concepts → Algorithm
RLHF
also called reinforcement learning from human feedback
tilt a pretrained model toward human preference — how raw prediction becomes behaviour
Where it sits
The prism has six jobs across and eight layers down. Its primary cell islearn × L4, with a noted second cell. Hatched cells cannot exist — a GPU does not learn, an institution does not infer.
Overlays: Safety & alignment. act×L4 — it is a policy-optimisation method, not just supervised fitting
What must come first — and what it unlocks
Left to right is reading order, derived from the prerequisite_of edges. Nothing here is hand-ordered: the diagram is the graph.
Before it: Fine-tuning · Reinforcement learning
Unlocks: GRPO
What kind of thing it is — and what it is made of
It is made of: Reward model.
Where to read it
The chapter that introduces it, and any chapter that uses it again.
Where it comes from
Every connection
All 7 edges touching this node, grouped by relation family — the sections above are highlights from this list. Colours match the relation families inthe atlas.
This page is a projection of one node in src/data/concepts.ts. It has no prose file of its own — 230 declared edges produce all 236 of these pages. Edit an edge and both endpoints change.