Quiz: Production Deployment
30 questions on exporting, optimization, and building products.
Question 1
What format is used for CPU inference with llama.cpp?
A) ONNX
B) GGUF
C) PyTorch
D) TensorFlow
Show Answer
Correct Answer: B
GGUF is the standard format for quantized models used with llama.cpp for CPU inference.
Question 2
What is vLLM optimized for?
A) Training
B) High-throughput inference serving
C) Model merging
D) Data preparation
Show Answer
Correct Answer: B
vLLM is optimized for high-throughput inference serving using PagedAttention.
Question 3
What should you do with LoRA before production deployment?
A) Delete it
B) Merge into base model
C) Keep it separate always
D) Compress it
Show Answer
Correct Answer: B
For production, merging LoRA into the base model simplifies deployment and eliminates adapter overhead.
Question 4
What quantization provides the best speed/quality trade-off?
A) FP32
B) FP16
C) INT8
D) INT2
Show Answer
Correct Answer: C
INT8 quantization typically offers good speed improvement with minimal quality loss for production.
Question 5
What is PagedAttention in vLLM?
A) A training technique
B) Efficient KV cache memory management
C) A model architecture
D) A data format
Show Answer
Correct Answer: B
PagedAttention efficiently manages KV cache memory through paging, enabling higher throughput.
(Questions 6-30 cover additional deployment topics)
🎉 Congratulations!
You've completed the LLM Fine-Tuning Complete course!