SKU PRF-3003 · Sold by External
Stanford Human Preferences (SHP)
Product specifications
| SKU | PRF-3003 |
|---|---|
| Data type | Preference data |
| Volume | 385,563 preference pairs |
| Size on disk | ~827 MB (827,255,243 bytes, original JSON files; ~167 MB as Parquet) |
| Format | JSON (one JSONL/JSON file per subreddit split; Parquet auto-conversion on HF) |
| Access model | PUBLIC LICENSE |
| Pricing | Free · open dataset (HF) |
| Quality score | — |
| License | No standard/SPDX license; Reddit-scraped under the Reddit API Terms of Use (User Content owned by Reddit users; Reddit grants a non-exclusive, non-transferable, revocable license) |
SHP is a dataset of 385K collective human preferences over responses to questions/instructions across 18 subject-area subreddits (from cooking to legal advice), intended for training RLHF reward models and NLG evaluation models such as SteamSHP. Each example is a Reddit self-post plus a pair of top-level comments, where one comment is labeled more preferred by Reddit users collectively. The preference is inferred structurally, not by an annotator: if comment A was written no later than comment B yet still has a higher Reddit score, A is taken to be more preferred (writing earlier would otherwise confer a visibility advantage). All responses are naturally occurring and human-written, distinguishing SHP from machine-written preference sets like Anthropic HH-RLHF. Pairs are only included when the post score is at least 10 and each comment score is at least 2, and labels are randomized to a roughly 50/50 A/B distribution. It was created at Stanford using the techniques from the ICML 2022 paper 'Understanding Dataset Difficulty with V-Usable Information'.