Evan Pu·Aug 26, 2024why models (such as llms) shouldn’t train on their own generated dataa small example
Evan Pu·Apr 3, 2024Chinese Characters as EmojisI get asked to explain the Chinese writing system a lot . . .
Evan Pu·Mar 27, 2023LLM self-play on 20 Questionsgpt-3.5-turbo has a score of 68/1823 playing 20 questions with itselfA response icon1A response icon1
Evan Pu·Dec 28, 2022Probing Compositional Understanding of ChatGPT with SVGI ask ChatGPT to draw and explain programmatic SVG imagesA response icon2A response icon2
Evan Pu·Dec 5, 2022The Language of Modificationsa short study on the languages of descriptions and modifications