8.9 KiB
8.9 KiB
| 1 | layer,module,loss,samples,damp,time |
|---|---|
| 2 | 0,self_attn.v_proj,0.0000008398,0.05000,3.626 |
| 3 | 0,self_attn.k_proj,0.0000009675,0.05000,3.894 |
| 4 | 0,self_attn.q_proj,0.0000020267,0.05000,3.921 |
| 5 | 0,self_attn.o_proj,0.0000021352,0.05000,0.764 |
| 6 | 0,mlp.gate_proj,0.0000197972,0.05000,2.009 |
| 7 | 0,mlp.up_proj,0.0000128524,0.05000,2.049 |
| 8 | 0,mlp.down_proj,0.0000043071,0.05000,2.590 |
| 9 | 1,self_attn.v_proj,0.0000006024,0.05000,4.008 |
| 10 | 1,self_attn.q_proj,0.0000012551,0.05000,4.125 |
| 11 | 1,self_attn.k_proj,0.0000006186,0.05000,4.172 |
| 12 | 1,self_attn.o_proj,0.0000006818,0.05000,1.591 |
| 13 | 1,mlp.gate_proj,0.0002898050,0.05000,2.585 |
| 14 | 1,mlp.up_proj,0.0001188112,0.05000,2.701 |
| 15 | 1,mlp.down_proj,0.0000065027,0.05000,3.925 |
| 16 | 2,self_attn.q_proj,0.0000027077,0.05000,4.020 |
| 17 | 2,self_attn.v_proj,0.0000012527,0.05000,4.084 |
| 18 | 2,self_attn.k_proj,0.0000012390,0.05000,4.077 |
| 19 | 2,self_attn.o_proj,0.0000011485,0.05000,1.080 |
| 20 | 2,mlp.gate_proj,0.0002410469,0.05000,1.939 |
| 21 | 2,mlp.up_proj,0.0000994754,0.05000,1.991 |
| 22 | 2,mlp.down_proj,0.0052349447,0.05000,2.563 |
| 23 | 3,self_attn.k_proj,0.0000090444,0.05000,3.422 |
| 24 | 3,self_attn.q_proj,0.0000174781,0.05000,3.473 |
| 25 | 3,self_attn.v_proj,0.0000092185,0.05000,3.435 |
| 26 | 3,self_attn.o_proj,0.0000022031,0.05000,1.213 |
| 27 | 3,mlp.gate_proj,0.0002953793,0.05000,2.184 |
| 28 | 3,mlp.up_proj,0.0001297567,0.05000,2.222 |
| 29 | 3,mlp.down_proj,0.0000122510,0.05000,2.570 |
| 30 | 4,self_attn.v_proj,0.0000088477,0.05000,3.349 |
| 31 | 4,self_attn.k_proj,0.0000081188,0.05000,3.382 |
| 32 | 4,self_attn.q_proj,0.0000165550,0.05000,3.424 |
| 33 | 4,self_attn.o_proj,0.0000034003,0.05000,0.853 |
| 34 | 4,mlp.gate_proj,0.0002626366,0.05000,1.896 |
| 35 | 4,mlp.up_proj,0.0001246561,0.05000,1.932 |
| 36 | 4,mlp.down_proj,0.0000179021,0.05000,2.540 |
| 37 | 5,self_attn.v_proj,0.0000134986,0.05000,3.336 |
| 38 | 5,self_attn.k_proj,0.0000108491,0.05000,3.354 |
| 39 | 5,self_attn.q_proj,0.0000251860,0.05000,3.433 |
| 40 | 5,self_attn.o_proj,0.0000050791,0.05000,0.768 |
| 41 | 5,mlp.gate_proj,0.0001303575,0.05000,2.189 |
| 42 | 5,mlp.up_proj,0.0000894665,0.05000,2.193 |
| 43 | 5,mlp.down_proj,0.0000214953,0.05000,3.112 |
| 44 | 6,self_attn.k_proj,0.0000100134,0.05000,3.482 |
| 45 | 6,self_attn.v_proj,0.0000103902,0.05000,3.516 |
| 46 | 6,self_attn.q_proj,0.0000207434,0.05000,3.553 |
| 47 | 6,self_attn.o_proj,0.0000053416,0.05000,0.858 |
| 48 | 6,mlp.up_proj,0.0001180585,0.05000,2.169 |
| 49 | 6,mlp.gate_proj,0.0001675667,0.05000,2.197 |
| 50 | 6,mlp.down_proj,0.0000290387,0.05000,2.959 |
| 51 | 7,self_attn.q_proj,0.0000449065,0.05000,3.457 |
| 52 | 7,self_attn.k_proj,0.0000192821,0.05000,3.492 |
| 53 | 7,self_attn.v_proj,0.0000234869,0.05000,3.489 |
| 54 | 7,self_attn.o_proj,0.0000085582,0.05000,0.825 |
| 55 | 7,mlp.gate_proj,0.0002227390,0.05000,2.001 |
| 56 | 7,mlp.up_proj,0.0001480158,0.05000,2.032 |
| 57 | 7,mlp.down_proj,0.0000303750,0.05000,2.681 |
| 58 | 8,self_attn.q_proj,0.0000543600,0.05000,3.252 |
| 59 | 8,self_attn.k_proj,0.0000271505,0.05000,3.266 |
| 60 | 8,self_attn.v_proj,0.0000257857,0.05000,3.332 |
| 61 | 8,self_attn.o_proj,0.0000072512,0.05000,0.798 |
| 62 | 8,mlp.up_proj,0.0001467434,0.05000,1.923 |
| 63 | 8,mlp.gate_proj,0.0001986378,0.05000,1.961 |
| 64 | 8,mlp.down_proj,0.0000381824,0.05000,2.596 |
| 65 | 9,self_attn.k_proj,0.0000496773,0.05000,3.310 |
| 66 | 9,self_attn.q_proj,0.0001079280,0.05000,3.415 |
| 67 | 9,self_attn.v_proj,0.0000501777,0.05000,3.450 |
| 68 | 9,self_attn.o_proj,0.0000210494,0.05000,0.810 |
| 69 | 9,mlp.up_proj,0.0001489672,0.05000,2.018 |
| 70 | 9,mlp.gate_proj,0.0001845221,0.05000,2.074 |
| 71 | 9,mlp.down_proj,0.0000494351,0.05000,2.579 |
| 72 | 10,self_attn.k_proj,0.0000368918,0.05000,3.372 |
| 73 | 10,self_attn.v_proj,0.0000397413,0.05000,3.451 |
| 74 | 10,self_attn.q_proj,0.0000817303,0.05000,3.489 |
| 75 | 10,self_attn.o_proj,0.0000214367,0.05000,0.834 |
| 76 | 10,mlp.gate_proj,0.0001894916,0.05000,2.030 |
| 77 | 10,mlp.up_proj,0.0001573987,0.05000,2.045 |
| 78 | 10,mlp.down_proj,0.0000817361,0.05000,2.763 |
| 79 | 11,self_attn.q_proj,0.0001917333,0.05000,3.454 |
| 80 | 11,self_attn.k_proj,0.0000857933,0.05000,3.472 |
| 81 | 11,self_attn.v_proj,0.0000807375,0.05000,3.529 |
| 82 | 11,self_attn.o_proj,0.0000535566,0.05000,1.224 |
| 83 | 11,mlp.gate_proj,0.0001578729,0.05000,1.960 |
| 84 | 11,mlp.up_proj,0.0001494230,0.05000,2.039 |
| 85 | 11,mlp.down_proj,0.0000893180,0.05000,2.668 |
| 86 | 12,self_attn.v_proj,0.0000763319,0.05000,3.301 |
| 87 | 12,self_attn.k_proj,0.0000745997,0.05000,3.364 |
| 88 | 12,self_attn.q_proj,0.0001640630,0.05000,3.380 |
| 89 | 12,self_attn.o_proj,0.0000343813,0.05000,1.200 |
| 90 | 12,mlp.up_proj,0.0001588971,0.05000,2.040 |
| 91 | 12,mlp.gate_proj,0.0001566968,0.05000,2.088 |
| 92 | 12,mlp.down_proj,0.0001093978,0.05000,2.760 |
| 93 | 13,self_attn.q_proj,0.0002423986,0.05000,3.553 |
| 94 | 13,self_attn.v_proj,0.0001063420,0.05000,3.661 |
| 95 | 13,self_attn.k_proj,0.0000977330,0.05000,3.718 |
| 96 | 13,self_attn.o_proj,0.0000535365,0.05000,0.863 |
| 97 | 13,mlp.gate_proj,0.0001755430,0.05000,2.180 |
| 98 | 13,mlp.up_proj,0.0001757205,0.05000,2.202 |
| 99 | 13,mlp.down_proj,0.0001403776,0.05000,3.126 |
| 100 | 14,self_attn.q_proj,0.0003975555,0.05000,3.663 |
| 101 | 14,self_attn.k_proj,0.0001722038,0.05000,3.679 |
| 102 | 14,self_attn.v_proj,0.0001753240,0.05000,3.693 |
| 103 | 14,self_attn.o_proj,0.0000595465,0.05000,0.862 |
| 104 | 14,mlp.gate_proj,0.0002011682,0.05000,2.305 |
| 105 | 14,mlp.up_proj,0.0002059663,0.05000,2.407 |
| 106 | 14,mlp.down_proj,0.0001984588,0.05000,3.394 |
| 107 | 15,self_attn.q_proj,0.0008955670,0.05000,3.697 |
| 108 | 15,self_attn.k_proj,0.0003116054,0.05000,3.737 |
| 109 | 15,self_attn.v_proj,0.0004181195,0.05000,3.791 |
| 110 | 15,self_attn.o_proj,0.0001117162,0.05000,0.872 |
| 111 | 15,mlp.gate_proj,0.0002431306,0.05000,2.044 |
| 112 | 15,mlp.up_proj,0.0002530560,0.05000,2.098 |
| 113 | 15,mlp.down_proj,0.0003042576,0.05000,2.786 |
| 114 | 16,self_attn.q_proj,0.0012468852,0.05000,3.625 |
| 115 | 16,self_attn.v_proj,0.0004789103,0.05000,3.631 |
| 116 | 16,self_attn.k_proj,0.0004600213,0.05000,3.666 |
| 117 | 16,self_attn.o_proj,0.0001508937,0.05000,0.845 |
| 118 | 16,mlp.up_proj,0.0003300065,0.05000,2.094 |
| 119 | 16,mlp.gate_proj,0.0002871720,0.05000,2.138 |
| 120 | 16,mlp.down_proj,0.0008724837,0.05000,2.769 |
| 121 | 17,self_attn.k_proj,0.0012925876,0.05000,3.544 |
| 122 | 17,self_attn.q_proj,0.0035529846,0.05000,3.629 |
| 123 | 17,self_attn.v_proj,0.0016027982,0.05000,3.645 |
| 124 | 17,self_attn.o_proj,0.0004630940,0.05000,0.837 |
| 125 | 17,mlp.up_proj,0.0005086830,0.05000,2.186 |
| 126 | 17,mlp.gate_proj,0.0004439794,0.05000,2.247 |
| 127 | 17,mlp.down_proj,0.0012258750,0.05000,2.896 |
| 128 | 18,self_attn.q_proj,0.0045345503,0.05000,3.490 |
| 129 | 18,self_attn.k_proj,0.0015577570,0.05000,3.584 |
| 130 | 18,self_attn.v_proj,0.0018117692,0.05000,3.587 |
| 131 | 18,self_attn.o_proj,0.0003796638,0.05000,0.831 |
| 132 | 18,mlp.up_proj,0.0006954494,0.05000,2.228 |
| 133 | 18,mlp.gate_proj,0.0006154586,0.05000,2.271 |
| 134 | 18,mlp.down_proj,0.0019921769,0.05000,2.766 |
| 135 | 19,self_attn.q_proj,0.0082794433,0.05000,3.540 |
| 136 | 19,self_attn.v_proj,0.0035734420,0.05000,3.562 |
| 137 | 19,self_attn.k_proj,0.0027089883,0.05000,3.621 |
| 138 | 19,self_attn.o_proj,0.0007345936,0.05000,0.883 |
| 139 | 19,mlp.up_proj,0.0010398286,0.05000,2.173 |
| 140 | 19,mlp.gate_proj,0.0009062103,0.05000,2.236 |
| 141 | 19,mlp.down_proj,0.0047007723,0.05000,2.741 |
| 142 | 20,self_attn.k_proj,0.0049409603,0.05000,3.462 |
| 143 | 20,self_attn.v_proj,0.0062833487,0.05000,3.546 |
| 144 | 20,self_attn.q_proj,0.0141710609,0.05000,3.616 |
| 145 | 20,self_attn.o_proj,0.0011029547,0.05000,0.901 |
| 146 | 20,mlp.up_proj,0.0013506484,0.05000,2.137 |
| 147 | 20,mlp.gate_proj,0.0011972403,0.05000,2.187 |
| 148 | 20,mlp.down_proj,0.0063867690,0.05000,2.863 |
| 149 | 21,self_attn.v_proj,0.0098904379,0.05000,3.438 |
| 150 | 21,self_attn.k_proj,0.0073228886,0.05000,3.531 |
| 151 | 21,self_attn.q_proj,0.0200067527,0.05000,3.600 |
| 152 | 21,self_attn.o_proj,0.0015405809,0.05000,0.902 |
| 153 | 21,mlp.gate_proj,0.0014323293,0.05000,2.231 |
| 154 | 21,mlp.up_proj,0.0016531138,0.05000,2.248 |
| 155 | 21,mlp.down_proj,0.0089285798,0.05000,2.775 |
| 156 | 22,self_attn.k_proj,0.0102904142,0.05000,3.497 |
| 157 | 22,self_attn.q_proj,0.0279091075,0.05000,3.558 |
| 158 | 22,self_attn.v_proj,0.0136463096,0.05000,3.561 |
| 159 | 22,self_attn.o_proj,0.0017626955,0.05000,0.875 |
| 160 | 22,mlp.up_proj,0.0019068349,0.05000,2.061 |
| 161 | 22,mlp.gate_proj,0.0016667122,0.05000,2.118 |
| 162 | 22,mlp.down_proj,0.0102301358,0.05000,2.784 |
| 163 | 23,self_attn.k_proj,0.0126362919,0.05000,3.470 |
| 164 | 23,self_attn.v_proj,0.0155633528,0.05000,3.562 |
| 165 | 23,self_attn.q_proj,0.0293295363,0.05000,3.648 |
| 166 | 23,self_attn.o_proj,0.0018687698,0.05000,0.878 |
| 167 | 23,mlp.gate_proj,0.0019935730,0.05000,2.148 |
| 168 | 23,mlp.up_proj,0.0022690450,0.05000,2.207 |
| 169 | 23,mlp.down_proj,0.0120860929,0.05000,2.854 |
| 170 | 24,self_attn.q_proj,0.0559459186,0.05000,3.423 |
| 171 | 24,self_attn.v_proj,0.0249366372,0.05000,3.485 |
| 172 | 24,self_attn.k_proj,0.0213881481,0.05000,3.516 |
| 173 | 24,self_attn.o_proj,0.0023866374,0.05000,1.135 |
| 174 | 24,mlp.up_proj,0.0022699147,0.05000,2.120 |
| 175 | 24,mlp.gate_proj,0.0019399213,0.05000,2.160 |
| 176 | 24,mlp.down_proj,0.0147081203,0.05000,2.699 |
| 177 | 25,self_attn.q_proj,0.0651313592,0.05000,3.340 |
| 178 | 25,self_attn.v_proj,0.0362999953,0.05000,3.385 |
| 179 | 25,self_attn.k_proj,0.0240991880,0.05000,3.424 |
| 180 | 25,self_attn.o_proj,0.0027495860,0.05000,1.318 |
| 181 | 25,mlp.up_proj,0.0022078746,0.05000,2.105 |
| 182 | 25,mlp.gate_proj,0.0018362088,0.05000,2.142 |
| 183 | 25,mlp.down_proj,0.0172555837,0.05000,2.694 |
| 184 | 26,self_attn.v_proj,0.0339960720,0.05000,3.506 |
| 185 | 26,self_attn.q_proj,0.0760382617,0.05000,3.565 |
| 186 | 26,self_attn.k_proj,0.0228024484,0.05000,3.605 |
| 187 | 26,self_attn.o_proj,0.0054717644,0.05000,1.536 |
| 188 | 26,mlp.gate_proj,0.0018231207,0.05000,2.180 |
| 189 | 26,mlp.up_proj,0.0021639906,0.05000,2.183 |
| 190 | 26,mlp.down_proj,0.0193138215,0.05000,2.813 |
| 191 | 27,self_attn.k_proj,0.0116609379,0.05000,3.489 |
| 192 | 27,self_attn.v_proj,0.0142843226,0.05000,3.582 |
| 193 | 27,self_attn.q_proj,0.0253722723,0.05000,3.648 |
| 194 | 27,self_attn.o_proj,0.0031136294,0.05000,1.527 |
| 195 | 27,mlp.up_proj,0.0031771694,0.05000,2.073 |
| 196 | 27,mlp.gate_proj,0.0031102671,0.05000,2.147 |
| 197 | 27,mlp.down_proj,0.0421205361,0.05000,2.798 |