What metadata should be logged per experiment run?

medium

Answer

$(1) Git commit + diff + repo state.$
$(2) Data version / hash.$
$(3) Full config (all hyperparameters).$
$(4) Environment: Python + libs + CUDA + hardware.$
$(5) Random seeds used.$
$(6) Metrics (train + val per epoch).$
$(7) Artifacts (model + tokenizer + preprocessor).$
$(8) Command line invocation.$
$(9) User + timestamp.$
$(10) System metrics (GPU utilization, memory).$
$Store in tracking system (MLflow / W&B).$
$Enable full reproduction from any run.$

Check yourself — multiple choice

Just metrics
Git+diff + data version + config + env (Python/libs/CUDA/HW) + seeds + metrics + artifacts + command + user/time + system metrics; MLflow/W&B; enables full repro from run
Random
Not real

Experiment metadata: git + data + config + env + seed + metrics + artifacts + command.

#reproducibility

Practise MLOps & Data Quality

215 interview questions in this topic.

Read the lesson Flashcards Take the quiz

Related questions