Jev directory
CommunityJev Models & Reimplementations
Strands-Decider 2B Hobson v21
A 2B decision model from Strands Agents for agentic workflows: a LoRA adapter on Qwen/Qwen3.5-2B-Base with a readout head that answers noul, choice and score questions with calibrated confidence, reporting 0.7619 on JevBench at 4096 context.
- Reading time
- 2 min
- Published by
- Community
- Author
- Strands Agents
- Last verified
- Oct 2026
- Languages
- Shell
- Capabilities
- Open sourceSelf-hostedRuns locally
Highlights
- LoRA adapter on Qwen/Qwen3.5-2B-Base plus a readout head, released under Apache-2.0 with code, recipe and evals on GitHub.
- v21 is v19's recipe plus checked question paraphrases and distillation from Qwen/Qwen3.5-4B where it agrees with the gold label.
- Reports 0.7619 accuracy (176/231) on JevBench served at 4096, 0.865 on contractnli and 0.882 on musique.
- The example output returns noul 0.875, choice billing at 0.837 confidence and score 1.07 with 0.603 confidence.
- Install with pip install strands-decider; ask or serve over local HTTP with --device cuda, mps or cpu.
Quickstart
bash
strands-decider ask StrandsAgents/strands-decider-2B-hobson-v21 \
--state "Help! My payouts have been failing for 3 days! " \
--choice "Which team should handle this?=billing,sales,retail" \
--noul "Does this convey urgency?" \
--score "How frustrated is the writer?=calm,frustrated,depressed"Watch out
Card reviewed 2026-10-17; the JevBench 0.7619 and the hobson-internal accuracies (contractnli 0.865, musique 0.886, adequacy and boardgame figures) are self-reported on the card with varying sample sizes and no third-party reproduction. The bundled HTTP server binds to 127.0.0.1 without authentication and is intended for local experiments.
Questions about this entry
Short answers drawn from the linked project or page.
A small decision model for agentic AI from Strands Agents. Rather than generating text, it picks between options and rates items on a scale, and every answer carries a calibrated confidence. It targets model routing, tool selection, argument checking, triage, guardrails and evaluations.
It accepts typed questions about a state: noul (yes/no), choice (one of N options) and score (a level on an ordered scale), and returns a probability per option from a readout head instead of generated text, matching the shape of Jev's typed-question contract.
It is a LoRA adapter on Qwen/Qwen3.5-2B-Base with a small readout head. Install the strands-decider package, then run ask with --device cuda, mps or cpu; the base weights download from the Hub at first use.
The model index lists 0.7619 accuracy (176/231) on JevBench public served at 4096 tokens, plus internal evaluations including 0.865 on contractnli, 0.882 on musique and 0.65 on held-out short tasks (n=6000); all are self-reported.
It is one of six released seeds and was picked by a rule written before the seeds were ranked. The recipe is v19 plus checked question paraphrases and distillation from Qwen/Qwen3.5-4B where that model agrees with the gold label.
Yes, via an optional vision extra: pip install "strands-decider[vision]" (transformers 5.18 or later) adds serve --vision or ask --image FILE. An HTTP server mode exists too, binding to 127.0.0.1 with no authentication.
Apache-2.0, the same licence as the code repository, which holds the training recipe, data inventory and evaluations. The card notes the seed was chosen by a pre-written rule rather than by ranking the six releases.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
Hacker News: ASTGL replay chose local Laya routing over hosted Jev on one Mac Studio setup.
Jmeg8r15 pointsDoes an open-weight decision model beat a hosted one? Jev vs. LayaRead the thread on Hacker NewsReddit: A walkthrough of routing between models with Jev.
X: Standard One SH (StandardThinking)
Facebook: Intake Router keeps diagnosis out of the answer space at the front desk.
FacebookClinicnotes
More Jev resources like this
The Ollama build of iapp's OpenThai-SystemOne: a 0.8B Thai and English decision model whose v0.3 weights are fine-tuned for Ollama's /v1/systemone letter readout, reporting a 74.3 macro across 13 public subsets for Q8_0.
Jev Models & Reimplementations#community#open-models#local
Community
A local daemon that pulls open decision models and serves them on TypeSafe's /v1/systemone, so a Jev client switches hosts with TYPESAFE_BASE_URL.
Jev Models & Reimplementations#community#rust#open-models
Community
ollaya-dev/ollayaRust
ollaya-dev/ollayaRustA 27B decision model from torchcast-ai that answers choice, noul and score questions together in one forward pass over the TypeSafe-compatible /v1/systemone format; the card reports Decision Index 0.2.1 of 65.00, +7.09 over Jev 1.13, at 33 ms per yes/no question.
Jev Models & Reimplementations#community#open-models#local
Community
The 9B entry in vLLM Semantic Router's Decision 2.0 family: a 7.94B decision model with a 16,384-token context that answers choice, yes/no and score questions in one pass, reporting 68.1 JevArena and 18.4 ms median single-question latency.
Jev Models & Reimplementations#community#open-models#local
Community
The 4B model in vLLM Semantic Router's Decision 2.0 family: a 4.21B fine-tune of Qwen3.5-4B-Base that answers choice, yes/no and score questions in one forward pass, with a self-reported JevArena 63.6 and a 12.9 ms median per request.
Jev Models & Reimplementations#community#open-models#local
Community
A 27B decision model from StartLux Labs that takes text, JSON or images up to 262,144 tokens and returns a probability per option through the TypeSafe /v1/systemone format, reporting Decision Index 0.2.1 63.88 and 208 of 231 on JevBench public.
Jev Models & Reimplementations#community#open-models#local
Community
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.