tokmerge4bit / quant_log.csv
semran1's picture
Upload folder using huggingface_hub
6314a70 verified
layer,module,loss,samples,damp,time
0,self_attn.k_proj,0.0000010345,0.05000,0.890
0,self_attn.v_proj,0.0000002105,0.05000,0.713
0,self_attn.q_proj,0.0000028005,0.05000,0.734
0,self_attn.o_proj,0.0000011810,0.05000,0.708
0,mlp.gate_proj,0.0000014312,0.05000,0.728
0,mlp.up_proj,0.0000012446,0.05000,0.723
0,mlp.down_proj,0.0000018967,0.05000,2.076
1,self_attn.k_proj,0.0000000372,0.05000,0.703
1,self_attn.v_proj,0.0000000142,0.05000,0.698
1,self_attn.q_proj,0.0000001447,0.05000,0.712
1,self_attn.o_proj,0.0000000181,0.05000,0.718
1,mlp.gate_proj,0.0000151709,0.05000,0.726
1,mlp.up_proj,0.0000062952,0.05000,0.732
1,mlp.down_proj,0.0000002049,0.05000,2.205
2,self_attn.k_proj,0.0000001718,0.05000,0.770
2,self_attn.v_proj,0.0000000505,0.05000,0.741
2,self_attn.q_proj,0.0000004803,0.05000,0.745
2,self_attn.o_proj,0.0000001290,0.05000,0.760
2,mlp.gate_proj,0.0000314005,0.05000,0.793
2,mlp.up_proj,0.0000175286,0.05000,0.772
2,mlp.down_proj,0.0000007901,0.05000,2.255
3,self_attn.k_proj,0.0000010903,0.05000,0.767
3,self_attn.v_proj,0.0000002457,0.05000,0.777
3,self_attn.q_proj,0.0000028919,0.05000,0.757
3,self_attn.o_proj,0.0000003798,0.05000,0.765
3,mlp.gate_proj,0.0000401692,0.05000,0.784
3,mlp.up_proj,0.0000193787,0.05000,0.777
3,mlp.down_proj,0.0000016605,0.05000,2.277
4,self_attn.k_proj,0.0000012013,0.05000,0.750
4,self_attn.v_proj,0.0000003660,0.05000,0.773
4,self_attn.q_proj,0.0000032377,0.05000,0.754
4,self_attn.o_proj,0.0000006876,0.05000,0.760
4,mlp.gate_proj,0.0000665692,0.05000,0.797
4,mlp.up_proj,0.0000360474,0.05000,0.797
4,mlp.down_proj,0.0003383170,0.05000,2.296
5,self_attn.k_proj,0.0000039737,0.05000,0.788
5,self_attn.v_proj,0.0000016558,0.05000,0.761
5,self_attn.q_proj,0.0000126068,0.05000,0.768
5,self_attn.o_proj,0.0000008714,0.05000,0.770
5,mlp.gate_proj,0.0000997196,0.05000,0.799
5,mlp.up_proj,0.0000517149,0.05000,0.790
5,mlp.down_proj,0.0000093353,0.05000,2.252
6,self_attn.k_proj,0.0000039945,0.05000,0.760
6,self_attn.v_proj,0.0000021285,0.05000,0.760
6,self_attn.q_proj,0.0000136984,0.05000,0.768
6,self_attn.o_proj,0.0000009256,0.05000,0.769
6,mlp.gate_proj,0.0001461390,0.05000,0.787
6,mlp.up_proj,0.0000788555,0.05000,0.769
6,mlp.down_proj,0.0000059479,0.05000,2.259
7,self_attn.k_proj,0.0000054761,0.05000,0.791
7,self_attn.v_proj,0.0000031289,0.05000,0.755
7,self_attn.q_proj,0.0000191049,0.05000,0.787
7,self_attn.o_proj,0.0000010236,0.05000,0.785
7,mlp.gate_proj,0.0001826673,0.05000,0.807
7,mlp.up_proj,0.0000967049,0.05000,0.811
7,mlp.down_proj,0.0000050637,0.05000,2.332
8,self_attn.k_proj,0.0000061526,0.05000,0.743
8,self_attn.v_proj,0.0000026682,0.05000,0.749
8,self_attn.q_proj,0.0000203355,0.05000,0.743
8,self_attn.o_proj,0.0000009448,0.05000,0.760
8,mlp.gate_proj,0.0001155897,0.05000,0.764
8,mlp.up_proj,0.0000636666,0.05000,0.760
8,mlp.down_proj,0.0000066151,0.05000,2.208
9,self_attn.k_proj,0.0000054034,0.05000,0.764
9,self_attn.v_proj,0.0000030490,0.05000,0.754
9,self_attn.q_proj,0.0000183273,0.05000,0.763
9,self_attn.o_proj,0.0000017297,0.05000,0.790
9,mlp.gate_proj,0.0000433400,0.05000,0.786
9,mlp.up_proj,0.0000393703,0.05000,0.774
9,mlp.down_proj,0.0000097124,0.05000,2.207
10,self_attn.k_proj,0.0000075722,0.05000,0.764
10,self_attn.v_proj,0.0000041700,0.05000,0.763
10,self_attn.q_proj,0.0000266967,0.05000,0.763
10,self_attn.o_proj,0.0000018300,0.05000,0.766
10,mlp.gate_proj,0.0000491168,0.05000,0.782
10,mlp.up_proj,0.0000445106,0.05000,0.773
10,mlp.down_proj,0.0000119973,0.05000,2.279
11,self_attn.k_proj,0.0000058302,0.05000,0.794
11,self_attn.v_proj,0.0000028397,0.05000,0.766
11,self_attn.q_proj,0.0000204721,0.05000,0.774
11,self_attn.o_proj,0.0000026513,0.05000,0.793
11,mlp.gate_proj,0.0000638938,0.05000,0.786
11,mlp.up_proj,0.0000507965,0.05000,0.794
11,mlp.down_proj,0.0000096449,0.05000,2.229
12,self_attn.k_proj,0.0000069046,0.05000,0.783
12,self_attn.v_proj,0.0000030947,0.05000,0.878
12,self_attn.q_proj,0.0000244664,0.05000,0.777
12,self_attn.o_proj,0.0000026981,0.05000,0.767
12,mlp.gate_proj,0.0000555484,0.05000,0.784
12,mlp.up_proj,0.0000520922,0.05000,0.791
12,mlp.down_proj,0.0000114576,0.05000,2.340
13,self_attn.k_proj,0.0000074464,0.05000,0.780
13,self_attn.v_proj,0.0000037448,0.05000,0.782
13,self_attn.q_proj,0.0000257437,0.05000,0.803
13,self_attn.o_proj,0.0000030890,0.05000,0.791
13,mlp.gate_proj,0.0000620614,0.05000,0.793
13,mlp.up_proj,0.0000581586,0.05000,0.786
13,mlp.down_proj,0.0000134533,0.05000,2.318
14,self_attn.k_proj,0.0000088659,0.05000,0.753
14,self_attn.v_proj,0.0000039831,0.05000,0.768
14,self_attn.q_proj,0.0000300158,0.05000,0.798
14,self_attn.o_proj,0.0000031933,0.05000,0.777
14,mlp.gate_proj,0.0000633648,0.05000,0.812
14,mlp.up_proj,0.0000597147,0.05000,0.801
14,mlp.down_proj,0.0000143671,0.05000,2.256
15,self_attn.k_proj,0.0000074085,0.05000,0.790
15,self_attn.v_proj,0.0000044317,0.05000,0.786
15,self_attn.q_proj,0.0000262538,0.05000,0.808
15,self_attn.o_proj,0.0000041149,0.05000,0.786
15,mlp.gate_proj,0.0000688258,0.05000,0.913
15,mlp.up_proj,0.0000624603,0.05000,0.794
15,mlp.down_proj,0.0000152924,0.05000,2.393
16,self_attn.k_proj,0.0000077834,0.05000,0.797
16,self_attn.v_proj,0.0000031221,0.05000,0.810
16,self_attn.q_proj,0.0000243524,0.05000,0.829
16,self_attn.o_proj,0.0000036495,0.05000,0.725
16,mlp.gate_proj,0.0000598356,0.05000,0.742
16,mlp.up_proj,0.0000587959,0.05000,0.762
16,mlp.down_proj,0.0000148449,0.05000,2.141
17,self_attn.k_proj,0.0000084560,0.05000,0.740
17,self_attn.v_proj,0.0000037563,0.05000,0.748
17,self_attn.q_proj,0.0000286677,0.05000,0.753
17,self_attn.o_proj,0.0000037051,0.05000,0.741
17,mlp.gate_proj,0.0000578213,0.05000,0.745
17,mlp.up_proj,0.0000585330,0.05000,0.764
17,mlp.down_proj,0.0000138345,0.05000,2.159
18,self_attn.k_proj,0.0000084084,0.05000,0.727
18,self_attn.v_proj,0.0000041656,0.05000,0.730
18,self_attn.q_proj,0.0000288625,0.05000,0.732
18,self_attn.o_proj,0.0000046708,0.05000,0.738
18,mlp.gate_proj,0.0000558453,0.05000,0.763
18,mlp.up_proj,0.0000582194,0.05000,0.764
18,mlp.down_proj,0.0000137017,0.05000,2.182
19,self_attn.k_proj,0.0000092760,0.05000,0.737
19,self_attn.v_proj,0.0000043924,0.05000,0.738
19,self_attn.q_proj,0.0000349181,0.05000,0.742
19,self_attn.o_proj,0.0000035104,0.05000,0.724
19,mlp.gate_proj,0.0000577134,0.05000,0.801
19,mlp.up_proj,0.0000601438,0.05000,0.736
19,mlp.down_proj,0.0000141060,0.05000,2.143
20,self_attn.k_proj,0.0000099261,0.05000,0.717
20,self_attn.v_proj,0.0000040569,0.05000,0.732
20,self_attn.q_proj,0.0000341077,0.05000,0.730
20,self_attn.o_proj,0.0000044676,0.05000,0.750
20,mlp.gate_proj,0.0000520009,0.05000,0.741
20,mlp.up_proj,0.0000570655,0.05000,0.748
20,mlp.down_proj,0.0000139291,0.05000,2.276
21,self_attn.k_proj,0.0000087641,0.05000,0.760
21,self_attn.v_proj,0.0000039525,0.05000,0.765
21,self_attn.q_proj,0.0000297285,0.05000,0.765
21,self_attn.o_proj,0.0000044091,0.05000,0.807
21,mlp.gate_proj,0.0000513103,0.05000,0.797
21,mlp.up_proj,0.0000566967,0.05000,0.768
21,mlp.down_proj,0.0000128298,0.05000,2.276
22,self_attn.k_proj,0.0000097624,0.05000,0.741
22,self_attn.v_proj,0.0000058217,0.05000,0.717
22,self_attn.q_proj,0.0000330026,0.05000,0.733
22,self_attn.o_proj,0.0000053421,0.05000,0.730
22,mlp.gate_proj,0.0000517052,0.05000,0.765
22,mlp.up_proj,0.0000574648,0.05000,0.746
22,mlp.down_proj,0.0000163017,0.05000,2.181
23,self_attn.k_proj,0.0000091015,0.05000,0.735
23,self_attn.v_proj,0.0000058541,0.05000,0.755
23,self_attn.q_proj,0.0000333629,0.05000,0.771
23,self_attn.o_proj,0.0000059910,0.05000,0.743
23,mlp.gate_proj,0.0000558644,0.05000,0.747
23,mlp.up_proj,0.0000587899,0.05000,0.765
23,mlp.down_proj,0.0000146186,0.05000,2.239
24,self_attn.k_proj,0.0000124702,0.05000,0.735
24,self_attn.v_proj,0.0000050404,0.05000,0.722
24,self_attn.q_proj,0.0000385022,0.05000,0.740
24,self_attn.o_proj,0.0000055796,0.05000,0.765
24,mlp.gate_proj,0.0000534254,0.05000,0.765
24,mlp.up_proj,0.0000571216,0.05000,0.766
24,mlp.down_proj,0.0000131968,0.05000,2.224
25,self_attn.k_proj,0.0000123175,0.05000,0.781
25,self_attn.v_proj,0.0000069963,0.05000,0.761
25,self_attn.q_proj,0.0000430898,0.05000,0.782
25,self_attn.o_proj,0.0000045160,0.05000,0.762
25,mlp.gate_proj,0.0000526878,0.05000,0.785
25,mlp.up_proj,0.0000582908,0.05000,0.789
25,mlp.down_proj,0.0000140003,0.05000,2.231
26,self_attn.k_proj,0.0000107024,0.05000,0.763
26,self_attn.v_proj,0.0000044477,0.05000,0.751
26,self_attn.q_proj,0.0000376036,0.05000,0.771
26,self_attn.o_proj,0.0000049176,0.05000,0.769
26,mlp.gate_proj,0.0000533945,0.05000,0.783
26,mlp.up_proj,0.0000615110,0.05000,0.774
26,mlp.down_proj,0.0000153276,0.05000,2.146
27,self_attn.k_proj,0.0000109775,0.05000,0.719
27,self_attn.v_proj,0.0000054582,0.05000,0.718
27,self_attn.q_proj,0.0000362237,0.05000,0.731
27,self_attn.o_proj,0.0000067432,0.05000,0.747
27,mlp.gate_proj,0.0000555439,0.05000,0.743
27,mlp.up_proj,0.0000643386,0.05000,0.737
27,mlp.down_proj,0.0000185142,0.05000,2.128
28,self_attn.k_proj,0.0000090640,0.05000,0.728
28,self_attn.v_proj,0.0000076962,0.05000,0.719
28,self_attn.q_proj,0.0000360346,0.05000,0.731
28,self_attn.o_proj,0.0000078735,0.05000,0.733
28,mlp.gate_proj,0.0000557123,0.05000,0.753
28,mlp.up_proj,0.0000655429,0.05000,0.785
28,mlp.down_proj,0.0000212102,0.05000,2.175
29,self_attn.k_proj,0.0000124052,0.05000,0.731
29,self_attn.v_proj,0.0000079066,0.05000,0.725
29,self_attn.q_proj,0.0000431871,0.05000,0.749
29,self_attn.o_proj,0.0000082407,0.05000,0.759
29,mlp.gate_proj,0.0000578802,0.05000,0.764
29,mlp.up_proj,0.0000678786,0.05000,0.769
29,mlp.down_proj,0.0000245273,0.05000,2.176
30,self_attn.k_proj,0.0000104293,0.05000,0.738
30,self_attn.v_proj,0.0000085223,0.05000,0.729
30,self_attn.q_proj,0.0000395843,0.05000,0.741
30,self_attn.o_proj,0.0000108496,0.05000,0.743
30,mlp.gate_proj,0.0000602590,0.05000,0.760
30,mlp.up_proj,0.0000700332,0.05000,0.747
30,mlp.down_proj,0.0000283786,0.05000,2.199
31,self_attn.k_proj,0.0000104518,0.05000,0.767
31,self_attn.v_proj,0.0000081897,0.05000,0.726
31,self_attn.q_proj,0.0000394588,0.05000,0.732
31,self_attn.o_proj,0.0000096139,0.05000,0.743
31,mlp.gate_proj,0.0000703862,0.05000,0.777
31,mlp.up_proj,0.0000796129,0.05000,0.779
31,mlp.down_proj,0.0000355807,0.05000,2.291
32,self_attn.k_proj,0.0000115939,0.05000,0.766
32,self_attn.v_proj,0.0000111782,0.05000,0.795
32,self_attn.q_proj,0.0000435558,0.05000,0.795
32,self_attn.o_proj,0.0000094746,0.05000,0.768
32,mlp.gate_proj,0.0000780559,0.05000,0.801
32,mlp.up_proj,0.0000864127,0.05000,0.808
32,mlp.down_proj,0.0000409517,0.05000,2.330
33,self_attn.k_proj,0.0000120175,0.05000,0.767
33,self_attn.v_proj,0.0000108337,0.05000,0.782
33,self_attn.q_proj,0.0000466797,0.05000,0.830
33,self_attn.o_proj,0.0000109753,0.05000,0.781
33,mlp.gate_proj,0.0000965948,0.05000,0.784
33,mlp.up_proj,0.0001049083,0.05000,0.780
33,mlp.down_proj,0.0000591737,0.05000,2.203
34,self_attn.k_proj,0.0000126447,0.05000,0.759
34,self_attn.v_proj,0.0000134729,0.05000,0.751
34,self_attn.q_proj,0.0000514856,0.05000,0.787
34,self_attn.o_proj,0.0000105816,0.05000,0.763
34,mlp.gate_proj,0.0001163253,0.05000,0.802
34,mlp.up_proj,0.0001224971,0.05000,0.784
34,mlp.down_proj,0.0000742405,0.05000,2.217
35,self_attn.k_proj,0.0000122679,0.05000,0.753
35,self_attn.v_proj,0.0000121535,0.05000,0.764
35,self_attn.q_proj,0.0000477665,0.05000,0.735
35,self_attn.o_proj,0.0000167390,0.05000,0.764
35,mlp.gate_proj,0.0001359225,0.05000,0.801
35,mlp.up_proj,0.0001399457,0.05000,0.819
35,mlp.down_proj,0.0000927378,0.05000,2.227
36,self_attn.k_proj,0.0000139044,0.05000,0.758
36,self_attn.v_proj,0.0000217828,0.05000,0.760
36,self_attn.q_proj,0.0000595772,0.05000,0.770
36,self_attn.o_proj,0.0000149434,0.05000,0.777
36,mlp.gate_proj,0.0001476694,0.05000,0.785
36,mlp.up_proj,0.0001536171,0.05000,0.790
36,mlp.down_proj,0.0001145156,0.05000,2.285
37,self_attn.k_proj,0.0000154484,0.05000,0.750
37,self_attn.v_proj,0.0000210078,0.05000,0.748
37,self_attn.q_proj,0.0000604325,0.05000,0.749
37,self_attn.o_proj,0.0000172585,0.05000,0.778
37,mlp.gate_proj,0.0001625307,0.05000,0.744
37,mlp.up_proj,0.0001688472,0.05000,0.728
37,mlp.down_proj,0.0001271528,0.05000,2.139
38,self_attn.k_proj,0.0000146148,0.05000,0.733
38,self_attn.v_proj,0.0000206341,0.05000,0.746
38,self_attn.q_proj,0.0000587113,0.05000,0.750
38,self_attn.o_proj,0.0000170992,0.05000,0.754
38,mlp.gate_proj,0.0001785287,0.05000,0.787
38,mlp.up_proj,0.0001879436,0.05000,0.745
38,mlp.down_proj,0.0001442409,0.05000,2.204
39,self_attn.k_proj,0.0000137990,0.05000,0.755
39,self_attn.v_proj,0.0000218538,0.05000,0.739
39,self_attn.q_proj,0.0000598911,0.05000,0.750
39,self_attn.o_proj,0.0000223876,0.05000,0.759
39,mlp.gate_proj,0.0001947201,0.05000,0.781
39,mlp.up_proj,0.0002068335,0.05000,0.764
39,mlp.down_proj,0.0001697982,0.05000,2.225
40,self_attn.k_proj,0.0000156523,0.05000,0.744
40,self_attn.v_proj,0.0000300170,0.05000,0.758
40,self_attn.q_proj,0.0000653540,0.05000,0.760
40,self_attn.o_proj,0.0000205588,0.05000,0.760
40,mlp.gate_proj,0.0002070127,0.05000,0.784
40,mlp.up_proj,0.0002211764,0.05000,0.793
40,mlp.down_proj,0.0001941007,0.05000,2.267
41,self_attn.k_proj,0.0000143720,0.05000,0.749
41,self_attn.v_proj,0.0000298829,0.05000,0.775
41,self_attn.q_proj,0.0000652320,0.05000,0.767
41,self_attn.o_proj,0.0000170312,0.05000,0.763
41,mlp.gate_proj,0.0002186584,0.05000,0.778
41,mlp.up_proj,0.0002368328,0.05000,0.794
41,mlp.down_proj,0.0002126772,0.05000,2.215
42,self_attn.k_proj,0.0000154644,0.05000,0.748
42,self_attn.v_proj,0.0000335805,0.05000,0.742
42,self_attn.q_proj,0.0000643717,0.05000,0.783
42,self_attn.o_proj,0.0000202536,0.05000,0.774
42,mlp.gate_proj,0.0002305698,0.05000,0.819
42,mlp.up_proj,0.0002531587,0.05000,0.791
42,mlp.down_proj,0.0002493867,0.05000,2.257
43,self_attn.k_proj,0.0000155882,0.05000,0.767
43,self_attn.v_proj,0.0000471215,0.05000,0.767
43,self_attn.q_proj,0.0000713312,0.05000,0.766
43,self_attn.o_proj,0.0000335444,0.05000,0.763
43,mlp.gate_proj,0.0002482548,0.05000,0.798
43,mlp.up_proj,0.0002779884,0.05000,0.790
43,mlp.down_proj,0.0003057672,0.05000,2.334
44,self_attn.k_proj,0.0000140329,0.05000,0.801
44,self_attn.v_proj,0.0000511863,0.05000,0.807
44,self_attn.q_proj,0.0000693652,0.05000,0.801
44,self_attn.o_proj,0.0000460818,0.05000,0.818
44,mlp.gate_proj,0.0002590544,0.05000,0.802
44,mlp.up_proj,0.0002945287,0.05000,0.823
44,mlp.down_proj,0.0004848206,0.05000,2.348
45,self_attn.k_proj,0.0000146667,0.05000,0.809
45,self_attn.v_proj,0.0000620527,0.05000,0.796
45,self_attn.q_proj,0.0000714048,0.05000,0.802
45,self_attn.o_proj,0.0000510727,0.05000,0.828
45,mlp.gate_proj,0.0002783232,0.05000,0.829
45,mlp.up_proj,0.0003127684,0.05000,0.836
45,mlp.down_proj,0.0005779806,0.05000,2.360
46,self_attn.k_proj,0.0000135154,0.05000,0.756
46,self_attn.v_proj,0.0000692934,0.05000,0.770
46,self_attn.q_proj,0.0000701570,0.05000,0.782
46,self_attn.o_proj,0.0000890966,0.05000,0.776
46,mlp.gate_proj,0.0002948091,0.05000,0.778
46,mlp.up_proj,0.0003179033,0.05000,0.803
46,mlp.down_proj,0.0009762892,0.05000,2.270
47,self_attn.k_proj,0.0000116310,0.05000,0.781
47,self_attn.v_proj,0.0000427193,0.05000,0.780
47,self_attn.q_proj,0.0000536965,0.05000,0.783
47,self_attn.o_proj,0.0000413580,0.05000,0.800
47,mlp.gate_proj,0.0003372399,0.05000,0.795
47,mlp.up_proj,0.0003526463,0.05000,0.810
47,mlp.down_proj,0.0019484173,0.05000,2.288