{"id":19396,"date":"2026-09-23T05:41:31","date_gmt":"2026-09-23T05:41:31","guid":{"rendered":"https:\/\/www.examlabs.com\/certification\/?p=19396"},"modified":"2026-09-23T05:41:31","modified_gmt":"2026-09-23T05:41:31","slug":"snowflake-snowpro-specialty-gen-ai-ges-c01-practice-test-questions-and-exam-dumps-part-7-q121-140","status":"publish","type":"post","link":"https:\/\/www.examlabs.com\/certification\/snowflake-snowpro-specialty-gen-ai-ges-c01-practice-test-questions-and-exam-dumps-part-7-q121-140\/","title":{"rendered":"Snowflake SnowPro Specialty Gen AI GES-C01 Practice Test Questions and Exam Dumps Part 7 Q121-140"},"content":{"rendered":"

View Full <\/b>Snowflake SnowPro Specialty Gen AI GES-C01<\/b> Exam Dumps<\/b><\/a> and Practice Test Dumps<\/b><\/p>\n

 <\/p>\n

Question 121: Which Snowflake Cortex capability is specifically intended for working with unstructured documents to extract useful information?<\/b><\/p>\n

    \n
  1. <\/b> Cortex Analyst<\/span><\/li>\n
  2. <\/b> Cortex Search<\/span><\/li>\n
  3. <\/b> Document AI<\/span><\/li>\n
  4. <\/b> Semantic model<\/span><\/li>\n<\/ol>\n

    Correct Answer: 3. Document AI<\/b><\/p>\n

    Explanation:<\/b>
    \n<\/b>Document AI is intended for processing documents and extracting useful information from unstructured document content. It can support workflows where organizations need to identify information contained in documents such as invoices, forms, contracts, or other business records. The extracted information can subsequently be stored, analyzed, or incorporated into other data workflows. Cortex Analyst focuses on natural-language questions over structured data, while Cortex Search is focused on searching and retrieving relevant content. A semantic model provides business meaning for structured analytics. Selecting the appropriate Cortex capability depends on whether the workload involves document extraction, content retrieval, or structured-data analysis.<\/span><\/p>\n

    Question 122: What is the main benefit of using semantic search instead of relying exclusively on exact keyword matching?<\/b><\/p>\n

      \n
    1. <\/b> It can identify conceptually related content even when the wording differs<\/span><\/li>\n
    2. <\/b> It guarantees that every result is factually correct<\/span><\/li>\n
    3. <\/b> It removes the need to prepare source content<\/span><\/li>\n
    4. <\/b> It automatically grants users access to restricted data<\/span><\/li>\n<\/ol>\n

      Correct Answer: 1. It can identify conceptually related content even when the wording differs<\/b><\/p>\n

      Explanation:<\/b>
      \n<\/b>Semantic search uses representations of meaning to identify content that is conceptually related to a query. This allows the system to find relevant information even when the exact words in the query do not appear in the source content. This differs from purely lexical approaches that depend heavily on matching terms. Semantic search does not guarantee factual accuracy, remove the need for content preparation, or automatically grant access to restricted information. Embedding models are commonly used to support semantic retrieval by representing queries and documents as numerical vectors that can be compared for similarity.<\/span><\/p>\n

      Question 123: Which factor can most directly affect the relevance of results returned by a semantic retrieval system?<\/b><\/p>\n

        \n
      1. <\/b> The application’s screen resolution<\/span><\/li>\n
      2. <\/b> The quality and suitability of the embeddings used for the content and queries<\/span><\/li>\n
      3. <\/b> The number of users viewing the dashboard<\/span><\/li>\n
      4. <\/b> The color of the application’s buttons<\/span><\/li>\n<\/ol>\n

        Correct Answer: 2. The quality and suitability of the embeddings used for the content and queries<\/b><\/p>\n

        Explanation:<\/b>
        \n<\/b>Embeddings provide numerical representations of the semantic characteristics of queries and source content. If the embedding model is suitable for the application’s content and use case, semantically related items can be represented in a way that supports effective similarity-based retrieval. Poor or unsuitable embeddings can lead to irrelevant results even when the retrieval infrastructure is functioning correctly. Other factors, including chunking, indexing, filtering, and ranking, also affect retrieval quality. Application appearance and unrelated interface characteristics generally do not determine semantic similarity. Embedding quality should therefore be considered when diagnosing retrieval performance.<\/span><\/p>\n

        Question 124: What is the purpose of retrieval evaluation in a RAG system?<\/b><\/p>\n

          \n
        1. <\/b> To determine whether the retrieval component finds useful information for representative queries<\/span><\/li>\n
        2. <\/b> To permanently retrain the language model<\/span><\/li>\n
        3. <\/b> To eliminate the need for source documents<\/span><\/li>\n
        4. <\/b> To measure only the application’s visual appearance<\/span><\/li>\n<\/ol>\n

          Correct Answer: 1. To determine whether the retrieval component finds useful information for representative queries<\/b><\/p>\n

          Explanation:<\/b>
          \n<\/b>Retrieval evaluation focuses on whether the search or retrieval component returns information that is useful for answering representative user questions. This can involve assessing relevance, completeness, ranking, and other retrieval characteristics appropriate to the application. Evaluating retrieval separately from generation helps identify whether poor final responses are caused by inadequate source information or by the generation stage. Retrieval evaluation does not retrain the model and does not replace source data. Representative queries and evaluation datasets provide a more meaningful basis for assessment than testing only a small number of arbitrary examples.<\/span><\/p>\n

          Question 125: Which approach can help ensure that a RAG application does not retrieve information outside a user’s authorized scope?<\/b><\/p>\n

            \n
          1. <\/b> Applying appropriate access controls or authorization-aware filtering during retrieval<\/span><\/li>\n
          2. <\/b> Increasing model temperature<\/span><\/li>\n
          3. <\/b> Removing metadata from every document<\/span><\/li>\n
          4. <\/b> Sending all enterprise documents to the LLM<\/span><\/li>\n<\/ol>\n

            Correct Answer: 1. Applying appropriate access controls or authorization-aware filtering during retrieval<\/b><\/p>\n

            Explanation:<\/b>
            \n<\/b>Enterprise RAG applications may operate over information with different access requirements. Retrieval should therefore incorporate appropriate authorization controls so that users receive only information they are permitted to access. Depending on the architecture, this can involve permissions, metadata-based restrictions, row or document-level controls, or application-specific authorization logic. Increasing temperature does not address data access, and removing metadata can make filtering more difficult. Sending every enterprise document to the LLM creates unnecessary exposure and context. Security should be considered before retrieved information becomes part of the model’s context.<\/span><\/p>\n

            Question 126: What does the term “context window” generally refer to in an LLM?<\/b><\/p>\n

              \n
            1. <\/b> The amount of input and output information the model can process within a given interaction<\/span><\/li>\n
            2. <\/b> The number of users authorized to access a database<\/span><\/li>\n
            3. <\/b> The storage capacity of a Snowflake warehouse<\/span><\/li>\n
            4. <\/b> The number of documents stored in an index<\/span><\/li>\n<\/ol>\n

              Correct Answer: 1. The amount of input and output information the model can process within a given interaction<\/b><\/p>\n

              Explanation:<\/b>
              \n<\/b>A model’s context window refers broadly to the amount of information that can be considered within an interaction, including relevant input context and, depending on the model and API behavior, generated output. Context limits are important when designing RAG applications because retrieved documents, system instructions, user queries, and generated responses all consume available context. Supplying excessive information can reduce efficiency or exceed supported limits. Context-window capacity is different from database storage, warehouse compute capacity, or the number of indexed documents. Effective retrieval and chunking can help provide focused information within the available context.<\/span><\/p>\n

              Question 127: Why can prompt injection be a concern in a RAG application?<\/b><\/p>\n

                \n
              1. <\/b> Retrieved or user-supplied content may contain instructions that attempt to influence model behavior<\/span><\/li>\n
              2. <\/b> It automatically increases vector dimensions<\/span><\/li>\n
              3. <\/b> It guarantees that every document is correctly classified<\/span><\/li>\n
              4. <\/b> It changes the physical storage location of a warehouse<\/span><\/li>\n<\/ol>\n

                Correct Answer: 1. Retrieved or user-supplied content may contain instructions that attempt to influence model behavior<\/b><\/p>\n

                Explanation:<\/b>
                \n<\/b>Prompt injection is a security concern in applications where user input or retrieved content is supplied to a language model. Such content may contain instructions designed to influence the model in ways that conflict with the application’s intended behavior. RAG applications can be particularly relevant because retrieved documents become part of the model’s context. Developers should distinguish trusted instructions from untrusted content and apply appropriate controls for the application’s threat model. Prompt injection cannot be solved simply by changing vector dimensions or classification settings. Security testing and careful prompt and application design are important when deploying LLM-based systems.<\/span><\/p>\n

                Question 128: Which design choice can help reduce the impact of irrelevant retrieved documents on a RAG response?<\/b><\/p>\n

                  \n
                1. <\/b> Retrieve every available document<\/span><\/li>\n
                2. <\/b> Apply relevance ranking and appropriate filtering<\/span><\/li>\n
                3. <\/b> Remove the user’s question from the prompt<\/span><\/li>\n
                4. <\/b> Disable all indexing<\/span><\/li>\n<\/ol>\n

                  Correct Answer: 2. Apply relevance ranking and appropriate filtering<\/b><\/p>\n

                  Explanation:<\/b>
                  \n<\/b>Retrieval systems can improve the quality of supplied context by ranking candidate results according to relevance and applying appropriate filters. This helps the application provide the model with information that is more closely related to the user’s request. Metadata filters can further constrain results based on attributes such as source, date, category, or authorization. Returning every available document can overwhelm the model with unnecessary information. Disabling indexing would generally make retrieval less efficient. Relevance ranking and filtering are therefore important techniques for controlling the quality and quantity of context supplied to a generative model.<\/span><\/p>\n

                  Question 129: Which statement best describes zero-shot prompting?<\/b><\/p>\n

                    \n
                  1. <\/b> Asking the model to perform a task without providing task-specific examples<\/span><\/li>\n
                  2. <\/b> Retraining the model from scratch for every request<\/span><\/li>\n
                  3. <\/b> Providing hundreds of labeled examples in every prompt<\/span><\/li>\n
                  4. <\/b> Converting the task into a database permission<\/span><\/li>\n<\/ol>\n

                    Correct Answer: 1. Asking the model to perform a task without providing task-specific examples<\/b><\/p>\n

                    Explanation:<\/b>
                    \n<\/b>Zero-shot prompting asks a model to perform a task using instructions without supplying task-specific examples in the prompt. The model relies on its existing capabilities and the instructions provided to understand the requested task. This differs from few-shot prompting, where examples are included to demonstrate the expected behavior or output pattern. Zero-shot prompting does not involve retraining the model for each request. Its effectiveness depends on the clarity of the instructions, the model’s capabilities, and the complexity of the task. For more specialized tasks, examples or additional context may be useful.<\/span><\/p>\n

                    Question 130: What is the primary purpose of few-shot prompting?<\/b><\/p>\n

                      \n
                    1. <\/b> To provide examples that demonstrate the desired task or output pattern<\/span><\/li>\n
                    2. <\/b> To permanently change the model’s parameters<\/span><\/li>\n
                    3. <\/b> To remove all retrieved context<\/span><\/li>\n
                    4. <\/b> To configure Snowflake user authentication<\/span><\/li>\n<\/ol>\n

                      Correct Answer: 1. To provide examples that demonstrate the desired task or output pattern<\/b><\/p>\n

                      Explanation:<\/b>
                      \n<\/b>Few-shot prompting provides one or more examples within the prompt to demonstrate how the model should approach a task or format its output. Examples can clarify classification labels, formatting requirements, expected reasoning patterns, or other task-specific behavior. The examples influence the current interaction but do not permanently modify the model’s learned parameters. Few-shot prompting is different from model fine-tuning, which involves modifying model parameters through a training process. Well-chosen examples should be relevant and consistent because confusing or contradictory examples can negatively affect the model’s response.<\/span><\/p>\n

                      Question 131: Which characteristic is important when selecting a generative AI model for an enterprise workload?<\/b><\/p>\n

                        \n
                      1. <\/b> The model’s suitability for the application’s task, quality requirements, cost, and operational constraints<\/span><\/li>\n
                      2. <\/b> Whether the model automatically changes database ownership<\/span><\/li>\n
                      3. <\/b> Whether the model removes all governance requirements<\/span><\/li>\n
                      4. <\/b> Whether the model always produces identical responses regardless of configuration<\/span><\/li>\n<\/ol>\n

                        Correct Answer: 1. The model’s suitability for the application’s task, quality requirements, cost, and operational constraints<\/b><\/p>\n

                        Explanation:<\/b>
                        \n<\/b>Model selection should be based on the requirements of the intended workload. Relevant considerations can include task capability, response quality, latency, cost, supported languages, context requirements, output characteristics, and operational constraints. There is not necessarily one model that is optimal for every application. Model selection also does not eliminate governance, access controls, or evaluation requirements. Organizations should evaluate candidate models against representative workloads and application-specific criteria. This helps ensure that the selected model provides an appropriate balance between performance, functionality, cost, and other requirements.<\/span><\/p>\n

                        Question 132: What is one reason an enterprise may choose a smaller language model for a particular task?<\/b><\/p>\n

                          \n
                        1. <\/b> A smaller model may provide sufficient quality with lower latency or cost for the task<\/span><\/li>\n
                        2. <\/b> Smaller models automatically have access to more enterprise data<\/span><\/li>\n
                        3. <\/b> Smaller models eliminate hallucination risk<\/span><\/li>\n
                        4. <\/b> Smaller models always outperform larger models on every task<\/span><\/li>\n<\/ol>\n

                          Correct Answer: 1. A smaller model may provide sufficient quality with lower latency or cost for the task<\/b><\/p>\n

                          Explanation:<\/b>
                          \n<\/b>Model selection involves balancing quality with practical requirements such as latency, cost, and workload scale. For a task that does not require the capabilities of a larger model, a smaller model may provide sufficient quality while offering operational advantages. This does not mean smaller models always outperform larger models or eliminate hallucination risk. The appropriate choice should be based on evaluation against the actual application requirements. Organizations can compare candidate models using representative tasks and metrics that measure both quality and operational characteristics before selecting a model for deployment.<\/span><\/p>\n

                          Question 133: Which issue can occur when a prompt contains contradictory instructions?<\/b><\/p>\n

                            \n
                          1. <\/b> The model may produce an output that does not consistently follow the intended requirements<\/span><\/li>\n
                          2. <\/b> The model automatically creates a new database<\/span><\/li>\n
                          3. <\/b> Vector embeddings become permanently corrupted<\/span><\/li>\n
                          4. <\/b> User permissions are automatically increased<\/span><\/li>\n<\/ol>\n

                            Correct Answer: 1. The model may produce an output that does not consistently follow the intended requirements<\/b><\/p>\n

                            Explanation:<\/b>
                            \n<\/b>Contradictory instructions create ambiguity about what the model should do. When different parts of a prompt specify incompatible requirements, the resulting output may fail to satisfy the application’s intended behavior consistently. Clear, prioritized, and internally consistent instructions can reduce this problem. Prompt design should distinguish system-level requirements, user requests, retrieved content, and output constraints where appropriate. Contradictory instructions do not directly alter database permissions or permanently corrupt embeddings. Testing prompts against representative examples can help identify ambiguity and improve the reliability of the application’s responses.<\/span><\/p>\n

                            Question 134: What is an important consideration when using unstructured enterprise data for AI applications?<\/b><\/p>\n

                              \n
                            1. <\/b> The content may require preprocessing, extraction, chunking, indexing, or other preparation before effective retrieval<\/span><\/li>\n
                            2. <\/b> Every document is automatically ready for semantic search<\/span><\/li>\n
                            3. <\/b> Unstructured data never requires access controls<\/span><\/li>\n
                            4. <\/b> Document content cannot be used in generative AI applications<\/span><\/li>\n<\/ol>\n

                              Correct Answer: 1. The content may require preprocessing, extraction, chunking, indexing, or other preparation before effective retrieval<\/b><\/p>\n

                              Explanation:<\/b>
                              \n<\/b>Unstructured enterprise content often requires preparation before it can be effectively searched or incorporated into AI workflows. Depending on the source, preparation may involve extracting text, identifying useful fields, cleaning content, chunking documents, generating embeddings, indexing information, or attaching metadata. Access controls and governance remain important because unstructured documents can contain sensitive information. Not every document is automatically ready for effective semantic retrieval. Proper preparation helps the retrieval system identify useful content and provide focused context to downstream AI components.<\/span><\/p>\n

                              Question 135: Which technique can help an application constrain a model to return only values from an approved set of categories?<\/b><\/p>\n

                                \n
                              1. <\/b> Explicitly define the allowed categories and required output format<\/span><\/li>\n
                              2. <\/b> Remove all task instructions<\/span><\/li>\n
                              3. <\/b> Increase irrelevant retrieved context<\/span><\/li>\n
                              4. <\/b> Disable response validation<\/span><\/li>\n<\/ol>\n

                                Correct Answer: 1. Explicitly define the allowed categories and required output format<\/b><\/p>\n

                                Explanation:<\/b>
                                \n<\/b>Applications that require classification into a predefined set of categories can improve consistency by clearly specifying the allowed values and expected output format. The prompt can explain the classification task and establish constraints that the model should follow. Where supported, structured output mechanisms can further reinforce the expected response structure. Application-level validation can then check whether the returned value is actually one of the approved categories. Simply removing instructions or disabling validation makes it harder to control output quality. Explicit constraints are therefore useful for workflows that require predictable categorical responses.<\/span><\/p>\n

                                Question 136: What is the role of validation after an LLM generates a business-critical response?<\/b><\/p>\n

                                  \n
                                1. <\/b> To check whether the output satisfies relevant correctness, format, policy, or business requirements<\/span><\/li>\n
                                2. <\/b> To guarantee that the underlying model has been retrained<\/span><\/li>\n
                                3. <\/b> To automatically increase the model’s context window<\/span><\/li>\n
                                4. <\/b> To replace all source data<\/span><\/li>\n<\/ol>\n

                                  Correct Answer: 1. To check whether the output satisfies relevant correctness, format, policy, or business requirements<\/b><\/p>\n

                                  Explanation:<\/b>
                                  \n<\/b>Validation provides an additional control between model generation and downstream use. Depending on the application, validation can check output format, required fields, allowed values, business rules, policy requirements, or other characteristics. This is particularly important when generated information is used in business-critical workflows. Validation does not retrain the model or guarantee that every statement is factually correct. Its purpose is to detect outputs that fail defined requirements so the application can handle them appropriately. The level of validation should reflect the importance and risk of the intended use case.<\/span><\/p>\n

                                  Question 137: Which statement best describes the relationship between AI governance and AI security?<\/b><\/p>\n

                                    \n
                                  1. <\/b> Security can be one component of broader governance for responsible AI deployment<\/span><\/li>\n
                                  2. <\/b> Governance eliminates the need for access controls<\/span><\/li>\n
                                  3. <\/b> Security is relevant only during model training<\/span><\/li>\n
                                  4. <\/b> Governance applies only to user interface design<\/span><\/li>\n<\/ol>\n

                                    Correct Answer: 1. Security can be one component of broader governance for responsible AI deployment<\/b><\/p>\n

                                    Explanation:<\/b>
                                    \n<\/b>AI governance encompasses policies, processes, oversight, controls, and practices used to manage AI systems responsibly. Security is an important component because enterprise AI applications may process sensitive data and interact with business systems. Governance can also include evaluation, monitoring, responsible use, data management, human oversight, and procedures for addressing risks. Governance does not eliminate the need for technical access controls, and security is relevant throughout the application’s lifecycle rather than only during model training. Effective governance brings these considerations together according to the application’s risks and organizational requirements.<\/span><\/p>\n

                                    Question 138: Why should an AI application distinguish trusted instructions from retrieved document content?<\/b><\/p>\n

                                      \n
                                    1. <\/b> Retrieved content may contain text that should be treated as data rather than as instructions to the model<\/span><\/li>\n
                                    2. <\/b> All retrieved content is automatically more authoritative than system instructions<\/span><\/li>\n
                                    3. <\/b> Retrieved content always contains executable database commands<\/span><\/li>\n
                                    4. <\/b> The distinction is needed only to improve screen resolution<\/span><\/li>\n<\/ol>\n

                                      Correct Answer: 1. Retrieved content may contain text that should be treated as data rather than as instructions to the model<\/b><\/p>\n

                                      Explanation:<\/b>
                                      \n<\/b>Retrieved documents are generally source information for the model to use, but their content may also contain text that resembles instructions. Treating untrusted retrieved content as authoritative instructions can create security and reliability problems, including prompt-injection risks. Application designers should establish clear boundaries between trusted instructions and external or retrieved data. Appropriate prompt construction, content handling, access controls, and security testing can help manage these risks. Retrieved information should contribute evidence or context for the task rather than automatically gaining authority to change how the application operates.<\/span><\/p>\n

                                      Question 139: Which evaluation approach is most useful for comparing two candidate models for the same enterprise task?<\/b><\/p>\n

                                        \n
                                      1. <\/b> Test both models on the same representative evaluation set using consistent criteria<\/span><\/li>\n
                                      2. <\/b> Compare them using unrelated tasks<\/span><\/li>\n
                                      3. <\/b> Select the model based only on its name<\/span><\/li>\n
                                      4. <\/b> Evaluate only one model and assume the other behaves similarly<\/span><\/li>\n<\/ol>\n

                                        Correct Answer: 1. Test both models on the same representative evaluation set using consistent criteria<\/b><\/p>\n

                                        Explanation:<\/b>
                                        \n<\/b>A fair model comparison requires testing candidate models under comparable conditions. Using the same representative evaluation dataset and consistent criteria allows an organization to compare characteristics such as answer quality, relevance, format adherence, latency, and cost where applicable. Testing unrelated tasks or evaluating only one model does not provide a reliable basis for comparison. Model names or general reputation are also insufficient substitutes for application-specific testing. Evaluation should reflect the actual workload because a model that performs well on one task may not provide the same results on another.<\/span><\/p>\n

                                        Question 140: Which architecture best separates responsibilities in an enterprise RAG application?<\/b><\/p>\n

                                          \n
                                        1. <\/b> Retrieval handles relevant source selection, the LLM handles generation, and application controls handle authorization and validation<\/span><\/li>\n
                                        2. <\/b> The LLM alone performs retrieval, authorization, and validation<\/span><\/li>\n
                                        3. <\/b> The database alone generates all natural-language responses<\/span><\/li>\n
                                        4. <\/b> The user interface determines which confidential documents the model can access<\/span><\/li>\n<\/ol>\n

                                          Correct Answer: 1. Retrieval handles relevant source selection, the LLM handles generation, and application controls handle authorization and validation<\/b><\/p>\n

                                          Explanation:<\/b>
                                          \n<\/b>Separating responsibilities makes an enterprise RAG architecture easier to understand and control. Retrieval focuses on identifying relevant source information, while the language model uses the supplied context to generate a response. Authorization controls determine what information the requesting user is permitted to access, and validation or application logic can check the generated result against relevant requirements. These responsibilities should not be assumed to be handled entirely by the language model. A clear architecture allows organizations to evaluate, secure, and improve individual components while maintaining appropriate controls across the complete workflow.<\/span><\/p>\n

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

                                          View Full Snowflake SnowPro Specialty Gen AI GES-C01 Exam Dumps and Practice Test Dumps   Question 121: Which Snowflake Cortex capability is specifically intended for working with unstructured documents to extract useful information? Cortex Analyst Cortex Search Document AI Semantic model Correct Answer: 3. Document AI Explanation: Document AI is intended for processing documents and […]<\/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\/19396"}],"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=19396"}],"version-history":[{"count":1,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/19396\/revisions"}],"predecessor-version":[{"id":19397,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/19396\/revisions\/19397"}],"wp:attachment":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/media?parent=19396"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/categories?post=19396"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/tags?post=19396"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}