179 questions
No questions match those filters.
← All questions
EvaluationVery Hard· LLMOps· Diagnostics· Model Drift
How do you adjudicate a situation where context-relevan...
This is one of the questions in the full AI/ML interview bank. Pro unlocks all 1789 questions; Premium includes the same bank plus the highest daily Practice limit.
See plansBecause context relevance is computed by an LLM extraction step, a score drop can be caused by either the retriever or the extractor’s judgment shifting due to a model swap. You must isolate the variable by re-running the historical data through the old extraction model. If the scores recover, the issue is the evaluation pipeline; if they do not, the regression is real and independent of the model change.