A development team is building a conversational application with Snowflake Cortex Analyst to allow business users to ask follow-up questions about structured dat a. They are specifically designing the multi-turn conversation support and considering the underlying LLM choices for components like the summarization agent. Which of the following statements accurately reflects how Cortex Analyst handles conversational context and best practices for selecting an LLM for its summarization agent?

• A. Cortex Analyst directly passes the entire raw conversation history to every LLM call for all agents to ensure full context, which generally improves performance.
• B. A dedicated LLM summarization agent is introduced before the original workflow to distill conversation history into a concise context for subsequent agents, with Llama 3.1 70B identified as a suitable model due to its high summarization quality.
• C. Multi-turn conversation support primarily relies on caching previous SQL query results and re-executing them for follow-up questions, avoiding additional LLM calls for context summarization.
• D. The summarization agent in Cortex Analyst is primarily responsible for generating SQL queries from conversation history, thus requiring an LLM optimized for text- to-SQL tasks.
• E. To optimize for latency, Cortex Analyst recommends always using the smallest possible LLM for the summarization agent, such as Llama 3.1 8B, even if it has a slightly higher error rate in rewriting questions.

Explanation: Only visible for PassReview members. You can sign-up / login (it's free).

A data scientist is preparing to log a custom PyCaret classification model into the Snowflake Model Registry. The goal is to deploy this model on Snowpark Container Services (SPCS) for scalable inference. The PyCaret model relies on the 'pycaret' and 'scipy' Python libraries, and the data scientist has local 'sample data.csv' for inferring the model's signature. Which statements are crucial for successfully logging this custom model for eventual SPCS deployment?

• A. Option B
• B. Option E
• C. Option D
• D. Option C
• E. Option A

Explanation: Only visible for PassReview members. You can sign-up / login (it's free).

A Gen AI engineer is tasked with selecting the most suitable Large Language Model (LLM) from Snowflake Cortex AI for a new customer service chatbot. They need to rapidly prototype and compare different LLMs with varying parameters on a sample dataset before committing to a production deployment. Which of the following statements accurately describe how the Cortex Playground (Public Preview) can assist in this scenario?

• A. It enables side-by-side comparison of model outputs for different LLMs and model settings, facilitating an informed decision on model selection.
• B. It supports exporting the tested prompts and model configurations as Python code, ready for integration into a Snowpark ML pipeline.
• C. It allows direct fine-tuning of selected LLMs with custom datasets within the playground interface to improve model performance for specific tasks.
• D. It allows connection to a Snowflake table with textual data, processing up to 100 rows, to experiment with prompts directly on actual data.
• E. It provides a mechanism to deploy the chosen LLM directly into Snowpark Container Services (SPCS) compute pools from within the playground for immediate production use.

Explanation: Only visible for PassReview members. You can sign-up / login (it's free).

A data operations team is attempting to scale up their Document AI automated pipeline. They are using a Snowflake Task to process a large volume of daily scanned invoices and receipts, which are stored in an internal stage 'financial_docs stage'. The current processing involves documents that are frequently around 75 MB each, and often there are batches exceeding 1 ,200 documents in a single day. The pipeline is failing consistently. Which of the following factors could be contributing to the failures in this Document AI automated pipeline? (Select all that apply.)

• A. The number of documents in a single daily batch (exceeding 1 ,200) surpasses Document AI's processing limit per query.
• B. The Snowflake Task is configured as a 'SERVERLESS TASK, which is not supported by Document AI.
• C. The individual document size (75 MB) exceeds the maximum supported file size for Document AI.
• D. The was created without specifying 'ENCRYPTION = (TYPE = 'SNOWFLAKE SSE')'.
• E. The account role executing the task lacks the 'SNOWFLAKCORTEX USER database role, which is a prerequisite for Document AI functions.

Explanation: Only visible for PassReview members. You can sign-up / login (it's free).

A data scientist wants to fine-tune a
mistral -7b
model to improve its ability to generate specific product descriptions based on brief input features. They have a table named PRODUCT_CATALOG with columns PRODUCT_FEATURES (text) and GENERATED_DESCRIPTION (text). Which of the following statements correctly describe the preparation and initiation of this fine-tuning job in Snowflake Cortex?
(Select all that apply)

• A. The SQL query for the training data must select columns aliased as
• B. O To generate highly structured
• C. The fine-tuning job must be created using a
• D. The
• E. Once a fine-tuned model is created, it is fully managed by the Snowflake Model Registry API, allowing for programmatic updates to its parameters and versions.

Explanation: Only visible for PassReview members. You can sign-up / login (it's free).