| layer,module,loss,samples,damp,time | |
| 0,self_attn.q_proj,0.0000000075,0.05000,3.008 | |
| 0,self_attn.k_proj,0.0000000030,0.05000,3.031 | |
| 0,self_attn.v_proj,0.0000000002,0.05000,3.052 | |
| 0,self_attn.o_proj,0.0000000000,0.05000,1.200 | |
| 0,mlp.up_proj,0.0000000054,0.05000,1.456 | |
| 0,mlp.gate_proj,0.0000000059,0.05000,1.473 | |
| 0,mlp.down_proj,0.0000000113,0.05000,2.276 | |
| 1,self_attn.q_proj,0.0000000300,0.05000,2.750 | |
| 1,self_attn.v_proj,0.0000000024,0.05000,2.774 | |
| 1,self_attn.k_proj,0.0000000121,0.05000,2.783 | |
| 1,self_attn.o_proj,0.0000000000,0.05000,1.217 | |
| 1,mlp.gate_proj,0.0000000097,0.05000,2.606 | |
| 1,mlp.up_proj,0.0000000087,0.05000,2.636 | |
| 1,mlp.down_proj,0.0000000003,0.05000,2.435 | |
| 2,self_attn.v_proj,0.0000000030,0.05000,2.871 | |
| 2,self_attn.k_proj,0.0000000112,0.05000,2.881 | |
| 2,self_attn.q_proj,0.0000000293,0.05000,2.888 | |
| 2,self_attn.o_proj,0.0000000001,0.05000,1.247 | |
| 2,mlp.gate_proj,0.0000000248,0.05000,1.672 | |
| 2,mlp.up_proj,0.0000000208,0.05000,1.680 | |
| 2,mlp.down_proj,0.0000000163,0.05000,2.173 | |
| 3,self_attn.q_proj,0.0000000690,0.05000,2.740 | |
| 3,self_attn.v_proj,0.0000000055,0.05000,2.745 | |
| 3,self_attn.k_proj,0.0000000241,0.05000,2.754 | |
| 3,self_attn.o_proj,0.0000000001,0.05000,1.247 | |
| 3,mlp.gate_proj,0.0000000285,0.05000,1.467 | |
| 3,mlp.up_proj,0.0000000232,0.05000,1.512 | |
| 3,mlp.down_proj,0.0000000009,0.05000,2.859 | |
| 4,self_attn.q_proj,0.0000000442,0.05000,2.484 | |
| 4,self_attn.v_proj,0.0000000051,0.05000,2.488 | |
| 4,self_attn.k_proj,0.0000000158,0.05000,2.498 | |
| 4,self_attn.o_proj,0.0000000002,0.05000,1.251 | |
| 4,mlp.up_proj,0.0000000274,0.05000,1.512 | |
| 4,mlp.gate_proj,0.0000000344,0.05000,1.520 | |
| 4,mlp.down_proj,0.0000000013,0.05000,2.869 | |
| 5,self_attn.q_proj,0.0000000603,0.05000,2.531 | |
| 5,self_attn.k_proj,0.0000000206,0.05000,2.571 | |
| 5,self_attn.v_proj,0.0000000055,0.05000,2.579 | |
| 5,self_attn.o_proj,0.0000000003,0.05000,1.312 | |
| 5,mlp.up_proj,0.0000000321,0.05000,2.469 | |
| 5,mlp.gate_proj,0.0000000390,0.05000,2.480 | |
| 5,mlp.down_proj,0.0000000016,0.05000,2.164 | |
| 6,self_attn.q_proj,0.0000000641,0.05000,3.809 | |
| 6,self_attn.k_proj,0.0000000242,0.05000,3.838 | |
| 6,self_attn.v_proj,0.0000000058,0.05000,3.922 | |
| 6,self_attn.o_proj,0.0000000006,0.05000,1.298 | |
| 6,mlp.up_proj,0.0000000361,0.05000,1.936 | |
| 6,mlp.gate_proj,0.0000000441,0.05000,1.941 | |
| 6,mlp.down_proj,0.0000000020,0.05000,1.926 | |
| 7,self_attn.q_proj,0.0000000627,0.05000,2.615 | |
| 7,self_attn.k_proj,0.0000000221,0.05000,2.627 | |
| 7,self_attn.v_proj,0.0000000070,0.05000,2.644 | |
| 7,self_attn.o_proj,0.0000000008,0.05000,0.956 | |
| 7,mlp.up_proj,0.0000000377,0.05000,1.592 | |
| 7,mlp.gate_proj,0.0000000447,0.05000,1.611 | |
| 7,mlp.down_proj,0.0000000021,0.05000,2.775 | |
| 8,self_attn.q_proj,0.0000000653,0.05000,3.218 | |
| 8,self_attn.k_proj,0.0000000243,0.05000,3.319 | |
| 8,self_attn.v_proj,0.0000000069,0.05000,3.373 | |
| 8,self_attn.o_proj,0.0000000009,0.05000,1.220 | |
| 8,mlp.up_proj,0.0000000404,0.05000,1.542 | |
| 8,mlp.gate_proj,0.0000000455,0.05000,1.555 | |
| 8,mlp.down_proj,0.0000000023,0.05000,2.556 | |
| 9,self_attn.q_proj,0.0000000673,0.05000,2.872 | |
| 9,self_attn.k_proj,0.0000000240,0.05000,2.893 | |
| 9,self_attn.v_proj,0.0000000082,0.05000,2.906 | |
| 9,self_attn.o_proj,0.0000000013,0.05000,1.218 | |
| 9,mlp.gate_proj,0.0000000456,0.05000,1.421 | |
| 9,mlp.up_proj,0.0000000422,0.05000,1.434 | |
| 9,mlp.down_proj,0.0000000027,0.05000,2.207 | |
| 10,self_attn.q_proj,0.0000000675,0.05000,2.962 | |
| 10,self_attn.k_proj,0.0000000243,0.05000,3.045 | |
| 10,self_attn.v_proj,0.0000000080,0.05000,3.081 | |
| 10,self_attn.o_proj,0.0000000014,0.05000,1.270 | |
| 10,mlp.gate_proj,0.0000000466,0.05000,2.136 | |
| 10,mlp.up_proj,0.0000000431,0.05000,2.152 | |
| 10,mlp.down_proj,0.0000000027,0.05000,2.642 | |
| 11,self_attn.v_proj,0.0000000084,0.05000,3.812 | |
| 11,self_attn.k_proj,0.0000000279,0.05000,3.848 | |
| 11,self_attn.q_proj,0.0000000745,0.05000,3.951 | |
| 11,self_attn.o_proj,0.0000000015,0.05000,1.160 | |
| 11,mlp.up_proj,0.0000000442,0.05000,1.585 | |
| 11,mlp.gate_proj,0.0000000537,0.05000,1.601 | |
| 11,mlp.down_proj,0.0000000028,0.05000,2.170 | |
| 12,self_attn.v_proj,0.0000000095,0.05000,2.501 | |
| 12,self_attn.k_proj,0.0000000247,0.05000,2.519 | |
| 12,self_attn.q_proj,0.0000000698,0.05000,2.529 | |
| 12,self_attn.o_proj,0.0000000014,0.05000,1.287 | |
| 12,mlp.up_proj,0.0000000480,0.05000,1.503 | |
| 12,mlp.gate_proj,0.0000000539,0.05000,1.529 | |
| 12,mlp.down_proj,0.0000000031,0.05000,2.669 | |
| 13,self_attn.q_proj,0.0000000884,0.05000,2.548 | |
| 13,self_attn.v_proj,0.0000000099,0.05000,2.572 | |
| 13,self_attn.k_proj,0.0000000310,0.05000,2.574 | |
| 13,self_attn.o_proj,0.0000000016,0.05000,1.187 | |
| 13,mlp.up_proj,0.0000000504,0.05000,2.487 | |
| 13,mlp.gate_proj,0.0000000544,0.05000,2.501 | |
| 13,mlp.down_proj,0.0000000052,0.05000,2.544 | |
| 14,self_attn.q_proj,0.0000000950,0.05000,3.766 | |
| 14,self_attn.k_proj,0.0000000318,0.05000,3.796 | |
| 14,self_attn.v_proj,0.0000000102,0.05000,3.879 | |
| 14,self_attn.o_proj,0.0000000017,0.05000,1.170 | |
| 14,mlp.gate_proj,0.0000000566,0.05000,1.556 | |
| 14,mlp.up_proj,0.0000000554,0.05000,1.584 | |
| 14,mlp.down_proj,0.0000000039,0.05000,2.197 | |
| 15,self_attn.v_proj,0.0000000092,0.05000,2.836 | |
| 15,self_attn.q_proj,0.0000000792,0.05000,2.842 | |
| 15,self_attn.k_proj,0.0000000302,0.05000,2.860 | |
| 15,self_attn.o_proj,0.0000000017,0.05000,1.298 | |
| 15,mlp.gate_proj,0.0000000563,0.05000,1.485 | |
| 15,mlp.up_proj,0.0000000563,0.05000,1.498 | |
| 15,mlp.down_proj,0.0000000041,0.05000,2.854 | |
| 16,self_attn.q_proj,0.0000000807,0.05000,2.401 | |
| 16,self_attn.k_proj,0.0000000299,0.05000,2.433 | |
| 16,self_attn.v_proj,0.0000000123,0.05000,2.447 | |
| 16,self_attn.o_proj,0.0000000022,0.05000,1.260 | |
| 16,mlp.gate_proj,0.0000000590,0.05000,1.909 | |
| 16,mlp.up_proj,0.0000000594,0.05000,1.936 | |
| 16,mlp.down_proj,0.0000000046,0.05000,2.199 | |
| 17,self_attn.q_proj,0.0000000812,0.05000,3.101 | |
| 17,self_attn.v_proj,0.0000000129,0.05000,3.159 | |
| 17,self_attn.k_proj,0.0000000285,0.05000,3.158 | |
| 17,self_attn.o_proj,0.0000000027,0.05000,1.260 | |
| 17,mlp.up_proj,0.0000000625,0.05000,1.718 | |
| 17,mlp.gate_proj,0.0000000618,0.05000,1.748 | |
| 17,mlp.down_proj,0.0000000059,0.05000,2.384 | |
| 18,self_attn.v_proj,0.0000000151,0.05000,2.679 | |
| 18,self_attn.k_proj,0.0000000320,0.05000,2.693 | |
| 18,self_attn.q_proj,0.0000000885,0.05000,2.722 | |
| 18,self_attn.o_proj,0.0000000029,0.05000,1.201 | |
| 18,mlp.up_proj,0.0000000651,0.05000,1.546 | |
| 18,mlp.gate_proj,0.0000000675,0.05000,1.563 | |
| 18,mlp.down_proj,0.0000000067,0.05000,2.212 | |
| 19,self_attn.q_proj,0.0000000979,0.05000,3.052 | |
| 19,self_attn.v_proj,0.0000000190,0.05000,3.059 | |
| 19,self_attn.k_proj,0.0000000297,0.05000,3.083 | |
| 19,self_attn.o_proj,0.0000000024,0.05000,1.308 | |
| 19,mlp.gate_proj,0.0000000689,0.05000,1.671 | |
| 19,mlp.up_proj,0.0000000664,0.05000,1.682 | |
| 19,mlp.down_proj,0.0000000077,0.05000,2.353 | |
| 20,self_attn.q_proj,0.0000000935,0.05000,3.327 | |
| 20,self_attn.k_proj,0.0000000300,0.05000,3.385 | |
| 20,self_attn.v_proj,0.0000000177,0.05000,3.403 | |
| 20,self_attn.o_proj,0.0000000021,0.05000,1.283 | |
| 20,mlp.gate_proj,0.0000000806,0.05000,1.467 | |
| 20,mlp.up_proj,0.0000000756,0.05000,1.479 | |
| 20,mlp.down_proj,0.0000000089,0.05000,2.752 | |
| 21,self_attn.k_proj,0.0000000291,0.05000,3.551 | |
| 21,self_attn.v_proj,0.0000000160,0.05000,3.591 | |
| 21,self_attn.q_proj,0.0000000912,0.05000,3.676 | |
| 21,self_attn.o_proj,0.0000000016,0.05000,1.171 | |
| 21,mlp.up_proj,0.0000000838,0.05000,2.555 | |
| 21,mlp.gate_proj,0.0000000902,0.05000,2.602 | |
| 21,mlp.down_proj,0.0000000101,0.05000,2.382 | |
| 22,self_attn.q_proj,0.0000000906,0.05000,3.905 | |
| 22,self_attn.k_proj,0.0000000262,0.05000,3.912 | |
| 22,self_attn.v_proj,0.0000000213,0.05000,4.015 | |
| 22,self_attn.o_proj,0.0000000024,0.05000,0.851 | |
| 22,mlp.up_proj,0.0000000967,0.05000,1.559 | |
| 22,mlp.gate_proj,0.0000001041,0.05000,1.565 | |
| 22,mlp.down_proj,0.0000000127,0.05000,2.276 | |
| 23,self_attn.v_proj,0.0000000217,0.05000,3.784 | |
| 23,self_attn.k_proj,0.0000000295,0.05000,3.809 | |
| 23,self_attn.q_proj,0.0000000979,0.05000,3.914 | |
| 23,self_attn.o_proj,0.0000000019,0.05000,1.159 | |
| 23,mlp.up_proj,0.0000001080,0.05000,2.357 | |
| 23,mlp.gate_proj,0.0000001121,0.05000,2.384 | |
| 23,mlp.down_proj,0.0000000163,0.05000,2.817 | |
| 24,self_attn.q_proj,0.0000001070,0.05000,3.759 | |
| 24,self_attn.k_proj,0.0000000320,0.05000,3.819 | |
| 24,self_attn.v_proj,0.0000000248,0.05000,3.908 | |
| 24,self_attn.o_proj,0.0000000022,0.05000,1.191 | |
| 24,mlp.up_proj,0.0000001213,0.05000,1.383 | |
| 24,mlp.gate_proj,0.0000001228,0.05000,1.391 | |
| 24,mlp.down_proj,0.0000000182,0.05000,2.273 | |
| 25,self_attn.k_proj,0.0000000321,0.05000,3.559 | |
| 25,self_attn.v_proj,0.0000000275,0.05000,3.625 | |
| 25,self_attn.q_proj,0.0000001104,0.05000,3.706 | |
| 25,self_attn.o_proj,0.0000000023,0.05000,1.030 | |
| 25,mlp.up_proj,0.0000001367,0.05000,2.500 | |
| 25,mlp.gate_proj,0.0000001344,0.05000,2.523 | |
| 25,mlp.down_proj,0.0000000330,0.05000,2.737 | |
| 26,self_attn.q_proj,0.0000001115,0.05000,3.848 | |
| 26,self_attn.k_proj,0.0000000315,0.05000,3.908 | |
| 26,self_attn.v_proj,0.0000000338,0.05000,4.023 | |
| 26,self_attn.o_proj,0.0000000027,0.05000,0.856 | |
| 26,mlp.up_proj,0.0000001578,0.05000,1.510 | |
| 26,mlp.gate_proj,0.0000001507,0.05000,1.538 | |
| 26,mlp.down_proj,0.0000000300,0.05000,2.150 | |
| 27,self_attn.v_proj,0.0000000335,0.05000,3.903 | |
| 27,self_attn.k_proj,0.0000000346,0.05000,3.934 | |
| 27,self_attn.q_proj,0.0000001312,0.05000,4.007 | |
| 27,self_attn.o_proj,0.0000000041,0.05000,1.299 | |
| 27,mlp.up_proj,0.0000001809,0.05000,2.707 | |
| 27,mlp.gate_proj,0.0000001781,0.05000,2.739 | |
| 27,mlp.down_proj,0.0000000425,0.05000,2.745 | |
| 28,self_attn.q_proj,0.0000001430,0.05000,3.753 | |
| 28,self_attn.k_proj,0.0000000326,0.05000,3.801 | |
| 28,self_attn.v_proj,0.0000000781,0.05000,3.903 | |
| 28,self_attn.o_proj,0.0000000079,0.05000,1.300 | |
| 28,mlp.up_proj,0.0000002043,0.05000,2.586 | |
| 28,mlp.gate_proj,0.0000002063,0.05000,2.587 | |
| 28,mlp.down_proj,0.0000000514,0.05000,2.539 | |
| 29,self_attn.v_proj,0.0000000655,0.05000,3.764 | |
| 29,self_attn.k_proj,0.0000000319,0.05000,3.790 | |
| 29,self_attn.q_proj,0.0000001379,0.05000,3.878 | |
| 29,self_attn.o_proj,0.0000000067,0.05000,1.223 | |
| 29,mlp.up_proj,0.0000002308,0.05000,2.468 | |
| 29,mlp.gate_proj,0.0000002363,0.05000,2.504 | |
| 29,mlp.down_proj,0.0000000717,0.05000,2.485 | |
| 30,self_attn.q_proj,0.0000001404,0.05000,3.584 | |
| 30,self_attn.v_proj,0.0000000711,0.05000,3.688 | |
| 30,self_attn.k_proj,0.0000000310,0.05000,3.716 | |
| 30,self_attn.o_proj,0.0000000099,0.05000,1.164 | |
| 30,mlp.gate_proj,0.0000002582,0.05000,1.471 | |
| 30,mlp.up_proj,0.0000002543,0.05000,1.483 | |
| 30,mlp.down_proj,0.0000000925,0.05000,2.235 | |
| 31,self_attn.v_proj,0.0000000747,0.05000,3.657 | |
| 31,self_attn.k_proj,0.0000000291,0.05000,3.669 | |
| 31,self_attn.q_proj,0.0000001379,0.05000,3.747 | |
| 31,self_attn.o_proj,0.0000000234,0.05000,1.285 | |
| 31,mlp.up_proj,0.0000002673,0.05000,2.496 | |
| 31,mlp.gate_proj,0.0000002787,0.05000,2.525 | |
| 31,mlp.down_proj,0.0000001321,0.05000,2.262 | |
| 32,self_attn.q_proj,0.0000001691,0.05000,3.711 | |
| 32,self_attn.k_proj,0.0000000317,0.05000,3.835 | |
| 32,self_attn.v_proj,0.0000001289,0.05000,3.910 | |
| 32,self_attn.o_proj,0.0000000241,0.05000,1.301 | |
| 32,mlp.up_proj,0.0000002878,0.05000,1.836 | |
| 32,mlp.gate_proj,0.0000003177,0.05000,1.855 | |
| 32,mlp.down_proj,0.0000002104,0.05000,2.258 | |
| 33,self_attn.k_proj,0.0000000274,0.05000,3.998 | |
| 33,self_attn.q_proj,0.0000001259,0.05000,4.011 | |
| 33,self_attn.v_proj,0.0000000742,0.05000,4.066 | |
| 33,self_attn.o_proj,0.0000000296,0.05000,1.115 | |
| 33,mlp.up_proj,0.0000002813,0.05000,2.039 | |
| 33,mlp.gate_proj,0.0000003160,0.05000,2.050 | |
| 33,mlp.down_proj,0.0000006095,0.05000,2.117 | |