SciDER: Scientific Data-centric End-to-end Researcher

Ke Lin, Yilin Lu, Shreyas Bhat, Xuehang Guo, Junier Oliva, Qingyun Wang · 2026

Read SciDER: Scientific Data-centric End-to-end Researcher by Ke Lin, Yilin Lu, Shreyas Bhat, Xuehang Guo, Junier Oliva, Qingyun Wang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-01421.txt.

Description

Lin and five colleagues offer SciDER, a multi-agent system running the research life through a data-centric approach and a dynamic multi-mode skill system across four sub-agents. An idea agent makes guesses via Evolutionary Idea Search, a data agent shapes raw data, a lab agent writes runnable code from dataset traits, and a critic agent drives looping self-polish. They release OpenSciDER-SFT-8K, a high-quality run path dataset, with the OpenSciDER-27B fine-tuned model. Across six benchmarks, SciDER and OpenSciDER obtain leading results.

Catalog record

CREATORKe Lin, Yilin Lu, Shreyas Bhat, Xuehang Guo, Junier Oliva, Qingyun Wang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32217
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.01421

Browse the full library · About the Tower of Babel