# Jeva 4B decision model by Qi Li and Xinchao Wang (NUS)

> Jeva 4B is an open text decision model from NUS with an explicit none-of-the-above outcome. Read its candidate-scoring design and weight package.

Updated 2026-11-07 (Australia/Adelaide)

## Quick answers

### What is Jeva 4B by Qi Li and Xinchao Wang (NUS)?

[Jeva 4B](<https://decisionmodels.ai/models/jeva>): A text decision model with Qwen3.5-4B backbone, adapter and a separate decision head.

[Jeva 4B facts](<https://decisionmodels.ai/models/jeva>)

[Source](<https://huggingface.co/LIQIIIII/Jeva-4B>)

### Who makes Jeva 4B?

[Jeva 4B](<https://decisionmodels.ai/models/jeva>): Qi Li and Xinchao Wang (NUS)

[Jeva 4B facts](<https://decisionmodels.ai/models/jeva>)

[Source](<https://huggingface.co/LIQIIIII/Jeva-4B>)

By Qi Li and Xinchao Wang (NUS) · Weights upload: 2026-10-09 · OPEN

A text decision model with Qwen3.5-4B backbone, adapter and a separate decision head. It supports choices, yes/no and rubric scores, with an explicit none-of-the-above option. The authors design its candidate readout for option reuse and reordered candidate sets.

## Facts

- Base and size: Qwen3.5-4B
- Inputs: Text with supplied candidate options or scoring rubric
- Licence: Apache-2.0 weights and decision head; base licence retained
- Access: Public backbone, adapter and head.pt; use the Jeva project’s loading guidance

## Claims and sources

- **VENDOR-REPORTED**: The authors describe null-aware, permutation-invariant and cacheable candidate scoring. The checked model card has no numerical benchmark; these are design claims, not a fresh reproduction. [source](<https://huggingface.co/LIQIIIII/Jeva-4B>)

## Links

- [Source: Model card and weights](<https://huggingface.co/LIQIIIII/Jeva-4B>)

## Related projects

No related projects are listed in the registry.