Hugging Face's logo

ModelCloud
/
Qwen2.5-0.5B-Instruct-gptqmodel-w4a16

Text Generation
Safetensors
qwen2
gptqmodel
modelcloud
chat
instruct
int4
gptq
4bit
W4A16
conversational
4-bit precision
Model card Files
xet
Community
2
Qubitium commited on
Commit
7911278
·
verified ·
1 Parent(s): 931236e

Upload folder using huggingface_hub

Browse files
Files changed (4) hide show
  1. config.json +1 -1
  2. model.safetensors +2 -2
  3. quant_log.csv +168 -168
  4. quantize_config.json +1 -1
config.json CHANGED
@@ -47,7 +47,7 @@
47
  "bits": 4,
48
  "checkpoint_format": "gptq",
49
  "desc_act": false,
50
- "group_size": 128,
51
  "lm_head": false,
52
  "meta": {
53
  "act_group_aware": true,
 
47
  "bits": 4,
48
  "checkpoint_format": "gptq",
49
  "desc_act": false,
50
+ "group_size": 32,
51
  "lm_head": false,
52
  "meta": {
53
  "act_group_aware": true,
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:05d1802c2690026b6326f91a35e2d7740c44b526c602eb84d651c787ee358218
3
- size 459384459
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5431a0944d541d6043c2541ccdc041cc5abccb2148eb43433ea55ebe726d0afb
3
+ size 480351195
quant_log.csv CHANGED
@@ -1,169 +1,169 @@
1
  layer,module,loss,samples,damp,time
2
- 0,self_attn.k_proj,0.0000001846,0.05000,0.421
3
- 0,self_attn.v_proj,0.0000000044,0.05000,0.623
4
- 0,self_attn.q_proj,0.0000012105,0.05000,0.651
5
- 0,self_attn.o_proj,0.0000000008,0.05000,0.272
6
- 0,mlp.up_proj,0.0000175088,0.05000,0.287
7
- 0,mlp.gate_proj,0.0000274922,0.05000,0.304
8
- 0,mlp.down_proj,0.0000003773,0.05000,1.938
9
- 1,self_attn.k_proj,0.0000020258,0.05000,0.378
10
- 1,self_attn.v_proj,0.0000002284,0.05000,0.662
11
- 1,self_attn.q_proj,0.0000073043,0.05000,0.671
12
- 1,self_attn.o_proj,0.0000001099,0.05000,0.616
13
- 1,mlp.up_proj,0.0000257946,0.05000,0.674
14
- 1,mlp.gate_proj,0.0000495168,0.05000,0.693
15
- 1,mlp.down_proj,0.0000003927,0.05000,1.795
16
- 2,self_attn.k_proj,0.0000044331,0.05000,0.442
17
- 2,self_attn.v_proj,0.0000006006,0.05000,0.808
18
- 2,self_attn.q_proj,0.0000161915,0.05000,0.832
19
- 2,self_attn.o_proj,0.0000001450,0.05000,0.437
20
- 2,mlp.gate_proj,0.0000690778,0.05000,0.259
21
- 2,mlp.up_proj,0.0000374052,0.05000,0.706
22
- 2,mlp.down_proj,0.0000096427,0.05000,1.414
23
- 3,self_attn.k_proj,0.0000051747,0.05000,0.492
24
- 3,self_attn.v_proj,0.0000012535,0.05000,0.897
25
- 3,self_attn.q_proj,0.0000214496,0.05000,0.907
26
- 3,self_attn.o_proj,0.0000002033,0.05000,0.422
27
- 3,mlp.gate_proj,0.0001435908,0.05000,0.493
28
- 3,mlp.up_proj,0.0000814998,0.05000,0.537
29
- 3,mlp.down_proj,0.0000165190,0.05000,1.751
30
- 4,self_attn.v_proj,0.0000015579,0.05000,0.470
31
- 4,self_attn.q_proj,0.0000169404,0.05000,0.474
32
- 4,self_attn.k_proj,0.0000033608,0.05000,0.710
33
- 4,self_attn.o_proj,0.0000004269,0.05000,0.545
34
- 4,mlp.up_proj,0.0000605056,0.05000,0.257
35
- 4,mlp.gate_proj,0.0000924927,0.05000,0.268
36
- 4,mlp.down_proj,0.0000012452,0.05000,2.330
37
- 5,self_attn.k_proj,0.0000042852,0.05000,0.248
38
- 5,self_attn.v_proj,0.0000020662,0.05000,0.388
39
- 5,self_attn.q_proj,0.0000212661,0.05000,0.393
40
- 5,self_attn.o_proj,0.0000003989,0.05000,0.239
41
- 5,mlp.gate_proj,0.0001488598,0.05000,0.466
42
- 5,mlp.up_proj,0.0000753015,0.05000,0.479
43
- 5,mlp.down_proj,0.0000011859,0.05000,2.195
44
- 6,self_attn.k_proj,0.0000039774,0.05000,0.668
45
- 6,self_attn.q_proj,0.0000173516,0.05000,1.329
46
- 6,self_attn.v_proj,0.0000011332,0.05000,1.336
47
- 6,self_attn.o_proj,0.0000002758,0.05000,0.556
48
- 6,mlp.up_proj,0.0000618804,0.05000,0.244
49
- 6,mlp.gate_proj,0.0000890550,0.05000,0.289
50
- 6,mlp.down_proj,0.0000013619,0.05000,3.632
51
- 7,self_attn.k_proj,0.0000060954,0.05000,0.255
52
- 7,self_attn.q_proj,0.0000250038,0.05000,1.341
53
- 7,self_attn.v_proj,0.0000019706,0.05000,1.345
54
- 7,self_attn.o_proj,0.0000006489,0.05000,0.269
55
- 7,mlp.up_proj,0.0000707421,0.05000,0.258
56
- 7,mlp.gate_proj,0.0000907406,0.05000,0.683
57
- 7,mlp.down_proj,0.0000017962,0.05000,3.559
58
- 8,self_attn.k_proj,0.0000045065,0.05000,0.372
59
- 8,self_attn.v_proj,0.0000016776,0.05000,0.670
60
- 8,self_attn.q_proj,0.0000237909,0.05000,0.693
61
- 8,self_attn.o_proj,0.0000007376,0.05000,0.356
62
- 8,mlp.up_proj,0.0000616151,0.05000,0.469
63
- 8,mlp.gate_proj,0.0000799935,0.05000,0.557
64
- 8,mlp.down_proj,0.0000011966,0.05000,3.384
65
- 9,self_attn.k_proj,0.0000126081,0.05000,0.343
66
- 9,self_attn.q_proj,0.0000496791,0.05000,0.533
67
- 9,self_attn.v_proj,0.0000033377,0.05000,0.570
68
- 9,self_attn.o_proj,0.0000005403,0.05000,0.411
69
- 9,mlp.gate_proj,0.0000756505,0.05000,0.502
70
- 9,mlp.up_proj,0.0000649747,0.05000,0.518
71
- 9,mlp.down_proj,0.0000014291,0.05000,1.314
72
- 10,self_attn.k_proj,0.0000040391,0.05000,0.408
73
- 10,self_attn.q_proj,0.0000211797,0.05000,0.704
74
- 10,self_attn.v_proj,0.0000029188,0.05000,0.713
75
- 10,self_attn.o_proj,0.0000008407,0.05000,0.387
76
- 10,mlp.gate_proj,0.0000737496,0.05000,0.456
77
- 10,mlp.up_proj,0.0000574715,0.05000,0.465
78
- 10,mlp.down_proj,0.0000011740,0.05000,3.047
79
- 11,self_attn.q_proj,0.0000515761,0.05000,0.552
80
- 11,self_attn.v_proj,0.0000037881,0.05000,0.567
81
- 11,self_attn.k_proj,0.0000141380,0.05000,0.632
82
- 11,self_attn.o_proj,0.0000005721,0.05000,0.704
83
- 11,mlp.gate_proj,0.0000717162,0.05000,0.263
84
- 11,mlp.up_proj,0.0000662663,0.05000,0.535
85
- 11,mlp.down_proj,0.0000017422,0.05000,2.340
86
- 12,self_attn.q_proj,0.0000219778,0.05000,0.404
87
- 12,self_attn.v_proj,0.0000025610,0.05000,0.417
88
- 12,self_attn.k_proj,0.0000039857,0.05000,0.663
89
- 12,self_attn.o_proj,0.0000009195,0.05000,0.678
90
- 12,mlp.gate_proj,0.0000669683,0.05000,0.263
91
- 12,mlp.up_proj,0.0000601216,0.05000,0.676
92
- 12,mlp.down_proj,0.0000013678,0.05000,2.078
93
- 13,self_attn.q_proj,0.0000371196,0.05000,0.511
94
- 13,self_attn.v_proj,0.0000030710,0.05000,0.517
95
- 13,self_attn.k_proj,0.0000074217,0.05000,0.612
96
- 13,self_attn.o_proj,0.0000007384,0.05000,0.669
97
- 13,mlp.gate_proj,0.0000725489,0.05000,0.232
98
- 13,mlp.up_proj,0.0000715233,0.05000,0.550
99
- 13,mlp.down_proj,0.0000019356,0.05000,3.418
100
- 14,self_attn.k_proj,0.0000051945,0.05000,0.446
101
- 14,self_attn.q_proj,0.0000330241,0.05000,0.719
102
- 14,self_attn.v_proj,0.0000047070,0.05000,0.733
103
- 14,self_attn.o_proj,0.0000018053,0.05000,0.223
104
- 14,mlp.up_proj,0.0000730560,0.05000,0.249
105
- 14,mlp.gate_proj,0.0000833288,0.05000,0.251
106
- 14,mlp.down_proj,0.0000024172,0.05000,1.230
107
- 15,self_attn.k_proj,0.0000056382,0.05000,0.430
108
- 15,self_attn.v_proj,0.0000039362,0.05000,0.856
109
- 15,self_attn.q_proj,0.0000289588,0.05000,0.860
110
- 15,self_attn.o_proj,0.0000010586,0.05000,0.353
111
- 15,mlp.gate_proj,0.0000967614,0.05000,0.617
112
- 15,mlp.up_proj,0.0000926957,0.05000,0.702
113
- 15,mlp.down_proj,0.0000031419,0.05000,2.435
114
- 16,self_attn.k_proj,0.0000109810,0.05000,0.443
115
- 16,self_attn.v_proj,0.0000096741,0.05000,0.734
116
- 16,self_attn.q_proj,0.0000628785,0.05000,0.748
117
- 16,self_attn.o_proj,0.0000011271,0.05000,0.389
118
- 16,mlp.gate_proj,0.0001423836,0.05000,0.274
119
- 16,mlp.up_proj,0.0001109843,0.05000,0.730
120
- 16,mlp.down_proj,0.0000060371,0.05000,1.237
121
- 17,self_attn.k_proj,0.0000060039,0.05000,0.309
122
- 17,self_attn.q_proj,0.0000400760,0.05000,0.681
123
- 17,self_attn.v_proj,0.0000059069,0.05000,0.682
124
- 17,self_attn.o_proj,0.0000008255,0.05000,0.386
125
- 17,mlp.up_proj,0.0001233144,0.05000,0.384
126
- 17,mlp.gate_proj,0.0001934214,0.05000,0.412
127
- 17,mlp.down_proj,0.0000050482,0.05000,2.300
128
- 18,self_attn.q_proj,0.0000411873,0.05000,0.357
129
- 18,self_attn.v_proj,0.0000059423,0.05000,0.369
130
- 18,self_attn.k_proj,0.0000077203,0.05000,0.678
131
- 18,self_attn.o_proj,0.0000010869,0.05000,0.599
132
- 18,mlp.up_proj,0.0001260548,0.05000,0.258
133
- 18,mlp.gate_proj,0.0001582911,0.05000,0.264
134
- 18,mlp.down_proj,0.0000059834,0.05000,2.446
135
- 19,self_attn.k_proj,0.0000063490,0.05000,0.272
136
- 19,self_attn.q_proj,0.0000380589,0.05000,0.666
137
- 19,self_attn.v_proj,0.0000057355,0.05000,0.670
138
- 19,self_attn.o_proj,0.0000013602,0.05000,0.247
139
- 19,mlp.up_proj,0.0001832222,0.05000,0.367
140
- 19,mlp.gate_proj,0.0002190037,0.05000,0.372
141
- 19,mlp.down_proj,0.0000113082,0.05000,2.235
142
- 20,self_attn.k_proj,0.0000078660,0.05000,0.290
143
- 20,self_attn.q_proj,0.0000533177,0.05000,0.443
144
- 20,self_attn.v_proj,0.0000177706,0.05000,0.459
145
- 20,self_attn.o_proj,0.0000024812,0.05000,0.569
146
- 20,mlp.up_proj,0.0002066750,0.05000,0.240
147
- 20,mlp.gate_proj,0.0002273082,0.05000,0.717
148
- 20,mlp.down_proj,0.0000155725,0.05000,2.551
149
- 21,self_attn.k_proj,0.0000082325,0.05000,0.232
150
- 21,self_attn.q_proj,0.0000596139,0.05000,1.457
151
- 21,self_attn.v_proj,0.0000242640,0.05000,1.457
152
- 21,self_attn.o_proj,0.0000054829,0.05000,0.257
153
- 21,mlp.up_proj,0.0002139384,0.05000,0.277
154
- 21,mlp.gate_proj,0.0002226604,0.05000,0.695
155
- 21,mlp.down_proj,0.0000279960,0.05000,2.661
156
- 22,self_attn.k_proj,0.0000075570,0.05000,0.458
157
- 22,self_attn.v_proj,0.0000210302,0.05000,0.811
158
- 22,self_attn.q_proj,0.0000556213,0.05000,0.815
159
- 22,self_attn.o_proj,0.0000020455,0.05000,0.256
160
- 22,mlp.up_proj,0.0001812818,0.05000,0.505
161
- 22,mlp.gate_proj,0.0001728906,0.05000,0.536
162
- 22,mlp.down_proj,0.0000204518,0.05000,1.919
163
- 23,self_attn.k_proj,0.0000085684,0.05000,0.263
164
- 23,self_attn.v_proj,0.0000168603,0.05000,0.415
165
- 23,self_attn.q_proj,0.0000655303,0.05000,0.421
166
- 23,self_attn.o_proj,0.0000053465,0.05000,0.242
167
- 23,mlp.gate_proj,0.0001968066,0.05000,0.261
168
- 23,mlp.up_proj,0.0001895685,0.05000,0.267
169
- 23,mlp.down_proj,0.0000390435,0.05000,1.745
 
1
  layer,module,loss,samples,damp,time
2
+ 0,self_attn.k_proj,0.0000001374,0.05000,0.411
3
+ 0,self_attn.v_proj,0.0000000028,0.05000,0.584
4
+ 0,self_attn.q_proj,0.0000008549,0.05000,0.584
5
+ 0,self_attn.o_proj,0.0000000005,0.05000,0.248
6
+ 0,mlp.up_proj,0.0000120706,0.05000,0.277
7
+ 0,mlp.gate_proj,0.0000190739,0.05000,0.312
8
+ 0,mlp.down_proj,0.0000002370,0.05000,1.265
9
+ 1,self_attn.k_proj,0.0000014250,0.05000,0.297
10
+ 1,self_attn.q_proj,0.0000049810,0.05000,0.346
11
+ 1,self_attn.v_proj,0.0000001409,0.05000,0.369
12
+ 1,self_attn.o_proj,0.0000000737,0.05000,0.239
13
+ 1,mlp.gate_proj,0.0000336152,0.05000,0.245
14
+ 1,mlp.up_proj,0.0000177082,0.05000,0.262
15
+ 1,mlp.down_proj,0.0000002587,0.05000,1.278
16
+ 2,self_attn.k_proj,0.0000030212,0.05000,0.289
17
+ 2,self_attn.v_proj,0.0000003602,0.05000,0.443
18
+ 2,self_attn.q_proj,0.0000104272,0.05000,0.447
19
+ 2,self_attn.o_proj,0.0000000885,0.05000,0.270
20
+ 2,mlp.up_proj,0.0000254395,0.05000,0.270
21
+ 2,mlp.gate_proj,0.0000473820,0.05000,0.284
22
+ 2,mlp.down_proj,0.0000063855,0.05000,1.350
23
+ 3,self_attn.k_proj,0.0000036164,0.05000,0.276
24
+ 3,self_attn.v_proj,0.0000008465,0.05000,0.425
25
+ 3,self_attn.q_proj,0.0000145756,0.05000,0.441
26
+ 3,self_attn.o_proj,0.0000001321,0.05000,0.236
27
+ 3,mlp.gate_proj,0.0000963621,0.05000,0.259
28
+ 3,mlp.up_proj,0.0000551544,0.05000,0.288
29
+ 3,mlp.down_proj,0.0000118582,0.05000,1.231
30
+ 4,self_attn.k_proj,0.0000022631,0.05000,0.307
31
+ 4,self_attn.q_proj,0.0000111644,0.05000,0.412
32
+ 4,self_attn.v_proj,0.0000010152,0.05000,0.417
33
+ 4,self_attn.o_proj,0.0000002745,0.05000,0.240
34
+ 4,mlp.up_proj,0.0000394713,0.05000,0.259
35
+ 4,mlp.gate_proj,0.0000614217,0.05000,0.286
36
+ 4,mlp.down_proj,0.0000007736,0.05000,1.260
37
+ 5,self_attn.k_proj,0.0000028338,0.05000,0.261
38
+ 5,self_attn.v_proj,0.0000013757,0.05000,0.496
39
+ 5,self_attn.q_proj,0.0000140757,0.05000,0.508
40
+ 5,self_attn.o_proj,0.0000002477,0.05000,0.286
41
+ 5,mlp.gate_proj,0.0001037606,0.05000,0.313
42
+ 5,mlp.up_proj,0.0000515816,0.05000,0.318
43
+ 5,mlp.down_proj,0.0000008419,0.05000,1.268
44
+ 6,self_attn.k_proj,0.0000025804,0.05000,0.257
45
+ 6,self_attn.v_proj,0.0000007676,0.05000,0.399
46
+ 6,self_attn.q_proj,0.0000114309,0.05000,0.400
47
+ 6,self_attn.o_proj,0.0000002146,0.05000,0.251
48
+ 6,mlp.gate_proj,0.0000600376,0.05000,0.267
49
+ 6,mlp.up_proj,0.0000420924,0.05000,0.276
50
+ 6,mlp.down_proj,0.0000009019,0.05000,1.338
51
+ 7,self_attn.k_proj,0.0000039587,0.05000,0.267
52
+ 7,self_attn.q_proj,0.0000166899,0.05000,0.389
53
+ 7,self_attn.v_proj,0.0000013533,0.05000,0.391
54
+ 7,self_attn.o_proj,0.0000003905,0.05000,0.282
55
+ 7,mlp.gate_proj,0.0000611757,0.05000,0.305
56
+ 7,mlp.up_proj,0.0000478114,0.05000,0.308
57
+ 7,mlp.down_proj,0.0000011080,0.05000,1.370
58
+ 8,self_attn.k_proj,0.0000030156,0.05000,0.240
59
+ 8,self_attn.q_proj,0.0000151704,0.05000,0.440
60
+ 8,self_attn.v_proj,0.0000010115,0.05000,0.443
61
+ 8,self_attn.o_proj,0.0000004879,0.05000,0.233
62
+ 8,mlp.gate_proj,0.0000540080,0.05000,0.284
63
+ 8,mlp.up_proj,0.0000413425,0.05000,0.298
64
+ 8,mlp.down_proj,0.0000007874,0.05000,1.348
65
+ 9,self_attn.k_proj,0.0000090284,0.05000,0.236
66
+ 9,self_attn.v_proj,0.0000023909,0.05000,0.394
67
+ 9,self_attn.q_proj,0.0000352399,0.05000,0.423
68
+ 9,self_attn.o_proj,0.0000003621,0.05000,0.238
69
+ 9,mlp.gate_proj,0.0000505815,0.05000,0.237
70
+ 9,mlp.up_proj,0.0000433161,0.05000,0.265
71
+ 9,mlp.down_proj,0.0000008689,0.05000,1.370
72
+ 10,self_attn.k_proj,0.0000025689,0.05000,0.246
73
+ 10,self_attn.v_proj,0.0000018463,0.05000,0.368
74
+ 10,self_attn.q_proj,0.0000134080,0.05000,0.373
75
+ 10,self_attn.o_proj,0.0000004496,0.05000,0.223
76
+ 10,mlp.up_proj,0.0000373606,0.05000,0.249
77
+ 10,mlp.gate_proj,0.0000484065,0.05000,0.285
78
+ 10,mlp.down_proj,0.0000007205,0.05000,1.319
79
+ 11,self_attn.k_proj,0.0000096949,0.05000,0.277
80
+ 11,self_attn.q_proj,0.0000358525,0.05000,0.465
81
+ 11,self_attn.v_proj,0.0000026096,0.05000,0.469
82
+ 11,self_attn.o_proj,0.0000003160,0.05000,0.245
83
+ 11,mlp.gate_proj,0.0000471413,0.05000,0.256
84
+ 11,mlp.up_proj,0.0000432136,0.05000,0.275
85
+ 11,mlp.down_proj,0.0000010805,0.05000,1.266
86
+ 12,self_attn.k_proj,0.0000025656,0.05000,0.277
87
+ 12,self_attn.q_proj,0.0000143737,0.05000,0.480
88
+ 12,self_attn.v_proj,0.0000016514,0.05000,0.485
89
+ 12,self_attn.o_proj,0.0000005012,0.05000,0.252
90
+ 12,mlp.up_proj,0.0000396000,0.05000,0.251
91
+ 12,mlp.gate_proj,0.0000438130,0.05000,0.274
92
+ 12,mlp.down_proj,0.0000008642,0.05000,1.347
93
+ 13,self_attn.k_proj,0.0000047205,0.05000,0.264
94
+ 13,self_attn.q_proj,0.0000238298,0.05000,0.530
95
+ 13,self_attn.v_proj,0.0000020183,0.05000,0.534
96
+ 13,self_attn.o_proj,0.0000004878,0.05000,0.262
97
+ 13,mlp.gate_proj,0.0000470698,0.05000,0.246
98
+ 13,mlp.up_proj,0.0000468081,0.05000,0.260
99
+ 13,mlp.down_proj,0.0000011870,0.05000,1.242
100
+ 14,self_attn.k_proj,0.0000033385,0.05000,0.275
101
+ 14,self_attn.q_proj,0.0000209151,0.05000,0.441
102
+ 14,self_attn.v_proj,0.0000030890,0.05000,0.447
103
+ 14,self_attn.o_proj,0.0000010850,0.05000,0.270
104
+ 14,mlp.up_proj,0.0000475374,0.05000,0.226
105
+ 14,mlp.gate_proj,0.0000534707,0.05000,0.294
106
+ 14,mlp.down_proj,0.0000014841,0.05000,1.378
107
+ 15,self_attn.k_proj,0.0000037106,0.05000,0.280
108
+ 15,self_attn.q_proj,0.0000186038,0.05000,0.407
109
+ 15,self_attn.v_proj,0.0000024893,0.05000,0.410
110
+ 15,self_attn.o_proj,0.0000006463,0.05000,0.247
111
+ 15,mlp.gate_proj,0.0000625236,0.05000,0.269
112
+ 15,mlp.up_proj,0.0000601912,0.05000,0.285
113
+ 15,mlp.down_proj,0.0000019398,0.05000,1.263
114
+ 16,self_attn.k_proj,0.0000071184,0.05000,0.239
115
+ 16,self_attn.v_proj,0.0000057205,0.05000,0.528
116
+ 16,self_attn.q_proj,0.0000406430,0.05000,0.543
117
+ 16,self_attn.o_proj,0.0000005914,0.05000,0.268
118
+ 16,mlp.up_proj,0.0000715898,0.05000,0.255
119
+ 16,mlp.gate_proj,0.0000915251,0.05000,0.266
120
+ 16,mlp.down_proj,0.0000035746,0.05000,1.357
121
+ 17,self_attn.k_proj,0.0000038815,0.05000,0.244
122
+ 17,self_attn.q_proj,0.0000259958,0.05000,0.342
123
+ 17,self_attn.v_proj,0.0000036715,0.05000,0.342
124
+ 17,self_attn.o_proj,0.0000004955,0.05000,0.261
125
+ 17,mlp.gate_proj,0.0001284090,0.05000,0.227
126
+ 17,mlp.up_proj,0.0000846460,0.05000,0.239
127
+ 17,mlp.down_proj,0.0000032604,0.05000,1.224
128
+ 18,self_attn.k_proj,0.0000044824,0.05000,0.252
129
+ 18,self_attn.q_proj,0.0000252827,0.05000,0.333
130
+ 18,self_attn.v_proj,0.0000040620,0.05000,0.336
131
+ 18,self_attn.o_proj,0.0000007172,0.05000,0.221
132
+ 18,mlp.up_proj,0.0000873495,0.05000,0.270
133
+ 18,mlp.gate_proj,0.0001080927,0.05000,0.279
134
+ 18,mlp.down_proj,0.0000040154,0.05000,1.300
135
+ 19,self_attn.k_proj,0.0000040385,0.05000,0.257
136
+ 19,self_attn.v_proj,0.0000038138,0.05000,0.393
137
+ 19,self_attn.q_proj,0.0000248874,0.05000,0.409
138
+ 19,self_attn.o_proj,0.0000008094,0.05000,0.230
139
+ 19,mlp.gate_proj,0.0001446683,0.05000,0.232
140
+ 19,mlp.up_proj,0.0001231465,0.05000,0.265
141
+ 19,mlp.down_proj,0.0000073288,0.05000,1.241
142
+ 20,self_attn.k_proj,0.0000051520,0.05000,0.257
143
+ 20,self_attn.v_proj,0.0000102601,0.05000,0.331
144
+ 20,self_attn.q_proj,0.0000350694,0.05000,0.341
145
+ 20,self_attn.o_proj,0.0000015609,0.05000,0.217
146
+ 20,mlp.up_proj,0.0001401550,0.05000,0.235
147
+ 20,mlp.gate_proj,0.0001513926,0.05000,0.260
148
+ 20,mlp.down_proj,0.0000101059,0.05000,1.319
149
+ 21,self_attn.k_proj,0.0000052119,0.05000,0.259
150
+ 21,self_attn.v_proj,0.0000155353,0.05000,0.418
151
+ 21,self_attn.q_proj,0.0000392914,0.05000,0.429
152
+ 21,self_attn.o_proj,0.0000030117,0.05000,0.243
153
+ 21,mlp.gate_proj,0.0001529105,0.05000,0.236
154
+ 21,mlp.up_proj,0.0001498237,0.05000,0.251
155
+ 21,mlp.down_proj,0.0000403716,0.05000,1.181
156
+ 22,self_attn.k_proj,0.0000048279,0.05000,0.280
157
+ 22,self_attn.q_proj,0.0000367167,0.05000,0.423
158
+ 22,self_attn.v_proj,0.0000140020,0.05000,0.435
159
+ 22,self_attn.o_proj,0.0000014592,0.05000,0.233
160
+ 22,mlp.gate_proj,0.0001180850,0.05000,0.251
161
+ 22,mlp.up_proj,0.0001251401,0.05000,0.263
162
+ 22,mlp.down_proj,0.0000136729,0.05000,1.290
163
+ 23,self_attn.k_proj,0.0000055235,0.05000,0.273
164
+ 23,self_attn.q_proj,0.0000443739,0.05000,0.404
165
+ 23,self_attn.v_proj,0.0000110931,0.05000,0.414
166
+ 23,self_attn.o_proj,0.0000029716,0.05000,0.235
167
+ 23,mlp.gate_proj,0.0001360635,0.05000,0.258
168
+ 23,mlp.up_proj,0.0001328737,0.05000,0.277
169
+ 23,mlp.down_proj,0.0000243960,0.05000,1.267
quantize_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "bits": 4,
3
- "group_size": 128,
4
  "desc_act": false,
5
  "sym": true,
6
  "lm_head": false,
 
1
  {
2
  "bits": 4,
3
+ "group_size": 32,
4
  "desc_act": false,
5
  "sym": true,
6
  "lm_head": false,