DeepReviewer-14B: benchmark results
Provider: Other.
Aggregate Rating 1615 ± 43, rank #886 of 10916 rated models, from 28 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Beyond Rating - Rating MAE | 0.85 | Mean absolute error between the review's overall rating and | 100 |
| Beyond Rating - Strength Recall | 61 | Strength Max-Recall (from 0-1, times 100): share of a single | 100 |
| CoCoReviewBench - Thoroughness | 0.28 | Category-level thoroughness score (how completely the commen | 88.2 |
| CoCoReviewBench - Verifiability | 0.41 | Category-level verifiability score (whether a claim can be c | 88.2 |
| Beyond Rating - Weakness Focus Divergence | 0.07 | KL divergence of the distribution of the generated reviews' | 85.7 |
| CoCoReviewBench | 0.2 | Average category-level review score (the mean of correctness | 76.5 |
| CoCoReviewBench - Category Completeness | 81.98 | Category completeness (%): the average number of review cate | 76.5 |
| CoCoReviewBench - Paper Level | 0.36 | Paper-level review score (the same judge prompt applied once | 70.6 |
| CoCoReviewBench - Grounding | 0.41 | Category-level grounding score (how explicitly a comment poi | 64.7 |
| NovGauge - Pairwise Novelty - Method Verified F1 | 9.7 | Verified F1 (%) | 64.7 |
| NovGauge - Pairwise Novelty - Problem F1 | 73.1 | F1 (%) | 64.7 |
| NovBench - Correctness | 0.64 | Correctness (DistAcc, 0-1): one minus half the L1 distance b | 58.8 |
Interactive version: theaggregate.ai/model?slug=deepreviewer-14b · How It Works · Data refreshed daily, snapshot 2026-11-08.