Training-free inference techniques with decision models

Exact-target audit finding Jev's Choice probabilities are a step function on stated base rates while per-option Noul tracks them closely.

Research decoder that treats Jev as a language model, assembling replies from repeated choices over phrases, words and letters.

Playful experiments using Jev's Choice as a decoder: next character, next BPE token, and grammar-constrained SQL generation.

Reads Jev-style choice, yes/no and score probabilities from masked diffusion language models with shared Yes/No weights.

Research code using the Canopy-Jev decision model to route tokens to experts and extra recursion steps under explicit budgets.

Experiment generating chat replies character by character with Jev Choice over 98 next-character continuations.

Text generation with Jev via cascaded Choice questions: word type, WordNet category, embedding cluster, then the word itself.

Word-level language model whose output layer is Jev: next-word choices, Noul-verified drafted chunks and bits-per-token evaluation.

Character-level language model built from Jev Choice over 28 options, where each option is the full hypothesis text.

Research replicating Jev-style typed decisions with open LLMs via logprobs; Qwen3.5-9B reached 94% with good calibration in ~1 s.

Runnable yardstick that tests Jev's three headline claims on a small local model: typed output, no hallucination, calibration.

JevNext: decodes numbers from Jev by walking a multiway decision tree of interval or digit Choice calls, without logits or training.

Local reproduction of Jev-style inference that reads option-token logits from a small LLM in one pass, with an honest failure log.

Experiment that deliberately misuses Jev as a text generator by asking which symbol comes next and sampling from the returned distribution.

Chat client that makes Jev write by asking Next word? over a 20,000-word dictionary in 255-word buckets plus a run-off.

An LLM writes a task-specific Jev harness, improved with rewards and traces; a Pokemon harness rose from 25% to 75% wins.

Experimental engine that makes an existing LLM do Jev-like classification and scoring by reusing context and building JSON.

Inference engine for Jev-style decision models using segmented forwards with prefix reuse, serving the Jev wire contract.

Experiment that uses Jev as a sampler: DeepSeek proposes top-k next tokens and Jev chooses which comes next and when to stop.

Research fork of Laya that reads typed choice and yes/no decisions from an image plus optional text in one forward pass.

Open decision models that derive calibrated confidence from recurrent looped-layer dynamics instead of RL, with early exit and a Jev API.

Experiment that makes Jev produce prose by asking one choice question per word over a routed domain vocabulary.

Experiment testing whether a local Qwen read via label log-probabilities can replace Jev on synthetic AIOps alert decisions.

Inference-time loop where a System One judge like Jev scores a model's thought paths and picks which reasoning branch to keep.

Open Jev-class decision models on Qwen3.5 with candidate-logit readout; the 4B model reports 80.1% on JevBench's 231 items.

Local Jev-style decision model on Qwen3.5-4B with four stress tests that show where one-pass decisions break.

LogitTrail (formerly Mini Jev): local workbench that reads typed Choice, Noul and Score decisions from a frozen LLM's logits.

Reads typed decisions from one forward pass of PrismML's 1.75-bit Bonsai 27B under llama.cpp, using the OpenJev/SemIf logit method.

Java (llama.cpp via FFM) and Python engines that fill boolean/enum JSON schemas in two forward passes with per-field confidence.

Laptop-scale reimplementation of RL for Calibrated Decisions: parallel constrained decoding trained with GRPO and a Brier reward.