internvl3-5_tokenizer/quant_log.csv

layer,module,loss,samples,damp,time
0,self_attn.q_proj,0.0000044066,0.01000,0.476
0,self_attn.k_proj,0.0000019805,0.01000,0.236
0,self_attn.v_proj,0.0000015923,0.01000,0.235
0,self_attn.o_proj,0.0000012302,0.01000,0.467
0,mlp.gate_proj,0.0000475180,0.01000,0.230
0,mlp.up_proj,0.0000243452,0.01000,0.230
0,mlp.down_proj,0.0000015617,0.01000,0.697
1,self_attn.q_proj,0.0000019853,0.01000,0.342
1,self_attn.k_proj,0.0000008835,0.01000,0.291
1,self_attn.v_proj,0.0000008538,0.01000,0.266
1,self_attn.o_proj,0.0000003472,0.01000,0.471
1,mlp.gate_proj,0.0002938726,0.01000,0.234
1,mlp.up_proj,0.0000850689,0.01000,0.233
1,mlp.down_proj,0.0000021194,0.01000,0.699
2,self_attn.q_proj,0.0000038910,0.01000,0.279
2,self_attn.k_proj,0.0000016648,0.01000,0.231
2,self_attn.v_proj,0.0000016415,0.01000,0.231
2,self_attn.o_proj,0.0000005526,0.01000,0.462
2,mlp.gate_proj,0.0001997803,0.01000,0.232
2,mlp.up_proj,0.0000713199,0.01000,0.235
2,mlp.down_proj,0.0008174617,0.01000,0.696
3,self_attn.q_proj,0.0000274823,0.01000,0.270
3,self_attn.k_proj,0.0000134228,0.01000,0.237
3,self_attn.v_proj,0.0000133831,0.01000,0.237
3,self_attn.o_proj,0.0000007627,0.01000,0.471
3,mlp.gate_proj,0.0002638667,0.01000,0.251
3,mlp.up_proj,0.0001012415,0.01000,0.252
3,mlp.down_proj,0.0000053550,0.01000,0.701
4,self_attn.q_proj,0.0000261886,0.01000,0.255
4,self_attn.k_proj,0.0000123612,0.01000,0.229
4,self_attn.v_proj,0.0000132321,0.01000,0.230
4,self_attn.o_proj,0.0000017564,0.01000,0.463
4,mlp.gate_proj,0.0002446483,0.01000,0.234
4,mlp.up_proj,0.0001072414,0.01000,0.233
4,mlp.down_proj,0.0000070149,0.01000,0.696
5,self_attn.q_proj,0.0000465366,0.01000,0.254
5,self_attn.k_proj,0.0000193318,0.01000,0.231
5,self_attn.v_proj,0.0000204175,0.01000,0.231
5,self_attn.o_proj,0.0000022898,0.01000,0.462
5,mlp.gate_proj,0.0001662723,0.01000,0.231
5,mlp.up_proj,0.0000992902,0.01000,0.231
5,mlp.down_proj,0.0000078411,0.01000,0.695
6,self_attn.q_proj,0.0000313367,0.01000,0.256
6,self_attn.k_proj,0.0000139953,0.01000,0.229
6,self_attn.v_proj,0.0000134137,0.01000,0.230
6,self_attn.o_proj,0.0000019070,0.01000,0.462
6,mlp.gate_proj,0.0001977648,0.01000,0.229
6,mlp.up_proj,0.0001242954,0.01000,0.229
6,mlp.down_proj,0.0000100923,0.01000,0.691
7,self_attn.q_proj,0.0000582700,0.01000,0.255
7,self_attn.k_proj,0.0000240924,0.01000,0.229
7,self_attn.v_proj,0.0000268970,0.01000,0.229
7,self_attn.o_proj,0.0000028745,0.01000,0.460
7,mlp.gate_proj,0.0002205437,0.01000,0.231
7,mlp.up_proj,0.0001362631,0.01000,0.230
7,mlp.down_proj,0.0000117519,0.01000,0.694
8,self_attn.q_proj,0.0000694278,0.01000,0.255
8,self_attn.k_proj,0.0000311668,0.01000,0.227
8,self_attn.v_proj,0.0000296902,0.01000,0.229
8,self_attn.o_proj,0.0000033860,0.01000,0.460
8,mlp.gate_proj,0.0002141479,0.01000,0.230
8,mlp.up_proj,0.0001383425,0.01000,0.230
8,mlp.down_proj,0.0000117612,0.01000,0.693
9,self_attn.q_proj,0.0001289912,0.01000,0.253
9,self_attn.k_proj,0.0000529027,0.01000,0.228
9,self_attn.v_proj,0.0000542715,0.01000,0.226
9,self_attn.o_proj,0.0000050044,0.01000,0.459
9,mlp.gate_proj,0.0002396212,0.01000,0.231
9,mlp.up_proj,0.0001509409,0.01000,0.230
9,mlp.down_proj,0.0000166260,0.01000,0.687
10,self_attn.q_proj,0.0001100987,0.01000,0.253
10,self_attn.k_proj,0.0000461426,0.01000,0.226
10,self_attn.v_proj,0.0000473932,0.01000,0.226
10,self_attn.o_proj,0.0000052176,0.01000,0.462
10,mlp.gate_proj,0.0002404488,0.01000,0.229
10,mlp.up_proj,0.0001511854,0.01000,0.228
10,mlp.down_proj,0.0000251323,0.01000,0.700
11,self_attn.q_proj,0.0002361989,0.01000,0.231
11,self_attn.k_proj,0.0000922243,0.01000,0.227
11,self_attn.v_proj,0.0000842471,0.01000,0.227
11,self_attn.o_proj,0.0000151694,0.01000,0.459
11,mlp.gate_proj,0.0001903001,0.01000,0.229
11,mlp.up_proj,0.0001477866,0.01000,0.229
11,mlp.down_proj,0.0000251768,0.01000,0.694
12,self_attn.q_proj,0.0002058949,0.01000,0.256
12,self_attn.k_proj,0.0000757297,0.01000,0.230
12,self_attn.v_proj,0.0000796396,0.01000,0.228
12,self_attn.o_proj,0.0000055696,0.01000,0.460
12,mlp.gate_proj,0.0001705175,0.01000,0.231
12,mlp.up_proj,0.0001429938,0.01000,0.229
12,mlp.down_proj,0.0000244117,0.01000,0.694
13,self_attn.q_proj,0.0002131272,0.01000,0.254
13,self_attn.k_proj,0.0000752313,0.01000,0.228
13,self_attn.v_proj,0.0000870234,0.01000,0.226
13,self_attn.o_proj,0.0000073983,0.01000,0.458
13,mlp.gate_proj,0.0001896553,0.01000,0.227
13,mlp.up_proj,0.0001563918,0.01000,0.229
13,mlp.down_proj,0.0000264410,0.01000,0.687
14,self_attn.q_proj,0.0002928032,0.01000,0.256
14,self_attn.k_proj,0.0001088351,0.01000,0.229
14,self_attn.v_proj,0.0001156139,0.01000,0.228
14,self_attn.o_proj,0.0000082547,0.01000,0.462
14,mlp.gate_proj,0.0002012217,0.01000,0.227
14,mlp.up_proj,0.0001681767,0.01000,0.227
14,mlp.down_proj,0.0000352519,0.01000,0.690
15,self_attn.q_proj,0.0005602758,0.01000,0.232
15,self_attn.k_proj,0.0001800764,0.01000,0.229
15,self_attn.v_proj,0.0002261114,0.01000,0.229
15,self_attn.o_proj,0.0000099340,0.01000,0.461
15,mlp.gate_proj,0.0002238231,0.01000,0.230
15,mlp.up_proj,0.0001888835,0.01000,0.234
15,mlp.down_proj,0.0000422985,0.01000,0.696
16,self_attn.q_proj,0.0007166465,0.01000,0.240
16,self_attn.k_proj,0.0002515462,0.01000,0.238
16,self_attn.v_proj,0.0002401675,0.01000,0.228
16,self_attn.o_proj,0.0000150204,0.01000,0.461
16,mlp.gate_proj,0.0002338325,0.01000,0.231
16,mlp.up_proj,0.0002137495,0.01000,0.227
16,mlp.down_proj,0.0000872017,0.01000,0.693
17,self_attn.q_proj,0.0016075134,0.01000,0.242
17,self_attn.k_proj,0.0005267631,0.01000,0.229
17,self_attn.v_proj,0.0006408359,0.01000,0.230
17,self_attn.o_proj,0.0000315791,0.01000,0.464
17,mlp.gate_proj,0.0003384420,0.01000,0.229
17,mlp.up_proj,0.0003089405,0.01000,0.232
17,mlp.down_proj,0.0001076587,0.01000,0.700
18,self_attn.q_proj,0.0015861574,0.01000,0.254
18,self_attn.k_proj,0.0005132939,0.01000,0.228
18,self_attn.v_proj,0.0006109758,0.01000,0.228
18,self_attn.o_proj,0.0000254093,0.01000,0.461
18,mlp.gate_proj,0.0004072874,0.01000,0.228
18,mlp.up_proj,0.0003803636,0.01000,0.227
18,mlp.down_proj,0.0001712097,0.01000,0.690
19,self_attn.q_proj,0.0027800560,0.01000,0.252
19,self_attn.k_proj,0.0008529581,0.01000,0.226
19,self_attn.v_proj,0.0010568366,0.01000,0.226
19,self_attn.o_proj,0.0000415092,0.01000,0.457
19,mlp.gate_proj,0.0004365114,0.01000,0.227
19,mlp.up_proj,0.0004723935,0.01000,0.226
19,mlp.down_proj,0.0003228430,0.01000,0.693
20,self_attn.q_proj,0.0036227780,0.01000,0.254
20,self_attn.k_proj,0.0012287319,0.01000,0.228
20,self_attn.v_proj,0.0015251022,0.01000,0.226
20,self_attn.o_proj,0.0000489605,0.01000,0.462
20,mlp.gate_proj,0.0004963362,0.01000,0.228
20,mlp.up_proj,0.0005648018,0.01000,0.225
20,mlp.down_proj,0.0004068310,0.01000,0.702
21,self_attn.q_proj,0.0065157214,0.01000,0.239
21,self_attn.k_proj,0.0021859053,0.01000,0.226
21,self_attn.v_proj,0.0027217032,0.01000,0.226
21,self_attn.o_proj,0.0000782762,0.01000,0.462
21,mlp.gate_proj,0.0005434924,0.01000,0.227
21,mlp.up_proj,0.0006905879,0.01000,0.227
21,mlp.down_proj,0.0005955849,0.01000,0.687
22,self_attn.q_proj,0.0066680043,0.01000,0.239
22,self_attn.k_proj,0.0023850383,0.01000,0.228
22,self_attn.v_proj,0.0032360287,0.01000,0.228
22,self_attn.o_proj,0.0000882234,0.01000,0.463
22,mlp.gate_proj,0.0005762426,0.01000,0.229
22,mlp.up_proj,0.0007525707,0.01000,0.229
22,mlp.down_proj,0.0006313954,0.01000,0.699
23,self_attn.q_proj,0.0073616655,0.01000,0.235
23,self_attn.k_proj,0.0030943265,0.01000,0.226
23,self_attn.v_proj,0.0036315418,0.01000,0.227
23,self_attn.o_proj,0.0001049786,0.01000,0.460
23,mlp.gate_proj,0.0006277931,0.01000,0.230
23,mlp.up_proj,0.0008375677,0.01000,0.229
23,mlp.down_proj,0.0006724397,0.01000,0.693
24,self_attn.q_proj,0.0149952166,0.01000,0.240
24,self_attn.k_proj,0.0053064398,0.01000,0.230
24,self_attn.v_proj,0.0057998489,0.01000,0.229
24,self_attn.o_proj,0.0001213201,0.01000,0.455
24,mlp.gate_proj,0.0005788480,0.01000,0.230
24,mlp.up_proj,0.0008078051,0.01000,0.230
24,mlp.down_proj,0.0006965869,0.01000,0.703
25,self_attn.q_proj,0.0209445632,0.01000,0.240
25,self_attn.k_proj,0.0065191538,0.01000,0.233
25,self_attn.v_proj,0.0095592763,0.01000,0.231
25,self_attn.o_proj,0.0001847470,0.01000,0.466
25,mlp.gate_proj,0.0005431011,0.01000,0.231
25,mlp.up_proj,0.0007976848,0.01000,0.232
25,mlp.down_proj,0.0008871590,0.01000,0.700
26,self_attn.q_proj,0.0261648699,0.01000,0.238
26,self_attn.k_proj,0.0069253001,0.01000,0.233
26,self_attn.v_proj,0.0095860545,0.01000,0.232
26,self_attn.o_proj,0.0007540099,0.01000,0.463
26,mlp.gate_proj,0.0006003001,0.01000,0.228
26,mlp.up_proj,0.0008642617,0.01000,0.228
26,mlp.down_proj,0.0012626921,0.01000,0.693
27,self_attn.q_proj,0.0119673502,0.01000,0.230
27,self_attn.k_proj,0.0052954315,0.01000,0.227
27,self_attn.v_proj,0.0064633451,0.01000,0.226
27,self_attn.o_proj,0.0004531447,0.01000,0.460
27,mlp.gate_proj,0.0014340435,0.01000,0.227
27,mlp.up_proj,0.0016221918,0.01000,0.227
27,mlp.down_proj,0.0019386779,0.01000,0.691