research-paper-writing

Follow the phased pipeline: identify a one-sentence contribution, map each claim to an experiment, run experiments with incremental result saving, then draft using the venue's official LaTeX template. Finish with a simulated self-review round and the venue's mandatory checklist before submitting.

Never generate BibTeX from memory; fetch it programmatically via DOI content negotiation from CrossRef or the Semantic Scholar API. Verify each paper exists in at least two sources, confirm the cited claim appears in the abstract, and mark unverifiable entries as [CITATION NEEDED].

Use McNemar's test for paired binary outcomes between two methods, bootstrapped 95% confidence intervals for key metrics, and Cohen's h for effect size. Always report sample sizes, number of runs, and whether error bars show standard deviation or standard error.

Yes, it bundles official templates for ICML 2026, ICLR 2026, NeurIPS 2025, ACL, AAAI 2026, and COLM 2025, each with the correct style file, bibliography style, and example document. Page limits and anonymization rules per venue are documented in the checklists reference.

Use autoreason for mid-tier models on constrained tasks, where the generation-evaluation gap is largest; use single pass for very weak models or template-filling tasks. For frontier models on unconstrained tasks, critique-and-revise or single pass performs better because autoreason suffers synthesis drift.

LLMs hallucinate roughly 40% of citations without programmatic verification and can fabricate results when drafting from raw data without a structured experiment log. Human evaluation design, IRB approval, and final framing decisions still require researcher judgment.