- The Epistemic Intelligence in Machine Learning (EIML) Workshop at NeurIPS 2026 had not publicly disclosed a page limit for submissions as of August 22, 2026, leaving authors without clear guidance ahead of the August 29 deadline.
- NeurIPS workshops set their own submission rules independently: the “New In ML” workshop at NeurIPS 2026 specifies 2-8 pages, while the related 2025 EurIPS EIML workshop required 4-6 pages excluding references and appendices.
A week before the August 29, 2026, submission deadline, researchers targeting the NeurIPS Epistemic Intelligence in Machine Learning (EIML) Workshop still had no publicly confirmed page limit to work from. The organizers had not posted submission guidelines as of August 22, and at least one researcher reported unanswered inquiries to the workshop committee.
Missing Submission Guidelines
The EIML Workshop’s official channels had not published a page limit for contributed papers as of August 22, 2026. The gap surfaced publicly on Reddit on August 21, where a community member noted the workshop’s website offered no figure and that outreach to organizers had gone unanswered. Eight days from close, that silence is an operational problem.
How Workshop Limits Vary
NeurIPS workshops set their own submission rules, independent of the main conference’s standards, and the spread is wide. The “New In ML” workshop at NeurIPS 2026 accepts papers between 2 and 8 pages, excluding references. The 2025 EurIPS Workshop on Epistemic Intelligence in Machine Learning, a related but distinct event, required non-archival submissions of 4-6 pages, excluding references and appendices. Neither figure is a reliable proxy for what EIML 2026 will accept, but together they mark the plausible range researchers are currently working within. As the NeurIPS 2026 Evaluations track illustrates, individual workshops can diverge substantially in their requirements even within the same conference.
The Practical Problem for Authors
Without a stated limit, authors are making educated guesses. Some are reportedly benchmarking against a 6-page limit from a prior ICML workshop with a similar focus; others are preparing for the 9-page limit associated with the main NeurIPS conference. Either choice carries risk. A paper drafted at nine pages may need significant cutting if a tighter limit is announced days before the deadline. One drafted short may under-serve the research if more space was always available. There is no clean workaround until the organizers publish their guidelines.
