Databricks Generative AI Engineer Associate Question 154
Single answerYour team is developing an AI-powered customer service chatbot for a global e-commerce platform. The chatbot needs to understand and respond in multiple languages, handle domain-specific queries about product details and order status, and operate efficiently in a production environment with high traffic. Which LLM would be the best choice for this application?
- A
A general-purpose open-source model such as GPT-NeoX-20B
- B
A domain-tuned LLM fine-tuned on e-commerce data and multilingual tasks
- C
A small-scale LLM optimized for low-latency responses, but trained only on English data
- D
A proprietary LLM with high accuracy in multilingual tasks but limited to low query throughput
Show answer and explanation
Correct answer: B
Explanation
The best LLM for the application should meet all the requirements: understanding multilingual queries, handling domain-specific data, and operating efficiently in a production environment with high traffic. A domain-tuned LLM that has been fine-tuned on e-commerce data and multilingual tasks satisfies all these criteria, making it the optimal choice for the customer service chatbot.
- A. Incorrect.
A general-purpose open-source model like GPT-NeoX-20B may lack the pre-training or fine-tuning required for domain-specific tasks and multilingual support, making it less effective for the chatbot's needs.
- B. Correct.
A domain-tuned LLM fine-tuned on e-commerce data and multilingual tasks is the best choice as it is tailored to handle domain-specific queries and multiple languages, while also being efficient for production use.
- C. Incorrect.
A small-scale LLM optimized for low-latency responses but trained only on English data would not meet the multilingual requirements of the chatbot, even if it performs well in terms of response speed.
- D. Incorrect.
A proprietary LLM with high accuracy in multilingual tasks but limited to low query throughput would not scale well in a high-traffic production environment, making it unsuitable for this use case.