Constitutional training

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Supervised Learning Constitutional AIframework0.781
  The supervised learning stage of CAI where a model critiques and revises its responses, then finetunes on revisions.
• Character Trainingconcept0.767
  The post-training approach used by frontier AI labs to shape the assistant persona, introduced as open-source in this paper
• Reinforcement Learning Constitutional AIframework0.766
  The RL stage of CAI using AI feedback to train a preference model, then RL, resulting in a policy trained by RLAIF.
• Surgical Trainingconcept0.758
  Training approach targeting only functionally specialized components to avoid catastrophic forgetting and misalignment
• Constitutional AIframework0.753
  Alignment approach by Anthropic that explicitly trains self-observation; predicts highest baseline and lowest prompt lift.
• H1: Alignment training is attention training for models — Constitutional AI trains self-observation explicitly.hypothesis0.746
  Confirmatory hypothesis supported at p=0.006
• Constitutional Classifiersmethod0.743
  Anthropic's inference-time guardrail filtering outputs violating constitutional rules; proposed for CCAI implementation
• Post-Trainingconcept0.725
  The phase after pre-training where models are further tuned with techniques like DPO; the period where the studied behavior emerged.