teaching_llm_applications
Advanced topics
- JEPA Yann Le Cun
- Video of explanation of JEPA by Vizuara
- Cache sharing between models and share experience between models
-
World model: can have consequence of action
- Reasoning models video by Vizuara
Mixture of experts (MoE)
- Instead of one massive network, the model is divided into a Gating Network (the router) and a set of eight smaller, specialized “Experts” (for Word Choice, Code Gen, Emotional Tone, etc.).
.svg)