7767517 87 108 Input in0 0 1 in0 Input in1 0 1 in1 Split splitncnn_0 1 2 in1 2 3 Input in2 0 1 in2 Input kv_cache 0 4 cache_k0 cache_v0 cache_k1 cache_v1 Split splitncnn_1 1 2 in2 5 6 Gemm gemm_0 1 1 in0 7 10=4 2=0 3=1 4=0 5=1 6=1 7=144 8=384 9=2048 Split splitncnn_2 1 4 7 8 9 10 11 LayerNorm ln_22 1 1 3 12 0=384 1=1.000000e-5 2=1 Split splitncnn_3 1 3 12 13 14 15 Gemm gemm_1 1 1 15 16 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 Gemm gemm_2 1 1 14 17 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 Gemm gemm_3 1 1 13 18 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 Reshape reshape_47 1 1 17 19 0=24 1=16 2=-1 Permute permute_31 1 1 19 20 0=2 Reshape reshape_48 1 1 18 21 0=24 1=16 2=-1 Permute permute_32 1 1 21 22 0=2 Reshape reshape_49 1 1 16 23 0=24 1=16 2=-1 Permute permute_33 1 1 23 24 0=2 SDPA sdpa_77 6 3 24 20 22 6 cache_k0 cache_v0 25 out_cache_k0 out_cache_v0 5=1 6=0.204124 7=1 Permute permute_34 1 1 25 26 0=2 Reshape reshape_50 1 1 26 27 0=384 1=-1 Gemm gemm_4 1 1 27 28 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 BinaryOp add_0 2 1 2 28 29 0=0 Split splitncnn_4 1 2 29 30 31 LayerNorm ln_23 1 1 31 32 0=384 1=1.000000e-5 2=1 Gemm gemm_5 1 1 32 33 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 Gemm gemm_6 1 1 11 34 10=4 2=0 3=1 4=0 5=1 6=1 7=144 8=384 9=384 Gemm gemm_7 1 1 10 35 10=4 2=0 3=1 4=0 5=1 6=1 7=144 8=384 9=384 Reshape reshape_51 1 1 34 36 0=24 1=16 2=144 Permute permute_35 1 1 36 37 0=2 Reshape reshape_52 1 1 35 38 0=24 1=16 2=144 Permute permute_36 1 1 38 39 0=2 Reshape reshape_53 1 1 33 40 0=24 1=16 2=-1 Permute permute_37 1 1 40 41 0=2 SDPA sdpa_75 3 1 41 37 39 42 5=0 Permute permute_38 1 1 42 43 0=2 Reshape reshape_54 1 1 43 44 0=384 1=-1 Gemm gemm_8 1 1 44 45 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 BinaryOp add_1 2 1 30 45 46 0=0 Split splitncnn_5 1 2 46 47 48 LayerNorm ln_24 1 1 48 49 0=384 1=1.000000e-5 2=1 Gemm gemm_9 1 1 49 50 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=1536 9=384 GELU gelu_29 1 1 50 51 Gemm gemm_10 1 1 51 52 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=1536 BinaryOp add_2 2 1 47 52 53 0=0 Split splitncnn_6 1 2 53 54 55 LayerNorm ln_25 1 1 55 56 0=384 1=1.000000e-5 2=1 Split splitncnn_7 1 3 56 57 58 59 Gemm gemm_11 1 1 59 60 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 Gemm gemm_12 1 1 58 61 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 Gemm gemm_13 1 1 57 62 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 Reshape reshape_55 1 1 61 63 0=24 1=16 2=-1 Permute permute_39 1 1 63 64 0=2 Reshape reshape_56 1 1 62 65 0=24 1=16 2=-1 Permute permute_40 1 1 65 66 0=2 Reshape reshape_57 1 1 60 67 0=24 1=16 2=-1 Permute permute_41 1 1 67 68 0=2 SDPA sdpa_78 6 3 68 64 66 5 cache_k1 cache_v1 69 out_cache_k1 out_cache_v1 5=1 6=0.204124 7=1 Permute permute_42 1 1 69 70 0=2 Reshape reshape_58 1 1 70 71 0=384 1=-1 Gemm gemm_14 1 1 71 72 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 BinaryOp add_3 2 1 54 72 73 0=0 Split splitncnn_8 1 2 73 74 75 LayerNorm ln_26 1 1 75 76 0=384 1=1.000000e-5 2=1 Gemm gemm_15 1 1 76 77 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 Gemm gemm_16 1 1 9 78 10=4 2=0 3=1 4=0 5=1 6=1 7=144 8=384 9=384 Gemm gemm_17 1 1 8 79 10=4 2=0 3=1 4=0 5=1 6=1 7=144 8=384 9=384 Reshape reshape_59 1 1 78 80 0=24 1=16 2=144 Permute permute_43 1 1 80 81 0=2 Reshape reshape_60 1 1 79 82 0=24 1=16 2=144 Permute permute_44 1 1 82 83 0=2 Reshape reshape_61 1 1 77 84 0=24 1=16 2=-1 Permute permute_45 1 1 84 85 0=2 SDPA sdpa_76 3 1 85 81 83 86 5=0 Permute permute_46 1 1 86 87 0=2 Reshape reshape_62 1 1 87 88 0=384 1=-1 Gemm gemm_18 1 1 88 89 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=384 BinaryOp add_4 2 1 74 89 90 0=0 Split splitncnn_9 1 2 90 91 92 LayerNorm ln_27 1 1 92 93 0=384 1=1.000000e-5 2=1 Gemm gemm_19 1 1 93 94 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=1536 9=384 GELU gelu_30 1 1 94 95 Gemm gemm_20 1 1 95 96 10=4 2=0 3=1 4=0 5=1 6=1 7=0 8=384 9=1536 BinaryOp add_5 2 1 91 96 97 0=0 LayerNorm ln_28 1 1 97 98 0=384 1=1.000000e-5 2=1 Gemm gemm_21 1 1 98 out0 10=-1 2=0 3=1 4=0 5=1 6=1 7=0 8=50000 9=384