CommunityJev Models & Reimplementations
AgentJev-0.6B — huggingface.co
AgentJev-0.6B scores coding completion with a Qwen3-0.6B backbone stripped of its LM head: 57.8% accuracy (606/1048) on the coding set and 87.2% agreement with the teacher on invoice questions.
- Published by
- Community
- Author
- malevrigns
- Last verified
- Oct 2026
This is one of several Jev listings for the same project. The canonical entry is AgentJev-0.6B — malevrigns.
- Languages
- Python
- Capabilities
- Open sourceSelf-hostedRuns locally
Highlights
- Coding accuracy 57.8% (606/1048); recall 57.8%, up from 35.7%; AUROC 0.589, up from 0.516.
- Correct patch ranked first in 348 of 524 cases, up from 305; invoice processing agrees with the teacher on 87.2%.
- Shared-prefix caching cuts a 33,547-token load from 609.65 ms to 298.91 ms with a 0.000508 largest probability gap.
- One FP32 model.safetensors carries the Qwen3-0.6B backbone with the LM head removed plus the candidate head.
- The local server binds 127.0.0.1 and refuses input past its 2,048-token ceiling rather than cropping it.
Watch out
Apache-2.0, built on Qwen3-0.6B. Both headline scores are agreement measures on the card's own sets (executed labels for coding completion, a teacher argmax for invoice questions), and the card warns a win does not say whether a pull request merged or an invoice was paid. RLCD step 12 came from a frozen dev set, and the timing comparison is one warmed load.
Questions about this entry
Short answers drawn from the linked project or page.
A coding-completion one: it reads the task, the code and the checks that were shown, then returns a probability that the work is actually finished. Boolean, Choice and Score questions each come back as a full distribution over the supplied options.
Coding accuracy is agreement with the executed label on the coding-completion set at 57.8% (606/1048); invoice processing is 87.2% agreement with the public teacher argmax on 500 questions. The card warns neither score says whether a pull request merged or an invoice was paid.
Download model.safetensors, wrap it once as a torch checkpoint with torch.save({"state_dict": load_file(src)}), download temperatures.json, then run python -m jev_service.server with the checkpoint, the Qwen/Qwen3-0.6B model path, temperatures and port 8149.
On a fixed load of 64 choice options plus one boolean (33,547 path tokens), the unshared median was 609.65 ms against 298.91 ms shared, cutting backbone token ops from 33,547 to 2,551 with a 0.000508 largest probability gap and the same chosen option.
The server ceiling is 2,048 tokens and input past it is refused, so a long diff is an error instead of a silent crop of the question. The process binds 127.0.0.1 and exposes a workbench on port 8149.
The weights are hosted under the aimeigaoshou account, the citation names malevrigns as author, and the code repository lives at github.com/malevrigns/agent-jev. The release is Apache-2.0 and built on Qwen3-0.6B with the language-model head removed.
Supervised training and RLCD used executed coding pairs where both sides pass the same public checks and the label comes from the checks that were not shown; the frozen development set selected RLCD step 12 before the test split was opened.
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
Hacker News: Decision Index 0.2.1 scores open reproductions of Jev on a frozen suite.
jackbravo4 pointsDecision Index 0.2.1 – 70 open reproductions of TypeSafe Jev's Decision ModelRead the thread on Hacker NewsHacker News: Pointer to a DiffusionGemma plus vLLM tutorial for a Jev-shaped classifier.
1 pointHow to build a Jev-style classifier with DiffusionGemma and vLLMRead the thread on Hacker NewsHacker News: A one-token logprob wrapper that also classifies webcam frames.
allanrbo151 pointsA single function Jev-like wrapper for LLMs, including vision modelsRead the thread on Hacker NewsMore Jev resources like this
The v3 Jev-Style decision model at 2B in GGUF: F16 3.78 GB and Q8_0 2.01 GB match PyTorch FP32 on 100% and 99.7% of parity rows, 73.6% on JevBench v1.4.1 public items.
Jev Models & Reimplementations#community#open-models#local
Community
Mithril's open, Jev-shaped typed-decision model on DeBERTa-v3-large: choice up to 255 options, score and noul from one forward pass, 0.854 in-domain with ECE 0.022, trained on public gold labels only.
Jev Models & Reimplementations#community#open-models#local
Community
A Qwen2.5-1.5B-Instruct fine-tune that emits typed dependency verdicts, safe, caution, or block with a recommendation, for Dependabot and Renovate pipelines, reporting 95.7% F1 on the card's 4,800-update OSV-Bench.
Jev Models & Reimplementations#community#open-models#local
Community
The vision sibling: the same text-trained System 1 head on Qwen3.8-27B's image encoder, taking images zero-shot, ranked first of 20 vision decision models on the community Jev Decision Index 0.3.
Jev Models & Reimplementations#community#open-models#local
Community
1.3kGitHub stars
A local yes/no decision model that speaks Jev's wire format on a laptop CPU, refusing Choice and Score until those question types ship.
Jev Models & Reimplementations#community#python#open-models
Community
feder-cr/jevC++
feder-cr/jevC++58GitHub stars
An Apache-2.0 decision engine with a TypeSafe-compatible endpoint, document answers with source passages, and a ViZDoom demo that picks actions from structured game state.
Jev Models & Reimplementations#community#python#local
Community
deepanwadhwa/OpenDecisionPython
deepanwadhwa/OpenDecisionPythonFrom the Jev guides
Original write-ups on what this entry is for.
Training recipes, looped architectures, and language ports: what the open side of the typed-decision class shipped this month, and how to pick by runtime and calibration instead of the leaderboard.
Jev Models & Reimplementations#open-models#training#local
Read article
Taking a decision checkpoint out of Python: what the GGUF and browser WebGPU exports preserve, what they change, and the parity evidence the model cards publish.
Jev Tools & CLIs#open-models#local#onnx
Read article
Interface, size, license, claims — in that order. A selection guide across the system-one catalog, from a 706K form scorer to 27B Jev students.
Jev Practices & Patterns#open-models#evaluation#benchmarks
Read article
From the Jev community
Posts from builders shipping with Jev right now.