Skip to content
JevDirectory.orgTraffic count unavailable. View site traffic
API & Models

Option marker

Scoring each answer option at its own marker token, then softmaxing over the question's options so the answer space is defined at request time.

All 164 terms
Category
API & Models
Also known as
—
Related terms
4
Directory entries
4
Docs
—
Added
2026-10-10

Definition

Laya scores every option at its own [MASK] token and softmaxes over that question's options, so new schemas need no retraining. Julia 1, bev-decider, and Gero-4B use related readouts — one branch or slot per option, scored in a single forward pass.

The mechanism explains the limits. Options share a fixed head budget, 192 tokens on Laya's English checkpoint and 256 on the multilingual build, which is why a 77-option question allocates three to four tokens per label and why Laya scores 0.425 on Banking77 where Jev's hosted 255-option Choice reaches 0.870.

Tagsmodelsarchitecture

Related terms

Definitions that connect to this one.

Choice optionOne key in a Choice criteria map, with a description that can be a string, a structured object, or null.PrimitivesPrimitive limitsThe hard caps to design around: 255 Choice options, 2 to 10 Score levels, and a 64k context covering state plus questions.PrimitivesDecision headThe small trainable module added to a frozen backbone that turns hidden states into one probability per option.API & ModelsLogit readoutReading the next-token probability of each declared label, then renormalising, so no answer text is generated.Patterns

From the Jev directory

Laya on Hugging Face
The Hugging Face hub for Laya's open System 1 checkpoints: the 421M English model at the repo root, a 322M multilingual build in a subfolder, and the typed-decisions fine-tune, all Apache-2.0.
Jev Models & Reimplementations#community#open-models#local
Community
Julia 1
Supersonic Labs' 144.3M-parameter decision model on mmBERT-small: choice, score, and noul answers from supplied options in one pass, 73.15% on 2,000 typed decisions, Apache-2.0.
Jev Models & Reimplementations#community#open-models#local
Community
bev-decider 0.4B
A 0.4B Jev-compatible decision model with measured order invariance: 960 random option shuffles moved no probability by more than 1e-5, and bev-decider serve exposes /v1/systemone.
Jev Models & Reimplementations#community#open-models#local
Community
Gero-4B
A 4B scorer rebuilt from Qwen3-4B into a branching cross-encoder: one branch per option, no position bias, up to 256 options, and calibrated probabilities with no text to parse.
Jev Models & Reimplementations#community#open-models#local
Community

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai