179 questions
No questions match those filters.
← All questions
EvaluationHard· RAG· Metrics· Debugging
Your retrieval team says recall@k is 92%, clean. Your g...
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 plansThe discrepancy arises because the generation team is evaluating against ‘gold’ context, which is not necessarily what the retrieval system actually provides to the generator in the live pipeline. By evaluating the generation stage against the actual retrieved context, you can isolate whether the loss is occurring during the augmentation process, effectively resolving the blame game with data.