What does MATILDA-jev accept as input?

A text or JSON state, optionally an image, together with a choice, noul (yes/no), or score question; it then returns calibrated probabilities over the options you supply.

How large is the decision model?

The README describes Maincode/matilda-jev-v1 as a 26B one-pass decision model with a 255-option readout.

Is the runtime the same as the model repo's?

Yes. The maincode_jev_serve/ folder is the bundled runtime from the model repository, Apache-2.0, used unchanged for prompt construction, batching, readout, and stored temperature calibration.

Where do the example images come from?

From linoyts/repo-to-space-example-inputs, which the card notes is CC0; the Space ships two examples, a living-room photo and a sushi platter.

Does the demo support image questions?

Yes — an image can accompany the text or JSON state; the card does not document the vision path beyond that, so no image-specific measurements are available here.

What limits are documented?

None on the Space card. The README covers the model, inputs, runtime provenance, and image credits but publishes no latency, input-size, or accuracy numbers.

Who maintains the demo?

It is published under the hugging-apps account and demonstrates Maincode's matilda-jev-v1 model; the bundled serving code is attributed to the model repository.

CommunityJev Demos & Experiments

MATILDA-jev Demo

A Gradio demo of Maincode's 26B MATILDA-jev-v1 one-pass decision model: feed a text or JSON state, optionally an image, and a choice, noul, or score question to get calibrated probabilities over your options.

See highlights

Open huggingface.co

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

Tagscommunity playground demo gradio multimodal

Works forJev for Classification Jev for Scoring Jev for Verification

Used inJev for Software & DevTools

Platforms: Browser
Capabilities: Open sourceRuns locally

Highlights

Runs Maincode/matilda-jev-v1, a 26B one-pass decision model with a 255-option readout.
Accepts a text or JSON state, optionally an image, plus a choice, noul, or score question, and returns calibrated probabilities over your options.
The maincode_jev_serve/ folder is the Apache-2.0 runtime from the model repo, used unchanged for prompting, batching, readout, and temperature calibration.
Example images come from linoyts/repo-to-space-example-inputs, which the card notes is CC0.

Watch out

The card is short: it confirms the Apache-2.0 runtime is reused unchanged, but the demo's own measurements, latency, and input-size limits are not documented here, and image handling is not described beyond the optional input. Example images are CC0 stock from a separate dataset repository.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

LinkedIn: What multimodal decision models actually change, and what they do not.

LinkedInJevDirectory.org

Related terms

Glossary definitions related to this entry.

Playground

More Jev resources like this

KaLM-Jev Small

A Gradio playground for KaLM-Jev Small structured judgments: choice, score, noul, and mixed JSON questions return probability tables and full responses without generating answer text on shared ZeroGPU.