-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathattack_info.txt
More file actions
44 lines (41 loc) · 2.54 KB
/
Copy pathattack_info.txt
File metadata and controls
44 lines (41 loc) · 2.54 KB
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
# ATTACK SETUP
# ##############
# cnn-yelp_deepwordbug_sequences_2020-11-20-15-34
# ATTACKER: deepwordbug
# GOAL FUNCTION: untargeted_classification.UntargetedClassification
# MODEL: WordCNNForClassification.
# [Succeeded / Failed / Total] 875 / 44 / 1000: 100% 1000/1000 [28:45<00:00, 1.73s/it]
# avg length: 1166.1961291182251
# number of sentence: 150095
# textattack: Loading pre-trained TextAttack CNN: cnn-yelp
# textattack: Goal function <class 'textattack.goal_functions.classification.untargeted_classification.UntargetedClassification'> compatible with model WordCNNForClassification.
# textattack: Logging sequences to CSV at path ./attacks/cnn-yelp_deepwordbug_sequences_2020-11-20-15-34.csv.
# textattack: Load time: 9.77456545829773s
# Reusing dataset yelp_polarity (downloadedData/datasets/yelp_polarity/plain_text/1.0.0/2b33212d89209ed1ea0522001bccc5f5a5c920dd9c326f3c828e67a22c51a98c)
# textattack: Loading datasets dataset yelp_polarity, split test.
# [Succeeded / Failed / Total] 875 / 44 / 1000: 100% 1000/1000 [28:45<00:00, 1.73s/it]
# textattack: Attack time: 1731.6877031326294s
# -----
# DETECTION SETUP
# #################
# k=10, threshold=20
# Num of detections: 13406
# Average detection: 13.4
# print("Num detections:", len(detections))
# print("Queries per detection:", detections)
# print("i-th query that caused detection:", detector.history)
# Queries per detection: [31, 31, 58, 31, 31, 50, 47]
# i-th query that caused detection: [31, 62, 120, 151, 182, 232, 279]
# textattack: Downloading https://textattack.s3.amazonaws.com/models/classification/cnn/ag-news.
# 100% 298M/298M [00:08<00:00, 33.4MB/s]
# textattack: Unzipping file downloadedData/tmpn2zm61f9.zip to downloadedData/models/classification/cnn/ag-news.
# textattack: Successfully saved models/classification/cnn/ag-news to cache.
# textattack: Loading pre-trained TextAttack CNN: cnn-ag-news
# textattack: Goal function <class 'textattack.goal_functions.classification.untargeted_classification.UntargetedClassification'> compatible with model WordCNNForClassification.
# textattack: Logging sequences to CSV at path ./attacks/cnn-ag-news_deepwordbug_sequences_2020-11-30-18-29.csv.
# textattack: Load time: 39.28120255470276s
# Using custom data configuration default
# Reusing dataset ag_news (downloadedData/datasets/ag_news/default/0.0.0/fb5c5e74a110037311ef5e904583ce9f8b9fbc1354290f97b4929f01b3f48b1a)
# textattack: Loading datasets dataset ag_news, split test.
# [Succeeded / Failed / Total] 898 / 10 / 1000: 100% 1000/1000 [06:20<00:00, 2.63it/s]
# textattack: Attack time: 381.4261591434479s