This Repository is an up-to-date version of Harvard nlp's Legacy code and a Refactoring of the jupyter notebook version as a shell script version.

Overview

2022_the_annotated_transformer

Goal

This Repository is an up-to-date version of Harvard nlp's Legacy code and a Refactoring of the jupyter notebook version as a shell script version.

Key points

  • We have re-factored Harvard NLP's Annotated Trasformer into a shell script version.

  • Dataset utilized Multi30K. (The dataset is small, so you can see the results quickly even on computers with low specifications.)

  • We provide the Colab version along with the shell script version, making it easy to modify the model and test the method.

    https://colab.research.google.com/drive/1SrRmC_Ti8IepeHFNBZBjNxl_wkTSJReC?usp=sharing

  • Loss Graph can be drawn.

  • BLEU Score can be measured.

file structure

├── models
│   ├── __init__.py
│   ├── blocks
│   │   ├── __init__.py
│   │   ├── decoder_layer.py
│   │   ├── encoder_layer.py
│   ├── embedding
│   │   ├── __init__.py
│   │   ├── positional_encoding.py
│   │   └── token_embedding.py
│   ├── layers
│   │   ├── __init__.py
│   │   ├── layer_norm.py
│   │   ├── multi_headed_attention.py
│   │   ├── position_wise_feed_forward.py
│   │   └── sublayer_connection.py
│   ├── model
│   │   ├── __init__.py
│   │   ├── decoder.py
│   │   ├── encoder_decoder.py
│   │   ├── encoder.py
│   │   ├── generator.py
│   └── util.py
├── result
│   ├── loss_graph.png
│   ├── train_loss.txt
│   └── valid_loss.txt
├── saved
├── utils
    ├── __init__.py
    ├── batch.py
    ├── batch_size_fn.py
    ├── bleu.py
    ├── data_loader.py
    ├── epoch_time.py
    ├── greedy_decode.py
    ├── label_smoothing.py
    ├── make_model.py
    ├── NoamOpt.py
    ├── run_epoch.py
    ├── simple_loss_compute.py
    └── tokenizer.py
├── README.md
├── test.py
├── train.py
├── config.py
├── data.py
└── graph.py

Training Result

Train Validation loss graph

image

Test set(unseen data) Translation Example

image

Test set(unseen data) BLEU Score Average: 35.870847920953594

Reference

https://nlp.seas.harvard.edu/2018/04/03/attention.html

https://jalammar.github.io/illustrated-transformer/

https://www.facebook.com/groups/TensorFlowKR/permalink/1618169785190740/

https://github.com/hyunwoongko/transformer

Owner
신재욱
신재욱
SubFind - Subdomain Finder Tools

SubFind (Subdomain Finder Tools) Info Tools Result Of Subdomain Command In Termi

LangMurpY 2 Jan 25, 2022
Microsoft Exchange Server SSRF漏洞(CVE-2021-26855)

Microsoft_Exchange_Server_SSRF_CVE-2021-26855 zoomeye dork:app:"Microsoft Exchange Server" 使用Seebug工具箱及pocsuite3编写的脚本Microsoft_Exchange_Server_SSRF_CV

conjojo 37 Nov 12, 2022
解密哥斯拉webshell管理工具流量

kingkong 解密哥斯拉Godzilla-V2.96 webshell管理工具流量 目前只支持jsp类型的webshell流量解密 Usage 获取攻击者上传到服务器的webshell样本 获取wireshark之类的流量包,一般甲方有科来之类的全流量镜像设备,联系运维人员获取,这里以test.

h4ck for fun 46 Dec 21, 2022
Buffer Overflow para SLmail5.5 32 bits

SLmail5.5-Exploit-BoF Buffer Overflow para SLmail5.5 32 bits con un par de utilidades para que puedas hacer el tuyo REQUISITOS PARA QUE FUNCIONE: Desa

Luis Javier 15 Jul 30, 2022
Natas teaches the basics of serverside web-security.

over-the-wire-natas Natas teaches the basics of serverside web-security. Each level of natas consists of its own website located at http://natasX.nata

Siddhant Chouhan 1 Nov 27, 2021
A tool for making python source difficult to read.

obscurepy Description A tool for obscuring, or making python source code difficult to read. Table of Contents Installation Limitations Usage Disclaime

Andrew Christiansen 10 Jul 31, 2022
对安卓APP注入MSF PAYLOAD,并且对手机管家进行BYPASS。

520_APK_HOOK 介绍 将msf生成的payload,注入到一个正常的apk文件中,重新打包后进行加固,bypass手机安全管家的检测。 项目地址: https://github.com/cleverbao/520apkhook 作者: BaoGuo 优点 相比于原始的msf远控,此版本ap

BaoGuo 368 Jan 02, 2023
Exploit for GitLab CVE-2021-22205 Unauthenticated Remote Code Execution

Vuln Impact An issue has been discovered in GitLab CE/EE affecting all versions starting from 11.9. GitLab was not properly validating image files tha

Hendrik Agung 2 Dec 30, 2021
Instagram brute force tool that uses tor as its proxy connections

Insta-crack This is a instagram brute force tool that uses tor as its proxy connections, keep in mind that you should not do anything illegal with thi

Liam 3 Jan 28, 2022
Just your basic port scanner - with multiprocessing capabilities & further nmap enumeration.

Just-Your-Basic-Port-Scanner Just your basic port scanner - with multiprocessing capabilities & further nmap enumeration. Use at your own discretion,

Edward Zhou 0 Nov 06, 2021
This enforces signatures for CVE-2021-44228 across all policies on a BIG-IP ASM device

f5-waf-enforce-sigs-CVE-2021-44228 This enforces signatures for CVE-2021-44228 across all policies on a BIG-IP ASM device Overview This script enforce

Ismael Gonçalves 5 Mar 31, 2022
Make your own huge Wordlist with advanced options

#It's my first tool i hope to be useful for everyone, Make your own huge Wordlist with advanced options, You need python3 to run this tool, If you hav

0.1Arafa 6 Dec 08, 2022
md5 hash cracking with python.

Python-Md5-Cracker- md5 hash cracking with python. Original files added First create a file called word.txt then run the wordCreate.py script The task

Nebil Sharifi 0 Aug 31, 2022
Statistical Random Number Generator Attack Against The Kirchhoff-law-johnson-noise (Kljn) Secure Key Exchange Protocol

Statistical Random Number Generator Attack Against The Kirchhoff-law-johnson-noise (Kljn) Secure Key Exchange Protocol

zeze 1 Jan 13, 2022
Python tool for enumerating directories and for fuzzing

Python tool for enumerating directories and for fuzzing

Gourab Roy 5 Feb 21, 2022
利用NTLM Hash读取Exchange邮件

GetMail 利用NTLM Hash读取Exchange邮件:在进行内网渗透时候,我们经常拿到的是账号的Hash凭据而不是明文口令。在这种情况下采用邮件客户端或者WEBMAIL的方式读取邮件就很麻烦,需要进行破解,NTLM的破解主要依靠字典强度,破解概率并不是很大。

<a href=[email protected]"> 388 Dec 27, 2022
Wordlist attacks on Bitwarden data.json files

BitwardenDecryptBrute This is a slightly modified version of BitwardenDecrypt. In addition to the decryption this version can do wordlist attacks for

42 Nov 09, 2022
Repository for a project of the course EP2520 Building Networked Systems Security

EP2520_ACME_Project Repository for a project of the course EP2520 Building Networked Systems Security in Royal Institute of Technology (KTH), Stockhol

1 Dec 11, 2021
Oh365UserFinder is used for identifying valid o365 accounts without the risk of account lockouts.

Oh365 User Finder Oh365UserFinder is used for identifying valid o365 accounts without the risk of account lockouts. The tool parses responses to ident

Joe Helle 414 Jan 02, 2023