bro jia 1030514181
  • Joined on Jul 18, 2022
  • Organization
Loading Heatmap…

1030514181 closed issue 1030514181/OpenAttack#6

使用OpenAttack对外部NLP-API进行攻击

1 hour ago

1030514181 commented on issue 1030514181/OpenAttack#6

使用OpenAttack对外部NLP-API进行攻击

![image](/attachments/81472c63-8b4a-499e-8bb9-4cae38454f33) IBM NLP API

1 hour ago

1030514181 pushed to master at 1030514181/atd

1 hour ago

1030514181 upload dataset t10k-labels-idx1-ubyte.gz

1 hour ago

1030514181 upload dataset train-labels-idx1-ubyte.gz

1 hour ago

1030514181 upload dataset train-images-idx3-ubyte.gz

1 hour ago

1030514181 closed issue 1030514181/OpenAttack#10

计算BLEU指标用于评估攻击效果

1 hour ago

1030514181 commented on issue 1030514181/OpenAttack#10

计算BLEU指标用于评估攻击效果

计算公式如下: ![image](/attachments/249d50eb-2ea0-4043-843a-39f2813ecd58)

1 hour ago

1030514181 commented on issue 1030514181/OpenAttack#10

计算BLEU指标用于评估攻击效果

Bleu是IBM在2002提出的,用于机器翻译任务的评价,发表在ACL,引用次数10000+,原文题目是“BLEU: a Method for Automatic Evaluation of Machine Translation”。 它的总体思想就是准确率,假如给定标准译文reference,神经网络生成的句子是candidate,句子长度为n,candidate中有m个单词出现在reference,m/n就是bleu的1-gram的计算公式。 BLEU还有许多变种。根据n-gram可以划分成多种评价指标,常见的指标有BLEU-1、BLEU-2、BLEU-3、BLEU-4四种,其中n-gram指的是连续的单词个数为n。 BLEU-1衡量的是单词级别的准确性,更高阶的bleu可以衡量句子的流畅性。

1 hour ago

1030514181 created repository 1030514181/atd

1 hour ago

1030514181 upload dataset train.txt

1 hour ago

1030514181 upload dataset dev.txt

1 hour ago

1030514181 upload dataset test.txt

1 hour ago

1030514181 opened issue 1030514181/NER#1

在流量调度数据集上实现命名实体识别的对抗性攻击

1 hour ago

1030514181 pushed to master at 1030514181/NER

1 hour ago

1030514181 created repository 1030514181/NER

3 hours ago

1030514181 pushed to master at 1030514181/Crawler

4 hours ago

1030514181 created repository 1030514181/Crawler

4 hours ago

1030514181 merged pull request 1030514181/DeepExploit#1

上传文件至 ''

4 hours ago

1030514181 pushed to master at 1030514181/DeepExploit

  • 6cc90314d3 Merge pull request '上传文件至 ''' (#1) from 1030514181-patch-1 into master Reviewed-on: https://git.openi.org.cn/1030514181/DeepExploit/pulls/1
  • f5d5eaa101 上传文件至 '' 添加了数据集模块
  • Compare 2 commits »

4 hours ago