10 KiB
10 KiB
| 1 | layer,module,loss,samples,damp,time |
|---|---|
| 2 | 0,self_attn.k_proj,0.0000000243,0.05000,1.346 |
| 3 | 0,self_attn.v_proj,0.0000000040,0.05000,1.044 |
| 4 | 0,self_attn.q_proj,0.0000000903,0.05000,1.080 |
| 5 | 0,self_attn.o_proj,0.0000000000,0.05000,1.098 |
| 6 | 0,mlp.gate_proj,0.0000000940,0.05000,1.349 |
| 7 | 0,mlp.up_proj,0.0000000836,0.05000,1.323 |
| 8 | 0,mlp.down_proj,0.0000000001,0.05000,4.570 |
| 9 | 1,self_attn.k_proj,0.0000001356,0.05000,1.074 |
| 10 | 1,self_attn.v_proj,0.0000000265,0.05000,1.093 |
| 11 | 1,self_attn.q_proj,0.0000002987,0.05000,1.117 |
| 12 | 1,self_attn.o_proj,0.0000000001,0.05000,1.154 |
| 13 | 1,mlp.gate_proj,0.0000003163,0.05000,1.123 |
| 14 | 1,mlp.up_proj,0.0000002789,0.05000,1.105 |
| 15 | 1,mlp.down_proj,0.0000000908,0.05000,4.407 |
| 16 | 2,self_attn.k_proj,0.0000009065,0.05000,1.114 |
| 17 | 2,self_attn.v_proj,0.0000000839,0.05000,1.056 |
| 18 | 2,self_attn.q_proj,0.0000016890,0.05000,1.104 |
| 19 | 2,self_attn.o_proj,0.0000000001,0.05000,1.082 |
| 20 | 2,mlp.gate_proj,0.0000006744,0.05000,1.100 |
| 21 | 2,mlp.up_proj,0.0000005911,0.05000,1.105 |
| 22 | 2,mlp.down_proj,0.0000000003,0.05000,4.560 |
| 23 | 3,self_attn.k_proj,0.0000006056,0.05000,1.147 |
| 24 | 3,self_attn.v_proj,0.0000000826,0.05000,1.076 |
| 25 | 3,self_attn.q_proj,0.0000011741,0.05000,1.094 |
| 26 | 3,self_attn.o_proj,0.0000000001,0.05000,1.081 |
| 27 | 3,mlp.gate_proj,0.0000011593,0.05000,1.104 |
| 28 | 3,mlp.up_proj,0.0000010041,0.05000,1.086 |
| 29 | 3,mlp.down_proj,0.0000000005,0.05000,4.424 |
| 30 | 4,self_attn.k_proj,0.0000008699,0.05000,1.165 |
| 31 | 4,self_attn.v_proj,0.0000001383,0.05000,1.057 |
| 32 | 4,self_attn.q_proj,0.0000018787,0.05000,1.104 |
| 33 | 4,self_attn.o_proj,0.0000000002,0.05000,1.063 |
| 34 | 4,mlp.gate_proj,0.0000017767,0.05000,1.102 |
| 35 | 4,mlp.up_proj,0.0000014464,0.05000,1.073 |
| 36 | 4,mlp.down_proj,0.0000000010,0.05000,4.394 |
| 37 | 5,self_attn.k_proj,0.0000011964,0.05000,1.125 |
| 38 | 5,self_attn.v_proj,0.0000001727,0.05000,1.074 |
| 39 | 5,self_attn.q_proj,0.0000026247,0.05000,1.063 |
| 40 | 5,self_attn.o_proj,0.0000000004,0.05000,1.098 |
| 41 | 5,mlp.gate_proj,0.0000026750,0.05000,1.144 |
| 42 | 5,mlp.up_proj,0.0000020422,0.05000,1.143 |
| 43 | 5,mlp.down_proj,0.0000000018,0.05000,4.481 |
| 44 | 6,self_attn.k_proj,0.0000012233,0.05000,1.092 |
| 45 | 6,self_attn.v_proj,0.0000001751,0.05000,1.063 |
| 46 | 6,self_attn.q_proj,0.0000025896,0.05000,1.096 |
| 47 | 6,self_attn.o_proj,0.0000000005,0.05000,1.093 |
| 48 | 6,mlp.gate_proj,0.0000033022,0.05000,1.108 |
| 49 | 6,mlp.up_proj,0.0000025730,0.05000,1.122 |
| 50 | 6,mlp.down_proj,0.0000000026,0.05000,4.475 |
| 51 | 7,self_attn.k_proj,0.0000015788,0.05000,1.092 |
| 52 | 7,self_attn.v_proj,0.0000002271,0.05000,1.053 |
| 53 | 7,self_attn.q_proj,0.0000032580,0.05000,1.086 |
| 54 | 7,self_attn.o_proj,0.0000000009,0.05000,1.101 |
| 55 | 7,mlp.gate_proj,0.0000040918,0.05000,1.105 |
| 56 | 7,mlp.up_proj,0.0000031199,0.05000,1.098 |
| 57 | 7,mlp.down_proj,0.0000000035,0.05000,4.402 |
| 58 | 8,self_attn.k_proj,0.0000013194,0.05000,1.116 |
| 59 | 8,self_attn.v_proj,0.0000002595,0.05000,1.085 |
| 60 | 8,self_attn.q_proj,0.0000028454,0.05000,1.102 |
| 61 | 8,self_attn.o_proj,0.0000000011,0.05000,1.071 |
| 62 | 8,mlp.gate_proj,0.0000046629,0.05000,1.125 |
| 63 | 8,mlp.up_proj,0.0000036305,0.05000,1.095 |
| 64 | 8,mlp.down_proj,0.0000000043,0.05000,4.342 |
| 65 | 9,self_attn.k_proj,0.0000019476,0.05000,1.046 |
| 66 | 9,self_attn.v_proj,0.0000003167,0.05000,1.053 |
| 67 | 9,self_attn.q_proj,0.0000040997,0.05000,1.078 |
| 68 | 9,self_attn.o_proj,0.0000000013,0.05000,1.072 |
| 69 | 9,mlp.gate_proj,0.0000050292,0.05000,1.109 |
| 70 | 9,mlp.up_proj,0.0000040537,0.05000,1.096 |
| 71 | 9,mlp.down_proj,0.0000000050,0.05000,4.360 |
| 72 | 10,self_attn.k_proj,0.0000017725,0.05000,1.063 |
| 73 | 10,self_attn.v_proj,0.0000002921,0.05000,1.088 |
| 74 | 10,self_attn.q_proj,0.0000036800,0.05000,1.225 |
| 75 | 10,self_attn.o_proj,0.0000000018,0.05000,1.099 |
| 76 | 10,mlp.gate_proj,0.0000052495,0.05000,1.118 |
| 77 | 10,mlp.up_proj,0.0000043227,0.05000,1.088 |
| 78 | 10,mlp.down_proj,0.0000000060,0.05000,4.511 |
| 79 | 11,self_attn.k_proj,0.0000021077,0.05000,1.058 |
| 80 | 11,self_attn.v_proj,0.0000004395,0.05000,1.063 |
| 81 | 11,self_attn.q_proj,0.0000045010,0.05000,1.069 |
| 82 | 11,self_attn.o_proj,0.0000000029,0.05000,1.066 |
| 83 | 11,mlp.gate_proj,0.0000057856,0.05000,1.111 |
| 84 | 11,mlp.up_proj,0.0000048466,0.05000,1.083 |
| 85 | 11,mlp.down_proj,0.0000000072,0.05000,4.406 |
| 86 | 12,self_attn.k_proj,0.0000027316,0.05000,1.064 |
| 87 | 12,self_attn.v_proj,0.0000004806,0.05000,1.100 |
| 88 | 12,self_attn.q_proj,0.0000059322,0.05000,1.121 |
| 89 | 12,self_attn.o_proj,0.0000000029,0.05000,1.075 |
| 90 | 12,mlp.gate_proj,0.0000064971,0.05000,1.098 |
| 91 | 12,mlp.up_proj,0.0000055545,0.05000,1.127 |
| 92 | 12,mlp.down_proj,0.0000000088,0.05000,4.425 |
| 93 | 13,self_attn.k_proj,0.0000024133,0.05000,1.064 |
| 94 | 13,self_attn.v_proj,0.0000005040,0.05000,1.064 |
| 95 | 13,self_attn.q_proj,0.0000049132,0.05000,1.111 |
| 96 | 13,self_attn.o_proj,0.0000000038,0.05000,1.058 |
| 97 | 13,mlp.gate_proj,0.0000072860,0.05000,1.325 |
| 98 | 13,mlp.up_proj,0.0000064150,0.05000,1.165 |
| 99 | 13,mlp.down_proj,0.0000000112,0.05000,4.474 |
| 100 | 14,self_attn.k_proj,0.0000024321,0.05000,1.054 |
| 101 | 14,self_attn.v_proj,0.0000008004,0.05000,1.082 |
| 102 | 14,self_attn.q_proj,0.0000056635,0.05000,1.112 |
| 103 | 14,self_attn.o_proj,0.0000000058,0.05000,1.097 |
| 104 | 14,mlp.gate_proj,0.0000083033,0.05000,1.100 |
| 105 | 14,mlp.up_proj,0.0000073240,0.05000,1.117 |
| 106 | 14,mlp.down_proj,0.0000000147,0.05000,4.616 |
| 107 | 15,self_attn.k_proj,0.0000032265,0.05000,1.105 |
| 108 | 15,self_attn.v_proj,0.0000009160,0.05000,1.047 |
| 109 | 15,self_attn.q_proj,0.0000071495,0.05000,1.086 |
| 110 | 15,self_attn.o_proj,0.0000000060,0.05000,1.086 |
| 111 | 15,mlp.gate_proj,0.0000098523,0.05000,1.120 |
| 112 | 15,mlp.up_proj,0.0000083736,0.05000,1.130 |
| 113 | 15,mlp.down_proj,0.0000000185,0.05000,4.463 |
| 114 | 16,self_attn.k_proj,0.0000031272,0.05000,1.091 |
| 115 | 16,self_attn.v_proj,0.0000009370,0.05000,1.104 |
| 116 | 16,self_attn.q_proj,0.0000068247,0.05000,1.105 |
| 117 | 16,self_attn.o_proj,0.0000000068,0.05000,1.062 |
| 118 | 16,mlp.gate_proj,0.0000123035,0.05000,1.140 |
| 119 | 16,mlp.up_proj,0.0000099590,0.05000,1.118 |
| 120 | 16,mlp.down_proj,0.0000000257,0.05000,4.493 |
| 121 | 17,self_attn.k_proj,0.0000025959,0.05000,1.102 |
| 122 | 17,self_attn.v_proj,0.0000008746,0.05000,1.019 |
| 123 | 17,self_attn.q_proj,0.0000061843,0.05000,1.097 |
| 124 | 17,self_attn.o_proj,0.0000000070,0.05000,1.062 |
| 125 | 17,mlp.gate_proj,0.0000143714,0.05000,1.106 |
| 126 | 17,mlp.up_proj,0.0000116310,0.05000,1.075 |
| 127 | 17,mlp.down_proj,0.0000000354,0.05000,4.374 |
| 128 | 18,self_attn.k_proj,0.0000031679,0.05000,1.108 |
| 129 | 18,self_attn.v_proj,0.0000010451,0.05000,1.063 |
| 130 | 18,self_attn.q_proj,0.0000077381,0.05000,1.101 |
| 131 | 18,self_attn.o_proj,0.0000000081,0.05000,1.116 |
| 132 | 18,mlp.gate_proj,0.0000163663,0.05000,1.183 |
| 133 | 18,mlp.up_proj,0.0000131605,0.05000,1.079 |
| 134 | 18,mlp.down_proj,0.0000000498,0.05000,4.473 |
| 135 | 19,self_attn.k_proj,0.0000032494,0.05000,1.097 |
| 136 | 19,self_attn.v_proj,0.0000012439,0.05000,1.052 |
| 137 | 19,self_attn.q_proj,0.0000074487,0.05000,1.106 |
| 138 | 19,self_attn.o_proj,0.0000000095,0.05000,1.069 |
| 139 | 19,mlp.gate_proj,0.0000187246,0.05000,1.102 |
| 140 | 19,mlp.up_proj,0.0000147143,0.05000,1.099 |
| 141 | 19,mlp.down_proj,0.0000000669,0.05000,4.367 |
| 142 | 20,self_attn.k_proj,0.0000034868,0.05000,1.079 |
| 143 | 20,self_attn.v_proj,0.0000013987,0.05000,1.067 |
| 144 | 20,self_attn.q_proj,0.0000081005,0.05000,1.068 |
| 145 | 20,self_attn.o_proj,0.0000000096,0.05000,1.068 |
| 146 | 20,mlp.gate_proj,0.0000216847,0.05000,1.121 |
| 147 | 20,mlp.up_proj,0.0000165116,0.05000,1.085 |
| 148 | 20,mlp.down_proj,0.0000000704,0.05000,4.444 |
| 149 | 21,self_attn.k_proj,0.0000034291,0.05000,1.102 |
| 150 | 21,self_attn.v_proj,0.0000015221,0.05000,1.075 |
| 151 | 21,self_attn.q_proj,0.0000080014,0.05000,1.108 |
| 152 | 21,self_attn.o_proj,0.0000000102,0.05000,1.090 |
| 153 | 21,mlp.gate_proj,0.0000261188,0.05000,1.127 |
| 154 | 21,mlp.up_proj,0.0000185324,0.05000,1.129 |
| 155 | 21,mlp.down_proj,0.0000000756,0.05000,4.455 |
| 156 | 22,self_attn.k_proj,0.0000032560,0.05000,1.095 |
| 157 | 22,self_attn.v_proj,0.0000016688,0.05000,1.114 |
| 158 | 22,self_attn.q_proj,0.0000078390,0.05000,1.093 |
| 159 | 22,self_attn.o_proj,0.0000000073,0.05000,1.072 |
| 160 | 22,mlp.gate_proj,0.0000279689,0.05000,1.113 |
| 161 | 22,mlp.up_proj,0.0000200703,0.05000,1.082 |
| 162 | 22,mlp.down_proj,0.0000000813,0.05000,4.367 |
| 163 | 23,self_attn.k_proj,0.0000031836,0.05000,1.087 |
| 164 | 23,self_attn.v_proj,0.0000016718,0.05000,1.033 |
| 165 | 23,self_attn.q_proj,0.0000079744,0.05000,1.077 |
| 166 | 23,self_attn.o_proj,0.0000000097,0.05000,1.138 |
| 167 | 23,mlp.gate_proj,0.0000311143,0.05000,1.119 |
| 168 | 23,mlp.up_proj,0.0000222295,0.05000,1.104 |
| 169 | 23,mlp.down_proj,0.0000000910,0.05000,4.489 |
| 170 | 24,self_attn.k_proj,0.0000037302,0.05000,1.076 |
| 171 | 24,self_attn.v_proj,0.0000019298,0.05000,1.046 |
| 172 | 24,self_attn.q_proj,0.0000092244,0.05000,1.073 |
| 173 | 24,self_attn.o_proj,0.0000000099,0.05000,1.096 |
| 174 | 24,mlp.gate_proj,0.0000349600,0.05000,1.102 |
| 175 | 24,mlp.up_proj,0.0000245547,0.05000,1.057 |
| 176 | 24,mlp.down_proj,0.0000000984,0.05000,4.411 |
| 177 | 25,self_attn.k_proj,0.0000035627,0.05000,1.108 |
| 178 | 25,self_attn.v_proj,0.0000022298,0.05000,1.061 |
| 179 | 25,self_attn.q_proj,0.0000092782,0.05000,1.120 |
| 180 | 25,self_attn.o_proj,0.0000000102,0.05000,1.109 |
| 181 | 25,mlp.gate_proj,0.0000381017,0.05000,1.124 |
| 182 | 25,mlp.up_proj,0.0000269863,0.05000,1.131 |
| 183 | 25,mlp.down_proj,0.0000001128,0.05000,4.506 |
| 184 | 26,self_attn.k_proj,0.0000033970,0.05000,1.119 |
| 185 | 26,self_attn.v_proj,0.0000023643,0.05000,1.134 |
| 186 | 26,self_attn.q_proj,0.0000088182,0.05000,1.124 |
| 187 | 26,self_attn.o_proj,0.0000000131,0.05000,1.094 |
| 188 | 26,mlp.gate_proj,0.0000397645,0.05000,1.222 |
| 189 | 26,mlp.up_proj,0.0000291584,0.05000,1.126 |
| 190 | 26,mlp.down_proj,0.0000001288,0.05000,4.435 |
| 191 | 27,self_attn.k_proj,0.0000032655,0.05000,1.078 |
| 192 | 27,self_attn.v_proj,0.0000021690,0.05000,1.054 |
| 193 | 27,self_attn.q_proj,0.0000090136,0.05000,1.098 |
| 194 | 27,self_attn.o_proj,0.0000000146,0.05000,1.074 |
| 195 | 27,mlp.gate_proj,0.0000431077,0.05000,1.092 |
| 196 | 27,mlp.up_proj,0.0000319081,0.05000,1.129 |
| 197 | 27,mlp.down_proj,0.0000001526,0.05000,4.389 |
| 198 | 28,self_attn.k_proj,0.0000033885,0.05000,1.079 |
| 199 | 28,self_attn.v_proj,0.0000031190,0.05000,1.050 |
| 200 | 28,self_attn.q_proj,0.0000090153,0.05000,1.100 |
| 201 | 28,self_attn.o_proj,0.0000000257,0.05000,1.092 |
| 202 | 28,mlp.gate_proj,0.0000462653,0.05000,1.106 |
| 203 | 28,mlp.up_proj,0.0000354446,0.05000,1.100 |
| 204 | 28,mlp.down_proj,0.0000002016,0.05000,4.429 |
| 205 | 29,self_attn.k_proj,0.0000032734,0.05000,1.080 |
| 206 | 29,self_attn.v_proj,0.0000044012,0.05000,1.043 |
| 207 | 29,self_attn.q_proj,0.0000103499,0.05000,1.089 |
| 208 | 29,self_attn.o_proj,0.0000000528,0.05000,1.084 |
| 209 | 29,mlp.gate_proj,0.0000452730,0.05000,1.120 |
| 210 | 29,mlp.up_proj,0.0000361531,0.05000,1.125 |
| 211 | 29,mlp.down_proj,0.0000002568,0.05000,4.422 |
| 212 | 30,self_attn.k_proj,0.0000033870,0.05000,1.224 |
| 213 | 30,self_attn.v_proj,0.0000048556,0.05000,1.072 |
| 214 | 30,self_attn.q_proj,0.0000102100,0.05000,1.077 |
| 215 | 30,self_attn.o_proj,0.0000000454,0.05000,1.065 |
| 216 | 30,mlp.gate_proj,0.0000465985,0.05000,1.123 |
| 217 | 30,mlp.up_proj,0.0000379377,0.05000,1.112 |
| 218 | 30,mlp.down_proj,0.0000003248,0.05000,4.327 |
| 219 | 31,self_attn.k_proj,0.0000031314,0.05000,1.107 |
| 220 | 31,self_attn.v_proj,0.0000050672,0.05000,1.026 |
| 221 | 31,self_attn.q_proj,0.0000093823,0.05000,1.097 |
| 222 | 31,self_attn.o_proj,0.0000000755,0.05000,1.084 |
| 223 | 31,mlp.gate_proj,0.0000427195,0.05000,1.119 |
| 224 | 31,mlp.up_proj,0.0000335756,0.05000,1.123 |
| 225 | 31,mlp.down_proj,0.0000004450,0.05000,4.457 |