Remote ArcREMOTE ARC
Train and evaluate a small model
Build a local classification or regression workflow with a baseline and an independent evaluation.
With Arc
Build a machine-learning baseline from the dataset I provide. Read the scikit-learn skill, confirm the target and evaluation metric, split data before fitting preprocessing, and avoid leakage. Train an appropriate baseline, evaluate it on held-out data and save reproducible code with the actual metrics and limitations. Check current connections, permissions and dependencies first. Use real tools and continue from observations; claim only results actually observed or verified. Inspect uncertain external effects before any repeat. Report specific missing capabilities.
Arc requirements
A suitable labeled dataset and the local scikit-learn stack.
Arc boundaries
Measured performance applies to the evaluation data, not a guarantee of future predictions. Large language models are outside this workflow.
With Connect: example request
Build a machine-learning baseline from the dataset I provide. Read the scikit-learn skill, confirm the target and evaluation metric, split data before fitting preprocessing, and avoid leakage. Train an appropriate baseline, evaluate it on held-out data and save reproducible code with the actual metrics and limitations.
With Connect: how the work proceeds
Define the target and split data without leakage. Compare a baseline and validate on held-out data.
Connect tools and permissions
read_skill · read_file · write_file · start_process. A suitable labeled dataset and the local scikit-learn stack.
What useful evidence looks like
Reproducible training code, held-out metrics and a clear model/data scope.
Know the boundary
Measured performance applies to the evaluation data, not a guarantee of future predictions. Large language models are outside this workflow.