Browse
reward model
-
RLHF Explained for Developers
Area What the evidence says Planning use Pipeline shape Three stages: pretraining, instruction fine-tuning, then preference alignment Start from an…
-
Constitutional AI and Model Alignment Basics
What is Constitutional AI? It is a method from Anthropic for training a model to be harmless, in which AI…
-
DPO vs RLHF Compared
Criterion RLHF (reward model + PPO) DPO Planning use Reward model Built from preference data first No explicit reward function;…