Safety, alignment & interpretability· 2020
Learning to summarize from human feedback
On the reading lists
- #53 in Safety, alignment & interpretability (2026 edition)
Why it matters
Demonstrated that optimizing human preferences improves language-generation quality beyond reference-based training.
Condensed from the expert lists' notes.
The record
- First released
- 2020 (from arXiv 2009.01325)
- Citations
- 3,563 citations, as counted by Papers With Code (CC BY-SA 4.0)
- Named by
- the chatgpt list only
- Also relevant to
- Reinforcement learning
- Score
- 0.319 · computed rank 560 · consensus 0.12, must-read 0, citations percentile 0.809 (2020s · safety), awards 0, existing lists 0. Method
Data sources
Catalogue data from OpenAlex (CC0), Crossref (CC0 metadata; no abstracts are shown), arXiv, and Papers With Code (CC BY-SA 4.0). Each count names its source. See the method.