What does Jev-Omni return for a question?

A probability for each supplied option, produced from one forward pass; the README states it does not generate text, so there is no answer prose to read.

What model is the Space running?

A Gemma 4 12B IT base with a 30,000-question fine-tune, according to the README, which points to the akhilaaa3/Jev-Omni model card for details.

Do I need to write a prompt template?

The README describes the interaction simply: give it a situation and a question with options. It does not document a template or input format you must follow.

Where are the accuracy results?

The README defers model card and results to the linked akhilaaa3/Jev-Omni model card; the Space page itself publishes no benchmark numbers.

Why is there no generated explanation?

By design: the model returns per-option probabilities from a single forward pass rather than writing an answer, per the card, so there is no reasoning trace to inspect.

What files make up the Space?

A README, app.py, packages.txt, requirements.txt, and git attributes; the digest lists gradio among the tags, and the app is a plain Gradio interface.

What does the card not tell me?

It does not state the licence, calibration measurements, supported languages, or input limits, so those need checking against the model card before relying on the demo.

CommunityJev Demos & Experiments

Jev-Omni Demo

A Gradio playground for Jev-Omni: give a situation plus a question with options and get one probability per option from a single forward pass, with no generated text, from a Gemma 4 12B IT fine-tune.

See highlights

Open huggingface.co

Format: Playground
Published by: Community
Author: akhilaaa3
Last verified: Oct 2026

Tagscommunity playground demo gradio

Works forJev for Classification Jev for Scoring

Used inJev for Software & DevTools

Platforms: Browser
Capabilities: Open sourceRuns locally

Highlights

Accepts a situation and a question with options, returning a probability for each option from one forward pass with no generated text.
Built on Gemma 4 12B IT with a 30,000-question fine-tune, according to the README.
The README points to the akhilaaa3/Jev-Omni model card for training details and results.
Ships as a Gradio app whose repository holds app.py, requirements.txt, and packages.txt alongside the README.

Watch out

Sparse card: the README states only the one-pass probability behavior, the Gemma 4 12B IT base, and the 30,000-question fine-tune, deferring model details and results to the akhilaaa3/Jev-Omni model card. It documents no licence, calibration measurements, supported languages, or input limits.

Questions about this entry

Short answers drawn from the linked project or page.

Related terms

Glossary definitions related to this entry.

Playground

More Jev resources like this

JEV-9B Decision Demo

An interactive ZeroGPU demo of AutoTrust's JEV-9B: System 1 answers noul, choice, and score questions in one prefill pass with calibrated probabilities, while an Escalate tab hands low-confidence cases to the untouched Qwen3.5-9B base.