STOP. READ THIS ENTIRE SKILL.MD BEFORE CALLING ANY ENDPOINT.

Regressor Lab

Iterative regression model benchmarking and development. Trains multiple model architectures against the same dataset, compares on holdout, tunes the winner, and optionally trains a small GPT for prediction rationale.

Quick Start

# Benchmark all architectures on a dataset
./run.sh benchmark data.jsonl --target duration_ms --name timeout-estimator

# Benchmark with specific models
./run.sh benchmark data.jsonl --target duration_ms --models "gbr,xgb,rf,ridge"

# Tune the winner's hyperparameters
./run.sh tune timeout-estimator --trials 50