フリー問題

問題文

What does the phrase describing a data flywheel in the study guide refer to in an operational pipeline?

選択肢

1. A cache of previous model versions kept for rollback, so a poor release can be undone at once and the loop of improvement keeps turning.
2. Records produced by the deployed model in production are fed back as labeled training data so later versions improve over time.
3. A scheduled job that retrains on the same fixed dataset, so each cycle lifts the scores even while the records stay exactly as they were.
4. The partitions of a distributed dataframe are rotated among the workers between stages so that no single worker becomes a bottleneck, which is what keeps the throughput of a long pipeline constant as the number of stages grows.