Dohnuts¶

D’oh! But dohnuts.

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

One message, several decisions¶

Route a support request and check whether it asks for a refund in the same call:

from dohnuts.predictor import Predictor

model = Predictor.from_checkpoint("PsiACE/Dohnuts-0.1.0-0.8B")
result = model.predict(
    {"message": "I was charged twice. Please refund the duplicate."},
    {
        "route": {
            "type": "choice",
            "instructions": "Which team should handle the request?",
            "criteria": ["billing", "technical support", "sales"],
        },
        "refund": {"type": "noul", "instructions": "Is a refund requested?"},
    },
)

print(result["answers"]["route"]["choice"])
print(result["answers"]["refund"]["noul"])

Each answer has its own distribution. Use choice to select a candidate, noul to estimate whether a statement is true, and score to rate something on an ordered scale. Add a PIL image to the state for visual questions.

See installation and inference for setup, checkpoint loading, and response fields. The loader downloads and caches the checkpoint and its pinned base model, then merges LoRA and applies the saved calibration.

Train, evaluate, integrate¶

One workflow prepares the public-data mixture, trains with joint RLCD and cross-entropy, selects a checkpoint, calibrates it, and runs the evaluations:

pdm run python scripts/run_experiment.py

The training guide covers setup and resuming a run. The model card reports quality, latency, and limitations, with comparisons against Jev, Laya multilingual, and Laya Vision. For agents, Bub integration exposes the same interface as one decision tool through the Bub SDK.

Dohnuts 0.1.0 model overview and benchmarks

Built at home¶

We developed, trained, calibrated, and evaluated Dohnuts on a home PC with one AMD Radeon RX 7900 XTX (24 GB). We used hardware we already owned and a $0 additional project budget. The main ingredient was time.

License¶

The code is licensed under Apache-2.0. The model weights are licensed under CC BY-NC-SA 4.0 for non-commercial research. See the model card for checkpoint terms and training-data restrictions.