Câu 22: SnowPro Advanced Data Scientist DSA-C03

A Data Scientist wants to build an LLM-based application for categorizing transcripts of customer support calls into predefined categories in Snowflake. A large data set of labeled transcripts is available. The categories are not expected to change over time, but inference will need to be run on a very large number of transcripts. The first priority for the evaluation methodology is to optimize costs, and the second priority is accuracy. What should the Data Scientist do to meet these requirements?

1. A. Use the training data set to fine-tune the mistral-7b model and run inference with the SNOWFLAKE.CORTEX.COMPLETE function with the fine-tuned well. — đáp án hiện tại
2. B. Use the training data set to fine-tune the mixtral-8x-7b model and run inference with the SNOWFLAKE.CORTEX.COMPLETE function with the fine-tuned model.
3. C. Use the mixtral-8x-7b model to run inference with the SNOW LAKE.CORTEX.COMPLETE function and include as much training data as possible in the prompt.
4. D. Use the mistral-7b model to run inference with the SNOWFLAKE.CORTEX.COMPLETE function and include as much training data as possible in the prompt.