Reproduce “CLEAR: Structured Specifications for Controllable Long-Horizon ML Research Engineering” 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
Xin Liu, Hongru Song, Ruqing Zhang, Jiafeng Guo, Xueqi Cheng
CLEAR turns research papers and task descriptions into executable ML systems by creating a persistent specification of requirements that guides planning, parallel implementation, and targeted repair from execution feedback. A same-model PaperBench ablation scores full CLEAR at 29.87, versus 24.14 without its specification-processing skills, and costs $7.91 versus $10.08 per paper across 20 tasks, supporting those skills in this configuration.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.