Practice
Learn
← Quit
0 / 0 correct
LLMs & AI
Hard
1 of 20
What is a semantic cache for LLM responses, and its main risk?
A
Storing prompts encrypted; risk: key loss
B
Reuse a previous answer when a new query's embedding is very similar; risk: returning a wrong answer for a subtly different question
C
Caching model weights; risk: staleness
D
Caching tokens on the GPU; risk: overflow