Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Sign up for free
Menu
Search
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Pricing
Search
Sign in
Sign up for free
CoCon
Search
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
Zhang Yixiao
December 16, 2020
Science
420
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
CoCon
Zhang Yixiao
December 16, 2020
More Decks by Zhang Yixiao
See All by Zhang Yixiao
vq-cpc
ldzhangyx
0
410
MixPoet
ldzhangyx
4
470
diora
ldzhangyx
0
310
drummernet
ldzhangyx
0
280
ON-LSTM
ldzhangyx
0
240
Other Decks in Science
See All in Science
Toward Causal Scientific Discovery with AI
sshimizu2006
0
180
Van Dare naar Durf
voginip
0
300
共生概念の整理と AIアライメントの構想
hiroakihamada
0
340
Conversation is the New Dashboard: 属人性を排除する第4世代BIツールの勢力図
shomaekawa
1
660
AlgorithAlgorihms for Decision Making
mickey_kubo
0
140
(CVPR2026) Back to Basics: Let Denoising Generative Models Denoise
shumpei777
0
350
[第67回 CV勉強会@関東] CV × Scientific Figures / kantoCV 67th CVPR 2026
lychee1223
0
210
機械学習 - K近傍法 & 機械学習のお作法
trycycle
PRO
1
1.6k
データベース02: データベースの概念
trycycle
PRO
2
1.4k
AIを用いた PID制御で部屋 の温度制御をしてみた
nearme_tech
PRO
0
270
データベース06: SQL (3/3) 副問い合わせ
trycycle
PRO
1
1.1k
ssmonline #51 ヤマサキ春のサメ祭り 2026 / ssmjp Yamasaki Spring JAWS Festival 2026
naospon
1
150
Featured
See All Featured
Chrome DevTools: State of the Union 2024 - Debugging React & Beyond
addyosmani
10
1.3k
For a Future-Friendly Web
brad_frost
183
10k
Building AI with AI
inesmontani
PRO
1
1.2k
Jamie Indigo - Trashchat’s Guide to Black Boxes: Technical SEO Tactics for LLMs
techseoconnect
PRO
0
670
The Language of Interfaces
destraynor
162
27k
brightonSEO & MeasureFest 2025 - Christian Goodrich - Winning strategies for Black Friday CRO & PPC
cargoodrich
3
840
Avoiding the “Bad Training, Faster” Trap in the Age of AI
tmiket
0
240
Java REST API Framework Comparison - PWX 2021
mraible
34
9.7k
Building a Scalable Design System with Sketch
lauravandoore
464
34k
Writing Fast Ruby
sferik
630
63k
The Organizational Zoo: Understanding Human Behavior Agility Through Metaphoric Constructive Conversations (based on the works of Arthur Shelley, Ph.D)
kimpetersen
PRO
0
450
Documentation Writing (for coders)
carmenintech
77
5.5k
Transcript
CoCon: A Self-Supervised Approach for Controlled Text Generation Presenter: Yixiao
Zhang
TL;DR • 任务:用文本指导文本,进行可控生成 • 亮点: • 让文本成为控制变量:更加灵活 • 漂亮的损失函数 •
模型是自监督训练的 • 结果表明显著加强了语言模型的可控性
Introduction • 基于Transformer的预训练LM成为了新的浪潮,但是从头训练LM 的成本巨大 • 问题:不改变预训练LM的情况下,LM如何进行控制? • 解决办法1:PPLM,通过属性模型控制生成文本 • 缺点:不够精细,可能造成巨大差异
Sumanth Dathathri, Andrea Madotto, Janice Lan, Jane Hung, Eric Frank, Piero Molino, Jason Yosinski, and Rosanne Liu. Plug and play language models: a simple approach to controlled text generation. arXiv preprint arXiv:1912.02164, 2019.
Related Work • 生成包含所需属性的文本 • 早期工作 • 条件生成模型,可通过RL或GAN训练 • 缺陷:对预定属性的要求限制了生成文本的可能类型
• CTRL • 使用control code(预置的metadata)生成文本 • 缺陷:control code也是预先设定的 • PPLM(最相似) • 在LM上插拔一个模块,不重新训练实现生成 • 区别: • 本文旨在在更局部的内容上控制 • CoCon自监督学习,免去了标签数据
Related Work • 文本风格迁移 • 少数研究采用AE以分离表示 • 另外一些模型能识别attribute markers •
一些特定风格相关的n-grams • 通过替换的方式编辑文本风格
CoCon • 模型目标: • 给定引导文本1:−1 和控制文本,模型 生成: • 过程: •
分别编码c和x • 自注意力交互,得 到新的特征 • 进行下一个词预测
CoCon • CoCon是一个单层 Transformer Block • 首先得到x和c的QKV • 将KV拼起来过self-attn
Loss Function • 自重构损失(Self Reconstruction Loss) • 令c = ,使得模型能够学习结合控制文本的内容
• 无文本损失(Null Content Loss) • 令 = ∅,使得模型退化成LM,以生成流畅的文本
Loss Function • 循环重构损失(Cycle Reconstruction Loss) • 在inference中,生成文本不太可能与引导文本共存 • 给定两个不同的文本(,
′)
Loss Function • 对抗损失(Adversarial Loss) • 总优化目标
实验 • 文本引导的文本生成评估指标有BLEU、NIST、METEOR、PPL和 Dist-1/2/3
实验 • 文本引导的文本生成评估指标有BLEU、NIST、METEOR、PPL和 Dist-1/2/3
例子
多个控制