198 lines
8.7 KiB
CSV
198 lines
8.7 KiB
CSV
|
|
layer,module,loss,samples,damp,time
|
||
|
|
0,self_attn.q_proj,0.0000000157,0.01000,0.453
|
||
|
|
0,self_attn.k_proj,0.0000000070,0.01000,0.218
|
||
|
|
0,self_attn.v_proj,0.0000000055,0.01000,0.220
|
||
|
|
0,self_attn.o_proj,0.0000000039,0.01000,0.433
|
||
|
|
0,mlp.gate_proj,0.0000001457,0.01000,0.216
|
||
|
|
0,mlp.up_proj,0.0000000745,0.01000,0.215
|
||
|
|
0,mlp.down_proj,0.0000000050,0.01000,0.652
|
||
|
|
1,self_attn.q_proj,0.0000000056,0.01000,0.286
|
||
|
|
1,self_attn.k_proj,0.0000000025,0.01000,0.217
|
||
|
|
1,self_attn.v_proj,0.0000000024,0.01000,0.217
|
||
|
|
1,self_attn.o_proj,0.0000000010,0.01000,0.434
|
||
|
|
1,mlp.gate_proj,0.0000008918,0.01000,0.216
|
||
|
|
1,mlp.up_proj,0.0000002667,0.01000,0.216
|
||
|
|
1,mlp.down_proj,0.0000000066,0.01000,0.659
|
||
|
|
2,self_attn.q_proj,0.0000000110,0.01000,0.255
|
||
|
|
2,self_attn.k_proj,0.0000000047,0.01000,0.219
|
||
|
|
2,self_attn.v_proj,0.0000000046,0.01000,0.218
|
||
|
|
2,self_attn.o_proj,0.0000000016,0.01000,0.435
|
||
|
|
2,mlp.gate_proj,0.0000006115,0.01000,0.219
|
||
|
|
2,mlp.up_proj,0.0000002231,0.01000,0.218
|
||
|
|
2,mlp.down_proj,0.0000078325,0.01000,0.661
|
||
|
|
3,self_attn.q_proj,0.0000000868,0.01000,0.260
|
||
|
|
3,self_attn.k_proj,0.0000000421,0.01000,0.220
|
||
|
|
3,self_attn.v_proj,0.0000000429,0.01000,0.219
|
||
|
|
3,self_attn.o_proj,0.0000000023,0.01000,0.442
|
||
|
|
3,mlp.gate_proj,0.0000008147,0.01000,0.218
|
||
|
|
3,mlp.up_proj,0.0000003186,0.01000,0.218
|
||
|
|
3,mlp.down_proj,0.0000000175,0.01000,0.701
|
||
|
|
4,self_attn.q_proj,0.0000000812,0.01000,0.263
|
||
|
|
4,self_attn.k_proj,0.0000000383,0.01000,0.217
|
||
|
|
4,self_attn.v_proj,0.0000000410,0.01000,0.219
|
||
|
|
4,self_attn.o_proj,0.0000000053,0.01000,0.437
|
||
|
|
4,mlp.gate_proj,0.0000007532,0.01000,0.219
|
||
|
|
4,mlp.up_proj,0.0000003315,0.01000,0.218
|
||
|
|
4,mlp.down_proj,0.0000000219,0.01000,0.659
|
||
|
|
5,self_attn.q_proj,0.0000001455,0.01000,0.262
|
||
|
|
5,self_attn.k_proj,0.0000000597,0.01000,0.223
|
||
|
|
5,self_attn.v_proj,0.0000000636,0.01000,0.226
|
||
|
|
5,self_attn.o_proj,0.0000000076,0.01000,0.452
|
||
|
|
5,mlp.gate_proj,0.0000005187,0.01000,0.223
|
||
|
|
5,mlp.up_proj,0.0000003102,0.01000,0.218
|
||
|
|
5,mlp.down_proj,0.0000000246,0.01000,0.664
|
||
|
|
6,self_attn.q_proj,0.0000000990,0.01000,0.257
|
||
|
|
6,self_attn.k_proj,0.0000000439,0.01000,0.222
|
||
|
|
6,self_attn.v_proj,0.0000000422,0.01000,0.218
|
||
|
|
6,self_attn.o_proj,0.0000000059,0.01000,0.438
|
||
|
|
6,mlp.gate_proj,0.0000006165,0.01000,0.216
|
||
|
|
6,mlp.up_proj,0.0000003886,0.01000,0.216
|
||
|
|
6,mlp.down_proj,0.0000000316,0.01000,0.661
|
||
|
|
7,self_attn.q_proj,0.0000001937,0.01000,0.282
|
||
|
|
7,self_attn.k_proj,0.0000000798,0.01000,0.249
|
||
|
|
7,self_attn.v_proj,0.0000000900,0.01000,0.246
|
||
|
|
7,self_attn.o_proj,0.0000000101,0.01000,0.490
|
||
|
|
7,mlp.gate_proj,0.0000007165,0.01000,0.243
|
||
|
|
7,mlp.up_proj,0.0000004434,0.01000,0.243
|
||
|
|
7,mlp.down_proj,0.0000000395,0.01000,0.700
|
||
|
|
8,self_attn.q_proj,0.0000002418,0.01000,0.257
|
||
|
|
8,self_attn.k_proj,0.0000001092,0.01000,0.221
|
||
|
|
8,self_attn.v_proj,0.0000001022,0.01000,0.217
|
||
|
|
8,self_attn.o_proj,0.0000000103,0.01000,0.435
|
||
|
|
8,mlp.gate_proj,0.0000007137,0.01000,0.218
|
||
|
|
8,mlp.up_proj,0.0000004578,0.01000,0.216
|
||
|
|
8,mlp.down_proj,0.0000000417,0.01000,0.659
|
||
|
|
9,self_attn.q_proj,0.0000004561,0.01000,0.291
|
||
|
|
9,self_attn.k_proj,0.0000001857,0.01000,0.219
|
||
|
|
9,self_attn.v_proj,0.0000001929,0.01000,0.221
|
||
|
|
9,self_attn.o_proj,0.0000000167,0.01000,0.439
|
||
|
|
9,mlp.gate_proj,0.0000007954,0.01000,0.220
|
||
|
|
9,mlp.up_proj,0.0000004992,0.01000,0.218
|
||
|
|
9,mlp.down_proj,0.0000000556,0.01000,0.661
|
||
|
|
10,self_attn.q_proj,0.0000003840,0.01000,0.289
|
||
|
|
10,self_attn.k_proj,0.0000001600,0.01000,0.220
|
||
|
|
10,self_attn.v_proj,0.0000001638,0.01000,0.220
|
||
|
|
10,self_attn.o_proj,0.0000000157,0.01000,0.446
|
||
|
|
10,mlp.gate_proj,0.0000008065,0.01000,0.221
|
||
|
|
10,mlp.up_proj,0.0000005084,0.01000,0.222
|
||
|
|
10,mlp.down_proj,0.0000000788,0.01000,0.671
|
||
|
|
11,self_attn.q_proj,0.0000008024,0.01000,0.283
|
||
|
|
11,self_attn.k_proj,0.0000003071,0.01000,0.239
|
||
|
|
11,self_attn.v_proj,0.0000002804,0.01000,0.220
|
||
|
|
11,self_attn.o_proj,0.0000000493,0.01000,0.443
|
||
|
|
11,mlp.gate_proj,0.0000006262,0.01000,0.220
|
||
|
|
11,mlp.up_proj,0.0000004808,0.01000,0.220
|
||
|
|
11,mlp.down_proj,0.0000000880,0.01000,0.668
|
||
|
|
12,self_attn.q_proj,0.0000007135,0.01000,0.266
|
||
|
|
12,self_attn.k_proj,0.0000002560,0.01000,0.218
|
||
|
|
12,self_attn.v_proj,0.0000002706,0.01000,0.215
|
||
|
|
12,self_attn.o_proj,0.0000000163,0.01000,0.443
|
||
|
|
12,mlp.gate_proj,0.0000005636,0.01000,0.215
|
||
|
|
12,mlp.up_proj,0.0000004659,0.01000,0.216
|
||
|
|
12,mlp.down_proj,0.0000000890,0.01000,0.659
|
||
|
|
13,self_attn.q_proj,0.0000007450,0.01000,0.255
|
||
|
|
13,self_attn.k_proj,0.0000002548,0.01000,0.215
|
||
|
|
13,self_attn.v_proj,0.0000002986,0.01000,0.215
|
||
|
|
13,self_attn.o_proj,0.0000000196,0.01000,0.432
|
||
|
|
13,mlp.gate_proj,0.0000006345,0.01000,0.214
|
||
|
|
13,mlp.up_proj,0.0000005136,0.01000,0.218
|
||
|
|
13,mlp.down_proj,0.0000000943,0.01000,0.656
|
||
|
|
14,self_attn.q_proj,0.0000010099,0.01000,0.258
|
||
|
|
14,self_attn.k_proj,0.0000003666,0.01000,0.215
|
||
|
|
14,self_attn.v_proj,0.0000003883,0.01000,0.215
|
||
|
|
14,self_attn.o_proj,0.0000000272,0.01000,0.435
|
||
|
|
14,mlp.gate_proj,0.0000006731,0.01000,0.215
|
||
|
|
14,mlp.up_proj,0.0000005531,0.01000,0.217
|
||
|
|
14,mlp.down_proj,0.0000001202,0.01000,0.653
|
||
|
|
15,self_attn.q_proj,0.0000019066,0.01000,0.286
|
||
|
|
15,self_attn.k_proj,0.0000005995,0.01000,0.218
|
||
|
|
15,self_attn.v_proj,0.0000007671,0.01000,0.215
|
||
|
|
15,self_attn.o_proj,0.0000000275,0.01000,0.434
|
||
|
|
15,mlp.gate_proj,0.0000007228,0.01000,0.215
|
||
|
|
15,mlp.up_proj,0.0000006042,0.01000,0.215
|
||
|
|
15,mlp.down_proj,0.0000001389,0.01000,0.656
|
||
|
|
16,self_attn.q_proj,0.0000023607,0.01000,0.288
|
||
|
|
16,self_attn.k_proj,0.0000008254,0.01000,0.216
|
||
|
|
16,self_attn.v_proj,0.0000007666,0.01000,0.215
|
||
|
|
16,self_attn.o_proj,0.0000000499,0.01000,0.436
|
||
|
|
16,mlp.gate_proj,0.0000007419,0.01000,0.213
|
||
|
|
16,mlp.up_proj,0.0000006750,0.01000,0.215
|
||
|
|
16,mlp.down_proj,0.0000002760,0.01000,0.653
|
||
|
|
17,self_attn.q_proj,0.0000051417,0.01000,0.286
|
||
|
|
17,self_attn.k_proj,0.0000016262,0.01000,0.216
|
||
|
|
17,self_attn.v_proj,0.0000019666,0.01000,0.215
|
||
|
|
17,self_attn.o_proj,0.0000001240,0.01000,0.433
|
||
|
|
17,mlp.gate_proj,0.0000010348,0.01000,0.216
|
||
|
|
17,mlp.up_proj,0.0000009428,0.01000,0.213
|
||
|
|
17,mlp.down_proj,0.0000003055,0.01000,0.656
|
||
|
|
18,self_attn.q_proj,0.0000046798,0.01000,0.266
|
||
|
|
18,self_attn.k_proj,0.0000014684,0.01000,0.216
|
||
|
|
18,self_attn.v_proj,0.0000017435,0.01000,0.216
|
||
|
|
18,self_attn.o_proj,0.0000000528,0.01000,0.432
|
||
|
|
18,mlp.gate_proj,0.0000011800,0.01000,0.214
|
||
|
|
18,mlp.up_proj,0.0000010920,0.01000,0.213
|
||
|
|
18,mlp.down_proj,0.0000005320,0.01000,0.651
|
||
|
|
19,self_attn.q_proj,0.0000081104,0.01000,0.262
|
||
|
|
19,self_attn.k_proj,0.0000023922,0.01000,0.216
|
||
|
|
19,self_attn.v_proj,0.0000029877,0.01000,0.215
|
||
|
|
19,self_attn.o_proj,0.0000001101,0.01000,0.433
|
||
|
|
19,mlp.gate_proj,0.0000012617,0.01000,0.214
|
||
|
|
19,mlp.up_proj,0.0000013705,0.01000,0.214
|
||
|
|
19,mlp.down_proj,0.0000010956,0.01000,0.654
|
||
|
|
20,self_attn.q_proj,0.0000104670,0.01000,0.278
|
||
|
|
20,self_attn.k_proj,0.0000034022,0.01000,0.243
|
||
|
|
20,self_attn.v_proj,0.0000042421,0.01000,0.243
|
||
|
|
20,self_attn.o_proj,0.0000001907,0.01000,0.493
|
||
|
|
20,mlp.gate_proj,0.0000013812,0.01000,0.242
|
||
|
|
20,mlp.up_proj,0.0000015767,0.01000,0.242
|
||
|
|
20,mlp.down_proj,0.0000016353,0.01000,0.748
|
||
|
|
21,self_attn.q_proj,0.0000185475,0.01000,0.276
|
||
|
|
21,self_attn.k_proj,0.0000059908,0.01000,0.242
|
||
|
|
21,self_attn.v_proj,0.0000075683,0.01000,0.242
|
||
|
|
21,self_attn.o_proj,0.0000003463,0.01000,0.487
|
||
|
|
21,mlp.gate_proj,0.0000014603,0.01000,0.239
|
||
|
|
21,mlp.up_proj,0.0000018684,0.01000,0.239
|
||
|
|
21,mlp.down_proj,0.0000023298,0.01000,0.744
|
||
|
|
22,self_attn.q_proj,0.0000192655,0.01000,0.280
|
||
|
|
22,self_attn.k_proj,0.0000066621,0.01000,0.248
|
||
|
|
22,self_attn.v_proj,0.0000090980,0.01000,0.250
|
||
|
|
22,self_attn.o_proj,0.0000002642,0.01000,0.503
|
||
|
|
22,mlp.gate_proj,0.0000016697,0.01000,0.243
|
||
|
|
22,mlp.up_proj,0.0000021728,0.01000,0.240
|
||
|
|
22,mlp.down_proj,0.0000026377,0.01000,0.750
|
||
|
|
23,self_attn.q_proj,0.0000223355,0.01000,0.282
|
||
|
|
23,self_attn.k_proj,0.0000091246,0.01000,0.251
|
||
|
|
23,self_attn.v_proj,0.0000111566,0.01000,0.250
|
||
|
|
23,self_attn.o_proj,0.0000002412,0.01000,0.487
|
||
|
|
23,mlp.gate_proj,0.0000018961,0.01000,0.243
|
||
|
|
23,mlp.up_proj,0.0000025225,0.01000,0.239
|
||
|
|
23,mlp.down_proj,0.0000023748,0.01000,0.733
|
||
|
|
24,self_attn.q_proj,0.0000478594,0.01000,0.277
|
||
|
|
24,self_attn.k_proj,0.0000162195,0.01000,0.244
|
||
|
|
24,self_attn.v_proj,0.0000187105,0.01000,0.244
|
||
|
|
24,self_attn.o_proj,0.0000003140,0.01000,0.490
|
||
|
|
24,mlp.gate_proj,0.0000018491,0.01000,0.242
|
||
|
|
24,mlp.up_proj,0.0000025646,0.01000,0.241
|
||
|
|
24,mlp.down_proj,0.0000022875,0.01000,0.743
|
||
|
|
25,self_attn.q_proj,0.0000736474,0.01000,0.261
|
||
|
|
25,self_attn.k_proj,0.0000219495,0.01000,0.218
|
||
|
|
25,self_attn.v_proj,0.0000348672,0.01000,0.220
|
||
|
|
25,self_attn.o_proj,0.0000005398,0.01000,0.441
|
||
|
|
25,mlp.gate_proj,0.0000018458,0.01000,0.217
|
||
|
|
25,mlp.up_proj,0.0000026985,0.01000,0.217
|
||
|
|
25,mlp.down_proj,0.0000027734,0.01000,0.647
|
||
|
|
26,self_attn.q_proj,0.0000973767,0.01000,0.251
|
||
|
|
26,self_attn.k_proj,0.0000243580,0.01000,0.241
|
||
|
|
26,self_attn.v_proj,0.0000364561,0.01000,0.243
|
||
|
|
26,self_attn.o_proj,0.0000044389,0.01000,0.485
|
||
|
|
26,mlp.gate_proj,0.0000018377,0.01000,0.240
|
||
|
|
26,mlp.up_proj,0.0000026512,0.01000,0.246
|
||
|
|
26,mlp.down_proj,0.0000062773,0.01000,0.744
|
||
|
|
27,self_attn.q_proj,0.0000334389,0.01000,0.290
|
||
|
|
27,self_attn.k_proj,0.0000149378,0.01000,0.243
|
||
|
|
27,self_attn.v_proj,0.0000184760,0.01000,0.243
|
||
|
|
27,self_attn.o_proj,0.0000022290,0.01000,0.489
|
||
|
|
27,mlp.gate_proj,0.0000068754,0.01000,0.242
|
||
|
|
27,mlp.up_proj,0.0000081715,0.01000,0.240
|
||
|
|
27,mlp.down_proj,0.0000067090,0.01000,0.735
|