Project 4 · Evidence and judgment0/5 completeExperiment Lab
The learner turns configs into reproducible runs, compares them honestly, classifies errors by root cause, and selects evidence.
requires artifacts from: classifier-lab
- L1Reproducible Run Configdefine the contract · ~12 minstart →
- L2Experiment Runnerpreserve an invariant · ~18 minlocked
- L3Ablation + Comparisonmeasure behavior · ~18 minlocked
- L4Error Taxonomybound a failure · ~20 minlocked
- L5Evidence-Based Selectionmake an operational decision · ~24 minlocked
Artifacts this project producesL1 · Reproducible Run Configoutputs: experiment-lab:l1-run-config:artifact
L2 · Experiment Runneroutputs: experiment-lab:l2-experiment-runner:artifact
L3 · Ablation + Comparisonoutputs: experiment-lab:l3-ablation-comparison:artifact
L4 · Error Taxonomyoutputs: experiment-lab:l4-error-taxonomy:artifact
L5 · Evidence-Based Selectionoutputs: experiment-lab:l5-evidence-selection:artifact