Skip to main content
T3 Chips and hardware Taichu Internship Round 1

Taichu AI Infra internship first round (1)

Real Taichu AI Infra internship first-round interview questions covering training optimization

Training optimization

Below is what Taichu covered in the AI Infra internship first round.

Project experience

  1. Detailed walkthrough of internship work

Fundamentals

  1. Overall Transformer structure
  2. Architecture differences between BERT and GPT
  3. Data parallelism: principle and implementation
  4. Core features of FasterTransformer
  5. Common optimizers (e.g. SGD, Adam) and their traits
  6. Random forest principle
  7. How GBDT works

Inference optimization

  1. How to raise throughput for online inference serving

Coding

  1. Add two numbers represented as linked lists