escalate-bench / OPENREVIEW_CHECKLIST.md
kashun
Add Escalate Bench dataset archive and metadata
0cbdbf1
# OpenReview Submission Checklist
## Upload / Form Fields
- Dataset URL: https://huggingface.co/datasets/Spedoske/escalate-bench
- Dataset archive URL: https://huggingface.co/datasets/Spedoske/escalate-bench/resolve/main/escalate-bench-0.1.0.tar.gz
- Code URL: https://huggingface.co/datasets/Spedoske/escalate-bench
- Croissant metadata file: `croissant.json`
- Archive SHA256: ab04189fd79c384810f55230c822770b70b8d2666493dc48b006f6cb69ab2b66
## Before Submission
- Replace all `example.com/anonymous` placeholder URLs.
- Confirm reviewer access works in an incognito browser without exposing reviewer identity.
- Run the official Croissant validator on `croissant.json`.
- Confirm Croissant metadata contains no author names, affiliations, emails, local paths, funding, acknowledgments, private bucket names, or lab server hostnames.
- Confirm the hosted repository/account is anonymous for double-blind review.
- Confirm top-level licensing and upstream benchmark redistribution terms are correct.
- Confirm OpenReview dataset URL and code URL point to reviewer-accessible pages, not local files.
## Optional Local Archive Command
```sh
python scripts/prepare_neurips_submission.py --make-archive --dataset-url DATASET_URL --archive-url ARCHIVE_URL --code-url CODE_URL
```