Practice
Learn
← Quit
0 / 0 correct
LLMs & AI
Hard
1 of 10
How should you evaluate a RAG system?
A
By model perplexity
B
Separately measure retrieval (did the right chunks come back, e.g. recall@k) and generation (is the answer correct and faithful to the sources)
C
Only by asking users if they like it
D
By counting tokens