Databricks Generative AI Engineer Associate Question 205
Select 4You are tasked with building a Retrieval-Augmented Generation (RAG) application on Databricks. Which of the following are essential elements required to create the application?
- A
Model flavor
- B
Embedding model
- C
Retriever
- D
Input prompts
- E
Model signature
- F
Training dataset
Show answer and explanation
Correct answers: A, B, C, E
Explanation
To create a RAG application, you need the basic components such as the model flavor to select the generative model, the embedding model to encode data for retrieval, the retriever to fetch relevant documents, and the model signature to define how the model interacts with inputs and outputs. Input prompts and training datasets are not core requirements for the RAG application setup.
- A. Correct.
Model flavor is essential as it specifies the type of generative model (e.g., OpenAI, Hugging Face) being used in the RAG pipeline.
- B. Correct.
Embedding model is crucial for converting documents and queries into vector representations for similarity search.
- C. Correct.
Retriever is necessary to fetch relevant documents or data from the knowledge base to enhance the generative model's responses.
- D. Incorrect.
Input prompts are not a mandatory element for building the RAG application itself; they are typically used during inference or testing.
- E. Correct.
Model signature is required to define the input-output structure of the model, ensuring compatibility within the RAG pipeline.
- F. Incorrect.
Training dataset is not an essential component for deploying a RAG application. Pretrained models and embeddings are often used.