research-paper-writing

Follow a phased pipeline: define a one-sentence contribution, map each claim to an experiment, run baselines with statistical tests, then draft using the venue's official LaTeX template. Complete the mandatory checklist (NeurIPS has 16 items) before submission to avoid desk rejection.

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 and confirm the cited claim appears in the abstract, marking failures as [CITATION NEEDED].

The templates directory includes official style files for ICML 2026, ICLR 2026, NeurIPS 2025, ACL, AAAI 2026, and COLM 2025, each with its .sty and .bst files plus example documents. Compile with pdflatex and bibtex, or upload to Overleaf.

NeurIPS requires a 16-item checklist covering claims, reproducibility, and ethics; ACL mandates a Limitations section outside the page limit; ICML requires a Broader Impact Statement; ICLR 2026 requires LLM usage disclosure when LLMs contributed significantly to research or writing.

Use pairwise comparisons over Likert scales for reliability, recruit at least 3 annotators per item, and report Krippendorff's alpha for inter-annotator agreement. Document annotator qualifications, compensation, and attention checks, since ACL venues explicitly require these details.

Use autoreason-style iterative refinement for mid-tier models on constrained tasks, where the generation-evaluation gap is largest. For frontier models on unconstrained tasks, single pass or critique-and-revise performs better because synthesis drift prevents convergence.