{"id":17503,"date":"2026-09-21T09:56:06","date_gmt":"2026-09-21T09:56:06","guid":{"rendered":"https:\/\/www.examlabs.com\/certification\/?p=17503"},"modified":"2026-09-21T09:56:06","modified_gmt":"2026-09-21T09:56:06","slug":"microsoft-ai-300-practice-test-questions-and-exam-dumps-part13-q241-260","status":"publish","type":"post","link":"https:\/\/www.examlabs.com\/certification\/microsoft-ai-300-practice-test-questions-and-exam-dumps-part13-q241-260\/","title":{"rendered":"Microsoft AI-300 Practice Test Questions and Exam Dumps Part13 Q241-260"},"content":{"rendered":"

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

 <\/p>\n

Question 241. Which practice helps ensure that a deployed ML model can be reproduced from a previous release?<\/b><\/p>\n

    \n
  1. <\/b> Delete previous deployment metadata<\/span>
    \n<\/span>2.<\/b> Version the model, code, dependencies, and configuration<\/span>
    \n<\/span>3.<\/b> Modify production settings manually<\/span>
    \n<\/span>4.<\/b> Store only the latest model file<\/span><\/li>\n<\/ol>\n

    Correct Answer: 2. Version the model, code, dependencies, and configuration<\/b><\/p>\n

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

    Reproducibility requires enough information to recreate the conditions under which a model was trained, evaluated, and deployed. Versioning the model artifact alone may not be sufficient because changes in source code, libraries, runtime environments, or deployment configuration can affect behavior. By versioning these related artifacts and maintaining links between them, an MLOps workflow can identify exactly what was used for a particular release. This also supports troubleshooting and rollback because a previous validated combination can be restored. Reproducibility is especially important when multiple model iterations are developed over time and when automated pipelines are responsible for moving artifacts between environments.<\/span><\/p>\n

    Question 242. Which stage should verify that input data meets expected requirements before model training or inference?<\/b><\/p>\n

      \n
    1. <\/b> Data validation<\/span>
      \n<\/span>2.<\/b> Traffic switching<\/span>
      \n<\/span>3.<\/b> Endpoint retirement<\/span>
      \n<\/span>4.<\/b> Rollback execution<\/span><\/li>\n<\/ol>\n

      Correct Answer: 1. Data validation<\/b><\/p>\n

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

      Data validation checks whether incoming or training data satisfies predefined expectations before it is used by an ML workflow. These expectations can include schema, data types, required fields, acceptable ranges, missing-value limits, category values, and other workload-specific constraints. Detecting invalid data early can prevent corrupted training runs and unexpected inference behavior. Validation can be implemented as an automated pipeline stage so that failures are detected consistently. The validation criteria should reflect the requirements of the model and application rather than simply checking whether a file exists. When combined with data-quality monitoring in production, validation helps establish stronger controls across the ML data lifecycle.<\/span><\/p>\n

      Question 243. Which capability allows a team to compare several trained model versions before selecting one for deployment?<\/b><\/p>\n

        \n
      1. <\/b> Endpoint scaling<\/span>
        \n<\/span>2.<\/b> Experiment tracking and model versioning<\/span>
        \n<\/span>3.<\/b> DNS configuration<\/span>
        \n<\/span>4.<\/b> Log deletion<\/span><\/li>\n<\/ol>\n

        Correct Answer: 2. Experiment tracking and model versioning<\/b><\/p>\n

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

        Experiment tracking records information about training runs, including parameters, metrics, artifacts, and other relevant metadata. When this information is combined with model versioning, teams can identify how different candidate models were produced and compare their evaluation results. This makes model selection more systematic and traceable. Instead of relying on an unnamed model file, the team can associate a candidate with a specific training run and documented results. Such records are also useful after deployment because engineers can determine which experiment and artifact correspond to a production version. Experiment tracking therefore supports both development decisions and operational traceability.<\/span><\/p>\n

        Question 244. What is the primary purpose of a model registry in an MLOps workflow?<\/b><\/p>\n

          \n
        1. <\/b> To provide centralized management and versioning of model artifacts<\/span>
          \n<\/span>2.<\/b> To replace endpoint monitoring<\/span>
          \n<\/span>3.<\/b> To automatically fix all model errors<\/span>
          \n<\/span>4.<\/b> To store user passwords<\/span><\/li>\n<\/ol>\n

          Correct Answer: 1. To provide centralized management and versioning of model artifacts<\/b><\/p>\n

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

          A model registry provides a structured location for managing trained model artifacts and their associated metadata and versions. It can help teams track which models were evaluated, approved, promoted, or deployed. Depending on the implementation, registry metadata can include model versions, descriptions, evaluation results, lineage information, and lifecycle status. Centralized model management makes it easier to identify a known-good version and supports controlled promotion between environments. A registry does not replace monitoring or automated testing; instead, it complements those controls by providing an organized record of the models moving through the lifecycle.<\/span><\/p>\n

          Question 245. Which deployment method allows a new version to be tested while the existing production version continues serving most users?<\/b><\/p>\n

            \n
          1. <\/b> Canary deployment<\/span>
            \n<\/span>2.<\/b> Full replacement<\/span>
            \n<\/span>3.<\/b> Uncontrolled deployment<\/span>
            \n<\/span>4.<\/b> Manual file copying<\/span><\/li>\n<\/ol>\n

            Correct Answer: 1. Canary deployment<\/b><\/p>\n

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

            Canary deployment introduces a new version to a limited amount of production traffic while the existing version continues serving the majority of requests. This allows the team to observe real-world behavior without exposing the entire workload to the new release immediately. Operational metrics such as latency, error rate, availability, resource usage, and prediction characteristics can be monitored during the canary phase. If the new version behaves as expected, traffic can be increased gradually. If unexpected behavior occurs, traffic can be reduced or redirected to the established version. This makes canary deployment useful for staged production releases where controlled exposure is required.<\/span><\/p>\n

            Question 246. Which information should be captured when registering a production model to improve traceability?<\/b><\/p>\n

              \n
            1. <\/b> Only the model filename<\/span>
              \n<\/span>2.<\/b> Model version, source revision, evaluation results, and relevant metadata<\/span>
              \n<\/span>3.<\/b> Only the endpoint URL<\/span>
              \n<\/span>4.<\/b> Only the number of predictions<\/span><\/li>\n<\/ol>\n

              Correct Answer: 2. Model version, source revision, evaluation results, and relevant metadata<\/b><\/p>\n

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

              Model traceability depends on maintaining enough metadata to connect a deployed artifact with its development and validation history. Useful information can include the model version, source-code revision, training run identifier, evaluation metrics, environment information, data references, and deployment metadata. These records help engineers determine how a model was created and why it was approved for deployment. They also support incident investigation because a production model can be traced back to the artifacts and validation evidence associated with it. A filename or endpoint address alone does not provide sufficient context when many model versions exist.<\/span><\/p>\n

              Question 247. What is a key benefit of separating development, testing, and production environments?<\/b><\/p>\n

                \n
              1. <\/b> It allows all changes to be made directly in production<\/span>
                \n<\/span>2.<\/b> It removes the need for testing<\/span>
                \n<\/span>3.<\/b> It provides controlled stages for validating changes before production<\/span>
                \n<\/span>4.<\/b> It prevents version control<\/span><\/li>\n<\/ol>\n

                Correct Answer: 3. It provides controlled stages for validating changes before production<\/b><\/p>\n

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

                Separating development, testing, and production environments provides a structured path for moving ML changes through progressively controlled stages. Developers can experiment without directly affecting production, while testing environments can validate application behavior, model quality, dependencies, and deployment configuration. Production can then receive artifacts that have already passed the required checks. Environment separation also reduces the chance that an experimental configuration or unvalidated model will affect users. To be effective, the environments should use controlled and versioned configurations so that differences are understood. This approach works well with CI\/CD pipelines and promotion gates that manage movement between stages.<\/span><\/p>\n

                Question 248. Which metric should be monitored to identify increasing delays in responses from an online endpoint?<\/b><\/p>\n

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

                  Correct Answer: 2. Request latency<\/b><\/p>\n

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

                  Request latency measures the time between receiving a request and returning a response from an online service. Monitoring this metric helps identify performance changes that could affect application responsiveness. Latency can increase because of higher traffic, insufficient compute resources, inefficient model execution, dependency problems, network conditions, or configuration changes. Operators can compare current latency with established baselines and investigate significant deviations. Latency monitoring should not be considered in isolation because a service can have acceptable response times while experiencing elevated errors or poor model quality. Combining latency with availability, error-rate, resource, and model-related metrics provides a more complete picture of endpoint health.<\/span><\/p>\n

                  Question 249. Which approach can reduce the risk of exposing all production traffic to a newly deployed model?<\/b><\/p>\n

                    \n
                  1. <\/b> Gradually increase traffic after validation<\/span>
                    \n<\/span>2.<\/b> Disable monitoring during deployment<\/span>
                    \n<\/span>3.<\/b> Delete the existing model immediately<\/span>
                    \n<\/span>4.<\/b> Skip health checks<\/span><\/li>\n<\/ol>\n

                    Correct Answer: 1. Gradually increase traffic after validation<\/b><\/p>\n

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

                    Gradually increasing traffic provides an opportunity to evaluate a new model under production conditions before it becomes responsible for the entire workload. The new version can initially receive a small portion of requests while monitoring captures relevant operational and model signals. Teams can evaluate error rates, latency, resource consumption, prediction behavior, data quality, and other requirements. If the model remains within acceptable boundaries, traffic can be increased in controlled stages. If problems emerge, the deployment can be paused or traffic can be redirected. This strategy is commonly associated with canary-style releases and provides a practical way to limit exposure during production transitions.<\/span><\/p>\n

                    Question 250. Which security principle should guide permissions assigned to an automated ML deployment identity?<\/b><\/p>\n

                      \n
                    1. <\/b> Grant every available permission<\/span>
                      \n<\/span>2.<\/b> Use shared administrator credentials<\/span>
                      \n<\/span>3.<\/b> Least privilege<\/span>
                      \n<\/span>4.<\/b> Allow anonymous access<\/span><\/li>\n<\/ol>\n

                      Correct Answer: 3. Least privilege<\/b><\/p>\n

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

                      The principle of least privilege means that an identity receives only the permissions required to perform its assigned tasks. An automated ML deployment identity may need to read model artifacts, access specific resources, deploy an endpoint, or retrieve secrets, but it should not automatically receive unrestricted administrative permissions. Limiting access reduces the potential impact of compromised credentials, accidental changes, and configuration mistakes. Managed identities can further reduce the need to store long-lived credentials in source code or pipeline definitions. Permissions should be reviewed periodically and adjusted as workflows change. Applying least privilege is an important part of securing automated deployment processes and supporting controlled access.<\/span><\/p>\n

                      Question 251. What should an ML pipeline do if data validation fails before model training?<\/b><\/p>\n

                        \n
                      1. <\/b> Continue training with invalid data<\/span>
                        \n<\/span>2.<\/b> Skip the failed validation result<\/span>
                        \n<\/span>3.<\/b> Stop or fail the pipeline stage<\/span>
                        \n<\/span>4.<\/b> Promote the previous dataset automatically without review<\/span><\/li>\n<\/ol>\n

                        Correct Answer: 3. Stop or fail the pipeline stage<\/b><\/p>\n

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

                        A failed data-validation stage indicates that the data does not satisfy one or more predefined requirements. Continuing automatically can allow malformed, incomplete, or unexpected data to enter model training and potentially produce unreliable results. A controlled pipeline should therefore stop or fail the relevant stage and provide information about the validation failure. The responsible team can then investigate the source data, schema, transformation logic, or validation rules. After the problem is corrected, validation can be rerun before training proceeds. Treating data validation as an enforced pipeline gate makes the workflow more predictable and reduces the risk of producing models from unsuitable inputs.<\/span><\/p>\n

                        Question 252. Which practice supports auditing changes made to production ML deployments?<\/b><\/p>\n

                          \n
                        1. <\/b> Record deployment changes and retain appropriate audit logs<\/span>
                          \n<\/span>2.<\/b> Allow undocumented manual changes<\/span>
                          \n<\/span>3.<\/b> Delete deployment history<\/span>
                          \n<\/span>4.<\/b> Share one account among all operators<\/span><\/li>\n<\/ol>\n

                          Correct Answer: 1. Record deployment changes and retain appropriate audit logs<\/b><\/p>\n

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

                          Audit logs provide a record of important actions performed against production resources, including deployment activities, configuration changes, identity usage, and other operational events. When combined with source control and deployment metadata, these records help establish who performed a change, what changed, and when it occurred. This information can be valuable during incident investigation and operational reviews. Logs should be protected against unauthorized modification and should avoid exposing sensitive information such as passwords or access tokens. Auditing is not simply about retaining large quantities of logs; the important goal is to preserve useful, relevant records that support traceability and accountability.<\/span><\/p>\n

                          Question 253. Which mechanism can automatically stop a deployment when model quality is below a required threshold?<\/b><\/p>\n

                            \n
                          1. <\/b> Quality gate<\/span>
                            \n<\/span>2.<\/b> DNS record<\/span>
                            \n<\/span>3.<\/b> Storage lifecycle rule<\/span>
                            \n<\/span>4.<\/b> Manual file rename<\/span><\/li>\n<\/ol>\n

                            Correct Answer: 1. Quality gate<\/b><\/p>\n

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

                            A quality gate can evaluate model metrics against predefined thresholds and prevent an artifact from progressing when the requirements are not satisfied. For example, an organization might require a candidate model to achieve a specified performance level before production promotion. The exact metric and threshold depend on the application and should be established before the pipeline executes. When the condition fails, the pipeline can stop or pause and record the result for investigation. This makes quality control repeatable and reduces the risk of inconsistent manual decisions. Quality gates can be combined with automated testing, security validation, and approval checkpoints as part of a broader deployment process.<\/span><\/p>\n

                            Question 254. Why is automated rollback useful in an ML production environment?<\/b><\/p>\n

                              \n
                            1. <\/b> It removes the need for model testing<\/span>
                              \n<\/span>2.<\/b> It can restore a known-good deployment when predefined failure conditions occur<\/span>
                              \n<\/span>3.<\/b> It guarantees that every new model is accurate<\/span>
                              \n<\/span>4.<\/b> It prevents all production incidents<\/span><\/li>\n<\/ol>\n

                              Correct Answer: 2. It can restore a known-good deployment when predefined failure conditions occur<\/b><\/p>\n

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

                              Automated rollback provides a predefined recovery mechanism when a newly deployed model or service fails important operational conditions. Trigger conditions might include severe error rates, failed health checks, availability problems, or other deployment-specific criteria. When the necessary previous version and configuration are retained, the system can switch back to a known-good state without waiting for extensive manual reconstruction. Automated rollback does not prevent every incident or determine whether a model is scientifically correct. Instead, it limits the operational impact of certain failures and helps restore service more quickly. The rollback process should be tested and documented so that recovery behavior is understood before an incident occurs.<\/span><\/p>\n

                              Question 255. Which monitoring combination provides broader visibility into an online ML endpoint?<\/b><\/p>\n

                                \n
                              1. <\/b> Model filename and dataset name only<\/span>
                                \n<\/span>2.<\/b> Latency, errors, availability, resource usage, and model-related metrics<\/span>
                                \n<\/span>3.<\/b> Source-code comments only<\/span>
                                \n<\/span>4.<\/b> Number of developers only<\/span><\/li>\n<\/ol>\n

                                Correct Answer: 2. Latency, errors, availability, resource usage, and model-related metrics<\/b><\/p>\n

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

                                A comprehensive monitoring approach combines operational and model-specific signals. Latency can indicate responsiveness, error rate can reveal failed requests, and availability can show whether the endpoint remains accessible. Resource metrics such as CPU and memory utilization can provide context for performance changes. Model-related measurements can include prediction distributions, data drift, data quality, and available performance indicators. No single metric can fully describe the health of an ML service. Correlating several signals can help teams distinguish infrastructure problems from data or model issues. This broader observability supports faster investigation and provides evidence for decisions about scaling, remediation, retraining, or rollback.<\/span><\/p>\n

                                Question 256. What is the purpose of maintaining deployment metadata across ML releases?<\/b><\/p>\n

                                  \n
                                1. <\/b> To identify and trace the artifacts and settings used for each release<\/span>
                                  \n<\/span>2.<\/b> To prevent future deployments<\/span>
                                  \n<\/span>3.<\/b> To eliminate testing requirements<\/span>
                                  \n<\/span>4.<\/b> To replace monitoring systems<\/span><\/li>\n<\/ol>\n

                                  Correct Answer: 1. To identify and trace the artifacts and settings used for each release<\/b><\/p>\n

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

                                  Deployment metadata records contextual information about a release, such as model version, source revision, environment, configuration, pipeline run, deployment time, and other relevant identifiers. This information helps teams connect production behavior with the exact artifacts and settings involved. During troubleshooting, engineers can use deployment metadata to determine whether a problem appeared after a specific release or configuration change. Metadata also supports reproducibility and rollback because it helps identify the components associated with a previously validated deployment. Maintaining accurate metadata is therefore an important operational practice, particularly when many automated deployments occur over the lifetime of an ML application.<\/span><\/p>\n

                                  Question 257. Which process helps ensure that only tested artifacts move from one environment to the next?<\/b><\/p>\n

                                    \n
                                  1. <\/b> Controlled promotion<\/span>
                                    \n<\/span>2.<\/b> Random deployment<\/span>
                                    \n<\/span>3.<\/b> Manual copying without validation<\/span>
                                    \n<\/span>4.<\/b> Immediate production access<\/span><\/li>\n<\/ol>\n

                                    Correct Answer: 1. Controlled promotion<\/b><\/p>\n

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

                                    Controlled promotion establishes defined conditions that an artifact must satisfy before moving from development to testing and eventually to production. These conditions can include automated tests, model evaluation, data validation, security checks, environment validation, and approval requirements. The process creates a predictable path through the lifecycle and prevents unvalidated artifacts from bypassing important controls. Controlled promotion also supports traceability because each transition can be associated with a specific version and set of validation results. When integrated into CI\/CD, promotion can be largely automated while still enforcing the organization’s required quality and security conditions before the artifact reaches a more sensitive environment.<\/span><\/p>\n

                                    Question 258. Which feature helps identify whether a model version has passed required evaluation before deployment?<\/b><\/p>\n

                                      \n
                                    1. <\/b> Model registry metadata and evaluation records<\/span>
                                      \n<\/span>2.<\/b> Endpoint hostname only<\/span>
                                      \n<\/span>3.<\/b> Storage account name only<\/span>
                                      \n<\/span>4.<\/b> Network port number<\/span><\/li>\n<\/ol>\n

                                      Correct Answer: 1. Model registry metadata and evaluation records<\/b><\/p>\n

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

                                      Model registry metadata and evaluation records can connect a specific model version with the evidence used to assess it. A team can record evaluation metrics, validation results, training information, and lifecycle status so that the candidate’s history is available before promotion. This reduces ambiguity when multiple model versions are stored or generated by automated pipelines. The registry itself does not necessarily determine whether a model is suitable; rather, it provides an organized mechanism for storing the artifact and associated information. Promotion rules can then use documented evaluation results and quality criteria to determine whether the model is ready for the next stage.<\/span><\/p>\n

                                      Question 259. Which practice helps reduce configuration drift between ML environments?<\/b><\/p>\n

                                        \n
                                      1. <\/b> Use version-controlled and repeatable configuration definitions<\/span>
                                        \n<\/span>2.<\/b> Configure each environment independently without records<\/span>
                                        \n<\/span>3.<\/b> Change production settings manually after every release<\/span>
                                        \n<\/span>4.<\/b> Keep different undocumented dependency versions<\/span><\/li>\n<\/ol>\n

                                        Correct Answer: 1. Use version-controlled and repeatable configuration definitions<\/b><\/p>\n

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

                                        Configuration drift occurs when environments gradually develop differences because settings are changed independently or without consistent management. Version-controlled configuration definitions provide a documented source of truth for how environments should be configured. Automated deployment processes can use these definitions to create or update environments consistently. Changes can be reviewed, tested, and associated with specific source-control revisions. This does not mean every environment must have identical settings, because development and production can have legitimate differences. Instead, the differences should be intentional, documented, and managed through controlled configuration. This approach improves reproducibility and makes unexpected environment-specific failures easier to investigate.<\/span><\/p>\n

                                        Question 260. Which combination best supports reliable continuous ML deployment operations?<\/b><\/p>\n

                                          \n
                                        1. <\/b> Manual changes, no monitoring, and unversioned artifacts<\/span>
                                          \n<\/span>2.<\/b> Direct production deployment without validation<\/span>
                                          \n<\/span>3.<\/b> Versioning, automated testing, controlled promotion, monitoring, and rollback<\/span>
                                          \n<\/span>4.<\/b> Deleting previous releases after deployment<\/span><\/li>\n<\/ol>\n

                                          Correct Answer: 3. Versioning, automated testing, controlled promotion, monitoring, and rollback<\/b><\/p>\n

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

                                          Reliable continuous ML deployment depends on a collection of complementary controls. Versioning provides traceability for code, models, dependencies, and configurations. Automated testing and validation identify defects or unacceptable artifacts before production promotion. Controlled promotion establishes a defined path through environments and can include approval or quality gates. Once deployed, monitoring provides visibility into operational and model-related behavior, including errors, latency, availability, data drift, and performance. Finally, rollback capability provides a recovery path when a release produces unacceptable behavior. These practices work together to create a repeatable lifecycle in which ML changes can be developed, validated, deployed, observed, and recovered in a controlled manner.<\/span><\/p>\n

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

                                          View Full Microsoft AI-300 Exam Dumps and Practice Test Dumps   Question 241. Which practice helps ensure that a deployed ML model can be reproduced from a previous release? Delete previous deployment metadata 2. Version the model, code, dependencies, and configuration 3. Modify production settings manually 4. Store only the latest model file Correct Answer: […]<\/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\/17503"}],"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=17503"}],"version-history":[{"count":1,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/17503\/revisions"}],"predecessor-version":[{"id":17504,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/17503\/revisions\/17504"}],"wp:attachment":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/media?parent=17503"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/categories?post=17503"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/tags?post=17503"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}