Skip to main content
T0 Big tech ByteDance Internship Round 2

ByteDance Doubao AI Infra internship second round

Real ByteDance AI Infra internship second-round interview questions covering inference optimization and training optimization

Inference optimization Training optimization

Fundamentals

  1. Why did decoder-only become the mainstream choice for today’s large models?
  2. What is Qwen2’s model structure? What improvements vs Qwen1?
  3. What are the core ideas of PPO and DPO respectively?
  4. Why do you still need an RLHF stage after SFT?

Inference optimization

  1. What common model training and inference optimization methods exist today?

Project experience

  1. Introduce your internship project and paper research.

Coding

  1. Merge K sorted lists (LeetCode 23).