Practise natural English collocations for evaluating, benchmarking, and reporting on machine learning models.
0 / 5 completed
1 / 5
The ML team used a held-out test set to ___ the model before deploying it to production.
Evaluate a model is the standard machine learning collocation for systematically assessing model performance. 'Check around' and 'examine over' are informal. 'Test out about' is not a valid phrase.
2 / 5
Before publishing results, researchers are expected to ___ benchmarks on standardised datasets.
Run benchmarks is the established ML and systems collocation for executing standardised performance tests. 'Perform out' and 'carry away' are not standard. 'Execute along' is informal and imprecise.
3 / 5
The engineering team used precision and recall to ___ accuracy across all model variants.
Measure accuracy is the natural collocation in machine learning evaluation. 'Calculate out' is redundant. 'Count along' doesn't fit this technical context. 'Quantify around' is not idiomatic.
4 / 5
The researchers needed to ___ baselines to determine whether the new architecture was an improvement.
Compare baselines is the standard experimental ML collocation for evaluating a model against reference systems. 'Contrast against all' and 'check beside' are informal. 'Match up about' is not standard.
5 / 5
The paper was required to ___ metrics such as F1 score and AUC-ROC for reproducibility.
Report metrics is the standard academic and industry ML collocation for presenting evaluation results. 'Publish out' and 'declare around' are not idiomatic. 'State along' is too informal for a research context.