Optimizing foundation modelsFull notesSummaryRingkasanStoriesPracticeBusiness Case5 exam-style questions on this lesson.Question 1 of 5A telecom company wants a support chatbot to reduce online tickets by 70% and reach a satisfaction score of 4 out of 5. Why is an LLM the right type of model?AThe chatbot must predict pricesBThe chatbot must generate imagesCThe chatbot must understand natural languageDThe chatbot must cluster ticketsCheck answerQuestion 2 of 5The telecom chatbot gives generic answers because the LLM was trained on public data and doesn't know the company's plans and services. Which approach adds that company knowledge without retraining?ARemove the system prompt so the model answers freelyBRaise the temperature so answers vary moreCUse a smaller model trained on public dataDRAG, using a knowledge base of company dataCheck answerQuestion 3 of 5Before the company's chat logs, tickets, and call recordings go into the knowledge base, what must happen to them?AThey must be collected, anonymized, and cleansedBThey must be deleted after a weekCThey must be converted into imagesDThey must be translated into five languagesCheck answerQuestion 4 of 5The chatbot must also order new phones and upgrade customers' plans. Which capability adds this?ABenchmark datasetsBAgentsCFine-tuningDA bigger context windowCheck answerQuestion 5 of 5How will the telecom company measure its satisfaction goal?AModel perplexity on the chat transcriptsBROUGE score of each chatbot responseCA survey after each issue, scored out of 5DNumber of tokens used per conversationCheck answerDeploying the Application6 exam-style questions on this lesson.Retrieval-Augmented Generation8 exam-style questions on this lesson.