Question 670 An on-call team attributes every 5xx answer failure to model generation, although endpoint logs show upstream tool 429 throttling. Which design change directly satisfies the stated constraints? Enable approved inference telemetry capturing timestamps, latency, token counts, status and trace IDs while respecting privacy constraints. Inspect per-stage inference records for search timeouts and missing-context flags before changing prompt text. Configure the supported scorer sampling/monitoring schedule and verify trace ingestion and service prerequisites. Correlate model-serving and tool-spans with inference table status and retry records before diagnosing the failing component. ExplanationCorrect Answer D:The action targets the stated cause: Layered agents can turn upstream rate-limit errors into user-visible failures even when the base model is healthy. Check the repair by: trace a synthetic tool-throttling incident and identify the 429 call and final error under one request ID. Other Answers A:Layered agents can turn upstream rate-limit errors into user-visible failures even when the base model is healthy. B:Layered agents can turn upstream rate-limit errors into user-visible failures even when the base model is healthy. C:Layered agents can turn upstream rate-limit errors into user-visible failures even when the base model is healthy.
Question 671 A correctness evaluator expects a reference answer for every question, but the production trace dataset lacks any trusted gold response. Which implementation decision best restores the intended behavior? Continue running reference-based exact-answer correctness on every unlabeled production trace and treat missing expected answers as passing. Monitor citation validation rate, groundedness and source freshness as separate production-quality signals. Enable approved inference/trace logging with query, index revision, retrieved document IDs and citations while redacting sensitive values. Use a reference-free groundedness or instruction-following judge for traces, reserving expected-answer correctness for curated labeled cases. ExplanationCorrect Answer D:The action targets the stated cause: Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. Check the repair by: check scorer input schemas on one unlabeled trace and one labeled benchmark case and confirm only supported judgments run. Other Answers A:A reference-based correctness scorer has no trustworthy target for these unlabeled traces. It cannot treat an absent gold answer as evidence of correctness. B:Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. C:Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context.
Question 672 A correctness evaluator expects a reference answer for every question, but the production trace dataset lacks any trusted gold response. Which underlying technical principle explains why this occurs? Groundedness can be judged against supplied source context even when no canonical final answer has been authored. Without evidence-version provenance, the team cannot separate retrieval mistakes from answer-generation mistakes. Fluent output does not establish that an answer is supported by a live authoritative passage. Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. ExplanationCorrect Answer D:Use a reference-free groundedness or instruction-following judge for traces, reserving expected-answer correctness for curated labeled cases. Other Answers A:Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. B:Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. C:That mechanism describes another problem: A production agent returns fluent answers but silently fails to resolve citations on several newly ingested policy versions. It does not explain this case, where reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context.
Question 673 A correctness evaluator expects a reference answer for every question, but the production trace dataset lacks any trusted gold response. After correction, what targeted test should demonstrate success? Sample live traces and check that citation IDs exist in the current authorized retrieval set. Construct a known unsupported claim and confirm the judge flags it against the retrieved passage. Check scorer input schemas on one unlabeled trace and one labeled benchmark case and confirm only supported judgments run. Reproduce a reported answer and find the exact document IDs and index revision in the recorded trace. ExplanationCorrect Answer C:Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. Other Answers A:Check scorer input schemas on one unlabeled trace and one labeled benchmark case and confirm only supported judgments run. B:Check scorer input schemas on one unlabeled trace and one labeled benchmark case and confirm only supported judgments run. D:Check scorer input schemas on one unlabeled trace and one labeled benchmark case and confirm only supported judgments run.
Question 674 A correctness evaluator expects a reference answer for every question, but the production trace dataset lacks any trusted gold response. Which release safeguard most directly addresses this failure class? Define explicit evidence requirements for each judge and label the results by methodology. Require source-version metadata and sanitized retrieval-span logging for grounded applications. Require source-provenance monitoring and freshness alerts after ingestion changes. Include reference-free evidence entailment tests and preserve source IDs for each judge input. ExplanationCorrect Answer A:Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. Other Answers B:Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. C:This control addresses a different exposure: A production agent returns fluent answers but silently fails to resolve citations on several newly ingested policy versions. It would not prevent this failure, because reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. D:That reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context.
Question 675 A document assistant cites a passage that is present in retrieved context, but it adds a claim not supported by that passage; no reference answer exists. Which design change directly satisfies the stated constraints? Apply a context-groundedness judge comparing generated claims with retrieved evidence, rather than requiring unavailable gold text. Enable approved inference/trace logging with query, index revision, retrieved document IDs and citations while redacting sensitive values. Use a scorer that checks only whether cited document IDs appeared in retrieval, without testing whether the generated claim follows from the passages. Judge only whether the response is fluent and citation IDs are syntactically well formed, without comparing the claim to retrieved passages. ExplanationCorrect Answer A:The action targets the stated cause: Groundedness can be judged against supplied source context even when no canonical final answer has been authored. Check the repair by: construct a known unsupported claim and confirm the judge flags it against the retrieved passage. Other Answers B:Groundedness can be judged against supplied source context even when no canonical final answer has been authored. C:Document-ID membership does not prove that the cited passage supports the generated claim. The scenario requires judging claim-to-context support, not only whether a citation refers to a retrieved item. D:Fluency and citation syntax do not establish entailment. Reference-free groundedness must compare the asserted claim against the actual retrieved context.
Question 676 A document assistant cites a passage that is present in retrieved context, but it adds a claim not supported by that passage; no reference answer exists. Which causal explanation best accounts for the failure? Without evidence-version provenance, the team cannot separate retrieval mistakes from answer-generation mistakes. Reference-based correctness requires an external target; reference-free judgments assess properties using prompts, answers and available context. Groundedness can be judged against supplied source context even when no canonical final answer has been authored. Citation syntax validity is weaker than provenance integrity; a fabricated identifier can still look well formed. ExplanationCorrect Answer C:Apply a context-groundedness judge comparing generated claims with retrieved evidence, rather than requiring unavailable gold text. Other Answers A:Groundedness can be judged against supplied source context even when no canonical final answer has been authored. B:That mechanism describes another problem: A correctness evaluator expects a reference answer for every question, but the production trace dataset lacks any trusted gold response. It does not explain this case, where groundedness can be judged against supplied source context even when no canonical final answer has been authored. D:Groundedness can be judged against supplied source context even when no canonical final answer has been authored.
Question 677 A document assistant cites a passage that is present in retrieved context, but it adds a claim not supported by that passage; no reference answer exists. Which concrete acceptance test best demonstrates the proposed change works? Reproduce a reported answer and find the exact document IDs and index revision in the recorded trace. Evaluate one fabricated and one genuinely grounded citation and inspect scorer feedback per trace. Check scorer input schemas on one unlabeled trace and one labeled benchmark case and confirm only supported judgments run. Construct a known unsupported claim and confirm the judge flags it against the retrieved passage. ExplanationCorrect Answer D:Groundedness can be judged against supplied source context even when no canonical final answer has been authored. Other Answers A:Construct a known unsupported claim and confirm the judge flags it against the retrieved passage. B:Construct a known unsupported claim and confirm the judge flags it against the retrieved passage. C:Construct a known unsupported claim and confirm the judge flags it against the retrieved passage.
Question 678 A document assistant cites a passage that is present in retrieved context, but it adds a claim not supported by that passage; no reference answer exists. Which operational practice should be made a durable requirement? Require source-version metadata and sanitized retrieval-span logging for grounded applications. Define explicit evidence requirements for each judge and label the results by methodology. Maintain provenance-aware scorers and negative citation fixtures across retrieval releases. Include reference-free evidence entailment tests and preserve source IDs for each judge input. ExplanationCorrect Answer D:Groundedness can be judged against supplied source context even when no canonical final answer has been authored. Other Answers A:This control addresses a different exposure: User reports say a RAG bot cites the wrong policy, but only final answers are stored and engineers cannot see which version was retrieved. It would not prevent this failure, because groundedness can be judged against supplied source context even when no canonical final answer has been authored. B:That groundedness can be judged against supplied source context even when no canonical final answer has been authored. C:Groundedness can be judged against supplied source context even when no canonical final answer has been authored.
Question 679 A team wants to limit one consumer application to a token budget but currently sees only aggregate endpoint usage. What should the implementation team change first to resolve the described defect? Record input/output tokens and per-application usage, then isolate cost drivers by model, tenant and workflow. Use approved Unity Gateway usage visibility and caller-scoped rate or budget policies where supported, with clear version/availability caveats. Apply rate limits and budget safeguards at the approved gateway/service boundary, with graceful error handling for excess traffic. Inspect gateway tool-usage telemetry for the MCP service and attribute calls, errors and latency to the tool and request identity. ExplanationCorrect Answer B:The action targets the stated cause: Central gateway controls can attribute usage to callers and enforce request-level policies beyond the raw model API. Check the repair by: generate traffic from two identities and verify the selected policy throttles only the configured caller. Other Answers A:Central gateway controls can attribute usage to callers and enforce request-level policies beyond the raw model API. C:Central gateway controls can attribute usage to callers and enforce request-level policies beyond the raw model API. D:Central gateway controls can attribute usage to callers and enforce request-level policies beyond the raw model API.
Question 680 A team wants to limit one consumer application to a token budget but currently sees only aggregate endpoint usage. Why does the current approach fail under these conditions? Cost visibility requires decomposing usage by request path and token consumption, not relying on a single aggregate invoice. Model serving cost is coupled to request rate and generated tokens; uncontrolled callers can exhaust capacity or spend. Agent spend and reliability include tool invocations; model-only tokens omit MCP service activity. Central gateway controls can attribute usage to callers and enforce request-level policies beyond the raw model API. ExplanationCorrect Answer D:Use approved Unity Gateway usage visibility and caller-scoped rate or budget policies where supported, with clear version/availability caveats. Other Answers A:That mechanism describes another problem: A team sees its monthly model bill grow, but metrics expose only total dollars without token counts or request attribution. It does not explain this case, where central gateway controls can attribute usage to callers and enforce request-level policies beyond the raw model API. B:Central gateway controls can attribute usage to callers and enforce request-level policies beyond the raw model API. C:Central gateway controls can attribute usage to callers and enforce request-level policies beyond the raw model API.
