_2F_Reshape_5F_1_5F_output_5F_0(0,1)*output:_2F_model_2F_model_2F_Less_5F_output_5F_0(0,0) /model/model/Where:Where*input:_2F_model_2F_model_2F_Less_5F_output_5F_0(0,0);_2F_model_2F_model_2F_Constant_5F_16_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_ConstantOfShape_5F_output_5F_0(0,0)*output:_2F_model_2F_model_2F_Where_5F_output_5F_0(0,0) /model/model/Unsqueeze_5:Unsqueeze*input:_2F_model_2F_model_2F_Gather_5F_2_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_Unsqueeze_5F_5_5F_output_5F_0(1) /model/model/Concat_3:Concat*input:_2F_model_2F_model_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_Unsqueeze_5F_5_5F_output_5F_0(1)*output:_2F_model_2F_model_2F_Concat_5F_3_5F_output_5F_0(2)*axis:0 /model/model/ConstantOfShape_1:ConstantOfShape*input:_2F_model_2F_model_2F_Concat_5F_3_5F_output_5F_0(2)*output:_2F_model_2F_model_2F_ConstantOfShape_5F_1_5F_output_5F_0(0,0)*value:0.0 /model/model/Concat_4:Concat*input:_2F_model_2F_model_2F_ConstantOfShape_5F_1_5F_output_5F_0(0,0);_2F_model_2F_model_2F_Where_5F_output_5F_0(0,0)*output:_2F_model_2F_model_2F_Concat_5F_4_5F_output_5F_0(0,0)*axis:-1 /model/model/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_Concat_5F_4_5F_output_5F_0(0,0);6390.bin(int64:2)*output:_2F_model_2F_model_2F_Unsqueeze_5F_7_5F_output_5F_0(1,1,0,0) /model/model/Add_1:Add*input:_2F_model_2F_model_2F_Gather_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Gather_5F_2_5F_output_5F_0()*output:_2F_model_2F_model_2F_Add_5F_1_5F_output_5F_0() /model/model/Unsqueeze_10:Unsqueeze*input:_2F_model_2F_model_2F_Add_5F_1_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_Unsqueeze_5F_10_5F_output_5F_0(1) /model/model/Concat_5:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_Unsqueeze_5F_10_5F_output_5F_0(1)*output:_2F_model_2F_model_2F_Concat_5F_5_5F_output_5F_0(4)*axis:0 /model/model/Equal:Equal*input:_2F_model_2F_model_2F_Concat_5F_5_5F_output_5F_0(4);_2F_model_2F_model_2F_Mul_5F_output_5F_0.bin(int64:4)*output:_2F_model_2F_model_2F_Equal_5F_output_5F_0(4) /model/model/Where_1:Where*input:_2F_model_2F_model_2F_Equal_5F_output_5F_0(4);_2F_model_2F_model_2F_ConstantOfShape_5F_2_5F_output_5F_0.bin(int64:4);_2F_model_2F_model_2F_Concat_5F_5_5F_output_5F_0(4)*output:_2F_model_2F_model_2F_Where_5F_1_5F_output_5F_0(4) /model/model/Expand:Expand*input:_2F_model_2F_model_2F_Unsqueeze_5F_7_5F_output_5F_0(1,1,0,0);_2F_model_2F_model_2F_Where_5F_1_5F_output_5F_0(4)*output:_2F_model_2F_model_2F_Expand_5F_output_5F_0(0,0,0,0) /model/model/Shape_7:Shape*input:attention_5F_mask(1,0)*output:_2F_model_2F_model_2F_Shape_5F_7_5F_output_5F_0(2) /model/model/Gather_4:Gather*input:_2F_model_2F_model_2F_Shape_5F_7_5F_output_5F_0(2);_2F_model_2F_model_2F_Constant_5F_1_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/Unsqueeze_12:Unsqueeze*input:attention_5F_mask(1,0);6478.bin(int64:2)*output:_2F_model_2F_model_2F_Unsqueeze_5F_12_5F_output_5F_0(1,1,1,0) /model/model/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_Gather_5F_4_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/Concat_6:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_Unsqueeze_5F_15_5F_output_5F_0(1)*output:_2F_model_2F_model_2F_Concat_5F_6_5F_output_5F_0(4)*axis:0 /model/model/Equal_1:Equal*input:_2F_model_2F_model_2F_Concat_5F_6_5F_output_5F_0(4);_2F_model_2F_model_2F_Mul_5F_output_5F_0.bin(int64:4)*output:_2F_model_2F_model_2F_Equal_5F_1_5F_output_5F_0(4) /model/model/Where_2:Where*input:_2F_model_2F_model_2F_Equal_5F_1_5F_output_5F_0(4);_2F_model_2F_model_2F_ConstantOfShape_5F_2_5F_output_5F_0.bin(int64:4);_2F_model_2F_model_2F_Concat_5F_6_5F_output_5F_0(4)*output:_2F_model_2F_model_2F_Where_5F_2_5F_output_5F_0(4) /model/model/Expand_1:Expand*input:_2F_model_2F_model_2F_Unsqueeze_5F_12_5F_output_5F_0(1,1,1,0);_2F_model_2F_model_2F_Where_5F_2_5F_output_5F_0(4)*output:_2F_model_2F_model_2F_Expand_5F_1_5F_output_5F_0(0,0,0,0) /model/model/Cast_4:Cast*input:_2F_model_2F_model_2F_Expand_5F_1_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_Cast_5F_4_5F_output_5F_0(0,0,0,0)*to:1 /model/model/Sub:Sub*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_Cast_5F_4_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_Sub_5F_output_5F_0(0,0,0,0) /model/model/Cast_5:Cast*input:_2F_model_2F_model_2F_Sub_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_Cast_5F_5_5F_output_5F_0(0,0,0,0)*to:9 /model/model/Where_3:Where*input:_2F_model_2F_model_2F_Cast_5F_5_5F_output_5F_0(0,0,0,0);_2F_model_2F_model_2F_Constant_5F_36_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_Sub_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_Where_5F_3_5F_output_5F_0(0,0,0,0) /model/model/Add_2:Add*input:_2F_model_2F_model_2F_Where_5F_3_5F_output_5F_0(0,0,0,0);_2F_model_2F_model_2F_Expand_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0) /model/model/layers.0/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_embed_5F_tokens_2F_Gather_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.0/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.0/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.0/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.0/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.0/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_embed_5F_tokens_2F_Gather_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.0/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_0_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.0/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6910.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.0/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6911.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.0/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6912.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.0/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.0/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.0/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.0/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.0/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.0/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.0/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.0/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.0/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Gather_5F_2_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.0/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.0/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.0/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.0/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.0/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.0/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.0/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.0/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.0/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.0/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.0/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.0/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.0/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.0/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.0/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.0/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.0/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.0/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.0/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.0/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.0/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.0/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.0/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.0/self_attn/Concat_5:Concat*input:pkv0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv0(1,4,0,64)*axis:2 /model/model/layers.0/self_attn/Concat_6:Concat*input:pkv1(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv1(1,4,0,64)*axis:2 /model/model/layers.0/self_attn/Shape_7:Shape*input:opkv0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Shape_5F_7_5F_output_5F_0(4) /model/model/layers.0/self_attn/Gather_9:Gather*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Shape_5F_7_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_9_5F_output_5F_0()*axis:0 /model/model/layers.0/self_attn/Unsqueeze_12:Unsqueeze*input:opkv0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.0/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_9_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.0/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.0/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.0/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.0/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.0/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.0/self_attn/Shape_12:Shape*input:opkv1(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Shape_5F_12_5F_output_5F_0(4) /model/model/layers.0/self_attn/Gather_13:Gather*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Shape_5F_12_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_13_5F_output_5F_0()*axis:0 /model/model/layers.0/self_attn/Unsqueeze_21:Unsqueeze*input:opkv1(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.0/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Gather_5F_13_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.0/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.0/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.0/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.0/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.0/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.0/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.0/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.0/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.0/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.0/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.0/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.0/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.0/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.0/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6940.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.0/Add:Add*input:_2F_model_2F_model_2F_embed_5F_tokens_2F_Gather_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_0_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.0/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_0_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.0/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.0/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.0/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.0/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.0/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_0_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.0/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_0_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.0/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6941.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.0/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.0/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.0/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_0_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6942.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.0/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.0/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_6943.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.0/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_0_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_0_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.1/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_0_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.1/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.1/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.1/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.1/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.1/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_0_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.1/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_1_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.1/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6944.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.1/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6945.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.1/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_1_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6946.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.1/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.1/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.1/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.1/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.1/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.1/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.1/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.1/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.1/self_attn/Shape_3:Shape*input:pkv2(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.1/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.1/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.1/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.1/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.1/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.1/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.1/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.1/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.1/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.1/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.1/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.1/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.1/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.1/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.1/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.1/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.1/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.1/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.1/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.1/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.1/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.1/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.1/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.1/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.1/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.1/self_attn/Concat_5:Concat*input:pkv2(1,4,0,64);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv2(1,4,0,64)*axis:2 /model/model/layers.1/self_attn/Concat_6:Concat*input:pkv3(1,4,0,64);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv3(1,4,0,64)*axis:2 /model/model/layers.1/self_attn/Shape_8:Shape*input:opkv2(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.1/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.1/self_attn/Unsqueeze_12:Unsqueeze*input:opkv2(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.1/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.1/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.1/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.1/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.1/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.1/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.1/self_attn/Shape_13:Shape*input:opkv3(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.1/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.1/self_attn/Unsqueeze_21:Unsqueeze*input:opkv3(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.1/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.1/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.1/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.1/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.1/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.1/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.1/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.1/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.1/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.1/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.1/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.1/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.1/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.1/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.1/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6974.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.1/Add:Add*input:_2F_model_2F_model_2F_layers_2E_0_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_1_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_1_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.1/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_1_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.1/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.1/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.1/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.1/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.1/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_1_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.1/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_1_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.1/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6975.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.1/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.1/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.1/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_1_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6976.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.1/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.1/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_6977.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.1/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_1_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_1_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_1_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.2/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_1_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.2/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.2/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.2/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.2/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.2/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_1_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.2/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_2_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.2/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6978.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.2/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6979.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.2/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_2_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_6980.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.2/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.2/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.2/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.2/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.2/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.2/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.2/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.2/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.2/self_attn/Shape_3:Shape*input:pkv4(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.2/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.2/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.2/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.2/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.2/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.2/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.2/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.2/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.2/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.2/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.2/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.2/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.2/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.2/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.2/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.2/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.2/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.2/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.2/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.2/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.2/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.2/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.2/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.2/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.2/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.2/self_attn/Concat_5:Concat*input:pkv4(1,4,0,64);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv4(1,4,0,64)*axis:2 /model/model/layers.2/self_attn/Concat_6:Concat*input:pkv5(1,4,0,64);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv5(1,4,0,64)*axis:2 /model/model/layers.2/self_attn/Shape_8:Shape*input:opkv4(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.2/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.2/self_attn/Unsqueeze_12:Unsqueeze*input:opkv4(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.2/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.2/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.2/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.2/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.2/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.2/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.2/self_attn/Shape_13:Shape*input:opkv5(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.2/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.2/self_attn/Unsqueeze_21:Unsqueeze*input:opkv5(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.2/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.2/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.2/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.2/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.2/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.2/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.2/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.2/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.2/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.2/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.2/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.2/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.2/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.2/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.2/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7008.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.2/Add:Add*input:_2F_model_2F_model_2F_layers_2E_1_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_2_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_2_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.2/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_2_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.2/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.2/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.2/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.2/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.2/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_2_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.2/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_2_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.2/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7009.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.2/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.2/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.2/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_2_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7010.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.2/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.2/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7011.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.2/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_2_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_2_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_2_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.3/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_2_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.3/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.3/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.3/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.3/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.3/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_2_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.3/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_3_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.3/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7012.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.3/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7013.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.3/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_3_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7014.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.3/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.3/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.3/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.3/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.3/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.3/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.3/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.3/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.3/self_attn/Shape_3:Shape*input:pkv6(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.3/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.3/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.3/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.3/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.3/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.3/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.3/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.3/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.3/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.3/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.3/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.3/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.3/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.3/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.3/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.3/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.3/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.3/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.3/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.3/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.3/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.3/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.3/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.3/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.3/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.3/self_attn/Concat_5:Concat*input:pkv6(1,4,0,64);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv6(1,4,0,64)*axis:2 /model/model/layers.3/self_attn/Concat_6:Concat*input:pkv7(1,4,0,64);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv7(1,4,0,64)*axis:2 /model/model/layers.3/self_attn/Shape_8:Shape*input:opkv6(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.3/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.3/self_attn/Unsqueeze_12:Unsqueeze*input:opkv6(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.3/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.3/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.3/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.3/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.3/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.3/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.3/self_attn/Shape_13:Shape*input:opkv7(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.3/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.3/self_attn/Unsqueeze_21:Unsqueeze*input:opkv7(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.3/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.3/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.3/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.3/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.3/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.3/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.3/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.3/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.3/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.3/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.3/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.3/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.3/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.3/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.3/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7042.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.3/Add:Add*input:_2F_model_2F_model_2F_layers_2E_2_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_3_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_3_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.3/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_3_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.3/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.3/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.3/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.3/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.3/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_3_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.3/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_3_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.3/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7043.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.3/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.3/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.3/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_3_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7044.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.3/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.3/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7045.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.3/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_3_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_3_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_3_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.4/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_3_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.4/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.4/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.4/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.4/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.4/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_3_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.4/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_4_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.4/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7046.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.4/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7047.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.4/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_4_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7048.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.4/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.4/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.4/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.4/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.4/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.4/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.4/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.4/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.4/self_attn/Shape_3:Shape*input:pkv8(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.4/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.4/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.4/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.4/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.4/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.4/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.4/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.4/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.4/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.4/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.4/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.4/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.4/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.4/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.4/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.4/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.4/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.4/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.4/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.4/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.4/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.4/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.4/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.4/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.4/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.4/self_attn/Concat_5:Concat*input:pkv8(1,4,0,64);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv8(1,4,0,64)*axis:2 /model/model/layers.4/self_attn/Concat_6:Concat*input:pkv9(1,4,0,64);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv9(1,4,0,64)*axis:2 /model/model/layers.4/self_attn/Shape_8:Shape*input:opkv8(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.4/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.4/self_attn/Unsqueeze_12:Unsqueeze*input:opkv8(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.4/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.4/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.4/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.4/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.4/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.4/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.4/self_attn/Shape_13:Shape*input:opkv9(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.4/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.4/self_attn/Unsqueeze_21:Unsqueeze*input:opkv9(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.4/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.4/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.4/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.4/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.4/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.4/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.4/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.4/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.4/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.4/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.4/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.4/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.4/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.4/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.4/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7076.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.4/Add:Add*input:_2F_model_2F_model_2F_layers_2E_3_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_4_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_4_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.4/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_4_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.4/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.4/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.4/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.4/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.4/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_4_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.4/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_4_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.4/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7077.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.4/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.4/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.4/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_4_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7078.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.4/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.4/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7079.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.4/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_4_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_4_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_4_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.5/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_4_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.5/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.5/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.5/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.5/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.5/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_4_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.5/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_5_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.5/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7080.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.5/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7081.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.5/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_5_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7082.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.5/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.5/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.5/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.5/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.5/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.5/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.5/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.5/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.5/self_attn/Shape_3:Shape*input:pkv10(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.5/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.5/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.5/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.5/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.5/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.5/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.5/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.5/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.5/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.5/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.5/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.5/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.5/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.5/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.5/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.5/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.5/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.5/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.5/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.5/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.5/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.5/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.5/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.5/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.5/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.5/self_attn/Concat_5:Concat*input:pkv10(1,4,0,64);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv10(1,4,0,64)*axis:2 /model/model/layers.5/self_attn/Concat_6:Concat*input:pkv11(1,4,0,64);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv11(1,4,0,64)*axis:2 /model/model/layers.5/self_attn/Shape_8:Shape*input:opkv10(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.5/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.5/self_attn/Unsqueeze_12:Unsqueeze*input:opkv10(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.5/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.5/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.5/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.5/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.5/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.5/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.5/self_attn/Shape_13:Shape*input:opkv11(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.5/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.5/self_attn/Unsqueeze_21:Unsqueeze*input:opkv11(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.5/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.5/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.5/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.5/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.5/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.5/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.5/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.5/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.5/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.5/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.5/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.5/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.5/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.5/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.5/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7110.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.5/Add:Add*input:_2F_model_2F_model_2F_layers_2E_4_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_5_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_5_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.5/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_5_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.5/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.5/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.5/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.5/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.5/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_5_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.5/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_5_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.5/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7111.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.5/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.5/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.5/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_5_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7112.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.5/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.5/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7113.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.5/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_5_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_5_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_5_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.6/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_5_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.6/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.6/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.6/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.6/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.6/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_5_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.6/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_6_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.6/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7114.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.6/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7115.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.6/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_6_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7116.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.6/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.6/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.6/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.6/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.6/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.6/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.6/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.6/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.6/self_attn/Shape_3:Shape*input:pkv12(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.6/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.6/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.6/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.6/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.6/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.6/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.6/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.6/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.6/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.6/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.6/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.6/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.6/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.6/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.6/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.6/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.6/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.6/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.6/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.6/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.6/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.6/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.6/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.6/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.6/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.6/self_attn/Concat_5:Concat*input:pkv12(1,4,0,64);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv12(1,4,0,64)*axis:2 /model/model/layers.6/self_attn/Concat_6:Concat*input:pkv13(1,4,0,64);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv13(1,4,0,64)*axis:2 /model/model/layers.6/self_attn/Shape_8:Shape*input:opkv12(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.6/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.6/self_attn/Unsqueeze_12:Unsqueeze*input:opkv12(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.6/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.6/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.6/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.6/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.6/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.6/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.6/self_attn/Shape_13:Shape*input:opkv13(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.6/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.6/self_attn/Unsqueeze_21:Unsqueeze*input:opkv13(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.6/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.6/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.6/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.6/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.6/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.6/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.6/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.6/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.6/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.6/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.6/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.6/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.6/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.6/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.6/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7144.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.6/Add:Add*input:_2F_model_2F_model_2F_layers_2E_5_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_6_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_6_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.6/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_6_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.6/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.6/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.6/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.6/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.6/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_6_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.6/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_6_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.6/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7145.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.6/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.6/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.6/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_6_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7146.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.6/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.6/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7147.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.6/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_6_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_6_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_6_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.7/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_6_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.7/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.7/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.7/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.7/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.7/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_6_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.7/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_7_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.7/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7148.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.7/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7149.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.7/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_7_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7150.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.7/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.7/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.7/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.7/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.7/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.7/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.7/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.7/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.7/self_attn/Shape_3:Shape*input:pkv14(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.7/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.7/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.7/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.7/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.7/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.7/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.7/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.7/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.7/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.7/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.7/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.7/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.7/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.7/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.7/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.7/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.7/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.7/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.7/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.7/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.7/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.7/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.7/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.7/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.7/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.7/self_attn/Concat_5:Concat*input:pkv14(1,4,0,64);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv14(1,4,0,64)*axis:2 /model/model/layers.7/self_attn/Concat_6:Concat*input:pkv15(1,4,0,64);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv15(1,4,0,64)*axis:2 /model/model/layers.7/self_attn/Shape_8:Shape*input:opkv14(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.7/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.7/self_attn/Unsqueeze_12:Unsqueeze*input:opkv14(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.7/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.7/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.7/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.7/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.7/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.7/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.7/self_attn/Shape_13:Shape*input:opkv15(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.7/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.7/self_attn/Unsqueeze_21:Unsqueeze*input:opkv15(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.7/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.7/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.7/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.7/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.7/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.7/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.7/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.7/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.7/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.7/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.7/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.7/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.7/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.7/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.7/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7178.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.7/Add:Add*input:_2F_model_2F_model_2F_layers_2E_6_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_7_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_7_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.7/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_7_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.7/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.7/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.7/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.7/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.7/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_7_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.7/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_7_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.7/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7179.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.7/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.7/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.7/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_7_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7180.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.7/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.7/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7181.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.7/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_7_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_7_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_7_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.8/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_7_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.8/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.8/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.8/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.8/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.8/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_7_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.8/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_8_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.8/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7182.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.8/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7183.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.8/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_8_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7184.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.8/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.8/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.8/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.8/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.8/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.8/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.8/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.8/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.8/self_attn/Shape_3:Shape*input:pkv16(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.8/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.8/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.8/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.8/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.8/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.8/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.8/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.8/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.8/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.8/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.8/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.8/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.8/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.8/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.8/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.8/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.8/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.8/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.8/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.8/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.8/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.8/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.8/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.8/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.8/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.8/self_attn/Concat_5:Concat*input:pkv16(1,4,0,64);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv16(1,4,0,64)*axis:2 /model/model/layers.8/self_attn/Concat_6:Concat*input:pkv17(1,4,0,64);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv17(1,4,0,64)*axis:2 /model/model/layers.8/self_attn/Shape_8:Shape*input:opkv16(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.8/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.8/self_attn/Unsqueeze_12:Unsqueeze*input:opkv16(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.8/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.8/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.8/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.8/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.8/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.8/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.8/self_attn/Shape_13:Shape*input:opkv17(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.8/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.8/self_attn/Unsqueeze_21:Unsqueeze*input:opkv17(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.8/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.8/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.8/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.8/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.8/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.8/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.8/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.8/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.8/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.8/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.8/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.8/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.8/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.8/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.8/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7212.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.8/Add:Add*input:_2F_model_2F_model_2F_layers_2E_7_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_8_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_8_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.8/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_8_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.8/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.8/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.8/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.8/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.8/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_8_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.8/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_8_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.8/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7213.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.8/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.8/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.8/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_8_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7214.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.8/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.8/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7215.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.8/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_8_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_8_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_8_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.9/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_8_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.9/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.9/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.9/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.9/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.9/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_8_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.9/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_9_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.9/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7216.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.9/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7217.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.9/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_9_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7218.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.9/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.9/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.9/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.9/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.9/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.9/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.9/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.9/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.9/self_attn/Shape_3:Shape*input:pkv18(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.9/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.9/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.9/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.9/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.9/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.9/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.9/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.9/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.9/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.9/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.9/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.9/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.9/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.9/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.9/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.9/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.9/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.9/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.9/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.9/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.9/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.9/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.9/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.9/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.9/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.9/self_attn/Concat_5:Concat*input:pkv18(1,4,0,64);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv18(1,4,0,64)*axis:2 /model/model/layers.9/self_attn/Concat_6:Concat*input:pkv19(1,4,0,64);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv19(1,4,0,64)*axis:2 /model/model/layers.9/self_attn/Shape_8:Shape*input:opkv18(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.9/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.9/self_attn/Unsqueeze_12:Unsqueeze*input:opkv18(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.9/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.9/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.9/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.9/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.9/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.9/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.9/self_attn/Shape_13:Shape*input:opkv19(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.9/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.9/self_attn/Unsqueeze_21:Unsqueeze*input:opkv19(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.9/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.9/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.9/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.9/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.9/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.9/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.9/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.9/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.9/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.9/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.9/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.9/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.9/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.9/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.9/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7246.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.9/Add:Add*input:_2F_model_2F_model_2F_layers_2E_8_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_9_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_9_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.9/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_9_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.9/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.9/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.9/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.9/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.9/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_9_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.9/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_9_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.9/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7247.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.9/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.9/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.9/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_9_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7248.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.9/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.9/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7249.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.9/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_9_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_9_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_9_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.10/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_9_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.10/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.10/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.10/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.10/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.10/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_9_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.10/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_10_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.10/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7250.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.10/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7251.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.10/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_10_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7252.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.10/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.10/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.10/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.10/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.10/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.10/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.10/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.10/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.10/self_attn/Shape_3:Shape*input:pkv20(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.10/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.10/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.10/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.10/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.10/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.10/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.10/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.10/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.10/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.10/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.10/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.10/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.10/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.10/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.10/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.10/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.10/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.10/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.10/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.10/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.10/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.10/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.10/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.10/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.10/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.10/self_attn/Concat_5:Concat*input:pkv20(1,4,0,64);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv20(1,4,0,64)*axis:2 /model/model/layers.10/self_attn/Concat_6:Concat*input:pkv21(1,4,0,64);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv21(1,4,0,64)*axis:2 /model/model/layers.10/self_attn/Shape_8:Shape*input:opkv20(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.10/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.10/self_attn/Unsqueeze_12:Unsqueeze*input:opkv20(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.10/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.10/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.10/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.10/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.10/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.10/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.10/self_attn/Shape_13:Shape*input:opkv21(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.10/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.10/self_attn/Unsqueeze_21:Unsqueeze*input:opkv21(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.10/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.10/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.10/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.10/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.10/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.10/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.10/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.10/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.10/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.10/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.10/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.10/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.10/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.10/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.10/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7280.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.10/Add:Add*input:_2F_model_2F_model_2F_layers_2E_9_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_10_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_10_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.10/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_10_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.10/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.10/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.10/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.10/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.10/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_10_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.10/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_10_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.10/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7281.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.10/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.10/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.10/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_10_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7282.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.10/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.10/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7283.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.10/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_10_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_10_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_10_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.11/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_10_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.11/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.11/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.11/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.11/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.11/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_10_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.11/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_11_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.11/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7284.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.11/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7285.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.11/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_11_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7286.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.11/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.11/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.11/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.11/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.11/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.11/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.11/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.11/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.11/self_attn/Shape_3:Shape*input:pkv22(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.11/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.11/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.11/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.11/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.11/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.11/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.11/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.11/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.11/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.11/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.11/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.11/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.11/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.11/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.11/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.11/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.11/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.11/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.11/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.11/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.11/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.11/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.11/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.11/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.11/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.11/self_attn/Concat_5:Concat*input:pkv22(1,4,0,64);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv22(1,4,0,64)*axis:2 /model/model/layers.11/self_attn/Concat_6:Concat*input:pkv23(1,4,0,64);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv23(1,4,0,64)*axis:2 /model/model/layers.11/self_attn/Shape_8:Shape*input:opkv22(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.11/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.11/self_attn/Unsqueeze_12:Unsqueeze*input:opkv22(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.11/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.11/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.11/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.11/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.11/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.11/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.11/self_attn/Shape_13:Shape*input:opkv23(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.11/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.11/self_attn/Unsqueeze_21:Unsqueeze*input:opkv23(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.11/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.11/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.11/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.11/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.11/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.11/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.11/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.11/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.11/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.11/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.11/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.11/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.11/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.11/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.11/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7314.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.11/Add:Add*input:_2F_model_2F_model_2F_layers_2E_10_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_11_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_11_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.11/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_11_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.11/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.11/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.11/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.11/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.11/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_11_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.11/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_11_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.11/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7315.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.11/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.11/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.11/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_11_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7316.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.11/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.11/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7317.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.11/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_11_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_11_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_11_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.12/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_11_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.12/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.12/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.12/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.12/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.12/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_11_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.12/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_12_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.12/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7318.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.12/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7319.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.12/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_12_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7320.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.12/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.12/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.12/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.12/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.12/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.12/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.12/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.12/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.12/self_attn/Shape_3:Shape*input:pkv24(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.12/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.12/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.12/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.12/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.12/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.12/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.12/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.12/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.12/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.12/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.12/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.12/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.12/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.12/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.12/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.12/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.12/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.12/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.12/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.12/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.12/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.12/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.12/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.12/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.12/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.12/self_attn/Concat_5:Concat*input:pkv24(1,4,0,64);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv24(1,4,0,64)*axis:2 /model/model/layers.12/self_attn/Concat_6:Concat*input:pkv25(1,4,0,64);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv25(1,4,0,64)*axis:2 /model/model/layers.12/self_attn/Shape_8:Shape*input:opkv24(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.12/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.12/self_attn/Unsqueeze_12:Unsqueeze*input:opkv24(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.12/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.12/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.12/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.12/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.12/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.12/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.12/self_attn/Shape_13:Shape*input:opkv25(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.12/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.12/self_attn/Unsqueeze_21:Unsqueeze*input:opkv25(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.12/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.12/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.12/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.12/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.12/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.12/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.12/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.12/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.12/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.12/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.12/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.12/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.12/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.12/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.12/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7348.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.12/Add:Add*input:_2F_model_2F_model_2F_layers_2E_11_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_12_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_12_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.12/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_12_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.12/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.12/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.12/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.12/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.12/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_12_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.12/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_12_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.12/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7349.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.12/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.12/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.12/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_12_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7350.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.12/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.12/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7351.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.12/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_12_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_12_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_12_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.13/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_12_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.13/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.13/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.13/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.13/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.13/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_12_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.13/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_13_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.13/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7352.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.13/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7353.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.13/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_13_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7354.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.13/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.13/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.13/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.13/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.13/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.13/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.13/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.13/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.13/self_attn/Shape_3:Shape*input:pkv26(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.13/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.13/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.13/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.13/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.13/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.13/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.13/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.13/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.13/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.13/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.13/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.13/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.13/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.13/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.13/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.13/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.13/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.13/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.13/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.13/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.13/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.13/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.13/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.13/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.13/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.13/self_attn/Concat_5:Concat*input:pkv26(1,4,0,64);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv26(1,4,0,64)*axis:2 /model/model/layers.13/self_attn/Concat_6:Concat*input:pkv27(1,4,0,64);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv27(1,4,0,64)*axis:2 /model/model/layers.13/self_attn/Shape_8:Shape*input:opkv26(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.13/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.13/self_attn/Unsqueeze_12:Unsqueeze*input:opkv26(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.13/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.13/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.13/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.13/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.13/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.13/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.13/self_attn/Shape_13:Shape*input:opkv27(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.13/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.13/self_attn/Unsqueeze_21:Unsqueeze*input:opkv27(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.13/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.13/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.13/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.13/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.13/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.13/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.13/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.13/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.13/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.13/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.13/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.13/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.13/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.13/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.13/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7382.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.13/Add:Add*input:_2F_model_2F_model_2F_layers_2E_12_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_13_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_13_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.13/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_13_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.13/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.13/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.13/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.13/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.13/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_13_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.13/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_13_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.13/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7383.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.13/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.13/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.13/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_13_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7384.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.13/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.13/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7385.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.13/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_13_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_13_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_13_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.14/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_13_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.14/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.14/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.14/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.14/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.14/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_13_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.14/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_14_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.14/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7386.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.14/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7387.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.14/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_14_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7388.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.14/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.14/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.14/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.14/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.14/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.14/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.14/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.14/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.14/self_attn/Shape_3:Shape*input:pkv28(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.14/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.14/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.14/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.14/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.14/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.14/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.14/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.14/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.14/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.14/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.14/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.14/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.14/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.14/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.14/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.14/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.14/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.14/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.14/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.14/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.14/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.14/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.14/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.14/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.14/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.14/self_attn/Concat_5:Concat*input:pkv28(1,4,0,64);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv28(1,4,0,64)*axis:2 /model/model/layers.14/self_attn/Concat_6:Concat*input:pkv29(1,4,0,64);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv29(1,4,0,64)*axis:2 /model/model/layers.14/self_attn/Shape_8:Shape*input:opkv28(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.14/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.14/self_attn/Unsqueeze_12:Unsqueeze*input:opkv28(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.14/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.14/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.14/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.14/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.14/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.14/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.14/self_attn/Shape_13:Shape*input:opkv29(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.14/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.14/self_attn/Unsqueeze_21:Unsqueeze*input:opkv29(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.14/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.14/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.14/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.14/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.14/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.14/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.14/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.14/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.14/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.14/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.14/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.14/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.14/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.14/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.14/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7416.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.14/Add:Add*input:_2F_model_2F_model_2F_layers_2E_13_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_14_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_14_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.14/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_14_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.14/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.14/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.14/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.14/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.14/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_14_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.14/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_14_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.14/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7417.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.14/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.14/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.14/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_14_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7418.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.14/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.14/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7419.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.14/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_14_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_14_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_14_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.15/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_14_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.15/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.15/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.15/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.15/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.15/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_14_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.15/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_15_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.15/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7420.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.15/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7421.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.15/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_15_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7422.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.15/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.15/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.15/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.15/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.15/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.15/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.15/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.15/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.15/self_attn/Shape_3:Shape*input:pkv30(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.15/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.15/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.15/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.15/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.15/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.15/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.15/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.15/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.15/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.15/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.15/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.15/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.15/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.15/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.15/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.15/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.15/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.15/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.15/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.15/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.15/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.15/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.15/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.15/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.15/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.15/self_attn/Concat_5:Concat*input:pkv30(1,4,0,64);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv30(1,4,0,64)*axis:2 /model/model/layers.15/self_attn/Concat_6:Concat*input:pkv31(1,4,0,64);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv31(1,4,0,64)*axis:2 /model/model/layers.15/self_attn/Shape_8:Shape*input:opkv30(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.15/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.15/self_attn/Unsqueeze_12:Unsqueeze*input:opkv30(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.15/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.15/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.15/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.15/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.15/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.15/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.15/self_attn/Shape_13:Shape*input:opkv31(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.15/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.15/self_attn/Unsqueeze_21:Unsqueeze*input:opkv31(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.15/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.15/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.15/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.15/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.15/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.15/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.15/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.15/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.15/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.15/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.15/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.15/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.15/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.15/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.15/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7450.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.15/Add:Add*input:_2F_model_2F_model_2F_layers_2E_14_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_15_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_15_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.15/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_15_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.15/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.15/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.15/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.15/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.15/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_15_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.15/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_15_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.15/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7451.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.15/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.15/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.15/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_15_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7452.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.15/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.15/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7453.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.15/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_15_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_15_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_15_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.16/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_15_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.16/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.16/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.16/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.16/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.16/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_15_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.16/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_16_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.16/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7454.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.16/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7455.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.16/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_16_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7456.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.16/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.16/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.16/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.16/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.16/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.16/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.16/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.16/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.16/self_attn/Shape_3:Shape*input:pkv32(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.16/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.16/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.16/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.16/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.16/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.16/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.16/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.16/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.16/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.16/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.16/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.16/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.16/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.16/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.16/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.16/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.16/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.16/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.16/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.16/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.16/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.16/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.16/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.16/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.16/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.16/self_attn/Concat_5:Concat*input:pkv32(1,4,0,64);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv32(1,4,0,64)*axis:2 /model/model/layers.16/self_attn/Concat_6:Concat*input:pkv33(1,4,0,64);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv33(1,4,0,64)*axis:2 /model/model/layers.16/self_attn/Shape_8:Shape*input:opkv32(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.16/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.16/self_attn/Unsqueeze_12:Unsqueeze*input:opkv32(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.16/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.16/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.16/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.16/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.16/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.16/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.16/self_attn/Shape_13:Shape*input:opkv33(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.16/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.16/self_attn/Unsqueeze_21:Unsqueeze*input:opkv33(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.16/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.16/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.16/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.16/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.16/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.16/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.16/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.16/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.16/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.16/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.16/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.16/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.16/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.16/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.16/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7484.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.16/Add:Add*input:_2F_model_2F_model_2F_layers_2E_15_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_16_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_16_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.16/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_16_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.16/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.16/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.16/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.16/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.16/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_16_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.16/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_16_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.16/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7485.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.16/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.16/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.16/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_16_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7486.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.16/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.16/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7487.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.16/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_16_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_16_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_16_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.17/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_16_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.17/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.17/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.17/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.17/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.17/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_16_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.17/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_17_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.17/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7488.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.17/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7489.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.17/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_17_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7490.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.17/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.17/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.17/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.17/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.17/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.17/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.17/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.17/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.17/self_attn/Shape_3:Shape*input:pkv34(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.17/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.17/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.17/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.17/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.17/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.17/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.17/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.17/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.17/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.17/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.17/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.17/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.17/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.17/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.17/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.17/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.17/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.17/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.17/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.17/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.17/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.17/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.17/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.17/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.17/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.17/self_attn/Concat_5:Concat*input:pkv34(1,4,0,64);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv34(1,4,0,64)*axis:2 /model/model/layers.17/self_attn/Concat_6:Concat*input:pkv35(1,4,0,64);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv35(1,4,0,64)*axis:2 /model/model/layers.17/self_attn/Shape_8:Shape*input:opkv34(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.17/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.17/self_attn/Unsqueeze_12:Unsqueeze*input:opkv34(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.17/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.17/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.17/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.17/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.17/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.17/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.17/self_attn/Shape_13:Shape*input:opkv35(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.17/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.17/self_attn/Unsqueeze_21:Unsqueeze*input:opkv35(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.17/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.17/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.17/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.17/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.17/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.17/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.17/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.17/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.17/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.17/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.17/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.17/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.17/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.17/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.17/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7518.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.17/Add:Add*input:_2F_model_2F_model_2F_layers_2E_16_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_17_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_17_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.17/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_17_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.17/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.17/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.17/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.17/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.17/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_17_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.17/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_17_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.17/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7519.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.17/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.17/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.17/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_17_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7520.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.17/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.17/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7521.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.17/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_17_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_17_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_17_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.18/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_17_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.18/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.18/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.18/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.18/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.18/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_17_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.18/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_18_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.18/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7522.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.18/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7523.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.18/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_18_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7524.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.18/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.18/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.18/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.18/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.18/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.18/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.18/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.18/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.18/self_attn/Shape_3:Shape*input:pkv36(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.18/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.18/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.18/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.18/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.18/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.18/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.18/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.18/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.18/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.18/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.18/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.18/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.18/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.18/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.18/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.18/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.18/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.18/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.18/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.18/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.18/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.18/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.18/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.18/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.18/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.18/self_attn/Concat_5:Concat*input:pkv36(1,4,0,64);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv36(1,4,0,64)*axis:2 /model/model/layers.18/self_attn/Concat_6:Concat*input:pkv37(1,4,0,64);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv37(1,4,0,64)*axis:2 /model/model/layers.18/self_attn/Shape_8:Shape*input:opkv36(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.18/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.18/self_attn/Unsqueeze_12:Unsqueeze*input:opkv36(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.18/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.18/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.18/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.18/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.18/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.18/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.18/self_attn/Shape_13:Shape*input:opkv37(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.18/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.18/self_attn/Unsqueeze_21:Unsqueeze*input:opkv37(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.18/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.18/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.18/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.18/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.18/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.18/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.18/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.18/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.18/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.18/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.18/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.18/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.18/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.18/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.18/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7552.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.18/Add:Add*input:_2F_model_2F_model_2F_layers_2E_17_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_18_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_18_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.18/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_18_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.18/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.18/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.18/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.18/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.18/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_18_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.18/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_18_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.18/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7553.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.18/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.18/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.18/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_18_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7554.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.18/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.18/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7555.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.18/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_18_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_18_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_18_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.19/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_18_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.19/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.19/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.19/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.19/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.19/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_18_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.19/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_19_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.19/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7556.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.19/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7557.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.19/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_19_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7558.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.19/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.19/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.19/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.19/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.19/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.19/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.19/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.19/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.19/self_attn/Shape_3:Shape*input:pkv38(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.19/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.19/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.19/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.19/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.19/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.19/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.19/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.19/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.19/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.19/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.19/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.19/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.19/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.19/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.19/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.19/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.19/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.19/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.19/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.19/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.19/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.19/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.19/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.19/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.19/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.19/self_attn/Concat_5:Concat*input:pkv38(1,4,0,64);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv38(1,4,0,64)*axis:2 /model/model/layers.19/self_attn/Concat_6:Concat*input:pkv39(1,4,0,64);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv39(1,4,0,64)*axis:2 /model/model/layers.19/self_attn/Shape_8:Shape*input:opkv38(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.19/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.19/self_attn/Unsqueeze_12:Unsqueeze*input:opkv38(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.19/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.19/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.19/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.19/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.19/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.19/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.19/self_attn/Shape_13:Shape*input:opkv39(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.19/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.19/self_attn/Unsqueeze_21:Unsqueeze*input:opkv39(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.19/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.19/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.19/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.19/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.19/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.19/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.19/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.19/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.19/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.19/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.19/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.19/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.19/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.19/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.19/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7586.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.19/Add:Add*input:_2F_model_2F_model_2F_layers_2E_18_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_19_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_19_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.19/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_19_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.19/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.19/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.19/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.19/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.19/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_19_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.19/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_19_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.19/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7587.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.19/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.19/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.19/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_19_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7588.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.19/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.19/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7589.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.19/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_19_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_19_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_19_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.20/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_19_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.20/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.20/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.20/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.20/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.20/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_19_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.20/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_20_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.20/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7590.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.20/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7591.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.20/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_20_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7592.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.20/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.20/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.20/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.20/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.20/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.20/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.20/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.20/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.20/self_attn/Shape_3:Shape*input:pkv40(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.20/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.20/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.20/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.20/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.20/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.20/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.20/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.20/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.20/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.20/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.20/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.20/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.20/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.20/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.20/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.20/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.20/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.20/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.20/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.20/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.20/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.20/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.20/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.20/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.20/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.20/self_attn/Concat_5:Concat*input:pkv40(1,4,0,64);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv40(1,4,0,64)*axis:2 /model/model/layers.20/self_attn/Concat_6:Concat*input:pkv41(1,4,0,64);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv41(1,4,0,64)*axis:2 /model/model/layers.20/self_attn/Shape_8:Shape*input:opkv40(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.20/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.20/self_attn/Unsqueeze_12:Unsqueeze*input:opkv40(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.20/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.20/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.20/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.20/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.20/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.20/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.20/self_attn/Shape_13:Shape*input:opkv41(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.20/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.20/self_attn/Unsqueeze_21:Unsqueeze*input:opkv41(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.20/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.20/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.20/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.20/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.20/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.20/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.20/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.20/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.20/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.20/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.20/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.20/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.20/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.20/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.20/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7620.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.20/Add:Add*input:_2F_model_2F_model_2F_layers_2E_19_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_20_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_20_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.20/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_20_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.20/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.20/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.20/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.20/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.20/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_20_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.20/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_20_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.20/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7621.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.20/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.20/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.20/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_20_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7622.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.20/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.20/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7623.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.20/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_20_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_20_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_20_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.21/input_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_20_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.21/input_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.21/input_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.21/input_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.21/input_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.21/input_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_20_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.21/input_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_21_2E_input_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.21/self_attn/q_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7624.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.21/self_attn/k_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7625.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.21/self_attn/v_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_21_2F_input_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7626.bin(float16:2048,256)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256) /model/model/layers.21/self_attn/Reshape:Reshape*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_q_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048);2646.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*allowzero:0 /model/model/layers.21/self_attn/Transpose:Transpose*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_output_5F_0(1,0,32,64)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64)*perm:0,2,1,3 /model/model/layers.21/self_attn/Reshape_1:Reshape*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_k_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.21/self_attn/Transpose_1:Transpose*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_1_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.21/self_attn/Reshape_2:Reshape*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_v_5F_proj_2F_MatMul_5F_output_5F_0(1,0,256);2648.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*allowzero:0 /model/model/layers.21/self_attn/Transpose_2:Transpose*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_2_5F_output_5F_0(1,0,4,64)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*perm:0,2,1,3 /model/model/layers.21/self_attn/Shape_2:Shape*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4) /model/model/layers.21/self_attn/Gather_2:Gather*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Shape_5F_2_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0()*axis:0 /model/model/layers.21/self_attn/Shape_3:Shape*input:pkv42(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4) /model/model/layers.21/self_attn/Gather_3:Gather*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Shape_5F_3_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*axis:0 /model/model/layers.21/self_attn/Add:Add*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Add_5F_output_5F_0() /model/model/layers.21/self_attn/rotary_emb/Unsqueeze:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Add_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1) /model/model/layers.21/self_attn/rotary_emb/Slice:Slice*input:onnx_3A__3A_Slice_5F_461.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0() /model/model/layers.21/self_attn/rotary_emb/Slice_1:Slice*input:onnx_3A__3A_Slice_5F_472.bin(float16:1,1,2048,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_rotary_5F_emb_2F_Unsqueeze_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0() /model/model/layers.21/self_attn/Squeeze_1:Squeeze*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0() /model/model/layers.21/self_attn/Squeeze_3:Squeeze*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_rotary_5F_emb_2F_Slice_5F_1_5F_output_5F_0();6390.bin(int64:2)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0() /model/model/layers.21/self_attn/Gather_4:Gather*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Squeeze_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0()*axis:0 /model/model/layers.21/self_attn/Unsqueeze_6:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_4_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0() /model/model/layers.21/self_attn/Gather_5:Gather*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Squeeze_5F_3_5F_output_5F_0();_2F_model_2F_model_2F_Reshape_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0()*axis:0 /model/model/layers.21/self_attn/Unsqueeze_7:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_5_5F_output_5F_0();_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0() /model/model/layers.21/self_attn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Mul_5F_output_5F_0() /model/model/layers.21/self_attn/Slice:Slice*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32) /model/model/layers.21/self_attn/Slice_1:Slice*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32) /model/model/layers.21/self_attn/Neg:Neg*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Slice_5F_1_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32) /model/model/layers.21/self_attn/Concat_3:Concat*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Neg_5F_output_5F_0(1,32,0,32);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Slice_5F_output_5F_0(1,32,0,32)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64)*axis:-1 /model/model/layers.21/self_attn/Mul_1:Mul*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_3_5F_output_5F_0(1,32,0,64);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0() /model/model/layers.21/self_attn/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Mul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Mul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0() /model/model/layers.21/self_attn/Mul_2:Mul*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_6_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0() /model/model/layers.21/self_attn/Slice_2:Slice*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32) /model/model/layers.21/self_attn/Slice_3:Slice*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_1_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_2_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_6_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_17_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32) /model/model/layers.21/self_attn/Neg_1:Neg*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Slice_5F_3_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32) /model/model/layers.21/self_attn/Concat_4:Concat*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Neg_5F_1_5F_output_5F_0(1,4,0,32);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Slice_5F_2_5F_output_5F_0(1,4,0,32)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64)*axis:-1 /model/model/layers.21/self_attn/Mul_3:Mul*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_4_5F_output_5F_0(1,4,0,64);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_7_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0() /model/model/layers.21/self_attn/Add_2:Add*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Mul_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Mul_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0() /model/model/layers.21/self_attn/Concat_5:Concat*input:pkv42(1,4,0,64);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Add_5F_2_5F_output_5F_0()*output:opkv42(1,4,0,64)*axis:2 /model/model/layers.21/self_attn/Concat_6:Concat*input:pkv43(1,4,0,64);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_2_5F_output_5F_0(1,4,0,64)*output:opkv43(1,4,0,64)*axis:2 /model/model/layers.21/self_attn/Shape_8:Shape*input:opkv42(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4) /model/model/layers.21/self_attn/Gather_10:Gather*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Shape_5F_8_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0()*axis:0 /model/model/layers.21/self_attn/Unsqueeze_12:Unsqueeze*input:opkv42(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64) /model/model/layers.21/self_attn/Unsqueeze_15:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_10_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1) /model/model/layers.21/self_attn/Concat_7:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_15_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*axis:0 /model/model/layers.21/self_attn/Equal:Equal*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5) /model/model/layers.21/self_attn/Where:Where*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Equal_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_7_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Where_5F_output_5F_0(5) /model/model/layers.21/self_attn/Expand:Expand*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_12_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Where_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64) /model/model/layers.21/self_attn/Reshape_4:Reshape*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Expand_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.21/self_attn/Shape_13:Shape*input:opkv43(1,4,0,64)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4) /model/model/layers.21/self_attn/Gather_14:Gather*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Shape_5F_13_5F_output_5F_0(4);_2F_model_2F_model_2F_Constant_5F_2_5F_output_5F_0.bin(int64:)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0()*axis:0 /model/model/layers.21/self_attn/Unsqueeze_21:Unsqueeze*input:opkv43(1,4,0,64);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_rotary_5F_emb_2F_Constant_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64) /model/model/layers.21/self_attn/Unsqueeze_24:Unsqueeze*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Gather_5F_14_5F_output_5F_0();onnx_3A__3A_Unsqueeze_5F_285.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1) /model/model/layers.21/self_attn/Concat_9:Concat*input:_2F_model_2F_model_2F_Constant_5F_15_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_4_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_42_5F_output_5F_0.bin(int64:1);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_24_5F_output_5F_0(1);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_3_5F_output_5F_0.bin(int64:1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*axis:0 /model/model/layers.21/self_attn/Equal_1:Equal*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Mul_5F_4_5F_output_5F_0.bin(int64:5)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5) /model/model/layers.21/self_attn/Where_1:Where*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Equal_5F_1_5F_output_5F_0(5);_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_ConstantOfShape_5F_output_5F_0.bin(int64:5);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Concat_5F_9_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5) /model/model/layers.21/self_attn/Expand_1:Expand*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Unsqueeze_5F_21_5F_output_5F_0(1,4,1,0,64);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Where_5F_1_5F_output_5F_0(5)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64) /model/model/layers.21/self_attn/Reshape_6:Reshape*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Expand_5F_1_5F_output_5F_0(0,4,0,0,64);2652.bin(int64:4)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*allowzero:0 /model/model/layers.21/self_attn/Transpose_3:Transpose*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_4_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*perm:0,1,3,2 /model/model/layers.21/self_attn/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Add_5F_1_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_3_5F_output_5F_0(1,32,64,0)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_MatMul_5F_output_5F_0() /model/model/layers.21/self_attn/Div_2:Div*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_MatMul_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_0_2F_self_5F_attn_2F_Constant_5F_61_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0() /model/model/layers.21/self_attn/Add_3:Add*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Div_5F_2_5F_output_5F_0();_2F_model_2F_model_2F_Add_5F_2_5F_output_5F_0(0,0,0,0)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0() /model/model/layers.21/self_attn/Softmax:Softmax*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Add_5F_3_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Softmax_5F_output_5F_0()*axis:-1 /model/model/layers.21/self_attn/MatMul_1:MatMul*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Softmax_5F_output_5F_0();_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_6_5F_output_5F_0(1,32,0,64)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0() /model/model/layers.21/self_attn/Transpose_4:Transpose*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_MatMul_5F_1_5F_output_5F_0()*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0()*perm:0,2,1,3 /model/model/layers.21/self_attn/Reshape_7:Reshape*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Transpose_5F_4_5F_output_5F_0();2656.bin(int64:3)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048)*allowzero:0 /model/model/layers.21/self_attn/o_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_Reshape_5F_7_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7654.bin(float16:2048,2048)*output:_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.21/Add:Add*input:_2F_model_2F_model_2F_layers_2E_20_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_21_2F_self_5F_attn_2F_o_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_21_2F_Add_5F_output_5F_0(1,0,2048) /model/model/layers.21/post_attention_layernorm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_21_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/layers.21/post_attention_layernorm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/layers.21/post_attention_layernorm/Add:Add*input:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1) /model/model/layers.21/post_attention_layernorm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/layers.21/post_attention_layernorm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1) /model/model/layers.21/post_attention_layernorm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_21_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/layers.21/post_attention_layernorm/Mul_1:Mul*input:model_2E_model_2E_layers_2E_21_2E_post_5F_attention_5F_layernorm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/model/layers.21/mlp/gate_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7655.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.21/mlp/act_fn/Sigmoid:Sigmoid*input:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632) /model/model/layers.21/mlp/act_fn/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_gate_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_act_5F_fn_2F_Sigmoid_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.21/mlp/up_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_21_2F_post_5F_attention_5F_layernorm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7656.bin(float16:2048,5632)*output:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632) /model/model/layers.21/mlp/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_act_5F_fn_2F_Mul_5F_output_5F_0(1,0,5632);_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_up_5F_proj_2F_MatMul_5F_output_5F_0(1,0,5632)*output:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632) /model/model/layers.21/mlp/down_proj/MatMul:MatMul*input:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_Mul_5F_output_5F_0(1,0,5632);onnx_3A__3A_MatMul_5F_7657.bin(float16:5632,2048)*output:_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048) /model/model/layers.21/Add_1:Add*input:_2F_model_2F_model_2F_layers_2E_21_2F_Add_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_21_2F_mlp_2F_down_5F_proj_2F_MatMul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_layers_2E_21_2F_Add_5F_1_5F_output_5F_0(1,0,2048) /model/model/norm/Pow:Pow*input:_2F_model_2F_model_2F_layers_2E_21_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_norm_2F_Pow_5F_output_5F_0(1,0,2048) /model/model/norm/ReduceMean:ReduceMean*input:_2F_model_2F_model_2F_norm_2F_Pow_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_norm_2F_ReduceMean_5F_output_5F_0(1,0,1)*axes:-1;keepdims:1 /model/model/norm/Add:Add*input:_2F_model_2F_model_2F_norm_2F_ReduceMean_5F_output_5F_0(1,0,1);_2F_model_2F_model_2F_layers_2E_0_2F_input_5F_layernorm_2F_Constant_5F_1_5F_output_5F_0.bin(float16:)*output:_2F_model_2F_model_2F_norm_2F_Add_5F_output_5F_0(1,0,1) /model/model/norm/Sqrt:Sqrt*input:_2F_model_2F_model_2F_norm_2F_Add_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_norm_2F_Sqrt_5F_output_5F_0(1,0,1) /model/model/norm/Div:Div*input:_2F_model_2F_model_2F_Constant_5F_35_5F_output_5F_0.bin(float16:);_2F_model_2F_model_2F_norm_2F_Sqrt_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_norm_2F_Div_5F_output_5F_0(1,0,1) /model/model/norm/Mul:Mul*input:_2F_model_2F_model_2F_layers_2E_21_2F_Add_5F_1_5F_output_5F_0(1,0,2048);_2F_model_2F_model_2F_norm_2F_Div_5F_output_5F_0(1,0,1)*output:_2F_model_2F_model_2F_norm_2F_Mul_5F_output_5F_0(1,0,2048) /model/model/norm/Mul_1:Mul*input:model_2E_model_2E_norm_2E_weight.bin(float16:2048);_2F_model_2F_model_2F_norm_2F_Mul_5F_output_5F_0(1,0,2048)*output:_2F_model_2F_model_2F_norm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048) /model/lm_head/MatMul:MatMul*input:_2F_model_2F_model_2F_norm_2F_Mul_5F_1_5F_output_5F_0(1,0,2048);onnx_3A__3A_MatMul_5F_7658.bin(float16:2048,32003)*output:logits(1,0,32003)