- AI grad school
- A new type of information theory
- Embeddings
-
Text Embeddings Reveal (Almost) As Much As Text
-
Contextual document embeddings https://arxiv.org/abs/2410.02525
-
Universality embeddings
Harnessing the Universal Geometry of Embeddings
- Language models
- GPT-style language models memorize 3.6 bits per param*
- Approximating Language Model Training Data from Weights
- Inversion
- "There Are No New Ideas In AI.... Only New Datasets"
—
https://junyanz.github.io/CycleGAN/