Practice free →
HomeGCP ML Engineercloudcomputingmleserving › Which ML model-optimisation technique should an …

Which ML model-optimisation technique should an ML Engineer use to REDUCE inference latency + memory footprint for a trained model — referenced in PMLE §4.2?

AQuantisation (and pruning + distillation) of the trained model
BIncreasing model size before serving
CCloud Memorystore
DCloud DNS
Answer & Solution
Correct answer: A. Quantisation (and pruning + distillation) of the trained model
Quantisation / pruning / distillation shrink the serving model (per PMLE §4.2 simplification techniques). The other options don't optimise.
Solve this in the app — GCP ML Engineer practice & 24k+ MCQs →
Related questions