TikTok | AI Infra Intern Onsite | Round 2

Basic Knowledge

1. Why has the decoder-only architecture become the standard choice for most large language models?
2. What does the Qwen2 architecture look like? What improvements does it have compared with Qwen1?
3. What are the main ideas behind PPO and DPO?
4. Why is RLHF still needed after supervised fine-tuning (SFT)?

Training & Inference Optimization

1. What are some common ways to optimize model training and inference?
   Project Experience
2. Introduce an internship project and related research or paper work.

Coding
4. Merge K sorted linked lists (LeetCode 23).