Datasets and language-model research
Princeton dates its foundation to 1746. Its AI contributions include the ImageNet dataset project and later research organized through Princeton Language and Intelligence, spanning methods, applications and social implications of large models. [1][2][3]
The 2009 ImageNet paper describes an image database organized using WordNet categories. Its reported snapshot contained 3.2 million images in 5,247 synsets; those figures belong to that paper, rather than the larger dataset versions used in later competitions. The authors describe collecting and organizing a database for large-scale visual recognition research. [2]
Princeton Language and Intelligence held its kickoff on September 26, 2023, as reported in the university's October 6 account. The initiative funded research and collaboration and described a compute purchase planned for later months. That plan should not be treated as a live GPU cluster at the launch. [3]