2025 USA-NA-AIO Round 2, Problem 3, Part 7

AI Olympiads 2025 USA-NA-AIO Round 2
1

Part 7 (5 points, non-coding task)

This part follows Part 6.

Do the following tasks.

  1. Explain why token lists of all samples begin with token ID 101.

  2. Explain why token lists of all samples end with token ID 102.

2

\color{green}{\text{### WRITE YOUR SOLUTION HERE ###}}

  1. In BERT, token ID 101 refers to a special token [CLS] that denotes the start of a sentence.

  2. In BERT, token ID 102 refers to a special token [SEP] that denotes the end of a sentence.

\color{red}{\text{""" END OF THIS PART """}}