cert
cert-1
cert-2

Pass Google Professional Machine Learning Engineer Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers to Pass the Actual Exam!
30 Days Free Updates, Instant Download!

cert-5
cert-6
Professional Machine Learning Engineer Exam - Verified By Experts
Professional Machine Learning Engineer Premium Bundle
$29.98

Professional Machine Learning Engineer Premium Bundle

$79.99
$109.97
  • Premium File 402 Questions & Answers. Last update: Nov 13, 2026
  • Training Course 69 Video Lectures
  • Study Guide 376 Pages
 
$109.97
$79.99
accept 47 downloads in last 7 days
block-screenshots
Purchase Google Professional Machine Learning Engineer Exam Training Products Individually
Google Professional Machine Learning Engineer Premium File
Premium File 402 Q&A
$76.99$69.99
Google Professional Machine Learning Engineer Video Training Course
Training Course 69 Lectures
$21.99 $19.99
Google Professional Machine Learning Engineer PDF Study Guide
Study Guide 376 Pages
$21.99 $19.99
Google Professional Machine Learning Engineer Practice Test 26

Question 250

During model-training validation on a Google Cloud experimental project, the following condition occurs: An engineer runs an unbounded grid across learning rates, depths and regularization values, using expensive accelerators without a metric-driven stopping rule. The change should preserve a meaningful independent model-quality comparison. Which diagnostic check should be performed FIRST to isolate the described cause?

  1. Review parameter ranges, total trial count, training budget and the metric used to rank completed experiments.
  2. Compare training and validation curves, learning-rate schedule and the checkpoint corresponding to minimum validation loss.
  3. Inspect region-zone capacity, quota limits, accelerator requirements and scheduling error events.
  4. Inspect shutdown events, checkpoint directory, durable artifact transfers and restoration logic after simulated interruptions.

Explanation

Correct Answer

A:

Review parameter ranges, total trial count, training budget and the metric used to rank completed experiments. This inspects evidence relevant to hyperparameter search budget limit. Search quality depends on comparable objectives and resource limits, not exhaustive combinations at any cost. After identifying the cause, the matching correction is: Use bounded hyperparameter tuning with an explicit objective metric, search space, concurrency and early stopping.

Incorrect Answers

B:

Compare training and validation curves, learning-rate schedule and the checkpoint corresponding to minimum validation loss. This examines early stopping on validation divergence rather than the specific mechanism reported here. It is not the decisive first check for hyperparameter search budget limit. The relevant evidence is: Review parameter ranges, total trial count, training budget and the metric used to rank completed experiments.

C:

Inspect region-zone capacity, quota limits, accelerator requirements and scheduling error events. This examines accelerator quota and regional availability rather than the specific mechanism reported here. It is not the decisive first check for hyperparameter search budget limit. The relevant evidence is: Review parameter ranges, total trial count, training budget and the metric used to rank completed experiments.

D:

Inspect shutdown events, checkpoint directory, durable artifact transfers and restoration logic after simulated interruptions. This examines training checkpoint persistence across interruptions rather than the specific mechanism reported here. It is not the decisive first check for hyperparameter search budget limit. The relevant evidence is: Review parameter ranges, total trial count, training budget and the metric used to rank completed experiments.

Question 251

During an independent acceptance rehearsal of a customer-facing AI capability, the following condition occurs: A small operations team with labeled tabular defect records and no custom training-code requirement begins maintaining a complex self-hosted neural framework. The owners require measurable quality and controlled operating cost. Which TWO selections represent the correct investigation and remediation together? Select TWO.

  1. Evaluate Agent Platform AutoML tabular training against a simple baseline and select it if quality and operating cost justify the managed path.
  2. Review row volume, model location, permissions and the current SQL report dependency on the local scoring script.
  3. Evaluate Cloud Vision API for supported generic image detection and use custom training only for missing domain-specific classes.
  4. Train an appropriately scaled BigQuery ML KMEANS model and assess cluster cohesion, size and business interpretability.
  5. Inspect label availability, feature types, tuning needs, governance requirements, and supported AutoML model interfaces.

Explanation

Correct Answers

A:

Evaluate Agent Platform AutoML tabular training against a simple baseline and select it if quality and operating cost justify the managed path. This is the targeted corrective response for automl tabular supervised model selection. Managed AutoML can train supported supervised tabular tasks without requiring custom neural-network code. Verify the diagnosis using: Inspect label availability, feature types, tuning needs, governance requirements, and supported AutoML model interfaces. Confirm success using: A trained candidate reaches the agreed quality threshold without maintaining unsupported bespoke training infrastructure.

E:

Inspect label availability, feature types, tuning needs, governance requirements, and supported AutoML model interfaces. This inspects evidence relevant to automl tabular supervised model selection. Managed AutoML can train supported supervised tabular tasks without requiring custom neural-network code. After identifying the cause, the matching correction is: Evaluate Agent Platform AutoML tabular training against a simple baseline and select it if quality and operating cost justify the managed path.

Incorrect Answers

B:

Review row volume, model location, permissions and the current SQL report dependency on the local scoring script. This is an action or diagnostic relevant to bigquery ml sql batch prediction integration, not one of the required two steps. In this scenario verify: Inspect label availability, feature types, tuning needs, governance requirements, and supported AutoML model interfaces. Then correct: Evaluate Agent Platform AutoML tabular training against a simple baseline and select it if quality and operating cost justify the managed path.

C:

Evaluate Cloud Vision API for supported generic image detection and use custom training only for missing domain-specific classes. This is an action or diagnostic relevant to vision api pretrained image labels versus custom detection, not one of the required two steps. In this scenario verify: Inspect label availability, feature types, tuning needs, governance requirements, and supported AutoML model interfaces. Then correct: Evaluate Agent Platform AutoML tabular training against a simple baseline and select it if quality and operating cost justify the managed path.

D:

Train an appropriately scaled BigQuery ML KMEANS model and assess cluster cohesion, size and business interpretability. This is an action or diagnostic relevant to bigquery ml kmeans customer segmentation, not one of the required two steps. In this scenario verify: Inspect label availability, feature types, tuning needs, governance requirements, and supported AutoML model interfaces. Then correct: Evaluate Agent Platform AutoML tabular training against a simple baseline and select it if quality and operating cost justify the managed path.

Question 252

During model-training validation on a Google Cloud experimental project, the following condition occurs: An eight-GPU job scales only slightly beyond two GPUs because each small batch requires excessive cross-worker gradient synchronization. The change should preserve a meaningful independent model-quality comparison. Which targeted engineering change most directly addresses the failure?

  1. Request justified quota or select an available compatible accelerator and region without compromising data residency controls.
  2. Evaluate supported tensor or pipeline model parallelism to split the model across compatible accelerators.
  3. Prefer suitable data parallel replicas with synchronized gradient aggregation, then benchmark scale efficiency.
  4. Tune supported distributed batch sizing, communication strategy or placement after measuring the bottleneck.

Explanation

Correct Answer

D:

Tune supported distributed batch sizing, communication strategy or placement after measuring the bottleneck. This is the targeted corrective response for distributed gradient communication bottleneck. Distributed speedups depend on communication-to-compute ratio, not merely the number of accelerators allocated. Verify the diagnosis using: Inspect all-reduce time, network bandwidth, effective batch size, worker stragglers and samples per second. Confirm success using: Scale efficiency improves and model quality remains stable without synchronization stalls dominating every step.

Incorrect Answers

A:

Request justified quota or select an available compatible accelerator and region without compromising data residency controls. This action addresses accelerator quota and regional availability, not the failure in the described distributed gradient communication bottleneck case. The targeted correction is: Tune supported distributed batch sizing, communication strategy or placement after measuring the bottleneck.

B:

Evaluate supported tensor or pipeline model parallelism to split the model across compatible accelerators. This action addresses model parallelism for memory-bound transformer, not the failure in the described distributed gradient communication bottleneck case. The targeted correction is: Tune supported distributed batch sizing, communication strategy or placement after measuring the bottleneck.

C:

Prefer suitable data parallel replicas with synchronized gradient aggregation, then benchmark scale efficiency. This action addresses distributed data parallel gradient synchronization, not the failure in the described distributed gradient communication bottleneck case. The targeted correction is: Tune supported distributed batch sizing, communication strategy or placement after measuring the bottleneck.

Question 253

As the team tests a new governed AI release against adversarial and normal data, the following condition occurs: A model keeps similar average scores after an upstream data migration, but feature importance shifts from stable signals toward a suspicious proxy variable. The fix must retain legitimate model usefulness and appropriate controls. Which targeted engineering change most directly addresses the failure?

  1. Present feature attributions as model-decision associations and use proper causal analysis before making intervention claims.
  2. Evaluate and address documented subgroup disparities using appropriate fairness metrics and approved data/model changes.
  3. Version a single feature contract and add automated parity tests to the preprocessing pipeline and serving workflow.
  4. Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation.

Explanation

Correct Answer

D:

Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation. This is the targeted corrective response for feature attribution drift after deployment. Feature attribution drift can indicate changing model reliance even when overall score distributions remain stable. Verify the diagnosis using: Examine explanation distributions, feature generation changes, protected attribute proxies and prediction outcomes by cohort. Confirm success using: Attribution changes are explained or remediated and decision drivers remain stable on known reference cases.

Incorrect Answers

A:

Present feature attributions as model-decision associations and use proper causal analysis before making intervention claims. This action addresses explainability versus causal claim, not the failure in the described feature attribution drift after deployment case. The targeted correction is: Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation.

B:

Evaluate and address documented subgroup disparities using appropriate fairness metrics and approved data/model changes. This action addresses model feature bias by subgroup, not the failure in the described feature attribution drift after deployment case. The targeted correction is: Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation.

C:

Version a single feature contract and add automated parity tests to the preprocessing pipeline and serving workflow. This action addresses training-serving feature skew detection, not the failure in the described feature attribution drift after deployment case. The targeted correction is: Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation.

Question 254

As the team tests a new governed AI release against adversarial and normal data, the following condition occurs: An LLM summarizer produces fluent answers containing unsupported policy details, but the team measures only response latency and vocabulary similarity. The fix must retain legitimate model usefulness and appropriate controls. Which evidence most directly establishes what is wrong with this AI or ML workflow?

  1. Inspect cohort conversion metrics, assignment validity, minimum sample requirements and the release guardrail configuration.
  2. Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions.
  3. Inspect retrieved evidence, claim-level factual consistency, reference answers and human review samples.
  4. Compare live and offline feature SQL, window definitions, sample entity values and prediction-time cutoffs.

Explanation

Correct Answer

C:

Inspect retrieved evidence, claim-level factual consistency, reference answers and human review samples. This inspects evidence relevant to generative ai factual grounding evaluation. Fluent text or semantic similarity is not proof an answer is supported by the retrieved source material. After identifying the cause, the matching correction is: Add groundedness and factual correctness checks to a representative gen AI evaluation suite and tune retrieval or prompting accordingly.

Incorrect Answers

A:

Inspect cohort conversion metrics, assignment validity, minimum sample requirements and the release guardrail configuration. This examines production incident rollback based on metrics rather than the specific mechanism reported here. It is not the decisive first check for generative ai factual grounding evaluation. The relevant evidence is: Inspect retrieved evidence, claim-level factual consistency, reference answers and human review samples.

B:

Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions. This examines prompt injection through retrieved documents rather than the specific mechanism reported here. It is not the decisive first check for generative ai factual grounding evaluation. The relevant evidence is: Inspect retrieved evidence, claim-level factual consistency, reference answers and human review samples.

D:

Compare live and offline feature SQL, window definitions, sample entity values and prediction-time cutoffs. This examines training-serving feature skew detection rather than the specific mechanism reported here. It is not the decisive first check for generative ai factual grounding evaluation. The relevant evidence is: Inspect retrieved evidence, claim-level factual consistency, reference answers and human review samples.

Question 255

During a regulated MLOps pipeline deployment, the following condition occurs: A continuous-training pipeline reads a mutable all-data table each run and never logs which rows produced its newest artifact. The workflow must be repeatable and protect the last approved production model. Which TWO selections represent the correct investigation and remediation together? Select TWO.

  1. Inspect snapshot identifiers, SQL filters, table partitions and the model version's recorded input manifest.
  2. Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger.
  3. Create immutable versioned input snapshots or partition manifests and record them in training metadata.
  4. Inspect training data freshness, drift signals, label availability, business cadence and automated promotion rules.
  5. Make the load idempotent with keyed merge or overwrite-per-version semantics and safe retry handling.

Explanation

Correct Answers

A:

Inspect snapshot identifiers, SQL filters, table partitions and the model version's recorded input manifest. This inspects evidence relevant to data version reproducibility in ct jobs. Model retraining reproducibility requires stable inputs rather than unversioned reads of an evolving table. After identifying the cause, the matching correction is: Create immutable versioned input snapshots or partition manifests and record them in training metadata.

C:

Create immutable versioned input snapshots or partition manifests and record them in training metadata. This is the targeted corrective response for data version reproducibility in ct jobs. Model retraining reproducibility requires stable inputs rather than unversioned reads of an evolving table. Verify the diagnosis using: Inspect snapshot identifiers, SQL filters, table partitions and the model version's recorded input manifest. Confirm success using: A previous model can be rebuilt from its exact documented training dataset and code revision.

Incorrect Answers

B:

Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger. This is an action or diagnostic relevant to eventbridge equivalent event trigger on storage, not one of the required two steps. In this scenario verify: Inspect snapshot identifiers, SQL filters, table partitions and the model version's recorded input manifest. Then correct: Create immutable versioned input snapshots or partition manifests and record them in training metadata.

D:

Inspect training data freshness, drift signals, label availability, business cadence and automated promotion rules. This is an action or diagnostic relevant to data drift triggered retraining policy, not one of the required two steps. In this scenario verify: Inspect snapshot identifiers, SQL filters, table partitions and the model version's recorded input manifest. Then correct: Create immutable versioned input snapshots or partition manifests and record them in training metadata.

E:

Make the load idempotent with keyed merge or overwrite-per-version semantics and safe retry handling. This is an action or diagnostic relevant to airflow task idempotency on retry, not one of the required two steps. In this scenario verify: Inspect snapshot identifiers, SQL filters, table partitions and the model version's recorded input manifest. Then correct: Create immutable versioned input snapshots or partition manifests and record them in training metadata.

Question 256

During a staged restoration test at a secondary serving deployment, the following condition occurs: A model server scales down to zero expensive workers overnight but cannot meet the first morning burst because loading weights takes several minutes. The team must prove an actual user-visible improvement without weakening access controls. Which targeted engineering change most directly addresses the failure?

  1. Roll out a small canary percentage first, monitor error and quality guardrails and expand only after acceptance.
  2. Schedule and monitor approved Feature Store synchronization at the needed frequency, with stale-feature handling.
  3. Set a justified warm minimum or scheduled capacity and balance initialization overhead against idle cost.
  4. Restrict invoke access to the necessary principal and endpoint, using appropriate short-lived authentication.

Explanation

Correct Answer

C:

Set a justified warm minimum or scheduled capacity and balance initialization overhead against idle cost. This is the targeted corrective response for autoscaling minimum capacity versus cold load. Reactive scaling cannot instantly serve requests when cold model initialization is long. Verify the diagnosis using: Inspect idle cost, model initialization time, min replicas, burst arrival, queueing and request deadlines. Confirm success using: The morning surge meets availability and p95 latency objectives while off-peak cost remains documented.

Incorrect Answers

A:

Roll out a small canary percentage first, monitor error and quality guardrails and expand only after acceptance. This action addresses canary gradual traffic rollout, not the failure in the described autoscaling minimum capacity versus cold load case. The targeted correction is: Set a justified warm minimum or scheduled capacity and balance initialization overhead against idle cost.

B:

Schedule and monitor approved Feature Store synchronization at the needed frequency, with stale-feature handling. This action addresses feature store freshness and online sync, not the failure in the described autoscaling minimum capacity versus cold load case. The targeted correction is: Set a justified warm minimum or scheduled capacity and balance initialization overhead against idle cost.

D:

Restrict invoke access to the necessary principal and endpoint, using appropriate short-lived authentication. This action addresses public endpoint iam oversharing, not the failure in the described autoscaling minimum capacity versus cold load case. The targeted correction is: Set a justified warm minimum or scheduled capacity and balance initialization overhead against idle cost.

Question 257

During a controlled replay of the data and model handoff to a second engineering team, the following condition occurs: A churn model joins historical feature rows to today's latest cancellation status, exposing a future outcome to prior prediction dates. The handoff must preserve privacy, lineage and testable behavior. Which TWO complementary choices jointly establish the evidence and appropriate correction? Select TWO.

  1. Inspect the labeling payload schema, data classification, annotator roles and fields needed to complete the labeling task.
  2. Use a narrowly scoped service identity or approved user access, least-privilege dataset views and separated development data.
  3. Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags.
  4. Revoke and rotate the exposed key and use managed workload identity or approved short-lived credentials with least privilege.
  5. Rebuild offline training data with point-in-time eligible joins and remove features available only after each prediction moment.

Explanation

Correct Answers

C:

Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags. This inspects evidence relevant to point-in-time feature leakage prevention. The globally latest value may contain future information that could not have informed the original prediction. After identifying the cause, the matching correction is: Rebuild offline training data with point-in-time eligible joins and remove features available only after each prediction moment.

E:

Rebuild offline training data with point-in-time eligible joins and remove features available only after each prediction moment. This is the targeted corrective response for point-in-time feature leakage prevention. The globally latest value may contain future information that could not have informed the original prediction. Verify the diagnosis using: Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags. Confirm success using: For every row the feature version predates its prediction cutoff and prospective holdout quality is realistic.

Incorrect Answers

A:

Inspect the labeling payload schema, data classification, annotator roles and fields needed to complete the labeling task. This is an action or diagnostic relevant to data minimization for labeling inputs, not one of the required two steps. In this scenario verify: Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags. Then correct: Rebuild offline training data with point-in-time eligible joins and remove features available only after each prediction moment.

B:

Use a narrowly scoped service identity or approved user access, least-privilege dataset views and separated development data. This is an action or diagnostic relevant to colab enterprise notebook data access scope, not one of the required two steps. In this scenario verify: Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags. Then correct: Rebuild offline training data with point-in-time eligible joins and remove features available only after each prediction moment.

D:

Revoke and rotate the exposed key and use managed workload identity or approved short-lived credentials with least privilege. This is an action or diagnostic relevant to workbench credentials embedded in repository, not one of the required two steps. In this scenario verify: Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags. Then correct: Rebuild offline training data with point-in-time eligible joins and remove features available only after each prediction moment.

Question 258

During a controlled replay of the data and model handoff to a second engineering team, the following condition occurs: A streaming feature pipeline aggregates transactions by processing arrival time, causing delayed events to land in the wrong customer-hour bucket. The handoff must preserve privacy, lineage and testable behavior. Which TWO selections form an appropriate before-and-after verification pair? Select TWO.

  1. Candidate scores can be reproduced on the same held-out data and their differences are attributable to model changes.
  2. Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer.
  3. The scheduled query produces equivalent daily features with less service orchestration and within the run window.
  4. Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract.
  5. Compare source event timestamps, processing timestamps, watermark settings, allowed lateness and actual window outputs from a replay.

Explanation

Correct Answers

D:

Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract. This is specific acceptance evidence for dataflow event-time window correctness. Arrival time and event time are different; event-time windows with lateness rules preserve business-time meaning. It confirms the controlled correction: Configure event-time windowing with an appropriate watermark and explicit late-data correction policy in the Dataflow pipeline.

E:

Compare source event timestamps, processing timestamps, watermark settings, allowed lateness and actual window outputs from a replay. This inspects evidence relevant to dataflow event-time window correctness. Arrival time and event time are different; event-time windows with lateness rules preserve business-time meaning. After identifying the cause, the matching correction is: Configure event-time windowing with an appropriate watermark and explicit late-data correction policy in the Dataflow pipeline.

Incorrect Answers

A:

Candidate scores can be reproduced on the same held-out data and their differences are attributable to model changes. This relates to experiment metric comparison with mismatched splits but does not establish the requested before-and-after evidence. The actual diagnostic is: Compare source event timestamps, processing timestamps, watermark settings, allowed lateness and actual window outputs from a replay. The specific acceptance result is: Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract.

B:

Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer. This relates to feature store online and offline entity alignment but does not establish the requested before-and-after evidence. The actual diagnostic is: Compare source event timestamps, processing timestamps, watermark settings, allowed lateness and actual window outputs from a replay. The specific acceptance result is: Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract.

C:

The scheduled query produces equivalent daily features with less service orchestration and within the run window. This relates to bigquery versus dataflow preparation selection but does not establish the requested before-and-after evidence. The actual diagnostic is: Compare source event timestamps, processing timestamps, watermark settings, allowed lateness and actual window outputs from a replay. The specific acceptance result is: Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract.

Question 259

During a controlled replay of the data and model handoff to a second engineering team, the following condition occurs: A credit service looks up a feature by account number online but the training table keys records by customer identifier, causing inconsistent histories. The handoff must preserve privacy, lineage and testable behavior. Which TWO selections represent the correct investigation and remediation together? Select TWO.

  1. Inspect the labeling payload schema, data classification, annotator roles and fields needed to complete the labeling task.
  2. Rewrite the filter to permit effective partition pruning while preserving its requested date boundaries.
  3. Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer.
  4. Standardize stable entity identifiers and validated feature definitions across training and online serving pathways.
  5. Refactor hidden state into explicit inputs and functions and add a clean-kernel end-to-end execution gate.

Explanation

Correct Answers

C:

Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer. This inspects evidence relevant to feature store online and offline entity alignment. Feature stores rely on consistent entity keys; incompatible identifiers lead to silent training-serving misalignment. After identifying the cause, the matching correction is: Standardize stable entity identifiers and validated feature definitions across training and online serving pathways.

D:

Standardize stable entity identifiers and validated feature definitions across training and online serving pathways. This is the targeted corrective response for feature store online and offline entity alignment. Feature stores rely on consistent entity keys; incompatible identifiers lead to silent training-serving misalignment. Verify the diagnosis using: Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer. Confirm success using: Known records retrieve corresponding historical and current feature values for the intended entity without mismatched accounts.

Incorrect Answers

A:

Inspect the labeling payload schema, data classification, annotator roles and fields needed to complete the labeling task. This is an action or diagnostic relevant to data minimization for labeling inputs, not one of the required two steps. In this scenario verify: Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer. Then correct: Standardize stable entity identifiers and validated feature definitions across training and online serving pathways.

B:

Rewrite the filter to permit effective partition pruning while preserving its requested date boundaries. This is an action or diagnostic relevant to bigquery partition filter cost optimization, not one of the required two steps. In this scenario verify: Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer. Then correct: Standardize stable entity identifiers and validated feature definitions across training and online serving pathways.

E:

Refactor hidden state into explicit inputs and functions and add a clean-kernel end-to-end execution gate. This is an action or diagnostic relevant to notebook transient state and execution order, not one of the required two steps. In this scenario verify: Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer. Then correct: Standardize stable entity identifiers and validated feature definitions across training and online serving pathways.

Question 260

During model-training validation on a Google Cloud experimental project, the following condition occurs: A GPU training job reports 10% utilization while CPU workers spend most time decoding tiny images from scattered Cloud Storage objects. The change should preserve a meaningful independent model-quality comparison. Which evidence most directly establishes what is wrong with this AI or ML workflow?

  1. Profile data-loader wait time, GCS read latency, prefetch queues and device utilization on a representative epoch.
  2. Inspect output schema, decoding configuration, structured-output support and malformed samples under representative prompts.
  3. Inspect region-zone capacity, quota limits, accelerator requirements and scheduling error events.
  4. Inspect model framework, accelerator kernel support, batch sizes, input pipeline speed and per-step cost.

Explanation

Correct Answer

A:

Profile data-loader wait time, GCS read latency, prefetch queues and device utilization on a representative epoch. This inspects evidence relevant to accelerator starvation from slow input pipeline. An accelerator can sit idle waiting for input even when its own memory and compute capacity are sufficient. After identifying the cause, the matching correction is: Use appropriately sharded data, parallel decode, caching or prefetching within memory limits to feed the accelerator.

Incorrect Answers

B:

Inspect output schema, decoding configuration, structured-output support and malformed samples under representative prompts. This examines constrained generative json outputs rather than the specific mechanism reported here. It is not the decisive first check for accelerator starvation from slow input pipeline. The relevant evidence is: Profile data-loader wait time, GCS read latency, prefetch queues and device utilization on a representative epoch.

C:

Inspect region-zone capacity, quota limits, accelerator requirements and scheduling error events. This examines accelerator quota and regional availability rather than the specific mechanism reported here. It is not the decisive first check for accelerator starvation from slow input pipeline. The relevant evidence is: Profile data-loader wait time, GCS read latency, prefetch queues and device utilization on a representative epoch.

D:

Inspect model framework, accelerator kernel support, batch sizes, input pipeline speed and per-step cost. This examines tpu versus gpu for tensorflow workload rather than the specific mechanism reported here. It is not the decisive first check for accelerator starvation from slow input pipeline. The relevant evidence is: Profile data-loader wait time, GCS read latency, prefetch queues and device utilization on a representative epoch.

Professional Machine Learning Engineer Practice Tests