User repository on GitHub
Dohnuts builds small multimodal models for direct decisions. -> System One model
- Forks
- 5
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.Homepage
CommunityJev Tools & CLIs
Dohnuts
Python toolkit plus 0.8B multimodal model that takes a message, document, or image and returns calibrated choice, yes/no, and score answers in one forward pass for local decision workloads.
- Stars
- 36
- Language
- Python
- License
- APACHE-2.0
- Published by
- Community
- Author
- PsiACE
- Last verified
- Oct 2026
About this project
Dohnuts builds small multimodal models for direct decisions. Give the model a message, a document, or an image, and ask it to choose, judge, or score. It returns probabilities in a single forward pass, with multiple questions sharing the same input. That is our take on System 1. The Python toolkit covers training, inference, evaluation, and agent integration. The 0.8B model runs locally on a consumer GPU. Model · Model card · Documentation · Benchmarks.
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- Python
Highlights
- One predict call with choice, noul, and score questions sharing state, including image-aware states.
- 0.8B checkpoint for local consumer-GPU inference with LoRA merge and saved calibration on load.
- End-to-end scripts for data mixture, training, checkpoint selection, calibration, and evaluation.
- Agent integration through the Bub SDK plus install, inference, training, and benchmark docs.
Watch out
Checkpoint loading downloads and caches weights plus the pinned base model, and model weights carry non-commercial research terms separate from the code license. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
Jev is an API model while Dohnuts is a small local model you run on a consumer GPU. The README compares them on quality and latency, with Dohnuts aimed at cheap single-pass local judgments.
Multiple questions share the same input state in one predict call, each with its own distribution, and a PIL image can be added to the state for visual questions.
One workflow prepares the public-data mixture, trains with joint RLCD and cross-entropy, selects and calibrates a checkpoint, and runs evaluations via pdm run python scripts/run_experiment.py.
Code is Apache-2.0 while weights are CC BY-NC-SA for non-commercial research, with checkpoint and training-data restrictions in the model card. Bub SDK integration exposes decisions as one tool.
Use Predictor.from_checkpoint("PsiACE/Dohnuts-0.1.0-0.8B"), then call model.predict with a state and typed questions. The loader downloads and caches the checkpoint and its pinned base model, merges the LoRA, and applies the saved calibration.
pdm run python scripts/run_experiment.py prepares the public-data mixture, trains with joint RLCD and cross-entropy, selects a checkpoint, calibrates it, and runs the evaluations. The training guide covers setup and resuming a run.
It was developed, trained, calibrated, and evaluated on a home PC with one AMD Radeon RX 7900 XTX (24 GB), using hardware they already owned and a $0 additional project budget.
More Jev resources like this
34GitHub stars
Pure-HTTP automation for the TypeSafe register, magic-link login, onboarding, and API-key creation flow, with pluggable temp-mail backends, concurrent batch runs, and key verification exports.
Jev Tools & CLIs#community#awesomejev#catalog-import
Community
2951461586/Jev-Register-ToolPython
2951461586/Jev-Register-ToolPython39GitHub stars
Video tool where Jev scores subtitle segments and code assembles a short highlights cut plus a de-watered full cut with timestamped summaries.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
cclank/jevclipPython
cclank/jevclipPython30GitHub stars
Forkable weekday radar that asks Jev one yes-or-no question per interest over each new arXiv announcement and publishes a page plus RSS feed.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
Eliot5566/JEV-Paper-RadarPython
Eliot5566/JEV-Paper-RadarPython38GitHub stars
Async supervisory layer for Hermes Agent that locks Definitions of Done, proves mechanical criteria locally, and calls Jev only for unresolved semantic checks under a token budget.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
keeltrace/hermes-nervePython
keeltrace/hermes-nervePython15GitHub stars
Two-tier decision engine pairing a 22.6M-parameter MiniLM encoder on the Apple Neural Engine with the on-device Apple Foundation Model as fallback for uncertain inputs.
Jev Tools & CLIs#community#awesomejev#catalog-import
Community
kerryrm/systemANEPython
kerryrm/systemANEPython19GitHub stars
Windows desktop aid that reads the local WeChat database for the foreground chat, strips identifiers, and asks Jev for stage, intent, tension, suggested moves, and a reply review.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
yushen100/wechat-jev-assistantPython
yushen100/wechat-jev-assistantPythonFrom the Jev guides
Original write-ups on what this entry is for.
The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.
Jev Tools & CLIs#gateway#vercel#cloudflare
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
Jev labels a record that already exists. Code moves, flags, or holds it. How the document, mail, and transfer tools keep the action out of the model.
Jev Tools & CLIs#classification#email#security
Read article
From the Jev community
Posts from builders shipping with Jev right now.