Contents

• Personality Traits in Large Language Models (Jul 2023, added 7/14/23)
• LLMZip: Lossless Text Compression using Large Language Models (Jun 2023, added 6/24/23)
• Eight Things to Know about Large Language Models (Apr 2023)
• Augmented Language Models: a Survey (Feb 2023)
• The Simplest Description Of The Classical Transformer Architecture (Jun 2023)
• Holistic Evaluation of Language Models (Nov 2022)
• Sparks of Artificial General Intelligence (Mar 2023)
• Explaining the Transformer
• Python Code Implementing a Transformer
• An Overview On Language Models: Recent Developments And Outlook (Mar 2023)
• Harnessing the Power of LLMs in Practice: A Survey on ChatGPT and Beyond (Apr 2023, added 5/7/23)

Personality Traits in Large Language Models (Jul 2023, added 7/14/23)

(link)

Good foundational insight: you can prompt LLMs into adopting a “personality”, and it will produce output that is consistent with the associated personality traits. To do that, the paper does the following:

1. Pick the right personality tests. For example, the Big Five Inventory, BFI: identifies five personality trait dimensions - extraversion (EXT), agreeableness (AGR), conscientiousness (CON), neuroticism (NEU), and openness to experience (OPE).
2. Ensure structural validity of a test. Is the test reliable across multiple measurements (i.e., its items)? In other words, do responses to the test’s items form consistent patterns? Do the test’s items reflect the variance of one underlying factor or construct?
3. Ensure external validity of a test. Are the test scores practically meaningful, outside (external to) the test context itself?