SaffalPoosh commited on
Commit
404d1da
1 Parent(s): abad0de

added quant logs

Browse files
Files changed (1) hide show
  1. README.md +554 -0
README.md ADDED
@@ -0,0 +1,554 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+
3
+ autogptq quant. logs
4
+
5
+
6
+ ```
7
+ >>> model.quantize(examples)
8
+ 2023-07-21 16:54:47 INFO [auto_gptq.modeling._base] Start quantizing layer 1/32
9
+ 2023-07-21 16:54:47 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 1/32...
10
+ 2023-07-21 16:54:48 INFO [auto_gptq.quantization.gptq] duration: 0.8171646595001221
11
+ 2023-07-21 16:54:48 INFO [auto_gptq.quantization.gptq] avg loss: 3.7546463012695312
12
+ 2023-07-21 16:54:48 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 1/32...
13
+ 2023-07-21 16:54:49 INFO [auto_gptq.quantization.gptq] duration: 0.8055715560913086
14
+ 2023-07-21 16:54:49 INFO [auto_gptq.quantization.gptq] avg loss: 0.2164316177368164
15
+ 2023-07-21 16:54:49 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 1/32...
16
+ 2023-07-21 16:54:50 INFO [auto_gptq.quantization.gptq] duration: 0.8417620658874512
17
+ 2023-07-21 16:54:50 INFO [auto_gptq.quantization.gptq] avg loss: 16.070518493652344
18
+ 2023-07-21 16:54:50 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 1/32...
19
+ 2023-07-21 16:54:53 INFO [auto_gptq.quantization.gptq] duration: 3.90244197845459
20
+ 2023-07-21 16:54:53 INFO [auto_gptq.quantization.gptq] avg loss: 0.5676069855690002
21
+ 2023-07-21 16:54:53 INFO [auto_gptq.modeling._base] Start quantizing layer 2/32
22
+ 2023-07-21 16:54:54 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 2/32...
23
+ 2023-07-21 16:54:54 INFO [auto_gptq.quantization.gptq] duration: 0.8373761177062988
24
+ 2023-07-21 16:54:54 INFO [auto_gptq.quantization.gptq] avg loss: 4.066518783569336
25
+ 2023-07-21 16:54:54 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 2/32...
26
+ 2023-07-21 16:54:55 INFO [auto_gptq.quantization.gptq] duration: 0.8285796642303467
27
+ 2023-07-21 16:54:55 INFO [auto_gptq.quantization.gptq] avg loss: 0.2558078169822693
28
+ 2023-07-21 16:55:25 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 2/32...
29
+ 2023-07-21 16:55:25 INFO [auto_gptq.quantization.gptq] duration: 0.8859198093414307
30
+ 2023-07-21 16:55:25 INFO [auto_gptq.quantization.gptq] avg loss: 16.571727752685547
31
+ 2023-07-21 16:55:26 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 2/32...
32
+ 2023-07-21 16:55:29 INFO [auto_gptq.quantization.gptq] duration: 3.86962890625
33
+ 2023-07-21 16:55:29 INFO [auto_gptq.quantization.gptq] avg loss: 0.34605544805526733
34
+ 2023-07-21 16:55:30 INFO [auto_gptq.modeling._base] Start quantizing layer 3/32
35
+ 2023-07-21 16:55:30 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 3/32...
36
+ 2023-07-21 16:55:30 INFO [auto_gptq.quantization.gptq] duration: 0.8118832111358643
37
+ 2023-07-21 16:55:30 INFO [auto_gptq.quantization.gptq] avg loss: 5.4185943603515625
38
+ 2023-07-21 16:55:30 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 3/32...
39
+ 2023-07-21 16:55:31 INFO [auto_gptq.quantization.gptq] duration: 0.8096959590911865
40
+ 2023-07-21 16:55:31 INFO [auto_gptq.quantization.gptq] avg loss: 0.22585009038448334
41
+ 2023-07-21 16:55:31 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 3/32...
42
+ 2023-07-21 16:55:32 INFO [auto_gptq.quantization.gptq] duration: 0.8473665714263916
43
+ 2023-07-21 16:55:32 INFO [auto_gptq.quantization.gptq] avg loss: 27.050426483154297
44
+ 2023-07-21 16:55:32 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 3/32...
45
+ 2023-07-21 16:55:36 INFO [auto_gptq.quantization.gptq] duration: 3.8430850505828857
46
+ 2023-07-21 16:55:36 INFO [auto_gptq.quantization.gptq] avg loss: 0.6839203834533691
47
+ 2023-07-21 16:55:36 INFO [auto_gptq.modeling._base] Start quantizing layer 4/32
48
+ 2023-07-21 16:55:36 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 4/32...
49
+ 2023-07-21 16:55:37 INFO [auto_gptq.quantization.gptq] duration: 0.7948899269104004
50
+ 2023-07-21 16:55:37 INFO [auto_gptq.quantization.gptq] avg loss: 6.523550987243652
51
+ 2023-07-21 16:55:37 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 4/32...
52
+ 2023-07-21 16:55:38 INFO [auto_gptq.quantization.gptq] duration: 0.7990512847900391
53
+ 2023-07-21 16:55:38 INFO [auto_gptq.quantization.gptq] avg loss: 0.21638213098049164
54
+ 2023-07-21 16:55:38 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 4/32...
55
+ 2023-07-21 16:55:39 INFO [auto_gptq.quantization.gptq] duration: 0.8403058052062988
56
+ 2023-07-21 16:55:39 INFO [auto_gptq.quantization.gptq] avg loss: 36.57025146484375
57
+ 2023-07-21 16:55:39 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 4/32...
58
+ 2023-07-21 16:55:43 INFO [auto_gptq.quantization.gptq] duration: 3.856529474258423
59
+ 2023-07-21 16:55:43 INFO [auto_gptq.quantization.gptq] avg loss: 9.424503326416016
60
+ 2023-07-21 16:55:43 INFO [auto_gptq.modeling._base] Start quantizing layer 5/32
61
+ 2023-07-21 16:55:43 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 5/32...
62
+ 2023-07-21 16:55:44 INFO [auto_gptq.quantization.gptq] duration: 0.7926647663116455
63
+ 2023-07-21 16:55:44 INFO [auto_gptq.quantization.gptq] avg loss: 6.277029037475586
64
+ 2023-07-21 16:55:44 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 5/32...
65
+ 2023-07-21 16:55:44 INFO [auto_gptq.quantization.gptq] duration: 0.7987856864929199
66
+ 2023-07-21 16:55:44 INFO [auto_gptq.quantization.gptq] avg loss: 0.1324760764837265
67
+ 2023-07-21 16:55:44 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 5/32...
68
+ 2023-07-21 16:55:45 INFO [auto_gptq.quantization.gptq] duration: 0.8394050598144531
69
+ 2023-07-21 16:55:45 INFO [auto_gptq.quantization.gptq] avg loss: 36.26388168334961
70
+ 2023-07-21 16:55:45 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 5/32...
71
+ 2023-07-21 16:55:49 INFO [auto_gptq.quantization.gptq] duration: 3.849104166030884
72
+ 2023-07-21 16:55:49 INFO [auto_gptq.quantization.gptq] avg loss: 2.376619338989258
73
+ 2023-07-21 16:55:49 INFO [auto_gptq.modeling._base] Start quantizing layer 6/32
74
+ 2023-07-21 16:55:49 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 6/32...
75
+ 2023-07-21 16:55:50 INFO [auto_gptq.quantization.gptq] duration: 0.7964150905609131
76
+ 2023-07-21 16:55:50 INFO [auto_gptq.quantization.gptq] avg loss: 8.479263305664062
77
+ 2023-07-21 16:55:50 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 6/32...
78
+ 2023-07-21 16:55:51 INFO [auto_gptq.quantization.gptq] duration: 0.7951827049255371
79
+ 2023-07-21 16:55:51 INFO [auto_gptq.quantization.gptq] avg loss: 0.14170163869857788
80
+ 2023-07-21 16:56:21 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 6/32...
81
+ 2023-07-21 16:56:22 INFO [auto_gptq.quantization.gptq] duration: 0.8720560073852539
82
+ 2023-07-21 16:56:22 INFO [auto_gptq.quantization.gptq] avg loss: 42.756919860839844
83
+ 2023-07-21 16:56:22 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 6/32...
84
+ 2023-07-21 16:56:25 INFO [auto_gptq.quantization.gptq] duration: 3.8685550689697266
85
+ 2023-07-21 16:56:25 INFO [auto_gptq.quantization.gptq] avg loss: 0.8117952346801758
86
+ 2023-07-21 16:56:26 INFO [auto_gptq.modeling._base] Start quantizing layer 7/32
87
+ 2023-07-21 16:56:26 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 7/32...
88
+ 2023-07-21 16:56:26 INFO [auto_gptq.quantization.gptq] duration: 0.7976808547973633
89
+ 2023-07-21 16:56:26 INFO [auto_gptq.quantization.gptq] avg loss: 7.019394397735596
90
+ 2023-07-21 16:56:26 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 7/32...
91
+ 2023-07-21 16:56:27 INFO [auto_gptq.quantization.gptq] duration: 0.803225040435791
92
+ 2023-07-21 16:56:27 INFO [auto_gptq.quantization.gptq] avg loss: 0.21443051099777222
93
+ 2023-07-21 16:56:27 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 7/32...
94
+ 2023-07-21 16:56:28 INFO [auto_gptq.quantization.gptq] duration: 0.8342931270599365
95
+ 2023-07-21 16:56:28 INFO [auto_gptq.quantization.gptq] avg loss: 39.33504104614258
96
+ 2023-07-21 16:56:28 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 7/32...
97
+ 2023-07-21 16:56:32 INFO [auto_gptq.quantization.gptq] duration: 3.8671581745147705
98
+ 2023-07-21 16:56:32 INFO [auto_gptq.quantization.gptq] avg loss: 0.9214520454406738
99
+ 2023-07-21 16:56:32 INFO [auto_gptq.modeling._base] Start quantizing layer 8/32
100
+ 2023-07-21 16:56:32 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 8/32...
101
+ 2023-07-21 16:56:33 INFO [auto_gptq.quantization.gptq] duration: 0.7989864349365234
102
+ 2023-07-21 16:56:33 INFO [auto_gptq.quantization.gptq] avg loss: 7.602280616760254
103
+ 2023-07-21 16:56:33 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 8/32...
104
+ 2023-07-21 16:56:34 INFO [auto_gptq.quantization.gptq] duration: 0.8112733364105225
105
+ 2023-07-21 16:56:34 INFO [auto_gptq.quantization.gptq] avg loss: 0.11391645669937134
106
+ 2023-07-21 16:56:34 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 8/32...
107
+ 2023-07-21 16:56:35 INFO [auto_gptq.quantization.gptq] duration: 0.8388988971710205
108
+ 2023-07-21 16:56:35 INFO [auto_gptq.quantization.gptq] avg loss: 34.74957275390625
109
+ 2023-07-21 16:56:35 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 8/32...
110
+ 2023-07-21 16:56:39 INFO [auto_gptq.quantization.gptq] duration: 3.8561182022094727
111
+ 2023-07-21 16:56:39 INFO [auto_gptq.quantization.gptq] avg loss: 1.1289432048797607
112
+ 2023-07-21 16:56:39 INFO [auto_gptq.modeling._base] Start quantizing layer 9/32
113
+ 2023-07-21 16:56:39 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 9/32...
114
+ 2023-07-21 16:56:40 INFO [auto_gptq.quantization.gptq] duration: 0.7969386577606201
115
+ 2023-07-21 16:56:40 INFO [auto_gptq.quantization.gptq] avg loss: 6.806826591491699
116
+ 2023-07-21 16:56:40 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 9/32...
117
+ 2023-07-21 16:56:41 INFO [auto_gptq.quantization.gptq] duration: 0.7953078746795654
118
+ 2023-07-21 16:56:41 INFO [auto_gptq.quantization.gptq] avg loss: 0.2318212240934372
119
+ 2023-07-21 16:56:41 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 9/32...
120
+ 2023-07-21 16:56:41 INFO [auto_gptq.quantization.gptq] duration: 0.8294937610626221
121
+ 2023-07-21 16:56:41 INFO [auto_gptq.quantization.gptq] avg loss: 35.324676513671875
122
+ 2023-07-21 16:56:41 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 9/32...
123
+ 2023-07-21 16:56:45 INFO [auto_gptq.quantization.gptq] duration: 3.8630259037017822
124
+ 2023-07-21 16:56:45 INFO [auto_gptq.quantization.gptq] avg loss: 1.4622347354888916
125
+ 2023-07-21 16:56:45 INFO [auto_gptq.modeling._base] Start quantizing layer 10/32
126
+ 2023-07-21 16:56:46 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 10/32...
127
+ 2023-07-21 16:56:46 INFO [auto_gptq.quantization.gptq] duration: 0.8029708862304688
128
+ 2023-07-21 16:56:46 INFO [auto_gptq.quantization.gptq] avg loss: 6.056252956390381
129
+ 2023-07-21 16:56:46 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 10/32...
130
+ 2023-07-21 16:56:47 INFO [auto_gptq.quantization.gptq] duration: 0.8028323650360107
131
+ 2023-07-21 16:56:47 INFO [auto_gptq.quantization.gptq] avg loss: 1.092197060585022
132
+ 2023-07-21 16:56:47 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 10/32...
133
+ 2023-07-21 16:56:48 INFO [auto_gptq.quantization.gptq] duration: 0.8335537910461426
134
+ 2023-07-21 16:56:48 INFO [auto_gptq.quantization.gptq] avg loss: 30.71457290649414
135
+ 2023-07-21 16:56:48 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 10/32...
136
+ 2023-07-21 16:56:52 INFO [auto_gptq.quantization.gptq] duration: 3.8703184127807617
137
+ 2023-07-21 16:56:52 INFO [auto_gptq.quantization.gptq] avg loss: 1.2208330631256104
138
+ 2023-07-21 16:56:52 INFO [auto_gptq.modeling._base] Start quantizing layer 11/32
139
+ 2023-07-21 16:56:52 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 11/32...
140
+ 2023-07-21 16:56:53 INFO [auto_gptq.quantization.gptq] duration: 0.814570426940918
141
+ 2023-07-21 16:56:53 INFO [auto_gptq.quantization.gptq] avg loss: 6.145627021789551
142
+ 2023-07-21 16:56:53 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 11/32...
143
+ 2023-07-21 16:56:54 INFO [auto_gptq.quantization.gptq] duration: 0.8268287181854248
144
+ 2023-07-21 16:56:54 INFO [auto_gptq.quantization.gptq] avg loss: 0.24324843287467957
145
+ 2023-07-21 16:56:54 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 11/32...
146
+ 2023-07-21 16:56:55 INFO [auto_gptq.quantization.gptq] duration: 0.8359119892120361
147
+ 2023-07-21 16:56:55 INFO [auto_gptq.quantization.gptq] avg loss: 30.847026824951172
148
+ 2023-07-21 16:56:55 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 11/32...
149
+ 2023-07-21 16:56:58 INFO [auto_gptq.quantization.gptq] duration: 3.831470489501953
150
+ 2023-07-21 16:56:58 INFO [auto_gptq.quantization.gptq] avg loss: 1.3961751461029053
151
+ 2023-07-21 16:57:26 INFO [auto_gptq.modeling._base] Start quantizing layer 12/32
152
+ 2023-07-21 16:57:26 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 12/32...
153
+ 2023-07-21 16:57:27 INFO [auto_gptq.quantization.gptq] duration: 0.7964096069335938
154
+ 2023-07-21 16:57:27 INFO [auto_gptq.quantization.gptq] avg loss: 6.053964614868164
155
+ 2023-07-21 16:57:27 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 12/32...
156
+ 2023-07-21 16:57:28 INFO [auto_gptq.quantization.gptq] duration: 0.799691915512085
157
+ 2023-07-21 16:57:28 INFO [auto_gptq.quantization.gptq] avg loss: 0.2671034336090088
158
+ 2023-07-21 16:57:28 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 12/32...
159
+ 2023-07-21 16:57:29 INFO [auto_gptq.quantization.gptq] duration: 0.8342888355255127
160
+ 2023-07-21 16:57:29 INFO [auto_gptq.quantization.gptq] avg loss: 29.729408264160156
161
+ 2023-07-21 16:57:29 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 12/32...
162
+ 2023-07-21 16:57:33 INFO [auto_gptq.quantization.gptq] duration: 3.8561949729919434
163
+ 2023-07-21 16:57:33 INFO [auto_gptq.quantization.gptq] avg loss: 1.495622158050537
164
+ 2023-07-21 16:57:33 INFO [auto_gptq.modeling._base] Start quantizing layer 13/32
165
+ 2023-07-21 16:57:33 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 13/32...
166
+ 2023-07-21 16:57:34 INFO [auto_gptq.quantization.gptq] duration: 0.7953364849090576
167
+ 2023-07-21 16:57:34 INFO [auto_gptq.quantization.gptq] avg loss: 5.408998489379883
168
+ 2023-07-21 16:57:34 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 13/32...
169
+ 2023-07-21 16:57:34 INFO [auto_gptq.quantization.gptq] duration: 0.7990250587463379
170
+ 2023-07-21 16:57:34 INFO [auto_gptq.quantization.gptq] avg loss: 0.5066410303115845
171
+ 2023-07-21 16:57:34 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 13/32...
172
+ 2023-07-21 16:57:35 INFO [auto_gptq.quantization.gptq] duration: 0.8330769538879395
173
+ 2023-07-21 16:57:35 INFO [auto_gptq.quantization.gptq] avg loss: 27.790515899658203
174
+ 2023-07-21 16:57:35 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 13/32...
175
+ 2023-07-21 16:57:39 INFO [auto_gptq.quantization.gptq] duration: 3.861015558242798
176
+ 2023-07-21 16:57:39 INFO [auto_gptq.quantization.gptq] avg loss: 1.3019633293151855
177
+ 2023-07-21 16:57:39 INFO [auto_gptq.modeling._base] Start quantizing layer 14/32
178
+ 2023-07-21 16:57:39 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 14/32...
179
+ 2023-07-21 16:57:40 INFO [auto_gptq.quantization.gptq] duration: 0.8011329174041748
180
+ 2023-07-21 16:57:40 INFO [auto_gptq.quantization.gptq] avg loss: 6.027165412902832
181
+ 2023-07-21 16:57:40 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 14/32...
182
+ 2023-07-21 16:57:41 INFO [auto_gptq.quantization.gptq] duration: 0.7977538108825684
183
+ 2023-07-21 16:57:41 INFO [auto_gptq.quantization.gptq] avg loss: 0.28969255089759827
184
+ 2023-07-21 16:57:41 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 14/32...
185
+ 2023-07-21 16:57:42 INFO [auto_gptq.quantization.gptq] duration: 0.8305981159210205
186
+ 2023-07-21 16:57:42 INFO [auto_gptq.quantization.gptq] avg loss: 28.996891021728516
187
+ 2023-07-21 16:57:42 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 14/32...
188
+ 2023-07-21 16:57:46 INFO [auto_gptq.quantization.gptq] duration: 3.874257802963257
189
+ 2023-07-21 16:57:46 INFO [auto_gptq.quantization.gptq] avg loss: 1.6258554458618164
190
+ 2023-07-21 16:57:46 INFO [auto_gptq.modeling._base] Start quantizing layer 15/32
191
+ 2023-07-21 16:57:46 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 15/32...
192
+ 2023-07-21 16:57:47 INFO [auto_gptq.quantization.gptq] duration: 0.7982082366943359
193
+ 2023-07-21 16:57:47 INFO [auto_gptq.quantization.gptq] avg loss: 5.937747001647949
194
+ 2023-07-21 16:57:47 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 15/32...
195
+ 2023-07-21 16:57:48 INFO [auto_gptq.quantization.gptq] duration: 0.8004462718963623
196
+ 2023-07-21 16:57:48 INFO [auto_gptq.quantization.gptq] avg loss: 0.3830963373184204
197
+ 2023-07-21 16:57:48 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 15/32...
198
+ 2023-07-21 16:57:48 INFO [auto_gptq.quantization.gptq] duration: 0.8347995281219482
199
+ 2023-07-21 16:57:48 INFO [auto_gptq.quantization.gptq] avg loss: 30.339778900146484
200
+ 2023-07-21 16:57:48 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 15/32...
201
+ 2023-07-21 16:57:52 INFO [auto_gptq.quantization.gptq] duration: 3.8794045448303223
202
+ 2023-07-21 16:57:52 INFO [auto_gptq.quantization.gptq] avg loss: 1.618453025817871
203
+ 2023-07-21 16:57:52 INFO [auto_gptq.modeling._base] Start quantizing layer 16/32
204
+ 2023-07-21 16:57:53 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 16/32...
205
+ 2023-07-21 16:57:53 INFO [auto_gptq.quantization.gptq] duration: 0.802685022354126
206
+ 2023-07-21 16:57:53 INFO [auto_gptq.quantization.gptq] avg loss: 5.992144584655762
207
+ 2023-07-21 16:57:53 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 16/32...
208
+ 2023-07-21 16:57:54 INFO [auto_gptq.quantization.gptq] duration: 0.8001143932342529
209
+ 2023-07-21 16:57:54 INFO [auto_gptq.quantization.gptq] avg loss: 0.3652211129665375
210
+ 2023-07-21 16:57:54 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 16/32...
211
+ 2023-07-21 16:57:55 INFO [auto_gptq.quantization.gptq] duration: 0.843254566192627
212
+ 2023-07-21 16:57:55 INFO [auto_gptq.quantization.gptq] avg loss: 29.359691619873047
213
+ 2023-07-21 16:57:55 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 16/32...
214
+ 2023-07-21 16:57:59 INFO [auto_gptq.quantization.gptq] duration: 3.8731229305267334
215
+ 2023-07-21 16:57:59 INFO [auto_gptq.quantization.gptq] avg loss: 1.8666539192199707
216
+ 2023-07-21 16:57:59 INFO [auto_gptq.modeling._base] Start quantizing layer 17/32
217
+ 2023-07-21 16:57:59 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 17/32...
218
+ 2023-07-21 16:58:00 INFO [auto_gptq.quantization.gptq] duration: 0.79642653465271
219
+ 2023-07-21 16:58:00 INFO [auto_gptq.quantization.gptq] avg loss: 6.463171482086182
220
+ 2023-07-21 16:58:00 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 17/32...
221
+ 2023-07-21 16:58:01 INFO [auto_gptq.quantization.gptq] duration: 0.8078687191009521
222
+ 2023-07-21 16:58:01 INFO [auto_gptq.quantization.gptq] avg loss: 0.24540238082408905
223
+ 2023-07-21 16:58:01 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 17/32...
224
+ 2023-07-21 16:58:02 INFO [auto_gptq.quantization.gptq] duration: 0.829270601272583
225
+ 2023-07-21 16:58:02 INFO [auto_gptq.quantization.gptq] avg loss: 30.825468063354492
226
+ 2023-07-21 16:58:02 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 17/32...
227
+ 2023-07-21 16:58:05 INFO [auto_gptq.quantization.gptq] duration: 3.855315923690796
228
+ 2023-07-21 16:58:05 INFO [auto_gptq.quantization.gptq] avg loss: 1.957414150238037
229
+ 2023-07-21 16:58:06 INFO [auto_gptq.modeling._base] Start quantizing layer 18/32
230
+ 2023-07-21 16:58:06 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 18/32...
231
+ 2023-07-21 16:58:07 INFO [auto_gptq.quantization.gptq] duration: 0.8099801540374756
232
+ 2023-07-21 16:58:07 INFO [auto_gptq.quantization.gptq] avg loss: 6.510787010192871
233
+ 2023-07-21 16:58:07 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 18/32...
234
+ 2023-07-21 16:58:07 INFO [auto_gptq.quantization.gptq] duration: 0.8008811473846436
235
+ 2023-07-21 16:58:07 INFO [auto_gptq.quantization.gptq] avg loss: 0.3201957941055298
236
+ 2023-07-21 16:58:07 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 18/32...
237
+ 2023-07-21 16:58:08 INFO [auto_gptq.quantization.gptq] duration: 0.8365602493286133
238
+ 2023-07-21 16:58:08 INFO [auto_gptq.quantization.gptq] avg loss: 31.26324462890625
239
+ 2023-07-21 16:58:08 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 18/32...
240
+ 2023-07-21 16:58:12 INFO [auto_gptq.quantization.gptq] duration: 3.8536572456359863
241
+ 2023-07-21 16:58:12 INFO [auto_gptq.quantization.gptq] avg loss: 2.0843615531921387
242
+ 2023-07-21 16:58:12 INFO [auto_gptq.modeling._base] Start quantizing layer 19/32
243
+ 2023-07-21 16:58:12 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 19/32...
244
+ 2023-07-21 16:58:13 INFO [auto_gptq.quantization.gptq] duration: 0.7980837821960449
245
+ 2023-07-21 16:58:13 INFO [auto_gptq.quantization.gptq] avg loss: 6.686659812927246
246
+ 2023-07-21 16:58:13 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 19/32...
247
+ 2023-07-21 16:58:14 INFO [auto_gptq.quantization.gptq] duration: 0.7951889038085938
248
+ 2023-07-21 16:58:14 INFO [auto_gptq.quantization.gptq] avg loss: 0.3053201138973236
249
+ 2023-07-21 16:58:14 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 19/32...
250
+ 2023-07-21 16:58:15 INFO [auto_gptq.quantization.gptq] duration: 0.8315420150756836
251
+ 2023-07-21 16:58:15 INFO [auto_gptq.quantization.gptq] avg loss: 31.97283935546875
252
+ 2023-07-21 16:58:15 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 19/32...
253
+ 2023-07-21 16:58:19 INFO [auto_gptq.quantization.gptq] duration: 3.868382215499878
254
+ 2023-07-21 16:58:19 INFO [auto_gptq.quantization.gptq] avg loss: 2.382962703704834
255
+ 2023-07-21 16:58:19 INFO [auto_gptq.modeling._base] Start quantizing layer 20/32
256
+ 2023-07-21 16:58:19 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 20/32...
257
+ 2023-07-21 16:58:20 INFO [auto_gptq.quantization.gptq] duration: 0.797062873840332
258
+ 2023-07-21 16:58:20 INFO [auto_gptq.quantization.gptq] avg loss: 6.721341133117676
259
+ 2023-07-21 16:58:20 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 20/32...
260
+ 2023-07-21 16:58:20 INFO [auto_gptq.quantization.gptq] duration: 0.806023120880127
261
+ 2023-07-21 16:58:20 INFO [auto_gptq.quantization.gptq] avg loss: 0.5635891556739807
262
+ 2023-07-21 16:58:20 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 20/32...
263
+ 2023-07-21 16:58:21 INFO [auto_gptq.quantization.gptq] duration: 0.841651201248169
264
+ 2023-07-21 16:58:21 INFO [auto_gptq.quantization.gptq] avg loss: 33.371273040771484
265
+ 2023-07-21 16:58:21 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 20/32...
266
+ 2023-07-21 16:58:25 INFO [auto_gptq.quantization.gptq] duration: 3.8724091053009033
267
+ 2023-07-21 16:58:25 INFO [auto_gptq.quantization.gptq] avg loss: 2.5540378093719482
268
+ 2023-07-21 16:58:25 INFO [auto_gptq.modeling._base] Start quantizing layer 21/32
269
+ 2023-07-21 16:58:25 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 21/32...
270
+ 2023-07-21 16:58:26 INFO [auto_gptq.quantization.gptq] duration: 0.8135292530059814
271
+ 2023-07-21 16:58:26 INFO [auto_gptq.quantization.gptq] avg loss: 7.383816242218018
272
+ 2023-07-21 16:58:26 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 21/32...
273
+ 2023-07-21 16:58:27 INFO [auto_gptq.quantization.gptq] duration: 0.8004577159881592
274
+ 2023-07-21 16:58:27 INFO [auto_gptq.quantization.gptq] avg loss: 0.2988166809082031
275
+ 2023-07-21 16:58:27 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 21/32...
276
+ 2023-07-21 16:58:28 INFO [auto_gptq.quantization.gptq] duration: 0.8346357345581055
277
+ 2023-07-21 16:58:28 INFO [auto_gptq.quantization.gptq] avg loss: 34.46820068359375
278
+ 2023-07-21 16:58:28 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 21/32...
279
+ 2023-07-21 16:58:32 INFO [auto_gptq.quantization.gptq] duration: 3.8698837757110596
280
+ 2023-07-21 16:58:32 INFO [auto_gptq.quantization.gptq] avg loss: 2.538421154022217
281
+ 2023-07-21 16:58:32 INFO [auto_gptq.modeling._base] Start quantizing layer 22/32
282
+ 2023-07-21 16:58:32 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 22/32...
283
+ 2023-07-21 16:58:33 INFO [auto_gptq.quantization.gptq] duration: 0.7975707054138184
284
+ 2023-07-21 16:58:33 INFO [auto_gptq.quantization.gptq] avg loss: 7.026803970336914
285
+ 2023-07-21 16:58:33 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 22/32...
286
+ 2023-07-21 16:58:34 INFO [auto_gptq.quantization.gptq] duration: 0.7988865375518799
287
+ 2023-07-21 16:58:34 INFO [auto_gptq.quantization.gptq] avg loss: 0.5440877079963684
288
+ 2023-07-21 16:58:34 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 22/32...
289
+ 2023-07-21 16:58:35 INFO [auto_gptq.quantization.gptq] duration: 0.847116231918335
290
+ 2023-07-21 16:58:35 INFO [auto_gptq.quantization.gptq] avg loss: 33.8814582824707
291
+ 2023-07-21 16:58:35 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 22/32...
292
+ 2023-07-21 16:58:38 INFO [auto_gptq.quantization.gptq] duration: 3.851823091506958
293
+ 2023-07-21 16:58:38 INFO [auto_gptq.quantization.gptq] avg loss: 2.612248182296753
294
+ 2023-07-21 16:58:39 INFO [auto_gptq.modeling._base] Start quantizing layer 23/32
295
+ 2023-07-21 16:58:39 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 23/32...
296
+ 2023-07-21 16:58:39 INFO [auto_gptq.quantization.gptq] duration: 0.7956225872039795
297
+ 2023-07-21 16:58:39 INFO [auto_gptq.quantization.gptq] avg loss: 7.3217453956604
298
+ 2023-07-21 16:58:39 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 23/32...
299
+ 2023-07-21 16:58:40 INFO [auto_gptq.quantization.gptq] duration: 0.8155944347381592
300
+ 2023-07-21 16:58:40 INFO [auto_gptq.quantization.gptq] avg loss: 0.3978100121021271
301
+ 2023-07-21 16:58:40 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 23/32...
302
+ 2023-07-21 16:58:41 INFO [auto_gptq.quantization.gptq] duration: 0.8472270965576172
303
+ 2023-07-21 16:58:41 INFO [auto_gptq.quantization.gptq] avg loss: 33.613494873046875
304
+ 2023-07-21 16:58:41 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 23/32...
305
+ 2023-07-21 16:58:45 INFO [auto_gptq.quantization.gptq] duration: 3.877121925354004
306
+ 2023-07-21 16:58:45 INFO [auto_gptq.quantization.gptq] avg loss: 3.0234107971191406
307
+ 2023-07-21 16:58:45 INFO [auto_gptq.modeling._base] Start quantizing layer 24/32
308
+ 2023-07-21 16:58:45 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 24/32...
309
+ 2023-07-21 16:58:46 INFO [auto_gptq.quantization.gptq] duration: 0.8478920459747314
310
+ 2023-07-21 16:58:46 INFO [auto_gptq.quantization.gptq] avg loss: 7.490325927734375
311
+ 2023-07-21 16:58:46 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 24/32...
312
+ 2023-07-21 16:58:47 INFO [auto_gptq.quantization.gptq] duration: 0.8023700714111328
313
+ 2023-07-21 16:58:47 INFO [auto_gptq.quantization.gptq] avg loss: 0.6462091207504272
314
+ 2023-07-21 16:58:47 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 24/32...
315
+ 2023-07-21 16:58:48 INFO [auto_gptq.quantization.gptq] duration: 0.8271210193634033
316
+ 2023-07-21 16:58:48 INFO [auto_gptq.quantization.gptq] avg loss: 35.156715393066406
317
+ 2023-07-21 16:58:48 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 24/32...
318
+ 2023-07-21 16:58:52 INFO [auto_gptq.quantization.gptq] duration: 3.8558664321899414
319
+ 2023-07-21 16:58:52 INFO [auto_gptq.quantization.gptq] avg loss: 3.4150047302246094
320
+ 2023-07-21 16:58:52 INFO [auto_gptq.modeling._base] Start quantizing layer 25/32
321
+ 2023-07-21 16:58:52 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 25/32...
322
+ 2023-07-21 16:58:53 INFO [auto_gptq.quantization.gptq] duration: 0.804887056350708
323
+ 2023-07-21 16:58:53 INFO [auto_gptq.quantization.gptq] avg loss: 7.842990875244141
324
+ 2023-07-21 16:58:53 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 25/32...
325
+ 2023-07-21 16:58:53 INFO [auto_gptq.quantization.gptq] duration: 0.7986440658569336
326
+ 2023-07-21 16:58:53 INFO [auto_gptq.quantization.gptq] avg loss: 0.5917433500289917
327
+ 2023-07-21 16:58:53 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 25/32...
328
+ 2023-07-21 16:58:54 INFO [auto_gptq.quantization.gptq] duration: 0.8256046772003174
329
+ 2023-07-21 16:58:54 INFO [auto_gptq.quantization.gptq] avg loss: 36.299095153808594
330
+ 2023-07-21 16:58:54 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 25/32...
331
+ 2023-07-21 16:58:58 INFO [auto_gptq.quantization.gptq] duration: 3.86680006980896
332
+ 2023-07-21 16:58:58 INFO [auto_gptq.quantization.gptq] avg loss: 4.292586326599121
333
+ 2023-07-21 16:58:58 INFO [auto_gptq.modeling._base] Start quantizing layer 26/32
334
+ 2023-07-21 16:58:58 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 26/32...
335
+ 2023-07-21 16:58:59 INFO [auto_gptq.quantization.gptq] duration: 0.7961215972900391
336
+ 2023-07-21 16:58:59 INFO [auto_gptq.quantization.gptq] avg loss: 8.335006713867188
337
+ 2023-07-21 16:58:59 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 26/32...
338
+ 2023-07-21 16:59:00 INFO [auto_gptq.quantization.gptq] duration: 0.7967922687530518
339
+ 2023-07-21 16:59:00 INFO [auto_gptq.quantization.gptq] avg loss: 0.5929185152053833
340
+ 2023-07-21 16:59:00 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 26/32...
341
+ 2023-07-21 16:59:01 INFO [auto_gptq.quantization.gptq] duration: 0.8355779647827148
342
+ 2023-07-21 16:59:01 INFO [auto_gptq.quantization.gptq] avg loss: 39.31059265136719
343
+ 2023-07-21 16:59:01 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 26/32...
344
+ 2023-07-21 16:59:05 INFO [auto_gptq.quantization.gptq] duration: 3.859668731689453
345
+ 2023-07-21 16:59:05 INFO [auto_gptq.quantization.gptq] avg loss: 5.2629475593566895
346
+ 2023-07-21 16:59:05 INFO [auto_gptq.modeling._base] Start quantizing layer 27/32
347
+ 2023-07-21 16:59:05 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 27/32...
348
+ 2023-07-21 16:59:06 INFO [auto_gptq.quantization.gptq] duration: 0.7974636554718018
349
+ 2023-07-21 16:59:06 INFO [auto_gptq.quantization.gptq] avg loss: 8.194433212280273
350
+ 2023-07-21 16:59:06 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 27/32...
351
+ 2023-07-21 16:59:07 INFO [auto_gptq.quantization.gptq] duration: 0.8030986785888672
352
+ 2023-07-21 16:59:07 INFO [auto_gptq.quantization.gptq] avg loss: 0.7090796828269958
353
+ 2023-07-21 16:59:07 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 27/32...
354
+ 2023-07-21 16:59:07 INFO [auto_gptq.quantization.gptq] duration: 0.8322622776031494
355
+ 2023-07-21 16:59:07 INFO [auto_gptq.quantization.gptq] avg loss: 39.4634895324707
356
+ 2023-07-21 16:59:07 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 27/32...
357
+ 2023-07-21 16:59:11 INFO [auto_gptq.quantization.gptq] duration: 3.878126859664917
358
+ 2023-07-21 16:59:11 INFO [auto_gptq.quantization.gptq] avg loss: 6.581557750701904
359
+ 2023-07-21 16:59:11 INFO [auto_gptq.modeling._base] Start quantizing layer 28/32
360
+ 2023-07-21 16:59:12 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 28/32...
361
+ 2023-07-21 16:59:12 INFO [auto_gptq.quantization.gptq] duration: 0.7974464893341064
362
+ 2023-07-21 16:59:12 INFO [auto_gptq.quantization.gptq] avg loss: 9.201988220214844
363
+ 2023-07-21 16:59:12 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 28/32...
364
+ 2023-07-21 16:59:13 INFO [auto_gptq.quantization.gptq] duration: 0.8018836975097656
365
+ 2023-07-21 16:59:13 INFO [auto_gptq.quantization.gptq] avg loss: 1.193915605545044
366
+ 2023-07-21 16:59:13 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 28/32...
367
+ 2023-07-21 16:59:14 INFO [auto_gptq.quantization.gptq] duration: 0.832056999206543
368
+ 2023-07-21 16:59:14 INFO [auto_gptq.quantization.gptq] avg loss: 39.874481201171875
369
+ 2023-07-21 16:59:14 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 28/32...
370
+ 2023-07-21 16:59:18 INFO [auto_gptq.quantization.gptq] duration: 3.8739585876464844
371
+ 2023-07-21 16:59:18 INFO [auto_gptq.quantization.gptq] avg loss: 7.8150634765625
372
+ 2023-07-21 16:59:18 INFO [auto_gptq.modeling._base] Start quantizing layer 29/32
373
+ 2023-07-21 16:59:18 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 29/32...
374
+ 2023-07-21 16:59:19 INFO [auto_gptq.quantization.gptq] duration: 0.7971282005310059
375
+ 2023-07-21 16:59:19 INFO [auto_gptq.quantization.gptq] avg loss: 8.788995742797852
376
+ 2023-07-21 16:59:19 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 29/32...
377
+ 2023-07-21 16:59:20 INFO [auto_gptq.quantization.gptq] duration: 0.8014233112335205
378
+ 2023-07-21 16:59:20 INFO [auto_gptq.quantization.gptq] avg loss: 0.9004578590393066
379
+ 2023-07-21 16:59:20 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 29/32...
380
+ 2023-07-21 16:59:21 INFO [auto_gptq.quantization.gptq] duration: 0.8585555553436279
381
+ 2023-07-21 16:59:21 INFO [auto_gptq.quantization.gptq] avg loss: 40.52891159057617
382
+ 2023-07-21 16:59:21 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 29/32...
383
+ 2023-07-21 16:59:24 INFO [auto_gptq.quantization.gptq] duration: 3.886247396469116
384
+ 2023-07-21 16:59:24 INFO [auto_gptq.quantization.gptq] avg loss: 7.627683639526367
385
+ 2023-07-21 16:59:25 INFO [auto_gptq.modeling._base] Start quantizing layer 30/32
386
+ 2023-07-21 16:59:25 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 30/32...
387
+ 2023-07-21 16:59:26 INFO [auto_gptq.quantization.gptq] duration: 0.8017170429229736
388
+ 2023-07-21 16:59:26 INFO [auto_gptq.quantization.gptq] avg loss: 7.885834217071533
389
+ 2023-07-21 16:59:26 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 30/32...
390
+ 2023-07-21 16:59:26 INFO [auto_gptq.quantization.gptq] duration: 0.8006551265716553
391
+ 2023-07-21 16:59:26 INFO [auto_gptq.quantization.gptq] avg loss: 1.0838208198547363
392
+ 2023-07-21 16:59:26 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 30/32...
393
+ 2023-07-21 16:59:27 INFO [auto_gptq.quantization.gptq] duration: 0.8757197856903076
394
+ 2023-07-21 16:59:27 INFO [auto_gptq.quantization.gptq] avg loss: 38.54998779296875
395
+ 2023-07-21 16:59:27 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 30/32...
396
+ 2023-07-21 16:59:31 INFO [auto_gptq.quantization.gptq] duration: 3.8700709342956543
397
+ 2023-07-21 16:59:31 INFO [auto_gptq.quantization.gptq] avg loss: 10.26675796508789
398
+ 2023-07-21 16:59:31 INFO [auto_gptq.modeling._base] Start quantizing layer 31/32
399
+ 2023-07-21 16:59:31 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 31/32...
400
+ 2023-07-21 16:59:32 INFO [auto_gptq.quantization.gptq] duration: 0.7995920181274414
401
+ 2023-07-21 16:59:32 INFO [auto_gptq.quantization.gptq] avg loss: 7.922703266143799
402
+ 2023-07-21 16:59:32 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 31/32...
403
+ 2023-07-21 16:59:33 INFO [auto_gptq.quantization.gptq] duration: 0.7997887134552002
404
+ 2023-07-21 16:59:33 INFO [auto_gptq.quantization.gptq] avg loss: 0.6395642757415771
405
+ 2023-07-21 16:59:33 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 31/32...
406
+ 2023-07-21 16:59:34 INFO [auto_gptq.quantization.gptq] duration: 0.8389708995819092
407
+ 2023-07-21 16:59:34 INFO [auto_gptq.quantization.gptq] avg loss: 38.0499153137207
408
+ 2023-07-21 16:59:34 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 31/32...
409
+ 2023-07-21 16:59:38 INFO [auto_gptq.quantization.gptq] duration: 3.8527672290802
410
+ 2023-07-21 16:59:38 INFO [auto_gptq.quantization.gptq] avg loss: 14.685250282287598
411
+ 2023-07-21 16:59:38 INFO [auto_gptq.modeling._base] Start quantizing layer 32/32
412
+ 2023-07-21 16:59:38 INFO [auto_gptq.modeling._base] Quantizing self_attention.query_key_value in layer 32/32...
413
+ 2023-07-21 16:59:39 INFO [auto_gptq.quantization.gptq] duration: 0.7899763584136963
414
+ 2023-07-21 16:59:39 INFO [auto_gptq.quantization.gptq] avg loss: 6.566901206970215
415
+ 2023-07-21 17:00:08 INFO [auto_gptq.modeling._base] Quantizing self_attention.dense in layer 32/32...
416
+ 2023-07-21 17:00:09 INFO [auto_gptq.quantization.gptq] duration: 0.890770673751831
417
+ 2023-07-21 17:00:09 INFO [auto_gptq.quantization.gptq] avg loss: 0.2703491747379303
418
+ 2023-07-21 17:00:09 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_h_to_4h in layer 32/32...
419
+ 2023-07-21 17:00:10 INFO [auto_gptq.quantization.gptq] duration: 0.8699018955230713
420
+ 2023-07-21 17:00:10 INFO [auto_gptq.quantization.gptq] avg loss: 33.582237243652344
421
+ 2023-07-21 17:00:10 INFO [auto_gptq.modeling._base] Quantizing mlp.dense_4h_to_h in layer 32/32...
422
+ 2023-07-21 17:00:14 INFO [auto_gptq.quantization.gptq] duration: 3.8666820526123047
423
+ 2023-07-21 17:00:14 INFO [auto_gptq.quantization.gptq] avg loss: 26.30276107788086
424
+ 2023-07-21 17:00:14 INFO [auto_gptq.modeling._utils] Packing model...
425
+ 2023-07-21 17:00:14 INFO [auto_gptq.modeling._utils] transformer.h.0.self_attention.dense
426
+ 2023-07-21 17:00:15 INFO [auto_gptq.modeling._utils] transformer.h.0.self_attention.query_key_value
427
+ 2023-07-21 17:00:15 INFO [auto_gptq.modeling._utils] transformer.h.0.mlp.dense_4h_to_h
428
+ 2023-07-21 17:00:18 INFO [auto_gptq.modeling._utils] transformer.h.0.mlp.dense_h_to_4h
429
+ 2023-07-21 17:00:19 INFO [auto_gptq.modeling._utils] transformer.h.1.self_attention.dense
430
+ 2023-07-21 17:00:19 INFO [auto_gptq.modeling._utils] transformer.h.1.self_attention.query_key_value
431
+ 2023-07-21 17:00:20 INFO [auto_gptq.modeling._utils] transformer.h.1.mlp.dense_4h_to_h
432
+ 2023-07-21 17:00:22 INFO [auto_gptq.modeling._utils] transformer.h.1.mlp.dense_h_to_4h
433
+ 2023-07-21 17:00:23 INFO [auto_gptq.modeling._utils] transformer.h.2.self_attention.dense
434
+ 2023-07-21 17:00:23 INFO [auto_gptq.modeling._utils] transformer.h.2.self_attention.query_key_value
435
+ 2023-07-21 17:00:24 INFO [auto_gptq.modeling._utils] transformer.h.2.mlp.dense_4h_to_h
436
+ 2023-07-21 17:00:26 INFO [auto_gptq.modeling._utils] transformer.h.2.mlp.dense_h_to_4h
437
+ 2023-07-21 17:00:27 INFO [auto_gptq.modeling._utils] transformer.h.3.self_attention.dense
438
+ 2023-07-21 17:00:28 INFO [auto_gptq.modeling._utils] transformer.h.3.self_attention.query_key_value
439
+ 2023-07-21 17:00:28 INFO [auto_gptq.modeling._utils] transformer.h.3.mlp.dense_4h_to_h
440
+ 2023-07-21 17:00:30 INFO [auto_gptq.modeling._utils] transformer.h.3.mlp.dense_h_to_4h
441
+ 2023-07-21 17:00:31 INFO [auto_gptq.modeling._utils] transformer.h.4.self_attention.dense
442
+ 2023-07-21 17:00:32 INFO [auto_gptq.modeling._utils] transformer.h.4.self_attention.query_key_value
443
+ 2023-07-21 17:00:32 INFO [auto_gptq.modeling._utils] transformer.h.4.mlp.dense_4h_to_h
444
+ 2023-07-21 17:00:34 INFO [auto_gptq.modeling._utils] transformer.h.4.mlp.dense_h_to_4h
445
+ 2023-07-21 17:00:35 INFO [auto_gptq.modeling._utils] transformer.h.5.self_attention.dense
446
+ 2023-07-21 17:00:35 INFO [auto_gptq.modeling._utils] transformer.h.5.self_attention.query_key_value
447
+ 2023-07-21 17:00:36 INFO [auto_gptq.modeling._utils] transformer.h.5.mlp.dense_4h_to_h
448
+ 2023-07-21 17:00:38 INFO [auto_gptq.modeling._utils] transformer.h.5.mlp.dense_h_to_4h
449
+ 2023-07-21 17:00:39 INFO [auto_gptq.modeling._utils] transformer.h.6.self_attention.dense
450
+ 2023-07-21 17:00:39 INFO [auto_gptq.modeling._utils] transformer.h.6.self_attention.query_key_value
451
+ 2023-07-21 17:00:40 INFO [auto_gptq.modeling._utils] transformer.h.6.mlp.dense_4h_to_h
452
+ 2023-07-21 17:00:41 INFO [auto_gptq.modeling._utils] transformer.h.6.mlp.dense_h_to_4h
453
+ 2023-07-21 17:00:42 INFO [auto_gptq.modeling._utils] transformer.h.7.self_attention.dense
454
+ 2023-07-21 17:00:43 INFO [auto_gptq.modeling._utils] transformer.h.7.self_attention.query_key_value
455
+ 2023-07-21 17:00:43 INFO [auto_gptq.modeling._utils] transformer.h.7.mlp.dense_4h_to_h
456
+ 2023-07-21 17:00:45 INFO [auto_gptq.modeling._utils] transformer.h.7.mlp.dense_h_to_4h
457
+ 2023-07-21 17:00:46 INFO [auto_gptq.modeling._utils] transformer.h.8.self_attention.dense
458
+ 2023-07-21 17:00:47 INFO [auto_gptq.modeling._utils] transformer.h.8.self_attention.query_key_value
459
+ 2023-07-21 17:00:47 INFO [auto_gptq.modeling._utils] transformer.h.8.mlp.dense_4h_to_h
460
+ 2023-07-21 17:00:49 INFO [auto_gptq.modeling._utils] transformer.h.8.mlp.dense_h_to_4h
461
+ 2023-07-21 17:00:50 INFO [auto_gptq.modeling._utils] transformer.h.9.self_attention.dense
462
+ 2023-07-21 17:00:50 INFO [auto_gptq.modeling._utils] transformer.h.9.self_attention.query_key_value
463
+ 2023-07-21 17:00:51 INFO [auto_gptq.modeling._utils] transformer.h.9.mlp.dense_4h_to_h
464
+ 2023-07-21 17:00:53 INFO [auto_gptq.modeling._utils] transformer.h.9.mlp.dense_h_to_4h
465
+ 2023-07-21 17:00:54 INFO [auto_gptq.modeling._utils] transformer.h.10.self_attention.dense
466
+ 2023-07-21 17:00:54 INFO [auto_gptq.modeling._utils] transformer.h.10.self_attention.query_key_value
467
+ 2023-07-21 17:00:55 INFO [auto_gptq.modeling._utils] transformer.h.10.mlp.dense_4h_to_h
468
+ 2023-07-21 17:00:56 INFO [auto_gptq.modeling._utils] transformer.h.10.mlp.dense_h_to_4h
469
+ 2023-07-21 17:00:57 INFO [auto_gptq.modeling._utils] transformer.h.11.self_attention.dense
470
+ 2023-07-21 17:00:58 INFO [auto_gptq.modeling._utils] transformer.h.11.self_attention.query_key_value
471
+ 2023-07-21 17:00:58 INFO [auto_gptq.modeling._utils] transformer.h.11.mlp.dense_4h_to_h
472
+ 2023-07-21 17:01:00 INFO [auto_gptq.modeling._utils] transformer.h.11.mlp.dense_h_to_4h
473
+ 2023-07-21 17:01:01 INFO [auto_gptq.modeling._utils] transformer.h.12.self_attention.dense
474
+ 2023-07-21 17:01:02 INFO [auto_gptq.modeling._utils] transformer.h.12.self_attention.query_key_value
475
+ 2023-07-21 17:01:02 INFO [auto_gptq.modeling._utils] transformer.h.12.mlp.dense_4h_to_h
476
+ 2023-07-21 17:01:04 INFO [auto_gptq.modeling._utils] transformer.h.12.mlp.dense_h_to_4h
477
+ 2023-07-21 17:01:05 INFO [auto_gptq.modeling._utils] transformer.h.13.self_attention.dense
478
+ 2023-07-21 17:01:06 INFO [auto_gptq.modeling._utils] transformer.h.13.self_attention.query_key_value
479
+ 2023-07-21 17:01:06 INFO [auto_gptq.modeling._utils] transformer.h.13.mlp.dense_4h_to_h
480
+ 2023-07-21 17:01:08 INFO [auto_gptq.modeling._utils] transformer.h.13.mlp.dense_h_to_4h
481
+ 2023-07-21 17:01:09 INFO [auto_gptq.modeling._utils] transformer.h.14.self_attention.dense
482
+ 2023-07-21 17:01:10 INFO [auto_gptq.modeling._utils] transformer.h.14.self_attention.query_key_value
483
+ 2023-07-21 17:01:10 INFO [auto_gptq.modeling._utils] transformer.h.14.mlp.dense_4h_to_h
484
+ 2023-07-21 17:01:12 INFO [auto_gptq.modeling._utils] transformer.h.14.mlp.dense_h_to_4h
485
+ 2023-07-21 17:01:13 INFO [auto_gptq.modeling._utils] transformer.h.15.self_attention.dense
486
+ 2023-07-21 17:01:13 INFO [auto_gptq.modeling._utils] transformer.h.15.self_attention.query_key_value
487
+ 2023-07-21 17:01:14 INFO [auto_gptq.modeling._utils] transformer.h.15.mlp.dense_4h_to_h
488
+ 2023-07-21 17:01:16 INFO [auto_gptq.modeling._utils] transformer.h.15.mlp.dense_h_to_4h
489
+ 2023-07-21 17:01:17 INFO [auto_gptq.modeling._utils] transformer.h.16.self_attention.dense
490
+ 2023-07-21 17:01:17 INFO [auto_gptq.modeling._utils] transformer.h.16.self_attention.query_key_value
491
+ 2023-07-21 17:01:18 INFO [auto_gptq.modeling._utils] transformer.h.16.mlp.dense_4h_to_h
492
+ 2023-07-21 17:01:19 INFO [auto_gptq.modeling._utils] transformer.h.16.mlp.dense_h_to_4h
493
+ 2023-07-21 17:01:21 INFO [auto_gptq.modeling._utils] transformer.h.17.self_attention.dense
494
+ 2023-07-21 17:01:21 INFO [auto_gptq.modeling._utils] transformer.h.17.self_attention.query_key_value
495
+ 2023-07-21 17:01:21 INFO [auto_gptq.modeling._utils] transformer.h.17.mlp.dense_4h_to_h
496
+ 2023-07-21 17:01:23 INFO [auto_gptq.modeling._utils] transformer.h.17.mlp.dense_h_to_4h
497
+ 2023-07-21 17:01:24 INFO [auto_gptq.modeling._utils] transformer.h.18.self_attention.dense
498
+ 2023-07-21 17:01:25 INFO [auto_gptq.modeling._utils] transformer.h.18.self_attention.query_key_value
499
+ 2023-07-21 17:01:25 INFO [auto_gptq.modeling._utils] transformer.h.18.mlp.dense_4h_to_h
500
+ 2023-07-21 17:01:27 INFO [auto_gptq.modeling._utils] transformer.h.18.mlp.dense_h_to_4h
501
+ 2023-07-21 17:01:28 INFO [auto_gptq.modeling._utils] transformer.h.19.self_attention.dense
502
+ 2023-07-21 17:01:29 INFO [auto_gptq.modeling._utils] transformer.h.19.self_attention.query_key_value
503
+ 2023-07-21 17:01:29 INFO [auto_gptq.modeling._utils] transformer.h.19.mlp.dense_4h_to_h
504
+ 2023-07-21 17:01:31 INFO [auto_gptq.modeling._utils] transformer.h.19.mlp.dense_h_to_4h
505
+ 2023-07-21 17:01:32 INFO [auto_gptq.modeling._utils] transformer.h.20.self_attention.dense
506
+ 2023-07-21 17:01:33 INFO [auto_gptq.modeling._utils] transformer.h.20.self_attention.query_key_value
507
+ 2023-07-21 17:01:33 INFO [auto_gptq.modeling._utils] transformer.h.20.mlp.dense_4h_to_h
508
+ 2023-07-21 17:01:35 INFO [auto_gptq.modeling._utils] transformer.h.20.mlp.dense_h_to_4h
509
+ 2023-07-21 17:01:36 INFO [auto_gptq.modeling._utils] transformer.h.21.self_attention.dense
510
+ 2023-07-21 17:01:37 INFO [auto_gptq.modeling._utils] transformer.h.21.self_attention.query_key_value
511
+ 2023-07-21 17:01:37 INFO [auto_gptq.modeling._utils] transformer.h.21.mlp.dense_4h_to_h
512
+ 2023-07-21 17:01:39 INFO [auto_gptq.modeling._utils] transformer.h.21.mlp.dense_h_to_4h
513
+ 2023-07-21 17:01:40 INFO [auto_gptq.modeling._utils] transformer.h.22.self_attention.dense
514
+ 2023-07-21 17:01:40 INFO [auto_gptq.modeling._utils] transformer.h.22.self_attention.query_key_value
515
+ 2023-07-21 17:01:41 INFO [auto_gptq.modeling._utils] transformer.h.22.mlp.dense_4h_to_h
516
+ 2023-07-21 17:01:43 INFO [auto_gptq.modeling._utils] transformer.h.22.mlp.dense_h_to_4h
517
+ 2023-07-21 17:01:44 INFO [auto_gptq.modeling._utils] transformer.h.23.self_attention.dense
518
+ 2023-07-21 17:01:44 INFO [auto_gptq.modeling._utils] transformer.h.23.self_attention.query_key_value
519
+ 2023-07-21 17:01:45 INFO [auto_gptq.modeling._utils] transformer.h.23.mlp.dense_4h_to_h
520
+ 2023-07-21 17:01:46 INFO [auto_gptq.modeling._utils] transformer.h.23.mlp.dense_h_to_4h
521
+ 2023-07-21 17:01:48 INFO [auto_gptq.modeling._utils] transformer.h.24.self_attention.dense
522
+ 2023-07-21 17:01:48 INFO [auto_gptq.modeling._utils] transformer.h.24.self_attention.query_key_value
523
+ 2023-07-21 17:01:49 INFO [auto_gptq.modeling._utils] transformer.h.24.mlp.dense_4h_to_h
524
+ 2023-07-21 17:01:51 INFO [auto_gptq.modeling._utils] transformer.h.24.mlp.dense_h_to_4h
525
+ 2023-07-21 17:01:52 INFO [auto_gptq.modeling._utils] transformer.h.25.self_attention.dense
526
+ 2023-07-21 17:01:52 INFO [auto_gptq.modeling._utils] transformer.h.25.self_attention.query_key_value
527
+ 2023-07-21 17:01:53 INFO [auto_gptq.modeling._utils] transformer.h.25.mlp.dense_4h_to_h
528
+ 2023-07-21 17:01:54 INFO [auto_gptq.modeling._utils] transformer.h.25.mlp.dense_h_to_4h
529
+ 2023-07-21 17:01:55 INFO [auto_gptq.modeling._utils] transformer.h.26.self_attention.dense
530
+ 2023-07-21 17:01:56 INFO [auto_gptq.modeling._utils] transformer.h.26.self_attention.query_key_value
531
+ 2023-07-21 17:01:56 INFO [auto_gptq.modeling._utils] transformer.h.26.mlp.dense_4h_to_h
532
+ 2023-07-21 17:01:58 INFO [auto_gptq.modeling._utils] transformer.h.26.mlp.dense_h_to_4h
533
+ 2023-07-21 17:02:00 INFO [auto_gptq.modeling._utils] transformer.h.27.self_attention.dense
534
+ 2023-07-21 17:02:00 INFO [auto_gptq.modeling._utils] transformer.h.27.self_attention.query_key_value
535
+ 2023-07-21 17:02:00 INFO [auto_gptq.modeling._utils] transformer.h.27.mlp.dense_4h_to_h
536
+ 2023-07-21 17:02:02 INFO [auto_gptq.modeling._utils] transformer.h.27.mlp.dense_h_to_4h
537
+ 2023-07-21 17:02:03 INFO [auto_gptq.modeling._utils] transformer.h.28.self_attention.dense
538
+ 2023-07-21 17:02:04 INFO [auto_gptq.modeling._utils] transformer.h.28.self_attention.query_key_value
539
+ 2023-07-21 17:02:04 INFO [auto_gptq.modeling._utils] transformer.h.28.mlp.dense_4h_to_h
540
+ 2023-07-21 17:02:06 INFO [auto_gptq.modeling._utils] transformer.h.28.mlp.dense_h_to_4h
541
+ 2023-07-21 17:02:07 INFO [auto_gptq.modeling._utils] transformer.h.29.self_attention.dense
542
+ 2023-07-21 17:02:08 INFO [auto_gptq.modeling._utils] transformer.h.29.self_attention.query_key_value
543
+ 2023-07-21 17:02:08 INFO [auto_gptq.modeling._utils] transformer.h.29.mlp.dense_4h_to_h
544
+ 2023-07-21 17:02:10 INFO [auto_gptq.modeling._utils] transformer.h.29.mlp.dense_h_to_4h
545
+ 2023-07-21 17:02:11 INFO [auto_gptq.modeling._utils] transformer.h.30.self_attention.dense
546
+ 2023-07-21 17:02:12 INFO [auto_gptq.modeling._utils] transformer.h.30.self_attention.query_key_value
547
+ 2023-07-21 17:02:12 INFO [auto_gptq.modeling._utils] transformer.h.30.mlp.dense_4h_to_h
548
+ 2023-07-21 17:02:14 INFO [auto_gptq.modeling._utils] transformer.h.30.mlp.dense_h_to_4h
549
+ 2023-07-21 17:02:15 INFO [auto_gptq.modeling._utils] transformer.h.31.self_attention.dense
550
+ 2023-07-21 17:02:16 INFO [auto_gptq.modeling._utils] transformer.h.31.self_attention.query_key_value
551
+ 2023-07-21 17:02:16 INFO [auto_gptq.modeling._utils] transformer.h.31.mlp.dense_4h_to_h
552
+ 2023-07-21 17:02:18 INFO [auto_gptq.modeling._utils] transformer.h.31.mlp.dense_h_to_4h
553
+ 2023-07-21 17:02:19 INFO [auto_gptq.modeling._utils] Model packed.
554
+ ```