What problem does it solve?
Producing a publication-ready ML/AI paper is difficult because it requires tightly connected research work, verified citations, reproducible experiments, and conference-specific formatting—without hallucinated claims or broken evidence chains.
Core Features & Use Cases
- End-to-end research paper pipeline: design→execute→analyze→draft→self-review→revise→submit for major ML venues.
- Citation hallucination prevention: fetch BibTeX programmatically and mark anything unverifiable as [CITATION NEEDED].
- Experiment-to-narrative alignment: map each claim to experiments, then generate a structured experiment log to ground the writeup.
- Reviewer-style guardrails: use venue checklists (e.g., NeurIPS/ICML/ACL/ICLR requirements) and human-eval and stats best practices where applicable.
- Iterative refinement loop: explicitly treat the workflow as an iterative research lifecycle, not a linear sequence.
Quick Start
Use the research-paper-writing skill to generate a draft ML paper outline and an initial section-by-section writing plan that is backed by an experiment log you will fill as results come in.