Preference learning without pairwise labels
Matches focus: RLHF + weak supervision
Built for 5–20 person R&D teams
The team research library that keeps itself alive.
scivein reads the open literature for your lab, sends a daily paper digest, and turns every save into shared institutional memory.
- 5 papers per teammate every weekday
- Why-it-matters summaries
- One-click save to the team library
Tomorrow's digest
RLHF systems group
Open weight evals for agentic search
Why it matters: benchmark drift
A survey of small-lab reproducibility
Saved by 3 similar teams
Digest
Library
Search
Join the private beta
We are onboarding teams now. Drop your email and we will add you as soon as there is capacity — typically within 48 hours.
Already invited? Sign in.
Who it is for
R&D teams
5–20 person labs drowning in preprints. scivein surfaces what matters daily.
Solo researchers
PhD students, postdocs, and principal scientists who want the literature curated.
Open-source labs
EleutherAI, HF, and decentralized research groups that live on GitHub + arXiv.