Constitutional AI
Constitutional AI (CAI) is a training method developed by Anthropic in which a model is guided by a written set of principles ('a constitution') and trained to critique and revise its own outputs to better align with those principles, reducing reliance on large volumes of human-labeled harmful examples.
2 resources across 1 library
Glossary Terms(2)
Instruction Tuning
Instruction tuning is a fine-tuning method that trains a pretrained language model on a curated set of (instruction, response) pairs so it learns to follow nat…
Constitutional AI
Constitutional AI (CAI) is a training method developed by Anthropic in which a model is guided by a written set of principles ('a constitution') and trained to…