EMNLP Camera Ready

Use this after the acceptance email. For EMNLP 2026 the notification landed August 20 and camera-ready was due September 20, for a hybrid conference in Budapest October 24-29. Main-conference and Findings papers both produce Anthology camera-readies on the same clock; what differs is the presentation package. Reopen the acceptance email and the current camera-ready instructions before acting — the conference, not ARR, owns this phase.

The three-way merge

A camera-ready is a merge of (a) the reviewed PDF, (b) the edits you promised during the response window, and (c) the de-anonymization diff. Do them in that order and keep the promises auditable:

1. Start from the exact submitted source, not a lab-internal draft that kept evolving.
2. Apply every edit committed in the response, and only contribution-preserving edits beyond that — SACs accepted a specific paper, not a direction.
3. Restore identity last: author block, acknowledgements, funding, and the real repository links that replaced anonymized placeholders.

# Keep the promise trail honest: diff reviewed vs camera-ready source
latexdiff submitted/main.tex camera/main.tex > promise-audit.tex
grep -n "TODO\|anonymous\|anon\.4open\|XXXX" camera/*.tex   # leftover anonymization
grep -rn "github.com\|huggingface.co" camera/*.tex          # links now real and public?

Page arithmetic after acceptance

Recent ACL-family practice grants accepted papers an extra content page to absorb reviewer-requested material — verify the 2026 allowance in the official instructions before using it (unconfirmed at 2026-07-08). Spend any extra space in value order: response-phase commitments first, then the error-analysis or scope clarifications reviewers wanted, and only then new polish. The Limitations section remains mandatory and uncounted; EMNLP 2025 additionally shipped the completed Responsible NLP checklist as a paper appendix — check whether 2026 repeats this and budget the pages.

Anthology metadata is forever

The ACL Anthology entry is generated from what you submit, and errors there outlive any typo in the PDF:

The Anthology is open access with no author fee; there is no APC anywhere in this pipeline — budget goes to registration and travel instead.

Artifact restoration

Acceptance dissolves the anonymity constraint, so the artifact story flips from hiding identity to maximizing findability:

• Move code and data from anonymized archives to permanent public homes; add licenses, model cards or data statements, and the exact configs behind each table.
• Make the repository README point at the paper (Anthology link once live, arXiv meanwhile) and state which commit reproduces the camera-ready numbers.
• If the paper promised data release "upon acceptance," this is the moment that promise becomes checkable — and NLP readers do check.

Findings-specific handling

A Findings camera-ready is a full peer-reviewed publication in its own Anthology volume (20XX.findings-emnlp), on the same deadline. Presentation options for Findings papers have varied by edition and were unconfirmed for 2026 at check time — ask the program chairs rather than assuming a poster exists or does not. Cite your own paper as "Findings of the Association for Computational Linguistics: EMNLP 2026" — miscrediting it to the main proceedings is a small dishonesty reviewers of your next paper will notice.

Presentation and logistics (hybrid, verify specifics)

• EMNLP 2026 is hybrid; the exact in-person versus virtual presentation obligations and registration tiers were 待核实 at check time — confirm which registration category the accepted paper requires and by when.
• Poster and talk formats, session assignment, and video or slide upload deadlines arrive by email between camera-ready and the conference; nominate one author to own that inbox.
• Budapest in late October: visa lead times for many passports exceed the notification-to-conference gap — the visa-chairs contact exists for exactly this, and early outreach beats airport improvisation.

Camera-ready week, ordered

1. Rebuild from the submitted source and apply the promise list (response commitments first) with the diff audit running.
2. Verify the 2026 page allowance and checklist-appendix requirement in the official instructions — both were unconfirmed at pack-check time and both change layout.
3. De-anonymize: author block, acknowledgements, funding, real links; then rerun the leftover-anonymization grep, because "anon.4open.science" surviving into the Anthology is a permanent embarrassment.
4. Each coauthor confirms their own name rendering and affiliation; the submitting author confirms title and abstract fields against the final PDF.
5. Publish the artifact repository, pin the reproducing commit, and test every link in the PDF from a logged-out browser.
6. Upload early enough to survive one round of publication-chair bounce-back (font embedding and metadata complaints are routine).

After Budapest

Two post-conference tasks close the loop cleanly: update the arXiv version (if one exists) to the camera-ready content with the Anthology citation on page one, so the version the field actually downloads credits the venue; and record the paper's final BibTeX from the Anthology itself rather than hand-writing it — Anthology BibTeX is the canonical form your citers will use, and your own reference lists should match it.

Output format

[Camera-ready state] merge pending / promises applied / metadata verified / uploaded
[Promise audit] <response commitment -> edit location>
[Anthology metadata] <names / title / abstract / fonts checked>
[Artifacts] <repo public, licensed, commit-pinned: yes/no per artifact>
[Logistics] <registration, presentation format, visa owner>