-
Notifications
You must be signed in to change notification settings - Fork 3
/
wikitext_2_april_s42_it1000_lr005_s2_20240313-001440.log
224 lines (224 loc) · 21.9 KB
/
wikitext_2_april_s42_it1000_lr005_s2_20240313-001440.log
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
[INFO][(2024-03-13) 00:14:40.950] [main.py:25]: Git branch: main, Git commit: 29e3433
[INFO][(2024-03-13) 00:14:40.951] [main.py:26]: Configuration:
{
"task": "text-generation",
"device": "cuda:0",
"global_seed": 42,
"datasource": {
"dataset": "wikitext-2",
"partition": "train",
"shuffle_seed": 10,
"num_points": 1,
"max_num_tokens": 20,
"full_name": [
"wikitext",
"wikitext-2-raw-v1"
],
"main_column": "text",
"unused_columns_to_remove": [
"text"
]
},
"tokenizer": {
"name": "gpt2"
},
"model": {
"name": "gpt2",
"from_pretrained": true,
"has_finetuned": false
},
"train": {
"batch_size": 1,
"use_cached": false
},
"attack": {
"name": "april",
"specific_args": {
"scale": 2.0,
"init_type": "randn-trunc",
"init_seed": 42,
"max_iterations": 1000,
"snapshot_interval": 100,
"print_interval": 10,
"d_model": 768,
"grad_clip": 1.0,
"optimizer": {
"name": "bert-adam",
"args": {
"lr": 0.05,
"betas": [
0.9,
0.999
],
"eps": 1e-06,
"weight_decay": 0.01
}
},
"scheduler": {
"name": "linear",
"args": {
"warmup": 50
}
}
}
},
"eval": {
"metrics": [
"accuracy",
"bleu",
"rouge"
]
},
"debug": {
"fix_input": true
},
"log_file": "1_gpt2_gen/wikitext_2_april_s42_it1000_lr005_s2_20240313-001440.log"
}
[INFO][(2024-03-13) 00:14:46.705] [utils.py:123]: Ground truth text: [' The Tower Building of the Little Rock Arsenal, also known as U.S.']
[INFO][(2024-03-13) 00:14:46.705] [utils.py:152]: Ground truth text length in tokens: 16
[INFO][(2024-03-13) 00:14:51.763] [attack.py:246]: | It: 1 | Atk. loss: 13860.91309 | Task loss: 20.70686 | T: 0.09s |
[INFO][(2024-03-13) 00:14:51.763] [attack.py:257]: Snapshots captured for It 1.
[INFO][(2024-03-13) 00:14:51.765] [eval.py:9]: Reconstructed text: [' 25 Quessel decisive Chill proposedagogue Ultraush Casda phone Mun Tier regretfet']
[INFO][(2024-03-13) 00:14:58.734] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:14:58.838] [eval.py:31]: Metric report: {"accuracy": 0.0, "bleu": {"bleu": 0.0, "precisions": [0.0, 0.0, 0.0, 0.0], "brevity_penalty": 0.6347364189402819, "length_ratio": 0.6875, "translation_length": 11, "reference_length": 16}, "rouge": {"rouge1": 0.0, "rouge2": 0.0, "rougeL": 0.0, "rougeLsum": 0.0}}
[INFO][(2024-03-13) 00:14:59.682] [attack.py:246]: | It: 11 | Atk. loss: 3167.31689 | Task loss: 19.07699 | T: 8.00s |
[INFO][(2024-03-13) 00:15:00.538] [attack.py:246]: | It: 21 | Atk. loss: 1488.63184 | Task loss: 16.69408 | T: 8.86s |
[INFO][(2024-03-13) 00:15:01.385] [attack.py:246]: | It: 31 | Atk. loss: 1271.15454 | Task loss: 14.99008 | T: 9.71s |
[INFO][(2024-03-13) 00:15:02.246] [attack.py:246]: | It: 41 | Atk. loss: 1138.26233 | Task loss: 13.77543 | T: 10.57s |
[INFO][(2024-03-13) 00:15:03.106] [attack.py:246]: | It: 51 | Atk. loss: 1079.43506 | Task loss: 13.26216 | T: 11.43s |
[INFO][(2024-03-13) 00:15:03.949] [attack.py:246]: | It: 61 | Atk. loss: 1054.14807 | Task loss: 13.08933 | T: 12.27s |
[INFO][(2024-03-13) 00:15:04.794] [attack.py:246]: | It: 71 | Atk. loss: 1022.06836 | Task loss: 12.31850 | T: 13.12s |
[INFO][(2024-03-13) 00:15:05.647] [attack.py:246]: | It: 81 | Atk. loss: 985.49792 | Task loss: 12.98967 | T: 13.97s |
[INFO][(2024-03-13) 00:15:06.501] [attack.py:246]: | It: 91 | Atk. loss: 910.21008 | Task loss: 12.33251 | T: 14.82s |
[INFO][(2024-03-13) 00:15:07.358] [attack.py:246]: | It: 101 | Atk. loss: 887.52551 | Task loss: 12.30311 | T: 15.68s |
[INFO][(2024-03-13) 00:15:07.359] [attack.py:257]: Snapshots captured for It 101.
[INFO][(2024-03-13) 00:15:07.361] [eval.py:9]: Reconstructed text: [' Theaelthat evictionacteria categor Cash Mass recruit Gina limitations upbeat become hue semanticfet']
[INFO][(2024-03-13) 00:15:13.250] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:15:13.352] [eval.py:31]: Metric report: {"accuracy": 0.0625, "bleu": {"bleu": 0.0, "precisions": [0.0, 0.0, 0.0, 0.0], "brevity_penalty": 0.7165313105737893, "length_ratio": 0.75, "translation_length": 12, "reference_length": 16}, "rouge": {"rouge1": 0.0, "rouge2": 0.0, "rougeL": 0.0, "rougeLsum": 0.0}}
[INFO][(2024-03-13) 00:15:14.143] [attack.py:246]: | It: 111 | Atk. loss: 820.07416 | Task loss: 12.13441 | T: 22.47s |
[INFO][(2024-03-13) 00:15:14.916] [attack.py:246]: | It: 121 | Atk. loss: 677.75415 | Task loss: 12.19991 | T: 23.24s |
[INFO][(2024-03-13) 00:15:15.692] [attack.py:246]: | It: 131 | Atk. loss: 460.68930 | Task loss: 12.23270 | T: 24.01s |
[INFO][(2024-03-13) 00:15:16.473] [attack.py:246]: | It: 141 | Atk. loss: 307.50443 | Task loss: 11.66310 | T: 24.79s |
[INFO][(2024-03-13) 00:15:17.254] [attack.py:246]: | It: 151 | Atk. loss: 292.93564 | Task loss: 11.60193 | T: 25.58s |
[INFO][(2024-03-13) 00:15:18.077] [attack.py:246]: | It: 161 | Atk. loss: 273.98123 | Task loss: 11.19479 | T: 26.40s |
[INFO][(2024-03-13) 00:15:18.929] [attack.py:246]: | It: 171 | Atk. loss: 254.10922 | Task loss: 10.92083 | T: 27.25s |
[INFO][(2024-03-13) 00:15:19.789] [attack.py:246]: | It: 181 | Atk. loss: 246.34825 | Task loss: 11.11664 | T: 28.11s |
[INFO][(2024-03-13) 00:15:20.648] [attack.py:246]: | It: 191 | Atk. loss: 224.75581 | Task loss: 10.74073 | T: 28.97s |
[INFO][(2024-03-13) 00:15:21.508] [attack.py:246]: | It: 201 | Atk. loss: 238.73843 | Task loss: 11.18038 | T: 29.83s |
[INFO][(2024-03-13) 00:15:21.508] [attack.py:257]: Snapshots captured for It 201.
[INFO][(2024-03-13) 00:15:21.510] [eval.py:9]: Reconstructed text: [' The lust building Sheriff revealing categor Chicken building recruitAnthony limitations skeptical guidelines procession semantic shift']
[INFO][(2024-03-13) 00:15:28.098] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:15:28.200] [eval.py:31]: Metric report: {"accuracy": 0.0625, "bleu": {"bleu": 0.0, "precisions": [0.06666666666666667, 0.0, 0.0, 0.0], "brevity_penalty": 0.9355069850316178, "length_ratio": 0.9375, "translation_length": 15, "reference_length": 16}, "rouge": {"rouge1": 0.14285714285714288, "rouge2": 0.0, "rougeL": 0.14285714285714288, "rougeLsum": 0.14285714285714288}}
[INFO][(2024-03-13) 00:15:29.048] [attack.py:246]: | It: 211 | Atk. loss: 225.17075 | Task loss: 10.64281 | T: 37.37s |
[INFO][(2024-03-13) 00:15:29.904] [attack.py:246]: | It: 221 | Atk. loss: 218.62659 | Task loss: 10.79064 | T: 38.23s |
[INFO][(2024-03-13) 00:15:30.742] [attack.py:246]: | It: 231 | Atk. loss: 275.68948 | Task loss: 9.93419 | T: 39.06s |
[INFO][(2024-03-13) 00:15:31.592] [attack.py:246]: | It: 241 | Atk. loss: 219.26222 | Task loss: 10.88620 | T: 39.91s |
[INFO][(2024-03-13) 00:15:32.445] [attack.py:246]: | It: 251 | Atk. loss: 190.63670 | Task loss: 10.19662 | T: 40.77s |
[INFO][(2024-03-13) 00:15:33.301] [attack.py:246]: | It: 261 | Atk. loss: 193.90359 | Task loss: 10.49473 | T: 41.62s |
[INFO][(2024-03-13) 00:15:34.153] [attack.py:246]: | It: 271 | Atk. loss: 179.27466 | Task loss: 9.99387 | T: 42.48s |
[INFO][(2024-03-13) 00:15:34.999] [attack.py:246]: | It: 281 | Atk. loss: 174.06937 | Task loss: 9.93274 | T: 43.32s |
[INFO][(2024-03-13) 00:15:35.822] [attack.py:246]: | It: 291 | Atk. loss: 170.91264 | Task loss: 10.16441 | T: 44.14s |
[INFO][(2024-03-13) 00:15:36.582] [attack.py:246]: | It: 301 | Atk. loss: 191.38870 | Task loss: 10.44592 | T: 44.90s |
[INFO][(2024-03-13) 00:15:36.583] [attack.py:257]: Snapshots captured for It 301.
[INFO][(2024-03-13) 00:15:36.585] [eval.py:9]: Reconstructed text: [' The Tower that Tower preparing backdrop Building Tower Peak interference mystic skeptical tune HY surgicalfet']
[INFO][(2024-03-13) 00:15:42.639] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:15:42.743] [eval.py:31]: Metric report: {"accuracy": 0.125, "bleu": {"bleu": 0.0, "precisions": [0.2, 0.07142857142857142, 0.0, 0.0], "brevity_penalty": 0.9355069850316178, "length_ratio": 0.9375, "translation_length": 15, "reference_length": 16}, "rouge": {"rouge1": 0.21428571428571427, "rouge2": 0.07692307692307691, "rougeL": 0.21428571428571427, "rougeLsum": 0.21428571428571427}}
[INFO][(2024-03-13) 00:15:43.585] [attack.py:246]: | It: 311 | Atk. loss: 173.20634 | Task loss: 10.03950 | T: 51.91s |
[INFO][(2024-03-13) 00:15:44.408] [attack.py:246]: | It: 321 | Atk. loss: 182.34618 | Task loss: 10.29733 | T: 52.73s |
[INFO][(2024-03-13) 00:15:45.261] [attack.py:246]: | It: 331 | Atk. loss: 170.29065 | Task loss: 9.69166 | T: 53.58s |
[INFO][(2024-03-13) 00:15:46.060] [attack.py:246]: | It: 341 | Atk. loss: 170.42142 | Task loss: 9.76432 | T: 54.38s |
[INFO][(2024-03-13) 00:15:46.838] [attack.py:246]: | It: 351 | Atk. loss: 159.73608 | Task loss: 9.84878 | T: 55.16s |
[INFO][(2024-03-13) 00:15:47.612] [attack.py:246]: | It: 361 | Atk. loss: 157.74411 | Task loss: 9.44861 | T: 55.93s |
[INFO][(2024-03-13) 00:15:48.389] [attack.py:246]: | It: 371 | Atk. loss: 149.41476 | Task loss: 9.60767 | T: 56.71s |
[INFO][(2024-03-13) 00:15:49.172] [attack.py:246]: | It: 381 | Atk. loss: 154.24371 | Task loss: 9.81971 | T: 57.49s |
[INFO][(2024-03-13) 00:15:49.989] [attack.py:246]: | It: 391 | Atk. loss: 148.92999 | Task loss: 9.34332 | T: 58.31s |
[INFO][(2024-03-13) 00:15:50.839] [attack.py:246]: | It: 401 | Atk. loss: 153.17601 | Task loss: 9.34183 | T: 59.16s |
[INFO][(2024-03-13) 00:15:50.840] [attack.py:257]: Snapshots captured for It 401.
[INFO][(2024-03-13) 00:15:50.842] [eval.py:9]: Reconstructed text: [' The Tower Building Tower job cannot Little Tower Peak Rock unleashed worried spokesman HY semicfet']
[INFO][(2024-03-13) 00:15:56.757] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:15:56.861] [eval.py:31]: Metric report: {"accuracy": 0.1875, "bleu": {"bleu": 0.0, "precisions": [0.3333333333333333, 0.14285714285714285, 0.07692307692307693, 0.0], "brevity_penalty": 0.9355069850316178, "length_ratio": 0.9375, "translation_length": 15, "reference_length": 16}, "rouge": {"rouge1": 0.3571428571428571, "rouge2": 0.15384615384615383, "rougeL": 0.3571428571428571, "rougeLsum": 0.3571428571428571}}
[INFO][(2024-03-13) 00:15:57.707] [attack.py:246]: | It: 411 | Atk. loss: 151.78200 | Task loss: 9.30972 | T: 66.03s |
[INFO][(2024-03-13) 00:15:58.552] [attack.py:246]: | It: 421 | Atk. loss: 143.84679 | Task loss: 9.73108 | T: 66.87s |
[INFO][(2024-03-13) 00:15:59.394] [attack.py:246]: | It: 431 | Atk. loss: 138.68341 | Task loss: 9.27218 | T: 67.72s |
[INFO][(2024-03-13) 00:16:00.255] [attack.py:246]: | It: 441 | Atk. loss: 138.35362 | Task loss: 9.33107 | T: 68.58s |
[INFO][(2024-03-13) 00:16:01.110] [attack.py:246]: | It: 451 | Atk. loss: 133.62865 | Task loss: 9.33694 | T: 69.43s |
[INFO][(2024-03-13) 00:16:01.961] [attack.py:246]: | It: 461 | Atk. loss: 136.52747 | Task loss: 9.33971 | T: 70.28s |
[INFO][(2024-03-13) 00:16:02.817] [attack.py:246]: | It: 471 | Atk. loss: 132.86765 | Task loss: 9.26339 | T: 71.14s |
[INFO][(2024-03-13) 00:16:03.672] [attack.py:246]: | It: 481 | Atk. loss: 130.83441 | Task loss: 9.40759 | T: 71.99s |
[INFO][(2024-03-13) 00:16:04.520] [attack.py:246]: | It: 491 | Atk. loss: 130.91093 | Task loss: 9.13527 | T: 72.84s |
[INFO][(2024-03-13) 00:16:05.378] [attack.py:246]: | It: 501 | Atk. loss: 127.85292 | Task loss: 9.27112 | T: 73.70s |
[INFO][(2024-03-13) 00:16:05.378] [attack.py:257]: Snapshots captured for It 501.
[INFO][(2024-03-13) 00:16:05.380] [eval.py:9]: Reconstructed text: [' The Tower Building Tower job redef Little Tower Peak Rock polled worried spokesman HY semicfet']
[INFO][(2024-03-13) 00:16:11.166] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:16:11.270] [eval.py:31]: Metric report: {"accuracy": 0.1875, "bleu": {"bleu": 0.0, "precisions": [0.3333333333333333, 0.14285714285714285, 0.07692307692307693, 0.0], "brevity_penalty": 0.9355069850316178, "length_ratio": 0.9375, "translation_length": 15, "reference_length": 16}, "rouge": {"rouge1": 0.3571428571428571, "rouge2": 0.15384615384615383, "rougeL": 0.3571428571428571, "rougeLsum": 0.3571428571428571}}
[INFO][(2024-03-13) 00:16:12.116] [attack.py:246]: | It: 511 | Atk. loss: 139.74399 | Task loss: 9.01856 | T: 80.44s |
[INFO][(2024-03-13) 00:16:12.974] [attack.py:246]: | It: 521 | Atk. loss: 196.17279 | Task loss: 8.83046 | T: 81.30s |
[INFO][(2024-03-13) 00:16:13.822] [attack.py:246]: | It: 531 | Atk. loss: 127.22018 | Task loss: 9.39889 | T: 82.14s |
[INFO][(2024-03-13) 00:16:14.674] [attack.py:246]: | It: 541 | Atk. loss: 127.72845 | Task loss: 9.40431 | T: 83.00s |
[INFO][(2024-03-13) 00:16:15.526] [attack.py:246]: | It: 551 | Atk. loss: 124.64494 | Task loss: 9.24670 | T: 83.85s |
[INFO][(2024-03-13) 00:16:16.376] [attack.py:246]: | It: 561 | Atk. loss: 118.92070 | Task loss: 9.07502 | T: 84.70s |
[INFO][(2024-03-13) 00:16:17.222] [attack.py:246]: | It: 571 | Atk. loss: 122.84193 | Task loss: 9.29637 | T: 85.54s |
[INFO][(2024-03-13) 00:16:18.000] [attack.py:246]: | It: 581 | Atk. loss: 123.87072 | Task loss: 9.27527 | T: 86.32s |
[INFO][(2024-03-13) 00:16:18.773] [attack.py:246]: | It: 591 | Atk. loss: 117.84701 | Task loss: 9.17473 | T: 87.09s |
[INFO][(2024-03-13) 00:16:19.549] [attack.py:246]: | It: 601 | Atk. loss: 117.34496 | Task loss: 8.87111 | T: 87.87s |
[INFO][(2024-03-13) 00:16:19.549] [attack.py:257]: Snapshots captured for It 601.
[INFO][(2024-03-13) 00:16:19.551] [eval.py:9]: Reconstructed text: [' The Tower Building Tower extremely redef Little Tower Rock antidepressants specializes worried spokesman HY semicfet']
[INFO][(2024-03-13) 00:16:25.341] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:16:25.446] [eval.py:31]: Metric report: {"accuracy": 0.1875, "bleu": {"bleu": 0.0, "precisions": [0.3333333333333333, 0.14285714285714285, 0.07692307692307693, 0.0], "brevity_penalty": 0.9355069850316178, "length_ratio": 0.9375, "translation_length": 15, "reference_length": 16}, "rouge": {"rouge1": 0.3571428571428571, "rouge2": 0.15384615384615383, "rougeL": 0.3571428571428571, "rougeLsum": 0.3571428571428571}}
[INFO][(2024-03-13) 00:16:26.324] [attack.py:246]: | It: 611 | Atk. loss: 116.40631 | Task loss: 8.97359 | T: 94.65s |
[INFO][(2024-03-13) 00:16:27.175] [attack.py:246]: | It: 621 | Atk. loss: 113.45232 | Task loss: 9.15462 | T: 95.50s |
[INFO][(2024-03-13) 00:16:28.028] [attack.py:246]: | It: 631 | Atk. loss: 119.58478 | Task loss: 9.29046 | T: 96.35s |
[INFO][(2024-03-13) 00:16:28.886] [attack.py:246]: | It: 641 | Atk. loss: 112.12913 | Task loss: 8.95409 | T: 97.21s |
[INFO][(2024-03-13) 00:16:29.744] [attack.py:246]: | It: 651 | Atk. loss: 109.60532 | Task loss: 8.91098 | T: 98.07s |
[INFO][(2024-03-13) 00:16:30.598] [attack.py:246]: | It: 661 | Atk. loss: 108.06034 | Task loss: 8.96307 | T: 98.92s |
[INFO][(2024-03-13) 00:16:31.444] [attack.py:246]: | It: 671 | Atk. loss: 114.14307 | Task loss: 8.86121 | T: 99.77s |
[INFO][(2024-03-13) 00:16:32.289] [attack.py:246]: | It: 681 | Atk. loss: 108.68069 | Task loss: 8.77719 | T: 100.61s |
[INFO][(2024-03-13) 00:16:33.137] [attack.py:246]: | It: 691 | Atk. loss: 105.70966 | Task loss: 8.77187 | T: 101.46s |
[INFO][(2024-03-13) 00:16:33.986] [attack.py:246]: | It: 701 | Atk. loss: 104.29579 | Task loss: 8.92379 | T: 102.31s |
[INFO][(2024-03-13) 00:16:33.986] [attack.py:257]: Snapshots captured for It 701.
[INFO][(2024-03-13) 00:16:33.988] [eval.py:9]: Reconstructed text: [' The Tower Building Tower the unsc Little Tower Rock antidepressants specializes worried spokesman HY semicfet']
[INFO][(2024-03-13) 00:16:39.985] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:16:40.087] [eval.py:31]: Metric report: {"accuracy": 0.25, "bleu": {"bleu": 0.0, "precisions": [0.4, 0.14285714285714285, 0.07692307692307693, 0.0], "brevity_penalty": 0.9355069850316178, "length_ratio": 0.9375, "translation_length": 15, "reference_length": 16}, "rouge": {"rouge1": 0.42857142857142855, "rouge2": 0.15384615384615383, "rougeL": 0.42857142857142855, "rougeLsum": 0.42857142857142855}}
[INFO][(2024-03-13) 00:16:40.950] [attack.py:246]: | It: 711 | Atk. loss: 103.46110 | Task loss: 8.66929 | T: 109.27s |
[INFO][(2024-03-13) 00:16:41.805] [attack.py:246]: | It: 721 | Atk. loss: 106.08562 | Task loss: 8.81652 | T: 110.13s |
[INFO][(2024-03-13) 00:16:42.664] [attack.py:246]: | It: 731 | Atk. loss: 102.14502 | Task loss: 8.66476 | T: 110.99s |
[INFO][(2024-03-13) 00:16:43.520] [attack.py:246]: | It: 741 | Atk. loss: 99.08666 | Task loss: 8.78207 | T: 111.84s |
[INFO][(2024-03-13) 00:16:44.376] [attack.py:246]: | It: 751 | Atk. loss: 102.63680 | Task loss: 8.46788 | T: 112.70s |
[INFO][(2024-03-13) 00:16:45.227] [attack.py:246]: | It: 761 | Atk. loss: 97.53892 | Task loss: 8.59803 | T: 113.55s |
[INFO][(2024-03-13) 00:16:46.080] [attack.py:246]: | It: 771 | Atk. loss: 93.79649 | Task loss: 8.57342 | T: 114.40s |
[INFO][(2024-03-13) 00:16:46.932] [attack.py:246]: | It: 781 | Atk. loss: 93.63765 | Task loss: 8.55741 | T: 115.25s |
[INFO][(2024-03-13) 00:16:47.772] [attack.py:246]: | It: 791 | Atk. loss: 92.77435 | Task loss: 8.51555 | T: 116.09s |
[INFO][(2024-03-13) 00:16:48.619] [attack.py:246]: | It: 801 | Atk. loss: 89.15296 | Task loss: 8.58908 | T: 116.94s |
[INFO][(2024-03-13) 00:16:48.620] [attack.py:257]: Snapshots captured for It 801.
[INFO][(2024-03-13) 00:16:48.622] [eval.py:9]: Reconstructed text: [' The Tower Building Tower the unsc Little Tower Rock hardcore specializes worried spokesman HY semicfet']
[INFO][(2024-03-13) 00:16:54.526] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:16:54.631] [eval.py:31]: Metric report: {"accuracy": 0.25, "bleu": {"bleu": 0.0, "precisions": [0.4, 0.14285714285714285, 0.07692307692307693, 0.0], "brevity_penalty": 0.9355069850316178, "length_ratio": 0.9375, "translation_length": 15, "reference_length": 16}, "rouge": {"rouge1": 0.42857142857142855, "rouge2": 0.15384615384615383, "rougeL": 0.42857142857142855, "rougeLsum": 0.42857142857142855}}
[INFO][(2024-03-13) 00:16:55.494] [attack.py:246]: | It: 811 | Atk. loss: 89.22855 | Task loss: 8.38420 | T: 123.82s |
[INFO][(2024-03-13) 00:16:56.348] [attack.py:246]: | It: 821 | Atk. loss: 83.64878 | Task loss: 8.51354 | T: 124.67s |
[INFO][(2024-03-13) 00:16:57.181] [attack.py:246]: | It: 831 | Atk. loss: 84.47688 | Task loss: 8.38267 | T: 125.50s |
[INFO][(2024-03-13) 00:16:58.015] [attack.py:246]: | It: 841 | Atk. loss: 87.83958 | Task loss: 8.47241 | T: 126.34s |
[INFO][(2024-03-13) 00:16:58.866] [attack.py:246]: | It: 851 | Atk. loss: 85.91773 | Task loss: 8.47077 | T: 127.19s |
[INFO][(2024-03-13) 00:16:59.705] [attack.py:246]: | It: 861 | Atk. loss: 81.99292 | Task loss: 8.36837 | T: 128.03s |
[INFO][(2024-03-13) 00:17:00.550] [attack.py:246]: | It: 871 | Atk. loss: 79.24536 | Task loss: 8.45014 | T: 128.87s |
[INFO][(2024-03-13) 00:17:01.403] [attack.py:246]: | It: 881 | Atk. loss: 78.97289 | Task loss: 8.41848 | T: 129.73s |
[INFO][(2024-03-13) 00:17:02.250] [attack.py:246]: | It: 891 | Atk. loss: 76.50019 | Task loss: 8.37570 | T: 130.57s |
[INFO][(2024-03-13) 00:17:03.106] [attack.py:246]: | It: 901 | Atk. loss: 79.63285 | Task loss: 8.26837 | T: 131.43s |
[INFO][(2024-03-13) 00:17:03.106] [attack.py:257]: Snapshots captured for It 901.
[INFO][(2024-03-13) 00:17:03.108] [eval.py:9]: Reconstructed text: [' The Tower Building Building the unsc Little Tower Rock aspiring specializes worried spokesman HY semicfet']
[INFO][(2024-03-13) 00:17:09.490] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:17:09.593] [eval.py:31]: Metric report: {"accuracy": 0.25, "bleu": {"bleu": 0.0, "precisions": [0.4, 0.14285714285714285, 0.07692307692307693, 0.0], "brevity_penalty": 0.9355069850316178, "length_ratio": 0.9375, "translation_length": 15, "reference_length": 16}, "rouge": {"rouge1": 0.42857142857142855, "rouge2": 0.15384615384615383, "rougeL": 0.42857142857142855, "rougeLsum": 0.42857142857142855}}
[INFO][(2024-03-13) 00:17:10.458] [attack.py:246]: | It: 911 | Atk. loss: 76.74858 | Task loss: 8.26511 | T: 138.78s |
[INFO][(2024-03-13) 00:17:11.310] [attack.py:246]: | It: 921 | Atk. loss: 74.00203 | Task loss: 8.32597 | T: 139.63s |
[INFO][(2024-03-13) 00:17:12.154] [attack.py:246]: | It: 931 | Atk. loss: 73.89743 | Task loss: 8.28601 | T: 140.48s |
[INFO][(2024-03-13) 00:17:13.008] [attack.py:246]: | It: 941 | Atk. loss: 72.96405 | Task loss: 8.31240 | T: 141.33s |
[INFO][(2024-03-13) 00:17:13.857] [attack.py:246]: | It: 951 | Atk. loss: 72.37440 | Task loss: 8.23065 | T: 142.18s |
[INFO][(2024-03-13) 00:17:14.700] [attack.py:246]: | It: 961 | Atk. loss: 71.56242 | Task loss: 8.28515 | T: 143.02s |
[INFO][(2024-03-13) 00:17:15.549] [attack.py:246]: | It: 971 | Atk. loss: 70.93457 | Task loss: 8.21237 | T: 143.87s |
[INFO][(2024-03-13) 00:17:16.407] [attack.py:246]: | It: 981 | Atk. loss: 70.36580 | Task loss: 8.25698 | T: 144.73s |
[INFO][(2024-03-13) 00:17:17.247] [attack.py:246]: | It: 991 | Atk. loss: 69.91876 | Task loss: 8.20674 | T: 145.57s |
[INFO][(2024-03-13) 00:17:18.014] [attack.py:246]: | It: 1000 | Atk. loss: 69.52119 | Task loss: 8.19744 | T: 146.34s |
[INFO][(2024-03-13) 00:17:18.015] [attack.py:257]: Snapshots captured for It 1000.
[INFO][(2024-03-13) 00:17:18.017] [eval.py:9]: Reconstructed text: [' The Tower Building Building the unsc Little Tower Rock aspiring specializes worried spokesman HY semicfet']
[INFO][(2024-03-13) 00:17:24.222] [rouge_scorer.py:83]: Using default tokenizer.
[INFO][(2024-03-13) 00:17:24.326] [eval.py:31]: Metric report: {"accuracy": 0.25, "bleu": {"bleu": 0.0, "precisions": [0.4, 0.14285714285714285, 0.07692307692307693, 0.0], "brevity_penalty": 0.9355069850316178, "length_ratio": 0.9375, "translation_length": 15, "reference_length": 16}, "rouge": {"rouge1": 0.42857142857142855, "rouge2": 0.15384615384615383, "rougeL": 0.42857142857142855, "rougeLsum": 0.42857142857142855}}
[INFO][(2024-03-13) 00:17:24.326] [main.py:57]: Done in 163.42s.