✳flâneur — a map of the web's best reading
Arjun Kocher
k-a.in · 48 words · saved by 1 readers
I study deep neural architectures, and their underlying mathematics. A brief introduction to the seminal Transformer paper. A groundbreaking training approach from DeepSeek. How reasoning emerges in large language models. A neural network optimizer. A new learning paradigm. Architecture behind Nested Learning. Understanding Manifold Constrained Hyper Connections. Deriving Manifold Constrained Hyper Connections.
Explore this link on the map →