int4/quant_log.csv

layer,module,loss,samples,damp,time
0,self_attn.o_proj,failsafe(rtn): 0.0018692,0.00000,0.182
0,self_attn.q_proj,0.0000025054,0.05000,2.730
0,self_attn.k_proj,0.0000001197,0.05000,2.752
0,self_attn.v_proj,0.0000002673,0.05000,2.800
0,self_attn.o_gate,0.0000053353,0.05000,0.484
0,mlp.gate_proj,0.0000019878,0.05000,2.842
0,mlp.up_proj,0.0000028122,0.05000,2.889
0,mlp.down_proj,0.0000014094,0.05000,4.165
1,self_attn.q_proj,0.0000198860,0.05000,4.233
1,self_attn.o_proj,0.0000005668,0.05000,4.310
1,self_attn.v_proj,0.0000256484,0.05000,4.375
1,self_attn.k_proj,0.0000202024,0.05000,4.389
1,mlp.gate_proj,0.0000017809,0.05000,2.665
1,mlp.up_proj,0.0000018768,0.05000,2.688
1,mlp.down_proj,0.0000005710,0.05000,3.985
2,self_attn.v_proj,0.0000119164,0.05000,3.095
2,self_attn.q_proj,0.0000101246,0.05000,3.331
2,self_attn.o_proj,0.0000002428,0.05000,3.356
2,self_attn.k_proj,0.0000100614,0.05000,3.373
2,mlp.up_proj,0.0000022416,0.05000,2.600
2,mlp.gate_proj,0.0000022318,0.05000,2.641
2,mlp.down_proj,0.0000009728,0.05000,3.868
3,self_attn.k_proj,0.0000128744,0.05000,3.483
3,self_attn.q_proj,0.0000125645,0.05000,3.493
3,self_attn.v_proj,0.0000135237,0.05000,3.539
3,self_attn.o_proj,0.0000003379,0.05000,3.548
3,mlp.gate_proj,0.0000045511,0.05000,2.430
3,mlp.up_proj,0.0000040209,0.05000,2.515
3,mlp.down_proj,0.0000014397,0.05000,3.780
4,self_attn.q_proj,0.0000099334,0.05000,3.741
4,self_attn.o_proj,0.0000002742,0.05000,3.821
4,self_attn.v_proj,0.0000120185,0.05000,3.886
4,self_attn.k_proj,0.0000101482,0.05000,3.891
4,mlp.gate_proj,0.0000059484,0.05000,2.076
4,mlp.up_proj,0.0000054010,0.05000,2.234
4,mlp.down_proj,0.0000015523,0.05000,3.455
5,self_attn.v_proj,0.0000121319,0.05000,4.791
5,self_attn.k_proj,0.0000113703,0.05000,4.900
5,self_attn.q_proj,0.0000107690,0.05000,4.938
5,self_attn.o_proj,0.0000004462,0.05000,4.940
5,mlp.up_proj,0.0000066999,0.05000,2.813
5,mlp.gate_proj,0.0000073629,0.05000,2.827
5,mlp.down_proj,0.0000016318,0.05000,4.160
6,self_attn.q_proj,0.0000104139,0.05000,3.041
6,self_attn.v_proj,0.0000109036,0.05000,3.087
6,self_attn.o_proj,0.0000004455,0.05000,3.204
6,self_attn.k_proj,0.0000124642,0.05000,3.206
6,mlp.up_proj,0.0000082335,0.05000,2.703
6,mlp.gate_proj,0.0000089300,0.05000,2.707
6,mlp.down_proj,0.0000018200,0.05000,3.931
7,self_attn.k_proj,0.0000113131,0.05000,3.565
7,self_attn.v_proj,0.0000101746,0.05000,3.589
7,self_attn.q_proj,0.0000092834,0.05000,3.591
7,self_attn.o_proj,0.0000004802,0.05000,3.600
7,mlp.up_proj,0.0000086123,0.05000,2.667
7,mlp.gate_proj,0.0000091405,0.05000,2.684
7,mlp.down_proj,0.0000021072,0.05000,3.948
8,self_attn.q_proj,0.0000064478,0.05000,4.042
8,self_attn.k_proj,0.0000079277,0.05000,4.251
8,self_attn.o_proj,0.0000007475,0.05000,4.252
8,self_attn.v_proj,0.0000070005,0.05000,4.259
8,mlp.up_proj,0.0000088850,0.05000,2.588
8,mlp.gate_proj,0.0000097536,0.05000,2.604
8,mlp.down_proj,0.0000020750,0.05000,3.844
9,self_attn.o_proj,failsafe(rtn): 0.0020905,0.00000,0.124
9,self_attn.k_proj,0.0000007585,0.05000,2.513
9,self_attn.q_proj,0.0000078441,0.05000,2.525
9,self_attn.v_proj,0.0000004848,0.05000,2.549
9,self_attn.o_gate,0.0000054848,0.05000,0.438
9,mlp.gate_proj,0.0000107061,0.05000,2.870
9,mlp.up_proj,0.0000102051,0.05000,2.907
9,mlp.down_proj,0.0000026359,0.05000,4.160
10,self_attn.o_proj,0.0000006941,0.05000,3.627
10,self_attn.q_proj,0.0000094431,0.05000,3.633
10,self_attn.v_proj,0.0000100697,0.05000,3.651
10,self_attn.k_proj,0.0000118222,0.05000,3.652
10,mlp.gate_proj,0.0000105735,0.05000,2.624
10,mlp.up_proj,0.0000102459,0.05000,2.646
10,mlp.down_proj,0.0000024634,0.05000,3.931
11,self_attn.o_proj,0.0000007806,0.05000,4.636
11,self_attn.k_proj,0.0000083771,0.05000,4.707
11,self_attn.v_proj,0.0000072992,0.05000,4.721
11,self_attn.q_proj,0.0000068717,0.05000,4.732
11,mlp.gate_proj,0.0000102914,0.05000,2.696
11,mlp.up_proj,0.0000103112,0.05000,2.704
11,mlp.down_proj,0.0000026183,0.05000,3.968
12,self_attn.o_proj,0.0000009991,0.05000,4.706
12,self_attn.v_proj,0.0000100696,0.05000,4.722
12,self_attn.q_proj,0.0000086598,0.05000,4.739
12,self_attn.k_proj,0.0000095307,0.05000,4.742
12,mlp.up_proj,0.0000101439,0.05000,2.681
12,mlp.gate_proj,0.0000099271,0.05000,2.694
12,mlp.down_proj,0.0000026657,0.05000,3.994
13,self_attn.q_proj,0.0000065795,0.05000,3.849
13,self_attn.k_proj,0.0000084106,0.05000,3.866
13,self_attn.v_proj,0.0000067338,0.05000,3.952
13,self_attn.o_proj,0.0000010317,0.05000,3.956
13,mlp.up_proj,0.0000104438,0.05000,2.705
13,mlp.gate_proj,0.0000100330,0.05000,2.712
13,mlp.down_proj,0.0000028064,0.05000,4.023
14,self_attn.k_proj,0.0000081929,0.05000,4.184
14,self_attn.q_proj,0.0000071310,0.05000,4.353
14,self_attn.v_proj,0.0000070400,0.05000,4.428
14,self_attn.o_proj,0.0000013027,0.05000,4.431
14,mlp.up_proj,0.0000104130,0.05000,2.625
14,mlp.gate_proj,0.0000094866,0.05000,2.634
14,mlp.down_proj,0.0000031477,0.05000,3.901
15,self_attn.q_proj,0.0000046234,0.05000,3.621
15,self_attn.k_proj,0.0000059547,0.05000,3.634
15,self_attn.v_proj,0.0000049011,0.05000,3.641
15,self_attn.o_proj,0.0000017505,0.05000,3.648
15,mlp.up_proj,0.0000101065,0.05000,2.664
15,mlp.gate_proj,0.0000098456,0.05000,2.667
15,mlp.down_proj,0.0000027747,0.05000,3.969
16,self_attn.o_proj,failsafe(rtn): 0.0021057,0.00000,0.141
16,self_attn.q_proj,0.0000072823,0.05000,2.258
16,self_attn.v_proj,0.0000004060,0.05000,2.387
16,self_attn.k_proj,0.0000007123,0.05000,2.430
16,self_attn.o_gate,0.0000042507,0.05000,0.555
16,mlp.gate_proj,0.0000129596,0.05000,2.717
16,mlp.up_proj,0.0000126712,0.05000,2.726
16,mlp.down_proj,0.0000040638,0.05000,3.928
17,self_attn.o_proj,failsafe(rtn): 0.0021667,0.00000,0.129
17,self_attn.q_proj,0.0000075214,0.05000,2.375
17,self_attn.v_proj,0.0000005942,0.05000,2.423
17,self_attn.k_proj,0.0000006580,0.05000,2.431
17,self_attn.o_gate,0.0000050420,0.05000,0.459
17,mlp.up_proj,0.0000138615,0.05000,2.717
17,mlp.gate_proj,0.0000151962,0.05000,2.723
17,mlp.down_proj,0.0000039890,0.05000,3.947
18,self_attn.v_proj,0.0000065547,0.05000,3.642
18,self_attn.o_proj,0.0000012725,0.05000,3.650
18,self_attn.k_proj,0.0000091951,0.05000,3.671
18,self_attn.q_proj,0.0000064936,0.05000,3.675
18,mlp.up_proj,0.0000126479,0.05000,2.744
18,mlp.gate_proj,0.0000131430,0.05000,2.762
18,mlp.down_proj,0.0000034282,0.05000,4.064
19,self_attn.q_proj,0.0000081404,0.05000,3.999
19,self_attn.v_proj,0.0000085492,0.05000,4.100
19,self_attn.k_proj,0.0000106452,0.05000,4.179
19,self_attn.o_proj,0.0000012228,0.05000,4.186
19,mlp.up_proj,0.0000129798,0.05000,2.711
19,mlp.gate_proj,0.0000127026,0.05000,2.724
19,mlp.down_proj,0.0000038460,0.05000,3.995
20,self_attn.v_proj,0.0000063993,0.05000,3.757
20,self_attn.q_proj,0.0000064077,0.05000,3.794
20,self_attn.k_proj,0.0000086538,0.05000,3.810
20,self_attn.o_proj,0.0000019678,0.05000,3.823
20,mlp.up_proj,0.0000133294,0.05000,2.705
20,mlp.gate_proj,0.0000126178,0.05000,2.707
20,mlp.down_proj,0.0000050352,0.05000,3.979
21,self_attn.q_proj,0.0000066632,0.05000,4.200
21,self_attn.o_proj,0.0000029935,0.05000,4.312
21,self_attn.k_proj,0.0000075829,0.05000,4.322
21,self_attn.v_proj,0.0000071241,0.05000,4.324
21,mlp.up_proj,0.0000126541,0.05000,2.674
21,mlp.gate_proj,0.0000117755,0.05000,2.685
21,mlp.down_proj,0.0000059720,0.05000,3.946
22,self_attn.o_proj,failsafe(rtn): 0.0022430,0.00000,0.123
22,self_attn.q_proj,0.0000090866,0.05000,2.371
22,self_attn.k_proj,0.0000005854,0.05000,2.376
22,self_attn.v_proj,0.0000010102,0.05000,2.393
22,self_attn.o_gate,0.0000048253,0.05000,0.449
22,mlp.up_proj,0.0000161680,0.05000,2.200
22,mlp.gate_proj,0.0000150658,0.05000,2.405
22,mlp.down_proj,0.0000087456,0.05000,3.550
23,self_attn.o_proj,0.0000027456,0.05000,3.176
23,self_attn.k_proj,0.0000113444,0.05000,3.380
23,self_attn.v_proj,0.0000085384,0.05000,3.400
23,self_attn.q_proj,0.0000088479,0.05000,3.409
23,mlp.up_proj,0.0000190031,0.05000,2.656
23,mlp.gate_proj,0.0000176874,0.05000,2.668
23,mlp.down_proj,0.0000105173,0.05000,3.867
24,self_attn.v_proj,0.0000076524,0.05000,3.537
24,self_attn.k_proj,0.0000093884,0.05000,3.547
24,self_attn.q_proj,0.0000075387,0.05000,3.574
24,self_attn.o_proj,0.0000046072,0.05000,3.576
24,mlp.up_proj,0.0000224395,0.05000,2.165
24,mlp.gate_proj,0.0000207282,0.05000,2.259
24,mlp.down_proj,0.0000171202,0.05000,3.466
25,self_attn.q_proj,0.0000078854,0.05000,4.434
25,self_attn.k_proj,0.0000100391,0.05000,4.466
25,self_attn.v_proj,0.0000077710,0.05000,4.482
25,self_attn.o_proj,0.0000059637,0.05000,4.484
25,mlp.gate_proj,0.0000239999,0.05000,2.532
25,mlp.up_proj,0.0000264045,0.05000,2.543
25,mlp.down_proj,0.0000301652,0.05000,3.783
26,self_attn.q_proj,0.0000126346,0.05000,3.699
26,self_attn.v_proj,0.0000123492,0.05000,3.766
26,self_attn.o_proj,0.0000112689,0.05000,3.774
26,self_attn.k_proj,0.0000180101,0.05000,3.777
26,mlp.gate_proj,0.0000267217,0.05000,2.533
26,mlp.up_proj,0.0000302445,0.05000,2.547
26,mlp.down_proj,0.0000262709,0.05000,3.800
27,self_attn.q_proj,0.0000118468,0.05000,3.496
27,self_attn.o_proj,0.0000070263,0.05000,3.517
27,self_attn.v_proj,0.0000110681,0.05000,3.535
27,self_attn.k_proj,0.0000130042,0.05000,3.540
27,mlp.up_proj,0.0000350091,0.05000,2.500
27,mlp.gate_proj,0.0000303793,0.05000,2.513
27,mlp.down_proj,0.0000345169,0.05000,3.770
28,self_attn.v_proj,0.0000156234,0.05000,3.067
28,self_attn.k_proj,0.0000210312,0.05000,3.080
28,self_attn.q_proj,0.0000151926,0.05000,3.082
28,self_attn.o_proj,0.0000215157,0.05000,3.085
28,mlp.gate_proj,0.0000345202,0.05000,2.215
28,mlp.up_proj,0.0000407514,0.05000,2.220
28,mlp.down_proj,0.0000481835,0.05000,3.498
29,self_attn.o_proj,failsafe(rtn): 0.0025024,0.00000,0.103
29,self_attn.q_proj,0.0000194512,0.05000,2.011
29,self_attn.v_proj,0.0000098854,0.05000,2.039
29,self_attn.k_proj,0.0000010127,0.05000,2.043
29,self_attn.o_gate,0.0000215202,0.05000,0.443
29,mlp.gate_proj,0.0000427544,0.05000,2.250
29,mlp.up_proj,0.0000519517,0.05000,2.254
29,mlp.down_proj,0.0000784172,0.05000,3.514
30,self_attn.o_proj,failsafe(rtn): 0.0025940,0.00000,0.128
30,self_attn.v_proj,0.0000307847,0.05000,2.067
30,self_attn.q_proj,0.0000298560,0.05000,2.068
30,self_attn.k_proj,0.0000011827,0.05000,2.090
30,self_attn.o_gate,0.0000342893,0.05000,0.434
30,mlp.up_proj,0.0000671554,0.05000,2.368
30,mlp.gate_proj,0.0000556245,0.05000,2.376
30,mlp.down_proj,0.0001689895,0.05000,3.640
31,self_attn.o_proj,failsafe(rtn): 0.0024261,0.00000,0.133
31,self_attn.q_proj,0.0000205567,0.05000,2.020
31,self_attn.k_proj,0.0000007648,0.05000,2.065
31,self_attn.v_proj,0.0000048051,0.05000,2.072
31,self_attn.o_gate,0.0000215028,0.05000,0.456
31,mlp.gate_proj,0.0000875096,0.05000,2.375
31,mlp.up_proj,0.0000976488,0.05000,2.377
31,mlp.down_proj,0.0006907178,0.05000,3.652