PrimaCons benchmark exposes cross-primitive inconsistency in Jev decision-making
Xu Liu · hf · 2026-11-03
A Zhejiang University-led paper studies cross-primitive consistency in Jev, where the same semantic state expressed via different query primitives (binary vs multiple-choice) yields inconsistent decisions. The team introduces PrimaCons, a four-axis benchmark preserving query semantics, finds systematic consistency violations across three Jev sizes, and proposes a residual calibrator that reduces inconsistency while boosting decision accuracy. Code: 67L1/PrimaCons-CRPR.
More from Research
- DAMI: dual-axis memory initialization improves long-context retrieval in gated linear attention — SpeedAI-Lab · 2026-11-03
- DR Pixel: open-source page-native deep research agent beats OCR-based pipelines — Berkeley-NLP · 2026-11-03
- Paper rejected with two reports: the AI referee report was excellent, the human one cited themselves — birchlse · 2026-11-03
- Skyfall AI's SKY-001 world model matches GPT-6 Astra CAD accuracy at 150x lower cost — rohanpaul_ai · 2026-11-03
- CU Boulder's Programmable Reality Lab shows five projects at UIST 2026 — ryosuzk · 2026-11-03
- Chunk document pages as horizontal image bands: visual-context retrieval with a dense index — tomaarsen · 2026-11-03