Series 2
Model Internals
What attention is actually doing. How an LLM knows anything. The geometry of embeddings. RLHF. Diffusion. The transformer not as a diagram to memorise, but as a structure to understand.
The Transformer Rebuilt
Series 2
What attention is actually doing. How an LLM knows anything. The geometry of embeddings. RLHF. Diffusion. The transformer not as a diagram to memorise, but as a structure to understand.