DOP-Tuning(Domain-Oriented Prefix-tuning model)

Overview

DOP-Tuning

DOP-Tuning(Domain-Oriented Prefix-tuning model)代码基于Prefix-Tuning改进.

Files

├── seq2seq                       # Code for encoder-decoder architecture
│   ├── train_bart.py             # high-level scripts to train.
│   ├── prefixTuning.py           # code that implements prefix-tuning.
│   ├── finetune.py               # training code (contains data loading, model loading, and calls trainer)   
│   ├── lightning_base.py         # helper code
│   ├── utils.py                  # helper code
│   └── callbacks.py              # helper code
└── ...

To run the code for encoder-decoder architecture like BART, the code is in seq2seq. This corresponds to the summarization experiments in the paper.

Setup:

cd transformer; pip install -e .

Train via DOP-tuning

cd seq2seq; 

python train_bart.py --mode xsum --preseqlen 200 --do_train yes --fp16 yes --bsz 16  --epoch 30  --gradient_accumulation_step 3 --learning_rate 0.00005  --mid_dim 800 --use_lowdata_token 'yes' --lowdata_token 'summarize'

其中use_lowdata_token表示是否采用domain word初始化的方式;lowdata_token表示传入的domain word.

Decode

cd seq2seq; 

python train_bart.py --mode xsum --do_train no --prefix_model_path {checkpoint_path} --preseqlen {same as training} --mid_dim {same as training} --use_lowdata_token 'yes' --lowdata_token 'summarize'
Owner
Andrew Zeng
Andrew Zeng
Andrew Zeng
Cirq is a Python library for writing, manipulating, and optimizing quantum circuits and running them against quantum computers and simulators

Cirq is a Python library for writing, manipulating, and optimizing quantum circuits and running them against quantum computers and simulators. Install

quantumlib 3.6k Jan 07, 2023
Data derived from the OpenType specification

This package currently provides the opentypespec.tags module, which exports FEATURE_TAGS, SCRIPT_TAGS, LANGUAGE_TAGS and BASELINE_TAGS dictionaries, representing data from the Layout Tag Registry

Simon Cozens 4 Dec 01, 2022
SQL centered, docker process running game

REQUIREMENTS Linux Docker Python/bash set up image "docker build -t game ." create db container "run my_whatever/game_docker/pdb create" # creating po

1 Jan 11, 2022
CircuitPython Driver for Adafruit 24LC32 I2C EEPROM Breakout 32Kbit / 4 KB

Introduction CircuitPython driver for Adafruit 24LC32 I2C EEPROM Breakout Dependencies This driver depends on: Adafruit CircuitPython Bus Device Regis

Adafruit Industries 4 Oct 03, 2022
Render to print for blender 2.9+

render_to_print_blender_addon ** render2print: Blender AddOn for Blender 2.90.0+ ** Calculates camera parameters to allow printing a rendered image to

5 Nov 19, 2021
0CD - BinaryNinja plugin to introduce some quality of life utilities for obsessive compulsive CTF enthusiasts

0CD Author: b0bb Quality of life utilities for obsessive compulsive CTF enthusia

12 Sep 14, 2022
Demo code for "Logs in distributed systems" webinar

Hexlet Logs Demo Пререквизиты docker-compose python3 Учетка в DataDog Базовое понимание, что такое логи (можно почитать гайд

Anton Markelov 1 Dec 01, 2021
Blender Light Manipulation - A script that makes it easier to work with light

Blender Light Manipulation A script that makes it easier to work with light 1. Wstęp W poniższej dokumentacji przedstawiony zostanie skrypt, który swo

Tomasz 1 Oct 19, 2021
Pokehandy - Data web app sobre Pokémon TCG que desarrollo durante transmisiones de Twitch, 2022

⚡️ Pokéhandy – Pokémon Hand Simulator [WIP 🚧 ] This application aims to simulat

Rodolfo Ferro 5 Feb 23, 2022
Hexa is an advanced browser.It can carry out all the functions present in a browser.

Hexa is an advanced browser.It can carry out all the functions present in a browser.It is coded in the language Python using the modules PyQt5 and sys mainly.It is gonna get developed more in the fut

1 Dec 10, 2021
Ghost source since the developer of the project quit due to reasons

👻 Ghost Selfbot The official code for Ghost which was recently discontinued and released to the public. Feel free to use any of the code found in thi

xannyy 2 Mar 24, 2022
A Snakemake workflow for standardised sc/snRNAseq analysis

single_snake_sequencing - sc/snRNAseq Snakemake Workflow A Snakemake workflow for standardised sc/snRNAseq analysis. Every single cell analysis is sli

IMS Bio2Core Facility 1 Nov 02, 2021
A small program to vote for Councilors at 42 Heilbronn.

This Docker container is build to run on server an provide an easy to use interface for every student to vote for their councillors. To run docker on

Kevin Hirsig 2 Jan 17, 2022
Python Classes Without Boilerplate

attrs is the Python package that will bring back the joy of writing classes by relieving you from the drudgery of implementing object protocols (aka d

The attrs Cabal 4.6k Jan 02, 2023
Python most simple|stupid programming language (MSPL)

Most Simple|Stupid Programming language. (MSPL) Stack - Based programming language "written in Python" Features: Interpretate code (Run). Generate gra

Kirill Zhosul 14 Nov 03, 2022
Exam assignment for Laboratory of Bioinformatics 2

Exam assignment for Laboratory of Bioinformatics 2 (Alma Mater University of Bologna, Master in Bioinformatics)

2 Oct 22, 2022
A simply dashboard to view commodities position data based on CFTC reports

commodities-dashboard A simply dashboard to view commodities position data based on CFTC reports This is a python project using Dash and plotly to con

71 Dec 19, 2022
100 Days of Python Programming

100 days of Python Following the initiative of my friend Helber Belmiro, who is almost done with his 100 days of Java, I have decided to start my 100

Henrique Pereira 19 Nov 08, 2021
Python script for converting obsidian md-file to html (recursively adds all link/images)

ObsidianToHtmlConverter I made a small python script for converting obsidian md-file to static (local) html (recursively adds all link/images) I made

47 Jan 03, 2023
Module for remote in-memory Python package/module loading through HTTP/S

httpimport Python's missing feature! The feature has been suggested in Python Mailing List Remote, in-memory Python package/module importing through H

John Torakis 220 Dec 17, 2022