OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Open Jev

Open Jev is an unofficial, open-source PyTorch research implementation that maps program state to typed probabilistic decisions using a transformer-based architecture.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Open Jev — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Open Jev to inspect or experiment with a typed-decision architecture when you can work with Python and PyTorch and understand that the included model is untrained.

Who it is for

  • ML researchers exploring system-one-style model architectures.
  • Python developers prototyping typed decisions from application or program state.
  • Readers studying transformer-based state encoding and probabilistic readout heads.

Documented uses

  • Estimate whether a statement about supplied state is true.
  • Choose among a runtime-provided set of options.
  • Score state against ordered labels and return a distribution with an expectation.
  • Experiment with probability-distribution targets and consistency training.
Before you choose

Requirements & limits

The repository is public and identifies its license as Apache-2.0. The README recommends Python 3.10 or newer and PyTorch 2.0 or newer.

  • The model has random weights; predictions in the example are random.
  • The included deterministic hash tokenizer is not intended for training a useful model.
  • A Choice readout is structurally limited to options supplied at runtime, but this does not ensure a correct answer.
  • The implementation's architecture and training details are hypotheses, not disclosed specifications of TypeSafe AI's Jev.
  • The repository describes calibration and distribution-shift evaluation as concerns before unattended deployment.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install Open Jev and prepare its documented model example

Clone the repository, install the stated PyTorch dependency, and prepare the example inputs for a local model forward pass.

Before you start

  • Python 3.10 or newer is recommended.
  • Git and Python package installer access.
  • PyTorch 2.0 or newer.

Steps

  1. Clone https://github.com/kyegomez/open-jev and change into the open-jev directory.

  2. Install PyTorch with the command below.

    Documented command
    python -m pip install "torch>=2.0"

  3. Use the README example to construct Jev with JevConfig, create a JSON-like state, and define Noul, Choice, or Score questions.

  4. Follow the README example to call the model in evaluation mode under torch.no_grad() and inspect the returned answers.

Expected outcome

A local setup and the documented code path for requesting typed outputs from supplied program state.

Guide limitations

  • The repository says its weights are random, so example predictions are not useful trained-model results.
  • The hash tokenizer is suitable for running the architecture but not for training a useful model.
  • Typed output constrains output form, not prediction correctness.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
kyegomez/open-jev · Official homepage title · retained in its source language
How we review sourcesSuggest a correction