Jev vs LLMs: typed probabilistic queries with parallel evaluation instead of token-by-token decoding
techNmak · x · 2026-09-22
- The author contrasts LLMs with Jev: LLMs generate autoregressively token by token, even for JSON output.
- Jev instead takes state plus typed questions and returns typed probabilistic answers: Choice (pick from a fixed set), Score (ordered levels), and Noul (probability a statement is true).
- Questions sharing the same state can be evaluated in parallel — the key architectural difference.
More from Models
- User presses Claude Code lead on whether usage resets will actually be banked — Angaisb_ · 2026-09-22
- Anthropic's 'Constitution' Is Just RLAIF, Argues Researcher — Its Real Benefit Is Cutting Human Labeling — WillRinehart · 2026-09-22
- Qwen4 training now: Alibaba teases Max/Flash/Plus variants, 5-10T params for Qwen4.5-5 — ChrisGPT · 2026-09-22
- Reverse-engineered Splash format ports Qwen3.8-27B to Mac, 88 tok/s at 64k on M5 Max — SeveralViolins · 2026-09-22
- Fable can interrupt itself: answer your new prompt, then resume the first — gleech · 2026-09-22
- Did OpenAI Solve the Wrong Navier-Stokes Problem? Experts Cry Loophole — joshgans · 2026-09-22