Fine-Grained RLHF
Free while signed in. Answers cite the passages they came from.
First page
The curator’s take
Key pointsTrains LMs with segment-level human feedback rather than whole-response preferences.
01
Segment-level rewards: Provides multiple reward models targeting specific dimensions (factuality, relevance, fluency) at the span level.
02
Long-form QA gains: Substantial improvements on long-form question answering where whole-response preferences are too coarse.
03
Toxicity reduction: Enables targeted reduction of toxic spans without degrading overall response quality.
04
Controllable RLHF: Enables model customization by emphasizing different reward dimensions at inference time.
Every Monday
Get next week’s papers.
The same picks and the same summaries, in your inbox. Free, and 176 issues deep.
Subscribe on Substack