{"id":17478,"date":"2026-09-21T09:44:55","date_gmt":"2026-09-21T09:44:55","guid":{"rendered":"https:\/\/www.examlabs.com\/certification\/?p=17478"},"modified":"2026-09-21T09:44:55","modified_gmt":"2026-09-21T09:44:55","slug":"microsoft-ai-300-practice-test-questions-and-exam-dumps-part2-q21-40","status":"publish","type":"post","link":"https:\/\/www.examlabs.com\/certification\/microsoft-ai-300-practice-test-questions-and-exam-dumps-part2-q21-40\/","title":{"rendered":"Microsoft AI-300 Practice Test Questions and Exam Dumps Part2 Q21-40"},"content":{"rendered":"

View Full <\/b>Microsoft AI-300 Exam Dumps<\/b><\/a> and Practice Test Dumps<\/b><\/p>\n

 <\/p>\n

Question 21. Which Azure service can be used to orchestrate and automate machine learning workflows in Azure?<\/b><\/p>\n

    \n
  1. <\/b> Azure Machine Learning Pipelines<\/span>
    \n<\/span>2.<\/b> Azure DNS<\/span>
    \n<\/span>3.<\/b> Azure Storage Explorer<\/span>
    \n<\/span>4.<\/b> Azure Bastion<\/span><\/li>\n<\/ol>\n

    Correct Answer: 1. Azure Machine Learning Pipelines<\/b><\/p>\n

    Explanation:<\/b><\/p>\n

    Azure Machine Learning Pipelines provide a structured way to orchestrate repeatable machine learning workflows. A pipeline can contain multiple steps such as data preparation, model training, evaluation, registration, and deployment. This approach is useful in MLOps because individual stages can be automated and reused rather than executed manually each time. Pipelines can also support parameterization, scheduling, and integration with other Azure Machine Learning capabilities. Azure DNS, Storage Explorer, and Azure Bastion serve different infrastructure or administration purposes and are not designed to orchestrate machine learning workflows. Using pipelines improves consistency and makes complex machine learning processes easier to automate and maintain.<\/span><\/p>\n

    Question 22. What is a primary benefit of using reusable components in an Azure Machine Learning workflow?<\/b><\/p>\n

      \n
    1. <\/b> They eliminate the need for source control<\/span>
      \n<\/span>2.<\/b> They make workflow steps reusable and consistent<\/span>
      \n<\/span>3.<\/b> They prevent all model failures<\/span>
      \n<\/span>4.<\/b> They automatically increase model accuracy<\/span><\/li>\n<\/ol>\n

      Correct Answer: 2. They make workflow steps reusable and consistent<\/b><\/p>\n

      Explanation:<\/b><\/p>\n

      Reusable components help standardize common tasks within machine learning workflows. For example, a component can define a data preparation, training, validation, or scoring operation that can be used across multiple pipelines. This reduces duplicated configuration and makes workflows easier to maintain. When a component is updated appropriately, improvements can be applied consistently wherever that component is used. Components do not eliminate the need for source control, guarantee that models will never fail, or automatically improve model accuracy. Their main value is repeatability and maintainability. In an MLOps environment, reusable components also help teams establish standardized processes and reduce differences between development and production workflows.<\/span><\/p>\n

      Question 23. Which practice best supports reproducibility when training machine learning models?<\/b><\/p>\n

        \n
      1. <\/b> Changing the training environment for every run<\/span>
        \n<\/span>2.<\/b> Manually installing dependencies on each machine<\/span>
        \n<\/span>3.<\/b> Versioning code, data references, dependencies, and configuration<\/span>
        \n<\/span>4.<\/b> Removing all training metadata<\/span><\/li>\n<\/ol>\n

        Correct Answer: 3. Versioning code, data references, dependencies, and configuration<\/b><\/p>\n

        Explanation:<\/b><\/p>\n

        Reproducibility requires an organization to retain enough information to recreate a machine learning experiment or production model. Versioning source code, data references, dependency definitions, environment specifications, and relevant configuration provides a reliable record of how a model was created. If only the final model file is stored, reproducing the exact training process may be difficult or impossible. Manually changing environments increases the risk of inconsistencies, while removing metadata makes investigation and recreation harder. In an MLOps workflow, reproducibility supports troubleshooting, auditing, validation, and reliable deployment. It also allows teams to compare model versions and understand what changed between different training runs.<\/span><\/p>\n

        Question 24. Which deployment strategy initially sends a small percentage of production traffic to a new model version?<\/b><\/p>\n

          \n
        1. <\/b> Blue-green deployment<\/span>
          \n<\/span>2.<\/b> Batch processing<\/span>
          \n<\/span>3.<\/b> Shadow deployment<\/span>
          \n<\/span>4.<\/b> Canary deployment<\/span><\/li>\n<\/ol>\n

          Correct Answer: 4. Canary deployment<\/b><\/p>\n

          Explanation:<\/b><\/p>\n

          Canary deployment introduces a new model version to a small portion of production traffic before expanding the rollout. This allows the team to observe important metrics such as prediction quality, latency, error rates, resource consumption, and business-related indicators while limiting the potential impact of problems. If the new model behaves as expected, traffic can gradually increase until the new version serves the full workload. If issues occur, traffic can be redirected to the previous stable version. This makes canary deployment particularly useful for reducing deployment risk. Blue-green deployment uses separate environments, while batch processing concerns how workloads are executed rather than how traffic is progressively shifted.<\/span><\/p>\n

          Question 25. What should be monitored after deploying a machine learning model to production?<\/b><\/p>\n

            \n
          1. <\/b> Only the model’s file size<\/span>
            \n<\/span>2.<\/b> Only the developer’s local environment<\/span>
            \n<\/span>3.<\/b> Latency, errors, data quality, and model performance<\/span>
            \n<\/span>4.<\/b> Only the number of source-code files<\/span><\/li>\n<\/ol>\n

            Correct Answer: 3. Latency, errors, data quality, and model performance<\/b><\/p>\n

            Explanation:<\/b><\/p>\n

            Production model monitoring should cover both technical and machine learning-specific indicators. Technical metrics can include request latency, availability, error rates, CPU usage, memory consumption, and throughput. Machine learning monitoring can include data quality, feature distributions, prediction distributions, data drift, and model performance when ground-truth outcomes become available. Monitoring these areas helps teams identify problems that may not be visible from infrastructure metrics alone. A model can remain operational while its input data changes significantly or its predictive quality declines. Monitoring therefore provides an important feedback mechanism for deciding when investigation, retraining, configuration changes, or rollback may be necessary.<\/span><\/p>\n

            Question 26. Which capability helps an organization compare different versions of a trained model?<\/b><\/p>\n

              \n
            1. <\/b> Model registry<\/span>
              \n<\/span>2.<\/b> Azure DNS<\/span>
              \n<\/span>3.<\/b> Network security group<\/span>
              \n<\/span>4.<\/b> Azure Bastion<\/span><\/li>\n<\/ol>\n

              Correct Answer: 1. Model registry<\/b><\/p>\n

              Explanation:<\/b><\/p>\n

              A model registry provides a centralized mechanism for organizing and managing trained model versions. Teams can register models along with metadata such as version information, descriptions, tags, training details, and other useful properties. This makes it easier to determine which model is currently approved for deployment and which earlier versions are available for comparison or rollback. Model registries are particularly valuable in MLOps because they provide structure around the transition from experimentation to production. Azure DNS, network security groups, and Azure Bastion address networking or administrative requirements and do not provide model lifecycle management. Proper model versioning improves traceability and helps teams maintain controlled deployment processes.<\/span><\/p>\n

              Question 27. Why are automated tests important in an MLOps pipeline?<\/b><\/p>\n

                \n
              1. <\/b> They remove the need for monitoring<\/span>
                \n<\/span>2.<\/b> They guarantee perfect predictions<\/span>
                \n<\/span>3.<\/b> They prevent developers from changing code<\/span>
                \n<\/span>4.<\/b> They identify problems before changes reach production<\/span><\/li>\n<\/ol>\n

                Correct Answer: 4. They identify problems before changes reach production<\/b><\/p>\n

                Explanation:<\/b><\/p>\n

                Automated tests provide an early validation mechanism within an MLOps workflow. They can evaluate source code, data-processing logic, pipeline components, configuration, model behavior, and other artifacts before deployment. Detecting problems earlier is generally less costly and less disruptive than discovering them after production release. Tests may include unit tests, integration tests, data validation checks, model validation tests, and deployment verification tests. Automated testing does not guarantee perfect predictions and does not eliminate the need for production monitoring. Instead, it creates repeatable quality gates that help prevent known classes of errors from moving through the deployment process without detection.<\/span><\/p>\n

                Question 28. Which Azure capability can help securely store secrets required by an application or machine learning workflow?<\/b><\/p>\n

                  \n
                1. <\/b> Azure Key Vault<\/span>
                  \n<\/span>2.<\/b> Azure Monitor<\/span>
                  \n<\/span>3.<\/b> Azure Load Balancer<\/span>
                  \n<\/span>4.<\/b> Azure DNS<\/span><\/li>\n<\/ol>\n

                  Correct Answer: 1. Azure Key Vault<\/b><\/p>\n

                  Explanation:<\/b><\/p>\n

                  Azure Key Vault is designed to securely store and manage sensitive information such as secrets, cryptographic keys, and certificates. In machine learning workflows, applications or automation processes may need credentials, connection strings, certificates, or other sensitive values. Keeping these values in source code or configuration files can expose them to unauthorized users and create security risks. Key Vault provides a centralized mechanism for managing these secrets and controlling access through identity and authorization mechanisms. Azure Monitor focuses on monitoring and diagnostics, while Azure Load Balancer and Azure DNS provide networking-related services. Integrating Key Vault into an MLOps architecture helps reduce the risk of exposing credentials and supports secure automation.<\/span><\/p>\n

                  Question 29. What is the purpose of an approval gate in a production deployment pipeline?<\/b><\/p>\n

                    \n
                  1. <\/b> To automatically delete previous models<\/span>
                    \n<\/span>2.<\/b> To require validation before allowing deployment to continue<\/span>
                    \n<\/span>3.<\/b> To disable monitoring after deployment<\/span>
                    \n<\/span>4.<\/b> To remove version history<\/span><\/li>\n<\/ol>\n

                    Correct Answer: 2. To require validation before allowing deployment to continue<\/b><\/p>\n

                    Explanation:<\/b><\/p>\n

                    An approval gate provides a controlled checkpoint in a deployment workflow. Before a model or application is promoted to a sensitive environment, required validation can be performed against predefined criteria. Depending on organizational requirements, the gate may involve automated quality checks, security validation, model performance thresholds, compliance requirements, or human approval. This helps prevent unvalidated changes from reaching production. Approval gates are especially useful when production deployments have significant operational or business impact. They do not delete previous model versions, disable monitoring, or remove version history. Instead, they establish governance around promotion between environments and help ensure that deployment decisions are based on established requirements.<\/span><\/p>\n

                    Question 30. Which metric is particularly useful for identifying slow response behavior in an online machine learning endpoint?<\/b><\/p>\n

                      \n
                    1. <\/b> Model name<\/span>
                      \n<\/span>2.<\/b> Deployment description<\/span>
                      \n<\/span>3.<\/b> Request latency<\/span>
                      \n<\/span>4.<\/b> Number of source files<\/span><\/li>\n<\/ol>\n

                      Correct Answer: 3. Request latency<\/b><\/p>\n

                      Explanation:<\/b><\/p>\n

                      Request latency measures the time required for an online endpoint to process a request and return a response. It is an important operational metric for machine learning services because users and downstream applications may depend on timely predictions. Increasing latency can indicate resource constraints, inefficient model execution, dependency problems, network issues, or configuration changes. Monitoring latency over time helps teams establish normal performance ranges and identify unusual behavior. Other information, such as a model name or deployment description, may be useful for administration but does not directly measure responsiveness. Combining latency monitoring with error rates, throughput, and resource metrics provides a more complete view of endpoint health.<\/span><\/p>\n

                      Question 31. What is the main purpose of data validation before model training?<\/b><\/p>\n

                        \n
                      1. <\/b> To increase storage capacity<\/span>
                        \n<\/span>2.<\/b> To replace source control<\/span>
                        \n<\/span>3.<\/b> To remove all historical datasets<\/span>
                        \n<\/span>4.<\/b> To detect quality and structural problems in input data<\/span><\/li>\n<\/ol>\n

                        Correct Answer: 4. To detect quality and structural problems in input data<\/b><\/p>\n

                        Explanation:<\/b><\/p>\n

                        Data validation helps identify problems that could negatively affect model training or produce unreliable results. Checks can include missing values, incorrect data types, unexpected ranges, duplicate records, invalid categories, schema changes, and unusual distributions. Detecting these issues before training allows teams to correct or investigate the underlying problem instead of producing a model from defective inputs. Data validation does not replace source control or eliminate the need to retain historical data. It is one part of a broader machine learning quality process. Automated validation can also be integrated into pipelines so that significant data-quality problems prevent an unreliable training run from progressing.<\/span><\/p>\n

                        Question 32. Which approach helps prevent credentials from being embedded directly in machine learning application code?<\/b><\/p>\n

                          \n
                        1. <\/b> Store credentials in Azure Key Vault and access them through managed identities<\/span>
                          \n<\/span>2.<\/b> Add credentials to source-code comments<\/span>
                          \n<\/span>3.<\/b> Save credentials in public documentation<\/span>
                          \n<\/span>4.<\/b> Place passwords directly in pipeline scripts<\/span><\/li>\n<\/ol>\n

                          Correct Answer: 1. Store credentials in Azure Key Vault and access them through managed identities<\/b><\/p>\n

                          Explanation:<\/b><\/p>\n

                          Secrets should not be hard-coded into application source code, pipeline scripts, comments, or documentation. A more secure approach is to store sensitive values in a dedicated secret-management service such as Azure Key Vault and use identity-based access to retrieve them when required. Managed identities can reduce the need to distribute long-lived credentials because Azure resources can authenticate using an identity managed by the platform. This improves security and makes credential rotation easier. It also reduces the chance that secrets will accidentally appear in source-control repositories or deployment logs. Proper secret management is an important part of building secure and maintainable MLOps pipelines.<\/span><\/p>\n

                          Question 33. What does model drift generally describe?<\/b><\/p>\n

                          **1. A change in the physical server location<\/span>
                          \n<\/span>**2. A change that can reduce model effectiveness over time as conditions or data change<\/span>
                          \n<\/span>**3. A decrease in storage capacity<\/span>
                          \n<\/span>**4. A change in the model’s file extension<\/span><\/p>\n

                          Correct Answer: 2. A change that can reduce model effectiveness over time as conditions or data change<\/b><\/p>\n

                          Explanation:<\/b><\/p>\n

                          Model drift refers to changes in the relationship between model inputs, outputs, and the real-world conditions represented by the training data. As production data or business conditions evolve, a model that previously performed well may become less effective. Different types of drift can affect feature distributions, predictions, or the relationship between features and outcomes. Monitoring can help identify these changes and determine whether investigation or retraining is necessary. Drift does not simply mean that a server moved, storage decreased, or a model file extension changed. Effective MLOps processes establish monitoring and evaluation mechanisms so that changing model behavior can be detected and addressed systematically.<\/span><\/p>\n

                          Question 34. Which practice improves traceability for machine learning experiments?<\/b><\/p>\n

                            \n
                          1. <\/b> Deleting experiment metadata<\/span>
                            \n<\/span>2.<\/b> Using undocumented manual changes<\/span>
                            \n<\/span>3.<\/b> Recording parameters, metrics, artifacts, and code versions<\/span>
                            \n<\/span>4.<\/b> Replacing all version identifiers with generic names<\/span><\/li>\n<\/ol>\n

                            Correct Answer: 3. Recording parameters, metrics, artifacts, and code versions<\/b><\/p>\n

                            Explanation:<\/b><\/p>\n

                            Experiment tracking improves traceability by recording important details about each training run. Useful information can include parameter values, evaluation metrics, dataset references, model artifacts, source-code versions, environment information, and execution details. With this information, teams can compare experiments and understand why one model performed differently from another. It also helps reproduce successful experiments and investigate unexpected results. Deleting metadata or relying on undocumented manual changes makes it harder to determine what happened during training. In an MLOps environment, experiment tracking provides an evidence-based history of model development and supports collaboration, debugging, governance, and controlled promotion of successful models.<\/span><\/p>\n

                            Question 35. What is a major advantage of infrastructure as code for MLOps environments?<\/b><\/p>\n

                              \n
                            1. <\/b> It makes infrastructure configuration repeatable and version-controlled<\/span>
                              \n<\/span>2.<\/b> It guarantees model accuracy<\/span>
                              \n<\/span>3.<\/b> It eliminates all security requirements<\/span>
                              \n<\/span>4.<\/b> It prevents infrastructure changes entirely<\/span><\/li>\n<\/ol>\n

                              Correct Answer: 1. It makes infrastructure configuration repeatable and version-controlled<\/b><\/p>\n

                              Explanation:<\/b><\/p>\n

                              Infrastructure as code allows infrastructure configuration to be represented in files that can be reviewed, versioned, tested, and reused. This is valuable for MLOps because development, testing, staging, and production environments can be created more consistently. Changes can be reviewed through standard development workflows and tracked over time, making it easier to understand how an environment evolved. Infrastructure as code does not guarantee model accuracy, eliminate security requirements, or prevent changes. Instead, it provides a controlled mechanism for defining and deploying infrastructure. This repeatability reduces configuration differences between environments and supports more reliable automation across the machine learning lifecycle.<\/span><\/p>\n

                              Question 36. Why should model deployment configurations be version-controlled?<\/b><\/p>\n

                                \n
                              1. <\/b> To prevent all future deployments<\/span>
                                \n<\/span>2.<\/b> To remove model monitoring<\/span>
                                \n<\/span>3.<\/b> To make deployment changes traceable and reproducible<\/span>
                                \n<\/span>4.<\/b> To avoid documenting infrastructure<\/span><\/li>\n<\/ol>\n

                                Correct Answer: 3. To make deployment changes traceable and reproducible<\/b><\/p>\n

                                Explanation:<\/b><\/p>\n

                                Deployment configuration can include endpoint settings, compute requirements, environment references, scaling parameters, routing rules, and other operational details. Version-controlling these configurations creates a historical record of changes and makes it possible to reproduce known deployment states. If a deployment introduces unexpected behavior, the team can inspect the configuration history and identify what changed. Version control also supports peer review and automated deployment processes. It does not prevent future deployments or eliminate monitoring. Instead, it creates greater consistency and traceability between environments. Keeping deployment configuration alongside other MLOps artifacts can significantly simplify troubleshooting and controlled release management.<\/span><\/p>\n

                                Question 37. Which practice can reduce deployment risk when introducing a new machine learning model?<\/b><\/p>\n

                                  \n
                                1. <\/b> Deploy directly to all users without monitoring<\/span>
                                  \n<\/span>2.<\/b> Remove the previous model immediately<\/span>
                                  \n<\/span>3.<\/b> Disable health checks<\/span>
                                  \n<\/span>4.<\/b> Gradually increase traffic while monitoring the new model<\/span><\/li>\n<\/ol>\n

                                  Correct Answer: 4. Gradually increase traffic while monitoring the new model<\/b><\/p>\n

                                  Explanation:<\/b><\/p>\n

                                  Gradually increasing traffic to a new model allows a team to validate real-world behavior before committing the entire workload to the new version. This approach can be implemented through strategies such as canary or controlled traffic deployment. During the rollout, operational and model-related metrics can be monitored to determine whether the new version meets predefined requirements. If unexpected errors, latency problems, or performance degradation appear, traffic can be reduced or redirected to the previous stable version. Immediately sending all users to an untested model increases exposure to potential problems. Gradual rollout therefore provides an additional safety mechanism within a mature deployment process.<\/span><\/p>\n

                                  Question 38. What is the purpose of defining model quality thresholds in a deployment pipeline?<\/b><\/p>\n

                                    \n
                                  1. <\/b> To automatically rename the model<\/span>
                                    \n<\/span>2.<\/b> To ensure a model meets predefined requirements before promotion<\/span>
                                    \n<\/span>3.<\/b> To remove experiment tracking<\/span>
                                    \n<\/span>4.<\/b> To disable automated testing<\/span><\/li>\n<\/ol>\n

                                    Correct Answer: 2. To ensure a model meets predefined requirements before promotion<\/b><\/p>\n

                                    Explanation:<\/b><\/p>\n

                                    Model quality thresholds establish measurable criteria that a candidate model must satisfy before it can progress to a later deployment stage. Depending on the application, thresholds might involve accuracy, precision, recall, F1 score, calibration, latency, fairness-related measurements, or other relevant metrics. These gates help transform subjective deployment decisions into repeatable processes. A candidate that fails the required criteria can be rejected, investigated, or sent back for additional development. Thresholds do not remove experiment tracking or disable automated tests. Instead, they work alongside testing and monitoring to create a more controlled path from model development to production deployment.<\/span><\/p>\n

                                    Question 39. What is the purpose of logging in a machine learning production system?<\/b><\/p>\n

                                      \n
                                    1. <\/b> To provide diagnostic and operational information about system behavior<\/span>
                                      \n<\/span>2.<\/b> To guarantee that models never fail<\/span>
                                      \n<\/span>3.<\/b> To replace all security controls<\/span>
                                      \n<\/span>4.<\/b> To eliminate the need for testing<\/span><\/li>\n<\/ol>\n

                                      Correct Answer: 1. To provide diagnostic and operational information about system behavior<\/b><\/p>\n

                                      Explanation:<\/b><\/p>\n

                                      Logging records useful information about application and service activity, allowing teams to investigate events and diagnose problems. In a production machine learning system, logs can help capture request outcomes, errors, deployment events, application messages, and other operational details, while avoiding unnecessary exposure of sensitive information. Effective logging supports troubleshooting and incident investigation when combined with monitoring and alerting. Logs do not guarantee that a model will never fail, replace security controls, or eliminate the need for testing. Instead, they provide historical evidence that helps engineers understand what occurred during an event and determine appropriate corrective actions.<\/span><\/p>\n

                                      Question 40. Which combination best represents a mature MLOps production workflow?<\/b><\/p>\n

                                        \n
                                      1. <\/b> Manual training, manual deployment, and no monitoring<\/span>
                                        \n<\/span>2.<\/b> Training without version control or testing<\/span>
                                        \n<\/span>3.<\/b> Version control, automated validation, reproducible environments, CI\/CD, monitoring, and rollback<\/span>
                                        \n<\/span>4.<\/b> Production deployment without quality gates<\/span><\/li>\n<\/ol>\n

                                        Correct Answer: 3. Version control, automated validation, reproducible environments, CI\/CD, monitoring, and rollback<\/b><\/p>\n

                                        Explanation:<\/b><\/p>\n

                                        A mature MLOps workflow combines engineering practices that make machine learning systems repeatable, observable, and controllable. Version control provides traceability for source code and configuration. Automated validation helps detect defects before deployment. Reproducible environments reduce differences between development and production. CI\/CD automates reliable movement of validated changes through deployment stages. Production monitoring identifies operational problems and changes in model behavior, while rollback mechanisms provide a way to recover when a release does not meet requirements. These practices work together rather than operating independently. A mature workflow therefore emphasizes automation, governance, observability, repeatability, and controlled recovery throughout the machine learning lifecycle.<\/span><\/p>\n

                                         <\/p>\n","protected":false},"excerpt":{"rendered":"

                                        View Full Microsoft AI-300 Exam Dumps and Practice Test Dumps   Question 21. Which Azure service can be used to orchestrate and automate machine learning workflows in Azure? Azure Machine Learning Pipelines 2. Azure DNS 3. Azure Storage Explorer 4. Azure Bastion Correct Answer: 1. Azure Machine Learning Pipelines Explanation: Azure Machine Learning Pipelines provide […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[1648,1647],"tags":[],"_links":{"self":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/17478"}],"collection":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/comments?post=17478"}],"version-history":[{"count":1,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/17478\/revisions"}],"predecessor-version":[{"id":17479,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/17478\/revisions\/17479"}],"wp:attachment":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/media?parent=17478"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/categories?post=17478"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/tags?post=17478"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}