Concept

Effectiveness of Sparse but Informative Human Feedback in RLHF

Although the reward signals in RLHF are sparse, typically provided only once per sequence, they are highly effective for training. This is because the feedback, originating from human judgment, is very informative and accurate. The combination of sparsity with high-quality signals allows for a learning process that is both robust and efficient.

0

1

Updated 2025-10-10

Contributors are:

Who are from:

Tags

Ch.4 Alignment - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences