Jev Pytorch

A reproducible PyTorch stack for cross-sectional factor research: 213 mask-aware factors, cost-aware backtests and auditable reports, plus a preview typed portfolio-decision adapter in the System-One/Jev style.

An open-source, non-autoregressive System One decision engine inspired by TypeSafe AI: it evaluates state context against typed questions in a single forward pass, returning probability distributions and calibrated confidence without generating text.

A small from-scratch alternative to Jev and Laya-style System 1 decision models: a roughly 22M-parameter shared encoder trained with per-example calibrated confidence labels in PyTorch and exported to ONNX for local, sub-second inference.

Independent training toolkit and open weights for System One decision models in the same category as Jev: 17M, 68M, and 400M models answering Choice, Noul, Score, and reranking questions, with Python and ONNX inference and a documented generalization gap to Jev 1.13 on the project's S1MB benchmark.

Jag, a Jev-inspired visual grounding model: a Qwen3.5-0.8B backbone plus a lightweight MLP predicts normalized cxcywh bounding boxes for an image and referring expression in a single forward pass, removing autoregressive coordinate generation from the base model.