Hola Adrakey, PhD·Nov 6When Preventing Collapse Destroys Useful Representations: The Hidden Role of Optimisation in LeJEPAAn experimental journey into LeJEPA, Gaussian regularisation, and the difference between geometric diversity and discriminative…
InAccredianbyAdiNagare·Oct 23Representation Learning: The Art of Learning What MattersTeaching Machines to be Self- Reliant!
Maneaditya·Oct 11What If AI Could Understand Concepts Without Storing Them as Words?Inside the hidden mathematical world where AI represents meaning, connects ideas, and processes information beyond language.
Joumana Makki·Oct 5Why Do Trees Feel More at Home in Hyperbolic Space?An intuitive and mathematical introduction to hyperbolic representations in machine learningA response icon1A response icon1
Maneaditya·Sep 29What If AI Learned on a Sphere?When we talk about AI representations, we usually imagine vectors.
Anupama Bidargaddi·Sep 26🚀 How Do We Know If an Embedding Is Good?In the previous articles, we explored how representations are built.A response icon1A response icon1
Eleventh Hour Enthusiast·Sep 12Symmetry from BelowHow knowledge graph embedding models searched their way toward a lesson that geometric deep learning later echoed
Enrico De Santis·Sep 12Beyond Next-Token Prediction. How Transformers architectures can learn rulesTransformers can learn internal rules and procedures, going beyond simple memorization and next-token prediction.
Enrico De Santis·Sep 12The Real Leap of Transformers Is Learning How to Represent ContextWhy next-token prediction became dramatically more powerful when language models learned the space in which probability is estimated
