ModernJEV-Decide-Preview
ModernJEV-Decide-Preview is an experimental ModernBERT-based cross-encoder that scores choices supplied with a conversation, policy, and available tools. It returns a selected label and scores for the alternatives, but does not generate tool arguments or execute tools.
What it does
Use it to rank supplied action or tool choices for an agent, then have your application execute the selected choice.
- Availability
- Experimental private preview; Hugging Face account must have access to the repository.
- Pricing model
- Not established
- Deployment
- Downloadable checkpoint; repository does not expose an inference endpoint.
- Privacy
- Private preview; account access to the repository is required.
- Licence
- Apache-2.0
Capability
Modality
Access
Deployment
Pricing
Target user
Evidence
6 claims- The model scores supplied choices and returns a selected label with scores for alternatives. [1]
ModernJEV-Decide-Preview reads a conversation, its policy and available tools, then scores the choices supplied with your question.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - It does not generate tool arguments or execute tools. [1]
The model does not generate tool arguments or execute tools.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - During the private preview, a Hugging Face account must have access to the repository. [1]
During the private preview, your Hugging Face account must have access to the repository.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - The repository contains model artifacts and does not expose an inference endpoint. [1]
This repository contains model artifacts; it does not expose an inference endpoint.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - Held-out accuracy was 73.40% on next-action decisions and 62.18% on tool-selection decisions. [1]
Held-out results cover **1,158 next-action**, **542 tool-selection**, and **3,652 unseen When2Call** decisions, reported separately.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - The model's When2Call accuracy was 34.39%, below the 35.46% majority reference. [1]
**Transfer limitation:** When2Call is below its majority reference.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
Sources
1
1
Hugging FaceOwner-controlled
https://huggingface.co/MaziyarPanahi/ModernJEV-Decide-Preview
Retrieved 2026-10-27 01:40 UTC
Verified
Versions
2| Version | Effective | Change | Changed fields | State |
|---|---|---|---|---|
| 2 | 2026-10-04 07:56 UTC | material change | capability, access, availability, deployment | Current |
| 1 | 2026-10-03 08:48 UTC | publication | None | Current |
In briefs
2- Backburner leads 25 AI developments in week 40, 2026 Weekly brief 2026-W40
- Microsoft introduces FrogNano-4B-2609 for repository-level coding workflows Daily brief 2026-10-03
Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.