JevAny Choice-Readout Matrix - Transfer Test: leaderboard
231 / 1046 / 2000 / 724 Readouts are run configurations of the same weights: emit only as configuration-qualified rows (pattern like "(Benchmark Sweep Run: …)"), never as the plain JevAny-4B/27B row. Native JevJudge-text 4B here is 58.01% vs 57.87% in the external cohort run: two runs, keep separate. Metric: accuracy.
Metric: accuracy. Source: raw.githubusercontent.com. Saturation forecast: Rough model projection: around 2026. 2 models tracked. Updated 14 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | JevAny-27B | Choice Readout run: Fixed 50-50 stack | 89.2 | |
| 2 | JevAny-4B | Choice Readout run: Tuned stack | 78.01 |
Interactive version: theaggregate.ai/benchmark?slug=jevany-choice-readout-matrix-transfer-test · How It Works · Data refreshed daily, snapshot 2026-11-16.