Write ML papers in Typst with instant preview. Anonymous mode, checklist support, export to LaTeX for OpenReview.
Matches official NeurIPS style guidelines exactly.
One-click toggle for blind review submission.
Built-in NeurIPS reproducibility checklist.
Convert to NeurIPS LaTeX for submission.
To convert a NeurIPS paper from LaTeX to Typst, replace the neurips_2024 LaTeX document class with the Typst neurips2024 package and let TypeTeX map the rest — sections, citations, math, figures, and bibliography convert automatically. You keep an export-to-LaTeX path, so you can draft in Typst and still submit through OpenReview.
A minimal NeurIPS paper in the official LaTeX style and the equivalent Typst. The converter produces this automatically — here is exactly what changes.
\documentclass{article}
\usepackage[final]{neurips_2024}
\title{Attention Is All You Need}
\author{%
Jane Doe \\
Dept. of Computer Science \\
Example University \\
\texttt{jane@example.edu}
}
\begin{document}
\maketitle
\begin{abstract}
We propose a simple architecture for
sequence modeling.
\end{abstract}
\section{Introduction}
Transformers rely on
self-attention~\cite{vaswani2017}.
\begin{equation}
\mathcal{L} = -\sum_i y_i \log \hat{y}_i
\end{equation}
\bibliographystyle{plainnat}
\bibliography{references}
\end{document}#import "@preview/neurips2024:0.1.0": neurips
#show: neurips.with(
title: [Attention Is All You Need],
authors: (
(
name: "Jane Doe",
department: "Dept. of Computer Science",
organization: "Example University",
email: "jane@example.edu",
),
),
accepted: true, // camera-ready; omit for blind review
abstract: [
We propose a simple architecture for
sequence modeling.
],
bibliography: bibliography("references.bib"),
)
= Introduction
Transformers rely on
self-attention @vaswani2017.
$ cal(L) = -sum_i y_i log hat(y)_i $| LaTeX | Typst | What it does |
|---|---|---|
| \usepackage{neurips_2024} | #import "@preview/neurips2024": neurips | Load the template |
| [final] / blind option | accepted: true / false | Toggle anonymous review |
| \section{Intro} | = Intro | Headings |
| \cite{key} | @key | Citations |
| \begin{equation} … \end{equation} | $ … $ | Display math |
| \mathcal{L} | cal(L) | Calligraphic |
| \hat{y} | hat(y) | Accents |
| \bibliography{references} | bibliography("references.bib") | Bibliography |
Toggle anonymous mode with one click. Author names and affiliations are automatically hidden for submission, then revealed for camera-ready.
Our template includes the NeurIPS reproducibility checklist. Check items as you complete them:
From draft to OpenReview
Use our NeurIPS template or convert your existing LaTeX paper.
Real-time preview with anonymous mode toggle for reviewing.
Export to LaTeX and upload to OpenReview.
Use our NeurIPS template or convert your existing LaTeX paper to Typst.
Paste your NeurIPS .tex into the TypeTeX LaTeX-to-Typst converter, or import the Overleaf project. It maps the neurips_2024 document class to the Typst neurips2024 package and converts your sections, citations, math, figures, and bibliography. You can then edit with live preview and export back to LaTeX for OpenReview.
Yes. The Typst neurips2024 package exposes an option (accepted: false) that hides author names and affiliations for blind submission, mirroring the [final] toggle in the official LaTeX style. Switch it to accepted: true for the camera-ready version.
Yes. NeurIPS submits through OpenReview, which accepts both LaTeX and PDF. TypeTeX exports a PDF directly and can produce LaTeX source, so you can draft in Typst and still submit in the format the venue expects.
Common math (subscripts, sums, calligraphic and hatted symbols) and \cite references convert automatically — for example \mathcal{L} becomes cal(L) and \cite{vaswani2017} becomes @vaswani2017. Verify a few equations and the reference list after converting, which the migration checklist walks through.