Glossary term
Glossary term
Agentic Systems
Using feedback from human raters to improve the quality of a model's responses. For example, an RLHF mechanism can ask users to rate the quality of a model's response with a 👍 or 👎 emoji. The system can then adjust its future responses based on that feedback.
Created for this library
An LLM team uses RLHF to align a base model with human preferences for helpfulness and safety.
A customer support team uses RLHF on its assistant to fine-tune responses toward styles that human reviewers prefer.
A research lab uses RLHF on its instruction-tuned model to align responses with human judgments at scale.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License