Reproduce “PANINI-DR: Controlled Deep Research through Generative Semantic Workspaces” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
M. Yigit Turali, Shreyas Rajesh, M. Efe Lorasdagi, Chenda Duan, Kartik Sharma, Vwani Roychowdhury
PANINI-DR organizes deep research as a dependency plan in a shared workspace, attaching findings and source evidence to subquestions and checking proposed answers before reuse. On BrowseComp-Plus CF-150, it achieves 73.3% accuracy, versus 57.3% for SLEUTH, and leads in saved-evidence recoverability across three benchmarks. However, it ranks fifth on official MoNaCo F1 and uses substantially more estimated computation than SLEUTH.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.