layer,module,loss,samples,damp,time
0,linear_attn.in_proj_qkv,0.0000512330,0.05000,0.753
0,linear_attn.in_proj_z,0.0000258486,0.05000,0.439
0,linear_attn.out_proj,0.0000000288,0.05000,0.438
0,mlp.gate_proj,0.0000006692,0.05000,1.283
0,mlp.up_proj,0.0000005881,0.05000,1.321
0,mlp.down_proj,0.0000000033,0.05000,2.023
1,linear_attn.in_proj_qkv,0.0000058743,0.05000,0.505
1,linear_attn.in_proj_z,0.0000030396,0.05000,0.441
1,linear_attn.out_proj,0.0000000073,0.05000,0.443
1,mlp.gate_proj,0.0000017216,0.05000,1.314
1,mlp.up_proj,0.0000015407,0.05000,1.320
1,mlp.down_proj,0.0000000096,0.05000,1.996
2,linear_attn.in_proj_qkv,0.0000102022,0.05000,0.508
2,linear_attn.in_proj_z,0.0000051198,0.05000,0.442
2,linear_attn.out_proj,0.0000000120,0.05000,0.438
2,mlp.up_proj,0.0000024092,0.05000,1.574
2,mlp.gate_proj,0.0000028889,0.05000,1.579
2,mlp.down_proj,0.0000000213,0.05000,1.979
3,self_attn.q_proj,0.0000457913,0.05000,1.437
3,self_attn.v_proj,0.0000040927,0.05000,1.443
3,self_attn.k_proj,0.0000045126,0.05000,1.448
3,self_attn.o_proj,0.0000000247,0.05000,0.442
3,mlp.up_proj,0.0000033437,0.05000,1.295
3,mlp.gate_proj,0.0000037084,0.05000,1.303
3,mlp.down_proj,0.0000000310,0.05000,1.934
4,linear_attn.in_proj_qkv,0.0000166285,0.05000,0.504
4,linear_attn.in_proj_z,0.0000084468,0.05000,0.441
4,linear_attn.out_proj,0.0000000281,0.05000,0.433
4,mlp.up_proj,0.0000054035,0.05000,1.541
4,mlp.gate_proj,0.0000060489,0.05000,1.560
4,mlp.down_proj,0.0000000529,0.05000,1.950
5,linear_attn.in_proj_qkv,0.0000234676,0.05000,0.497
5,linear_attn.in_proj_z,0.0000116633,0.05000,0.438
5,linear_attn.out_proj,0.0000000641,0.05000,0.438
5,mlp.gate_proj,0.0000087650,0.05000,1.280
5,mlp.up_proj,0.0000072798,0.05000,1.282
5,mlp.down_proj,0.0000001020,0.05000,1.931
6,linear_attn.in_proj_qkv,0.0000271829,0.05000,0.499
6,linear_attn.in_proj_z,0.0000114913,0.05000,0.442
6,linear_attn.out_proj,0.0000001114,0.05000,0.441
6,mlp.gate_proj,0.0000125882,0.05000,1.596
6,mlp.up_proj,0.0000099671,0.05000,1.624
6,mlp.down_proj,0.0000001583,0.05000,1.969
7,self_attn.q_proj,0.0000327816,0.05000,1.793
7,self_attn.k_proj,0.0000040194,0.05000,1.808
7,self_attn.v_proj,0.0000034953,0.05000,1.826
7,self_attn.o_proj,0.0000001741,0.05000,0.441
7,mlp.gate_proj,0.0000132600,0.05000,1.668
7,mlp.up_proj,0.0000104244,0.05000,1.672
7,mlp.down_proj,0.0000001733,0.05000,1.950
8,linear_attn.in_proj_qkv,0.0000301906,0.05000,0.507
8,linear_attn.in_proj_z,0.0000130997,0.05000,0.447
8,linear_attn.out_proj,0.0000001314,0.05000,0.437
8,mlp.up_proj,0.0000114845,0.05000,1.308
8,mlp.gate_proj,0.0000140776,0.05000,1.313
8,mlp.down_proj,0.0000001827,0.05000,1.964
9,linear_attn.in_proj_qkv,0.0000328663,0.05000,0.503
9,linear_attn.in_proj_z,0.0000136644,0.05000,0.436
9,linear_attn.out_proj,0.0000001533,0.05000,0.435
9,mlp.up_proj,0.0000123621,0.05000,1.654
9,mlp.gate_proj,0.0000141435,0.05000,1.675
9,mlp.down_proj,0.0000002136,0.05000,1.915
10,linear_attn.in_proj_qkv,0.0000313985,0.05000,0.500
10,linear_attn.in_proj_z,0.0000127113,0.05000,0.440
10,linear_attn.out_proj,0.0000001706,0.05000,0.435
10,mlp.gate_proj,0.0000139577,0.05000,1.244
10,mlp.up_proj,0.0000133629,0.05000,1.249
10,mlp.down_proj,0.0000002500,0.05000,1.933
11,self_attn.q_proj,0.0000255943,0.05000,1.750
11,self_attn.k_proj,0.0000034389,0.05000,1.761
11,self_attn.v_proj,0.0000033088,0.05000,1.785
11,self_attn.o_proj,0.0000003653,0.05000,0.439
11,mlp.gate_proj,0.0000142216,0.05000,1.447
11,mlp.up_proj,0.0000144107,0.05000,1.477
11,mlp.down_proj,0.0000002969,0.05000,1.908
12,linear_attn.in_proj_qkv,0.0000334196,0.05000,0.499
12,linear_attn.in_proj_z,0.0000123923,0.05000,0.439
12,linear_attn.out_proj,0.0000002414,0.05000,0.437
12,mlp.gate_proj,0.0000144687,0.05000,1.479
12,mlp.up_proj,0.0000152860,0.05000,1.505
12,mlp.down_proj,0.0000003427,0.05000,1.941
13,linear_attn.in_proj_qkv,0.0000387516,0.05000,0.497
13,linear_attn.in_proj_z,0.0000150452,0.05000,0.446
13,linear_attn.out_proj,0.0000002380,0.05000,0.434
13,mlp.gate_proj,0.0000151776,0.05000,1.609
13,mlp.up_proj,0.0000163668,0.05000,1.639
13,mlp.down_proj,0.0000003871,0.05000,1.940
14,linear_attn.in_proj_qkv,0.0000384952,0.05000,0.496
14,linear_attn.in_proj_z,0.0000148005,0.05000,0.444
14,linear_attn.out_proj,0.0000003604,0.05000,0.436
14,mlp.gate_proj,0.0000154822,0.05000,1.289
14,mlp.up_proj,0.0000175342,0.05000,1.295
14,mlp.down_proj,0.0000004424,0.05000,1.960
15,self_attn.k_proj,0.0000037890,0.05000,1.564
15,self_attn.q_proj,0.0000281934,0.05000,1.579
15,self_attn.v_proj,0.0000048000,0.05000,1.587
15,self_attn.o_proj,0.0000006070,0.05000,0.435
15,mlp.gate_proj,0.0000170689,0.05000,1.280
15,mlp.up_proj,0.0000197788,0.05000,1.285
15,mlp.down_proj,0.0000006178,0.05000,1.924
16,linear_attn.in_proj_qkv,0.0000414585,0.05000,0.497
16,linear_attn.in_proj_z,0.0000157437,0.05000,0.434
16,linear_attn.out_proj,0.0000004966,0.05000,0.435
16,mlp.gate_proj,0.0000189846,0.05000,1.235
16,mlp.up_proj,0.0000219061,0.05000,1.240
16,mlp.down_proj,0.0000008104,0.05000,1.911
17,linear_attn.in_proj_qkv,0.0000563444,0.05000,0.498
17,linear_attn.in_proj_z,0.0000180389,0.05000,0.437
17,linear_attn.out_proj,0.0000006251,0.05000,0.434
17,mlp.gate_proj,0.0000202587,0.05000,1.301
17,mlp.up_proj,0.0000241804,0.05000,1.309
17,mlp.down_proj,0.0000010840,0.05000,1.916
18,linear_attn.in_proj_qkv,0.0000529036,0.05000,0.503
18,linear_attn.in_proj_z,0.0000192245,0.05000,0.440
18,linear_attn.out_proj,0.0000010398,0.05000,0.437
18,mlp.up_proj,0.0000281983,0.05000,1.249
18,mlp.gate_proj,0.0000245166,0.05000,1.264
18,mlp.down_proj,0.0000020065,0.05000,1.944
19,self_attn.k_proj,0.0000058760,0.05000,1.776
19,self_attn.v_proj,0.0000109502,0.05000,1.784
19,self_attn.q_proj,0.0000395083,0.05000,1.813
19,self_attn.o_proj,0.0000026417,0.05000,0.442
19,mlp.gate_proj,0.0000300821,0.05000,1.619
19,mlp.up_proj,0.0000356534,0.05000,1.640
19,mlp.down_proj,0.0000029585,0.05000,1.944
20,linear_attn.in_proj_qkv,0.0000828392,0.05000,0.502
20,linear_attn.in_proj_z,0.0000297302,0.05000,0.434
20,linear_attn.out_proj,0.0000014146,0.05000,0.443
20,mlp.gate_proj,0.0000373262,0.05000,1.534
20,mlp.up_proj,0.0000371792,0.05000,1.553
20,mlp.down_proj,0.0000031579,0.05000,1.929
21,linear_attn.in_proj_qkv,0.0000811663,0.05000,0.509
21,linear_attn.in_proj_z,0.0000304235,0.05000,0.448
21,linear_attn.out_proj,0.0000018507,0.05000,0.433
21,mlp.up_proj,0.0000391224,0.05000,1.600
21,mlp.gate_proj,0.0000418820,0.05000,1.611
21,mlp.down_proj,0.0000040956,0.05000,1.939
22,linear_attn.in_proj_qkv,0.0000800213,0.05000,0.499
22,linear_attn.in_proj_z,0.0000346308,0.05000,0.434
22,linear_attn.out_proj,0.0000020605,0.05000,0.439
22,mlp.up_proj,0.0000500837,0.05000,1.472
22,mlp.gate_proj,0.0000592674,0.05000,1.478
22,mlp.down_proj,0.0000062517,0.05000,1.917
23,self_attn.k_proj,0.0000078460,0.05000,1.776
23,self_attn.v_proj,0.0000156147,0.05000,1.789
23,self_attn.q_proj,0.0000542049,0.05000,1.805
23,self_attn.o_proj,0.0000037408,0.05000,0.437
23,mlp.gate_proj,0.0000577355,0.05000,1.643
23,mlp.up_proj,0.0000519635,0.05000,1.649
23,mlp.down_proj,0.0000051992,0.05000,1.916
24,linear_attn.in_proj_qkv,0.0000934729,0.05000,0.496
24,linear_attn.in_proj_z,0.0000357742,0.05000,0.435
24,linear_attn.out_proj,0.0000029748,0.05000,0.444
24,mlp.gate_proj,0.0000614639,0.05000,1.246
24,mlp.up_proj,0.0000522009,0.05000,1.247
24,mlp.down_proj,0.0000049123,0.05000,1.916
25,linear_attn.in_proj_qkv,0.0000965361,0.05000,0.508
25,linear_attn.in_proj_z,0.0000383143,0.05000,0.433
25,linear_attn.out_proj,0.0000024083,0.05000,0.435
25,mlp.gate_proj,0.0000721219,0.05000,1.608
25,mlp.up_proj,0.0000587548,0.05000,1.623
25,mlp.down_proj,0.0000050382,0.05000,1.904
26,linear_attn.in_proj_qkv,0.0000897841,0.05000,0.500
26,linear_attn.in_proj_z,0.0000403678,0.05000,0.430
26,linear_attn.out_proj,0.0000030055,0.05000,0.441
26,mlp.gate_proj,0.0000864327,0.05000,1.388
26,mlp.up_proj,0.0000691804,0.05000,1.407
26,mlp.down_proj,0.0000069061,0.05000,1.903
27,self_attn.k_proj,0.0000103169,0.05000,1.737
27,self_attn.q_proj,0.0000655058,0.05000,1.744
27,self_attn.v_proj,0.0000380844,0.05000,1.761
27,self_attn.o_proj,0.0000103013,0.05000,0.432
27,mlp.gate_proj,0.0000873213,0.05000,1.345
27,mlp.up_proj,0.0000718302,0.05000,1.352
27,mlp.down_proj,0.0000074708,0.05000,1.926
28,linear_attn.in_proj_qkv,0.0001090263,0.05000,0.499
28,linear_attn.in_proj_z,0.0000431458,0.05000,0.435
28,linear_attn.out_proj,0.0000066057,0.05000,0.436
28,mlp.up_proj,0.0000736457,0.05000,1.536
28,mlp.gate_proj,0.0000876350,0.05000,1.543
28,mlp.down_proj,0.0000099431,0.05000,1.923
29,linear_attn.in_proj_qkv,0.0000821576,0.05000,0.498
29,linear_attn.in_proj_z,0.0000386130,0.05000,0.439
29,linear_attn.out_proj,0.0000073645,0.05000,0.438
29,mlp.gate_proj,0.0000951019,0.05000,1.261
29,mlp.up_proj,0.0000801801,0.05000,1.267
29,mlp.down_proj,0.0000127125,0.05000,1.922
30,linear_attn.in_proj_qkv,0.0000864302,0.05000,0.506
30,linear_attn.in_proj_z,0.0000373503,0.05000,0.436
30,linear_attn.out_proj,0.0000150087,0.05000,0.431
30,mlp.gate_proj,0.0000910522,0.05000,1.601
30,mlp.up_proj,0.0000784964,0.05000,1.615
30,mlp.down_proj,0.0000179220,0.05000,1.920
31,self_attn.v_proj,0.0000322213,0.05000,1.749
31,self_attn.k_proj,0.0000086426,0.05000,1.755
31,self_attn.q_proj,0.0000653408,0.05000,1.787
31,self_attn.o_proj,0.0000210914,0.05000,0.436
31,mlp.gate_proj,0.0000866345,0.05000,1.648
31,mlp.up_proj,0.0000729714,0.05000,1.655
31,mlp.down_proj,0.0000323485,0.05000,1.933