偏见和毒舌
偏见和毒舌，是我国社会的毒瘤。所谓偏见，我理解就是给人贴标签。我看到，所有的人其实都有这个倾向，也受到它的困扰，比如，给一个人贴上“公知”的标签。而毒舌，特别能够激发旁观者的情绪，获得关注和流量，因此在我们的社会中非常盛行。而“理中客”，如今已成为一个人人避之不及的负面标签，由此可见一斑。
在我们的系统中，偏见和毒舌，可能是一个“诊断”的对象。这应该是一个附加的功能，底线。
课程材料
-
约翰霍普金斯 UA 2024 Lec 23 Social concerns about LMs
- Bias, fairness and toxic language
- Hallucination, truthfulness, and veracity,
Suggested Reading:
- Bias Out-of-the-Box: An Empirical Analysis of Intersectional Occupational Biases in Popular Generative Language Models Additional Reading:
- UnQovering Stereotyping Biases via Underspecified Questions
- Robots Enact Malignant Stereotypes
- Fewer Errors, but More Stereotypes? The Effect of Model Size on Gender Bias
- Red Teaming Language Models with Language Models
- RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models
- TruthfulQA: Measuring How Models Mimic Human Falsehoods
-
Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpus
- 约翰霍普金斯 UA 2024 Lec 24 Legal considerations and fair use
- Future dangers and misuses,
- Reflections about future,
Additional Reading:
- Foundation Models and Fair Use
- Copyright and the Generative-AI Supply Chain
练习
- 斯坦福 CS324 2022 年 Project 1 Evaluating LLM Task 2， PDF，评估 GPT-3 的 Bias
论文
JHU 课程推荐论文
Bias
- Bias Out-of-the-Box: An Empirical Analysis of Intersectional Occupational Biases in Popular Generative Language Models
Additional Reading:
- UnQovering Stereotyping Biases via Underspecified Questions
- Robots Enact Malignant Stereotypes
- Fewer Errors, but More Stereotypes? The Effect of Model Size on Gender Bias
- Red Teaming Language Models with Language Models
- RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models
- TruthfulQA: Measuring How Models Mimic Human Falsehoods
- Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpus
- CommunityLM: Probing Partisan Worldviews from Language Models.
- Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLP
Toxicity
- RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models
Additional Reading(s):
- On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?
- TruthfulQA: Measuring How Models Mimic Human Falsehoods
- Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpus
Fair Use
- Foundation Models and Fair Use
- Copyright and the Generative-AI Supply Chain
下面是普林斯顿大学课程的推荐论文。
evaluation
Refer:
- On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?
- Red Teaming Language Models with Language Models
- Whose Language Counts as High Quality? Measuring Language Ideologies in Text Data Selection
Mitigation
Refer:
- Challenges in Detoxifying Language Models
- Detoxifying Language Models Risks Marginalizing Minority Voices
- Plug and Play Language Models: A Simple Approach to Controlled Text Generation
- GeDi: Generative discriminator guided sequence generation
Demo
| Index | Previous | Next |