All decision model projects

Chat copilot for Mac, Windows, Android and an iOS keyboard that reads chats, judges with Jev or DeepSeek and drafts replies.

Open Jev-style decision model built on Gemma 3 270M with a scoring head, scoring all options in one forward pass on CPU.

Self-hosted encoder model giving Jev-compatible choice, score and noul answers in one pass, pitched as a better-calibrated Laya successor.

Benchmark testing whether Jev or open Kev can replace an LLM judge in agent guardrail callbacks on 7,925 public examples.

Experimental protocol that checks the evidence behind a worker agent's PASS or BLOCKED report, with Jev answering narrow evidence questions.

33-chapter Chinese tutorial on agentic RL and GRPO, including a chapter on when Jev-style external decision models beat training with RL.

Language and VM for long-running agent programs that pause for approval, resume after restarts and replay, with typed decision cells.

Joke decision model trained to answer wrongly with total confidence: 1.9% accuracy at 0.957 mean confidence, via a Jev-style API.

Paper-only research tool for Polymarket BTC 5-minute Up/Down markets: Jev scores the direction, separate code decides entries.

Paper-only BTC experiment on Cloudflare Workers: code screens candidates every 15 minutes and Jev picks a fixed long policy or no trade.

Study of Jev's choices in multi-armed bandits, testing whether Bayesian instructions and supplied statistics improve its exploration.

Experiments testing whether Jev can spot positive-expected-value dice bets from payout tables, and whether recent wins or losses sway it.

Open clinical text classifiers (0.8B-9B on Qwen3.5) with labels set at inference time, run locally and compared with Jev 1.13.

Deterministic evidence selector for noisy Codex investigations; its earlier Jev subagent router is kept for the record, not recommended.

Reference framework where Jev picks a concrete computer-use action and its execution route from legal, verifiable candidates.

Pre-registered field test of Jev, GLiNER2.5-Decide, Laya and other decision models against small LLMs on fresh September 2026 data.

Research runtime where every agent step is a bounded typed decision, with a paged virtual option space over tools and memory.

Experiment using Jev Noul judgments on accumulated evidence to decide when Zilliz's DeepSearcher agent should stop searching.

Open 4B System I driving model that scores behaviours like cruise, stop, yield and brake from three camera frames each decision.

11 MB CPU-only model that answers choice, noul and score questions with fitted calibration and no LLM at inference.

Official code for the FastJEV paper: training-free reduction of redundant candidate computation in multimodal JEV-style models.

Tiny pre-registered experiment testing whether a confidence-scoring reward (RLCD-like) yields confidences safe to automate on at 0.9.

GoodWatch movie and TV recommendation monorepo, listed for an experiment comparing Jev question designs for attribute scoring.

Research code for an image-and-text System One model that reads option-letter logits from a multimodal LLM, with training and calibration.

Instruction dataset of 119 rows compiled from TypeSafe's public Jev documentation, split by Choice, Noul and Score.

Local server (xks) implementing Jev's /v1/systemone contract on an open-weight LLM, accelerated with Xeon Phi cards.

Bilingual research report on what Jev is, how its claims compare with independent evidence, and the open-source alternatives.

Zero-shot run of Jev 1.13 on all seven LexGLUE legal tasks (23,607 examples) versus GPT-5.6 Luna, open models and fine-tuned BERT.

Six recorded chess experiments with Jev via OpenRouter, candidly showing that unaided Jev still blunders pieces.

Experimental robot loop where Jev chooses the next verified action from perception-generated options for a language goal.