Jeva paper: decision-native architecture scales to thousands of candidates, near-perfect null detection
burny_tech · x · 2026-10-20
The arXiv paper "Jeva: System One Models from First Principles" (Qi Li, Xinchao Wang) argues that predicting answers from serialized text ignores the structure of decision problems, and proposes four structural requirements for decision-native models:
- null-aware: distinguish answer absence from uncertainty among candidates
- permutation-invariant: reordering alternatives shouldn't change the decision
- option-scalable: growing candidate sets shouldn't require redefining the output space
- cacheable: request-independent computation over recurring options should be reusable
Jeva uses a shared encoder to represent the request and each option independently, with a compact decision stack refining only the request state against immutable option memory. Symmetric aggregation yields permutation-equivariant probabilities, a separate coverage estimate enables explicit null outcomes, and the design unifies Choice, Null and Score outputs. In evaluations, Jeva achieves near-perfect null detection with no decision flips under any tested permutations, scales to thousands of candidates, and substantially reduces latency.
Related event: Jeva Paper Builds System One Decision Models from First Principles(2 posts)→
More from Research
- mjfit: open-source convex decomposition for tight-clearance parts, 10x faster than SDF — yacineMTB · 2026-10-20
- iSEE world model learns object permanence from video alone, no labels — CSProfKGD · 2026-10-20
- TasteBench tests model research taste on ever-fresh out-of-sample ML tasks — marktenenholtz · 2026-10-20
- "Neurons aren't already clean — interpretability just covers the dunghill with snow" — 1a3orn · 2026-10-20
- Single neuron steering makes Qwen3-8B insist it's an owl — Sauers_ · 2026-10-20
- Q-CTRL: over 99.5% of qubits sit idle in one fault-tolerant Shor's algorithm blueprint — MJBiercuk · 2026-10-20