{"id":18149,"date":"2026-09-22T05:40:18","date_gmt":"2026-09-22T05:40:18","guid":{"rendered":"https:\/\/www.examlabs.com\/certification\/?p=18149"},"modified":"2026-09-22T05:40:18","modified_gmt":"2026-09-22T05:40:18","slug":"databricks-certified-generative-ai-engineer-associate-practice-test-questions-and-exam-dumps-part14-q261-280","status":"publish","type":"post","link":"https:\/\/www.examlabs.com\/certification\/databricks-certified-generative-ai-engineer-associate-practice-test-questions-and-exam-dumps-part14-q261-280\/","title":{"rendered":"Databricks Certified Generative AI Engineer Associate Practice Test Questions and Exam Dumps Part14 Q261-280"},"content":{"rendered":"

View Full <\/b>Databricks Certified Generative AI Engineer Associate Exam Dumps<\/b><\/a> and Practice Test Dumps<\/b><\/p>\n

 <\/p>\n

Question 261. Which RAG component is responsible for converting retrieved documents into context that can be supplied to the language model?<\/b><\/p>\n

    \n
  1. <\/b> Context assembly or prompt construction<\/span><\/li>\n
  2. <\/b> Authentication service<\/span><\/li>\n
  3. <\/b> Vector index<\/span><\/li>\n
  4. <\/b> Monitoring dashboard<\/span><\/li>\n<\/ol>\n

    Correct Answer: 1. Context assembly or prompt construction<\/b><\/p>\n

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

    After relevant documents have been retrieved, the application needs to organize that information into a form that the language model can use. Context assembly or prompt construction combines the selected passages with the user’s question and the application’s instructions. This step may also include source identifiers, formatting rules, conversation history, or other relevant information. Good context construction helps the model distinguish instructions from supporting evidence and keeps the prompt within the available context window. Poorly assembled context can reduce answer quality even when retrieval itself is accurate. Therefore, context construction is an important stage between retrieval and generation in a RAG workflow.<\/span><\/p>\n

    Question 262. What is a major benefit of using a similarity search index for large knowledge collections?<\/b><\/p>\n

      \n
    1. <\/b> It automatically writes application code<\/span><\/li>\n
    2. <\/b> It enables efficient search over vector representations<\/span><\/li>\n
    3. <\/b> It eliminates the need for source documents<\/span><\/li>\n
    4. <\/b> It guarantees correct responses<\/span><\/li>\n<\/ol>\n

      Correct Answer: 2. It enables efficient search over vector representations<\/b><\/p>\n

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

      A similarity search index is designed to efficiently locate vectors that are close to a query vector according to a selected similarity measure. In a RAG system, document chunks can be converted into embeddings and stored in a vector-searchable index. A user’s query is then embedded and compared against those stored representations to identify potentially relevant content. Efficient indexing becomes increasingly important as the knowledge collection grows because a system should avoid performing unnecessarily expensive comparisons across every stored vector. The index supports the retrieval layer, while the language model remains responsible for generating the final response using the retrieved context.<\/span><\/p>\n

      Question 263. Which approach can improve retrieval for queries containing exact product identifiers or error codes?<\/b><\/p>\n

        \n
      1. <\/b> Keyword or lexical search<\/span><\/li>\n
      2. <\/b> Increasing temperature<\/span><\/li>\n
      3. <\/b> Removing metadata<\/span><\/li>\n
      4. <\/b> Summarizing all documents<\/span><\/li>\n<\/ol>\n

        Correct Answer: 1. Keyword or lexical search<\/b><\/p>\n

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

        Keyword or lexical search can be particularly effective for exact identifiers such as product codes, error codes, ticket numbers, configuration values, or specialized names. Semantic search is useful for understanding conceptual similarity, but exact strings may be important when the query contains a precise identifier that should appear directly in the source material. A hybrid retrieval architecture can combine lexical matching with semantic search to cover both exact-term and conceptual queries. Teams should evaluate the retrieval strategy using realistic examples from the application’s domain. For enterprise knowledge bases, combining complementary retrieval methods can improve the likelihood of finding the precise information users need.<\/span><\/p>\n

        Question 264. What is one purpose of adding citations or source references to RAG responses?<\/b><\/p>\n

          \n
        1. <\/b> To help users trace claims back to supporting information<\/span><\/li>\n
        2. <\/b> To increase the model’s parameter count<\/span><\/li>\n
        3. <\/b> To remove the retrieval stage<\/span><\/li>\n
        4. <\/b> To increase response randomness<\/span><\/li>\n<\/ol>\n

          Correct Answer: 1. To help users trace claims back to supporting information<\/b><\/p>\n

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

          Citations or source references can improve transparency by showing users where supporting information came from. In a RAG application, retrieved chunks can be associated with document identifiers, titles, URLs, sections, or other source metadata. The generated response can then present appropriate references alongside relevant claims when the application design supports this behavior. Source references can also assist developers during debugging and evaluation because they make it easier to determine which evidence was provided to the model. Citations do not automatically make a response correct, however. The underlying sources still need to be authoritative, current, relevant, and properly retrieved.<\/span><\/p>\n

          Question 265. Why should prompt instructions clearly distinguish retrieved context from user instructions?<\/b><\/p>\n

            \n
          1. <\/b> To reduce ambiguity about how retrieved information should be used<\/span><\/li>\n
          2. <\/b> To increase vector dimensions<\/span><\/li>\n
          3. <\/b> To disable semantic search<\/span><\/li>\n
          4. <\/b> To eliminate authentication<\/span><\/li>\n<\/ol>\n

            Correct Answer: 1. To reduce ambiguity about how retrieved information should be used<\/b><\/p>\n

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

            Clear prompt structure helps the model understand the different roles of system instructions, user requests, and retrieved evidence. In a RAG application, the prompt can explicitly identify the retrieved material as reference context and instruct the model on how to use it. This can reduce ambiguity and encourage grounded responses. Clear formatting may also make it easier to enforce rules such as answering only from supplied evidence or identifying when evidence is insufficient. Prompt organization should be tested because language models can still behave unexpectedly. Prompt instructions are one layer of control and should be combined with retrieval filtering, validation, monitoring, and other application-level safeguards.<\/span><\/p>\n

            Question 266. Which factor can influence the quality of document embeddings?<\/b><\/p>\n

              \n
            1. <\/b> The relevance of the embedding model to the document language and domain<\/span><\/li>\n
            2. <\/b> The user’s browser theme<\/span><\/li>\n
            3. <\/b> The number of application screenshots<\/span><\/li>\n
            4. <\/b> The monitor resolution<\/span><\/li>\n<\/ol>\n

              Correct Answer: 1. The relevance of the embedding model to the document language and domain<\/b><\/p>\n

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

              Embedding quality depends partly on how well the embedding model represents the language and concepts found in the application’s data. A model trained for broad general-purpose text may perform differently from one that supports particular languages or specialized terminology. If the knowledge base contains industry-specific vocabulary, abbreviations, or multilingual content, these characteristics should be considered during model selection. Teams can compare candidate embedding models using representative retrieval queries and metrics such as precision and recall. The embedding model is only one factor in retrieval quality; chunking, source quality, metadata, query formulation, indexing, and ranking strategies can also have substantial effects.<\/span><\/p>\n

              Question 267. What is the purpose of a document hash in an ingestion pipeline?<\/b><\/p>\n

                \n
              1. <\/b> It can help detect whether document content has changed<\/span><\/li>\n
              2. <\/b> It increases the model context window<\/span><\/li>\n
              3. <\/b> It replaces authorization<\/span><\/li>\n
              4. <\/b> It generates natural-language answers<\/span><\/li>\n<\/ol>\n

                Correct Answer: 1. It can help detect whether document content has changed<\/b><\/p>\n

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

                A document hash provides a compact representation of document content that can be compared with a previously stored value. If the content changes, the resulting hash will generally change, allowing an ingestion pipeline to identify documents that may require reprocessing. This can support incremental ingestion by avoiding unnecessary work on unchanged documents. When a document is detected as changed, the pipeline can extract and clean its content, recreate affected chunks and embeddings, update metadata, and refresh the relevant index entries. Hashes are therefore useful operational tools for tracking content changes, although they should be used alongside reliable document identifiers and source-system information.<\/span><\/p>\n

                Question 268. What is a potential disadvantage of creating chunks that are too small?<\/b><\/p>\n

                  \n
                1. <\/b> Important context may be separated across multiple chunks<\/span><\/li>\n
                2. <\/b> The model automatically becomes more accurate<\/span><\/li>\n
                3. <\/b> Authentication becomes unavailable<\/span><\/li>\n
                4. <\/b> The vector index is no longer required<\/span><\/li>\n<\/ol>\n

                  Correct Answer: 1. Important context may be separated across multiple chunks<\/b><\/p>\n

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

                  Very small chunks can make individual retrieval units overly narrow and may remove important contextual information. A definition might be separated from its explanation, a procedure from its prerequisites, or a question from the answer that follows it. Although small chunks can sometimes improve precision, excessive fragmentation can make it difficult for the retriever to return enough information for a complete answer. Teams can use moderate chunk sizes, meaningful document boundaries, and controlled overlap to preserve context. The appropriate strategy depends on document structure and query patterns. Evaluation should measure whether retrieved chunks provide sufficient evidence for accurate and grounded responses.<\/span><\/p>\n

                  Question 269. Which practice can help ensure that only current policy documents are retrieved?<\/b><\/p>\n

                    \n
                  1. <\/b> Applying active-status or effective-date metadata filters<\/span><\/li>\n
                  2. <\/b> Increasing temperature<\/span><\/li>\n
                  3. <\/b> Removing document identifiers<\/span><\/li>\n
                  4. <\/b> Increasing answer length<\/span><\/li>\n<\/ol>\n

                    Correct Answer: 1. Applying active-status or effective-date metadata filters<\/b><\/p>\n

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

                    Metadata filters can help restrict retrieval to documents that meet defined lifecycle conditions. For policy content, attributes such as active status, effective date, expiration date, and version can be used to identify the documents that should currently apply. This reduces the likelihood that obsolete policies are retrieved simply because they contain highly similar language to the user’s query. The metadata must be maintained accurately as documents are published, replaced, or retired. Filtering can be combined with semantic search and reranking to improve relevance. The application should also have a defined behavior for situations where no current document satisfies the retrieval criteria.<\/span><\/p>\n

                    Question 270. What is one reason to use an evaluation dataset containing expected answers or supporting documents?<\/b><\/p>\n

                      \n
                    1. <\/b> It provides a reference for measuring system performance<\/span><\/li>\n
                    2. <\/b> It removes the need for monitoring<\/span><\/li>\n
                    3. <\/b> It automatically retrains the model<\/span><\/li>\n
                    4. <\/b> It guarantees every production response<\/span><\/li>\n<\/ol>\n

                      Correct Answer: 1. It provides a reference for measuring system performance<\/b><\/p>\n

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

                      An evaluation dataset gives the team a consistent set of examples against which system behavior can be measured. Depending on the evaluation design, each example may include a user query, expected answer characteristics, relevant documents, reference answers, or other criteria. This allows developers to compare different prompts, retrieval configurations, models, and application versions under consistent conditions. Evaluation datasets can be used to measure retrieval precision and recall, groundedness, relevance, correctness, and other metrics. A strong dataset should represent important production scenarios and be maintained as the application evolves. It should not be treated as a guarantee that every future user request will be handled correctly.<\/span><\/p>\n

                      Question 271. What does grounded generation require from a RAG application?<\/b><\/p>\n

                        \n
                      1. <\/b> The generated response should be supported by relevant retrieved evidence<\/span><\/li>\n
                      2. <\/b> The model should ignore retrieved documents<\/span><\/li>\n
                      3. <\/b> The application should always increase temperature<\/span><\/li>\n
                      4. <\/b> The model should answer without context<\/span><\/li>\n<\/ol>\n

                        Correct Answer: 1. The generated response should be supported by relevant retrieved evidence<\/b><\/p>\n

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

                        Grounded generation means that the response is based on supporting information supplied to the model, typically through retrieval in a RAG system. The goal is to reduce unsupported claims by giving the model relevant evidence and clear instructions about how that evidence should be used. Grounding depends on multiple components working together: trustworthy source data, effective retrieval, appropriate context construction, and generation instructions that encourage evidence-based responses. Teams can evaluate groundedness to determine whether claims in responses are supported by retrieved material. Grounding is not equivalent to factual correctness in every situation because the retrieved source itself could be inaccurate or outdated.<\/span><\/p>\n

                        Question 272. What should a team consider when deciding whether to use RAG or fine-tuning for a use case?<\/b><\/p>\n

                          \n
                        1. <\/b> Whether the need is primarily changing knowledge or learned behavior<\/span><\/li>\n
                        2. <\/b> Only the number of application users<\/span><\/li>\n
                        3. <\/b> The screen size of the client device<\/span><\/li>\n
                        4. <\/b> The color of the interface<\/span><\/li>\n<\/ol>\n

                          Correct Answer: 1. Whether the need is primarily changing knowledge or learned behavior<\/b><\/p>\n

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

                          RAG and fine-tuning address different types of application requirements. RAG is often useful when the system needs access to external or frequently changing information because the knowledge can be updated through the retrieval layer without retraining the language model for every source change. Fine-tuning can be useful when the goal is to influence learned behavior, style, task performance, or other model characteristics using training examples. The two approaches can also be combined when appropriate. Teams should consider data freshness, task requirements, maintenance effort, evaluation results, cost, and operational complexity before selecting an architecture rather than assuming one method is universally suitable.<\/span><\/p>\n

                          Question 273. What is a key benefit of using a smaller language model when it meets application quality requirements?<\/b><\/p>\n

                            \n
                          1. <\/b> It may reduce latency and operational cost<\/span><\/li>\n
                          2. <\/b> It guarantees better reasoning<\/span><\/li>\n
                          3. <\/b> It removes the need for retrieval<\/span><\/li>\n
                          4. <\/b> It eliminates monitoring<\/span><\/li>\n<\/ol>\n

                            Correct Answer: 1. It may reduce latency and operational cost<\/b><\/p>\n

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

                            A smaller model can require fewer computational resources than a larger model and may therefore provide lower latency and lower serving costs when it is capable of meeting the application’s quality requirements. Model selection should be based on measured performance rather than size alone. A larger model may provide advantages for certain complex tasks, while a smaller model may be sufficient for straightforward generation, classification, extraction, or domain-specific workflows. Teams should evaluate response quality, groundedness, latency, throughput, and cost together. Choosing a model that satisfies the application’s requirements efficiently can help balance user experience and operational considerations in production.<\/span><\/p>\n

                            Question 274. Why can conversation-history summarization reduce costs in a multi-turn application?<\/b><\/p>\n

                              \n
                            1. <\/b> It can reduce the number of tokens sent to the model<\/span><\/li>\n
                            2. <\/b> It increases the number of retrieved documents<\/span><\/li>\n
                            3. <\/b> It removes the need for an embedding model<\/span><\/li>\n
                            4. <\/b> It increases model parameters<\/span><\/li>\n<\/ol>\n

                              Correct Answer: 1. It can reduce the number of tokens sent to the model<\/b><\/p>\n

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

                              As a conversation grows, repeatedly sending the entire history to the language model can increase token usage, latency, and cost. Summarizing older turns can preserve important conversational information while reducing the amount of text included in future prompts. The application can retain recent messages in detail and use a compact summary for older interactions. The summary should preserve facts, decisions, unresolved issues, and other information necessary for future turns. Because summarization can lose details, it should be evaluated for the application’s requirements. In some architectures, retrieval of relevant historical information can complement summarization and provide additional context when needed.<\/span><\/p>\n

                              Question 275. What is an important purpose of a similarity threshold when no retrieved result is sufficiently relevant?<\/b><\/p>\n

                                \n
                              1. <\/b> It allows the system to recognize that adequate evidence may be unavailable<\/span><\/li>\n
                              2. <\/b> It forces the model to answer anyway<\/span><\/li>\n
                              3. <\/b> It increases model randomness<\/span><\/li>\n
                              4. <\/b> It deletes the source data<\/span><\/li>\n<\/ol>\n

                                Correct Answer: 1. It allows the system to recognize that adequate evidence may be unavailable<\/b><\/p>\n

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

                                A similarity threshold can help a RAG application distinguish between meaningful retrieval matches and weak matches. If every candidate falls below the defined threshold, the system can treat the retrieval result as insufficient rather than passing unrelated documents to the language model. This can support safer behavior, such as asking for clarification, informing the user that the available knowledge does not contain enough information, or routing the request for further assistance. Threshold values should be tuned using evaluation data because an overly strict threshold can reduce recall, while an overly permissive threshold can introduce irrelevant context. The goal is to balance retrieval coverage and evidence quality.<\/span><\/p>\n

                                Question 276. Which operational metric is most directly associated with the number of requests a service can handle over time?<\/b><\/p>\n

                                  \n
                                1. <\/b> Throughput<\/span><\/li>\n
                                2. <\/b> Groundedness<\/span><\/li>\n
                                3. <\/b> Retrieval precision<\/span><\/li>\n
                                4. <\/b> Prompt relevance<\/span><\/li>\n<\/ol>\n

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

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

                                  Throughput measures the amount of work a service can process during a specified period. For a generative AI application, this may be represented as requests per second, tokens processed per second, or another workload-specific measurement. Throughput is useful for capacity planning because production systems may need to support many concurrent users. It is different from latency, which measures the time required to complete an individual request. A service can have low latency for one request but still have limited throughput under heavy concurrency. Monitoring both metrics helps teams understand whether the serving infrastructure and application architecture can meet expected workload requirements.<\/span><\/p>\n

                                  Question 277. Why should production logs include enough information to reconstruct an AI request without unnecessarily exposing sensitive data?<\/b><\/p>\n

                                    \n
                                  1. <\/b> To balance troubleshooting needs with security and privacy requirements<\/span><\/li>\n
                                  2. <\/b> To guarantee perfect model responses<\/span><\/li>\n
                                  3. <\/b> To eliminate authorization<\/span><\/li>\n
                                  4. <\/b> To increase embedding dimensions<\/span><\/li>\n<\/ol>\n

                                    Correct Answer: 1. To balance troubleshooting needs with security and privacy requirements<\/b><\/p>\n

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

                                    Production logging should provide enough information for developers and operators to understand application behavior while avoiding unnecessary exposure of sensitive information. Useful records may include request identifiers, model and prompt versions, retrieval metadata, latency measurements, error information, and other operational details. Depending on the application, storing complete prompts or retrieved documents may introduce privacy or security risks. Teams should therefore apply data minimization, masking, access controls, retention policies, and appropriate logging standards. The goal is to make troubleshooting and auditing possible without turning observability systems into unnecessary repositories of confidential content. Logging design should be reviewed as part of the application’s overall security architecture.<\/span><\/p>\n

                                    Question 278. What is the benefit of evaluating a new prompt against the same dataset used for the previous prompt version?<\/b><\/p>\n

                                      \n
                                    1. <\/b> It enables a more meaningful comparison between versions<\/span><\/li>\n
                                    2. <\/b> It guarantees the new prompt is better<\/span><\/li>\n
                                    3. <\/b> It eliminates the need for human review<\/span><\/li>\n
                                    4. <\/b> It changes the embedding model automatically<\/span><\/li>\n<\/ol>\n

                                      Correct Answer: 1. It enables a more meaningful comparison between versions<\/b><\/p>\n

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

                                      Using the same evaluation dataset for multiple prompt versions reduces one source of variability when comparing results. If the queries and evaluation criteria remain consistent, differences in measured performance can be more directly associated with changes to the prompt. Teams can compare metrics such as groundedness, response relevance, correctness, formatting compliance, latency, or cost depending on the application’s requirements. This approach is especially useful when testing system instructions, RAG context formatting, few-shot examples, or response constraints. The dataset should remain representative and should not be treated as the only evidence of production quality. Real-world monitoring and feedback remain important after deployment.<\/span><\/p>\n

                                      Question 279. What is one advantage of retaining a previous validated configuration after deploying a new RAG version?<\/b><\/p>\n

                                        \n
                                      1. <\/b> It provides a practical rollback option if the new version causes problems<\/span><\/li>\n
                                      2. <\/b> It guarantees the new version will succeed<\/span><\/li>\n
                                      3. <\/b> It eliminates the need for monitoring<\/span><\/li>\n
                                      4. <\/b> It prevents all source-data changes<\/span><\/li>\n<\/ol>\n

                                        Correct Answer: 1. It provides a practical rollback option if the new version causes problems<\/b><\/p>\n

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

                                        Keeping a previously validated configuration available gives a team a recovery path when a new deployment produces unexpected regressions. A RAG application may include multiple versioned components, including prompts, language models, embedding models, retrieval settings, rerankers, and data-processing configurations. If these components are tracked and the previous working configuration can be restored, production incidents can potentially be contained more quickly. Rollback does not replace testing or monitoring; it complements them. Teams should define deployment and rollback procedures in advance and verify that the required artifacts remain available so recovery does not depend on reconstructing an older system manually.<\/span><\/p>\n

                                        Question 280. Which sequence best represents a robust RAG request lifecycle?<\/b><\/p>\n

                                          \n
                                        1. <\/b> Generate an answer first, then search for supporting documents<\/span><\/li>\n
                                        2. <\/b> Retrieve relevant evidence, assemble context, generate a response, and evaluate or monitor the result<\/span><\/li>\n
                                        3. <\/b> Ignore the user query and retrieve random documents<\/span><\/li>\n
                                        4. <\/b> Fine-tune the model for every request<\/span><\/li>\n<\/ol>\n

                                          Correct Answer: 2. Retrieve relevant evidence, assemble context, generate a response, and evaluate or monitor the result<\/b><\/p>\n

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

                                          A typical RAG request begins when the application receives a user query and prepares it for retrieval. The retrieval layer searches the knowledge base and selects relevant evidence using methods such as semantic search, lexical search, metadata filtering, and reranking. The application then assembles the selected context with appropriate instructions and the user’s request before sending the prompt to the language model. The generated response can subsequently be checked, logged, evaluated, or monitored according to the application’s requirements. This workflow separates retrieval from generation while connecting them through evidence-based context, allowing teams to measure and improve each stage independently.<\/span><\/p>\n

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

                                          View Full Databricks Certified Generative AI Engineer Associate Exam Dumps and Practice Test Dumps   Question 261. Which RAG component is responsible for converting retrieved documents into context that can be supplied to the language model? Context assembly or prompt construction Authentication service Vector index Monitoring dashboard Correct Answer: 1. Context assembly or prompt construction […]<\/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\/18149"}],"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=18149"}],"version-history":[{"count":1,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/18149\/revisions"}],"predecessor-version":[{"id":18150,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/18149\/revisions\/18150"}],"wp:attachment":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/media?parent=18149"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/categories?post=18149"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/tags?post=18149"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}