research-paper-writing

To write an ML research paper, this Skill provides an end-to-end framework that guides authors from problem statement and contribution framing to a full draft with structured citations, reproducibility artifacts, and iterative refinement for submission.

For academic paper citation management, the framework integrates verified BibTeX entries using dependencies like semanticscholar and habanero, ensuring references are accurate and up-to-date throughout the drafting process.

To document reproducibility artifacts for NeurIPS or ICML, the framework provides scaffolding templates to capture data provenance, hyperparameters, experimental setup, and analysis artifacts required for submission-ready papers.

You can use this framework for theory and survey papers, as it applies to theory, survey, benchmark, and empirical ML papers across major venues including NeurIPS, ICML, ICLR, ACL, AAAI, and COLM.

The writing pipeline supports generating figures by including matplotlib and SciencePlots as dependencies, providing templates for figure and table generation within the reproducibility scaffolding.

To create a reproducibility appendix from project notes, the framework applies an iterative refinement workflow that captures data provenance and hyperparameter documentation, generating submission-ready reproducibility artifacts.