CommunityJev Models & Reimplementations
Cua-S1 Forms
A 706,048-parameter form-filling scorer from Cua AI: given a UI element and candidate options it returns one probability per option, including check, click, and skip actions.
- Published by
- Community
- Author
- Cua AI
- Last verified
- Oct 2026
Used inJev for Software & DevTools
- Platforms
- Browser
- Capabilities
- Open sourceSelf-hostedRuns locally
Highlights
- 706,048 parameters: a byte-level embedding plus a two-layer transformer encoder with an option-attention head.
- Reported 99.7% on its synthetic form task against 83.6% for the hosted Jev API — author-run, and not reproduced in the library repo.
- The 100% real-world figure rests on 196 decisions across three forms and three PDFs, which the card states plainly.
- A shuffled-context control drops to 37%, the card's own evidence it reads the element rather than option statistics.
Watch out
Trained entirely on synthetic forms and not calibrated with RLCD; it cannot invent a value, only choose among entities a document extractor already found. A pickle checkpoint was removed after a security report — load only the safetensors pair, and downstream code orders execution.
Questions about this entry
Short answers drawn from the linked project or page.
No. It chooses among Label: value entities a document extractor already found, plus check, click, and skip actions; downstream code orders execution as fills, then checkboxes, then one submit click. The card lists unsupervised production use and high-impact actions as out of scope.
A .pt pickle in older revisions was removed after security review (trycua/cua#3977); the cua_s1 package refuses pickle checkpoints by design. The card says to load only the safetensors pair, whose converted output is bit-for-bit identical.
Install the cua-s1 package from a checkout of trycua/cua with uv sync --project libs/cua-s1/python. load_checkpoint validates the safetensors pair's format, version, and SHA-256 tensor signature before returning the model; snapshot_download with allow_patterns for the .safetensors and .json files only.
MIT. The Cua-S1 research family checkpoint ships only as a safetensors pair with a JSON config, and the repository is licensed permissively even though it belongs to the wider Cua-S1 project.
On 10,000 synthetic episodes from cua_s1/synth.py: random forms of 2 to 16 fields from a 55-concept catalogue with synonyms and forced look-alike confuser pairs. Splits are disjoint by exact form field signature, and no real user submissions are included.
Each element has one context truncated to 224 bytes; options are one per document entity plus check, click, and skip, each truncated to 96 bytes. It returns one probability per option in a single forward pass, and the executor picks the argmax.
The author reported 99.95% on the synthetic test and 100% on a 196-decision real demo, but the supporting results document is not part of the trycua/cua repository and libs/cua-s1 does not reproduce them. The best validation metrics sit in cua-s1-forms.json.
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
A 400M-parameter Jev-compatible decision model with a browser build: 0.796 on typed decisions against Laya's 0.766, 17 ms per question on an L40S, and an ONNX file that runs in a tab.
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/OpenDecisionPython8GitHub stars
A Python server in front of llama-server that reads one-token label probabilities and exposes them as POST /v1/systemone.
Jev Models & Reimplementations#community#python#open-models
Community
khimaros/verdictPython
khimaros/verdictPythonA browser WebGPU export of Julia 1: the same 551 MB decision checkpoint, a Rust WASM tokenizer, and a JavaScript API that answers typed questions client-side with no server or key.
Jev Models & Reimplementations#community#open-models#local
Community
An open distillation of Jev 1.13 onto Qwen3.8-27B: a 108.9M-parameter decision block answers typed questions in one prefill pass and adds a POST /v1/decide route beside a full System 2.
Jev Models & Reimplementations#community#open-models#local
Community
From the Jev guides
Original write-ups on what this entry is for.
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
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
Hundreds of repositories carry the system-one tag, but the ecosystem is a dozen ideas in different sizes. A tour of the families, the licenses, and what a download does not include.
Jev Models & Reimplementations#open-models#hugging-face#local
Read article
From the Jev community
Posts from builders shipping with Jev right now.