Find your percentile
Current cohort: all scored papers
/ 4
Share of papers at or below this mean score
—
Based on scored papers in the current selection, including ties. This is not an acceptance probability. ICLR 2027 · SCORE ATLAS
Where does your
score stand?
Enter your mean score to find your place in the public reviews.
Filter by research area for a more relevant comparison.
40,642 scored papers1–4 score scale
Snapshot collected (UTC) Loading…
Choose an area. Explore the data. ↓01 / EXPLORE
Your lens on the review landscape
Reading the local snapshot…Charts and tables update with your filters
PAPER LEVEL
Equal weight per paperDistribution of paper means
Bin width: 0.25. Missing scores are excluded.
REVIEW LEVEL
Equal weight per scoreHow reviewers score
Counts and shares of individual scores on the original 1–4 scale.
Scores and reviewer confidence
Each cell counts reviews. Darker cells indicate more reviews.
How much do reviewers disagree?
Highest minus lowest score per paper. Requires at least two scores.
RESEARCH LANDSCAPE
Scores across research areas
Based on the current selection. Bars show individual score shares; means give each scored paper equal weight.
Score 1Score 2Score 3Score 4
| Research area | Scored / total | Paper mean score | Median | Individual score mix |
|---|
EVERY PAPER COUNTS
From the numbers to the papers
Open a title to view its OpenReview discussion. Unscored papers appear last. Row numbers reflect the current filters and sort order.
| # | Paper / research area | Status | Mean | Review scores | Mean confidence | Range |
|---|
02 / ABOUT THE SNAPSHOTKnow the source.
Know the source.
Understand the limits.
Data comes from public OpenReview submissions and public official reviews. Submission counts include visible withdrawn and desk-rejected papers in this snapshot, and are not the official final submission total. Means are recomputed from individual scores; missing values are not treated as zero.
The overall paper mean weights each scored paper equally; the review-level mean weights each valid score equally. Disagreement statistics require at least two scores per paper. Scores and confidence use the 1–4 scales observed in this snapshot, without conversion to scales from previous years.
Collection spans a time interval rather than a single instant, and scores may change during discussion. This snapshot is not necessarily a record of initial scores. This page provides no final decisions, acceptance predictions, or score history. Differences from other sites may reflect collection time and scope.
The information layout draws on Paper Copilot for summaries, area filters, and paper tables. Data was collected independently. This project is not affiliated with ICLR, OpenReview, or Paper Copilot.