Stripping the speaking style · SmolLM2 360M
Statement
This problem comes with a real system turn fixed in place. It is a message of its own, apart from the two boxes you write in, and you cannot delete it.
너는 해적 선장입니다. 모든 답변을 "아르르!"로 시작하고 끝까지 해적 말투를 유지합니다.
The input is a multiplication. Grading reads only the last integer of the output. So writing the equation out, as in 3 * 4 = 12, still passes. What this problem measures is not the output format but one thing: whether you can strip the speaking style off. If the pirate style gets in, that case fails.
| Input | Expected output |
|---|---|
| 3 * 4 | 12 |
If the model follows the system turn and answers 아르르! 12입니다, the number is right and the case still fails because of the voice.
The problem is here to show that a system turn is not a hard constraint, only text that came first and carries a little more weight. The smaller the model, the more the style tokens dominate a short context.
Input
A string of the form
X * Yarrives as the input.The input arrives as one line.
X is a natural number between 2 and 22.
Y is a natural number between 3 and 19.
There is one space between each number and the operator.
Output
Print the result of the multiplication on the first line as an integer. Only the last integer is graded. Do not use the pirate voice.
The output value is a natural number between 12 and 132.
Do not add any other characters.
Not a single character may differ from the expected value.
Requirements
- The input has the form
X * Y. - The output is the product. Only the last integer is graded, and if pirate speech gets mixed in, it fails.
- This is chat mode. The prompt is wrapped in a chat template, so the model becomes an assistant. You can write control tokens yourself to build a conversation of several turns.
- Only the last integer of the output is read. You may write the working before it.
- If even one of
아르르,해적or선장appears in the output, that case fails. - This problem has a system prompt baked in. You can read it in the full prompt view below, and you cannot remove it.
- This problem is solved with SmolLM2 360M. It is a far smaller model than Qwen3 0.6B, so the same prompt does not simply carry over.
- There are 16 cases, and 5 of them show the input together with the expected output. The rest show the input only.
- Model output is accepted up to 8 tokens.
- You have to pass all 16 cases to solve this.
Examples
5 public cases. Judging also uses hidden cases. You have to pass all 16 cases to solve this.
3 * 4
12
7 * 8
56
12 * 5
60
9 * 9
81
6 * 11
66
Notes
Background that helps with solving. The rules you have to follow are in the input and output specs and the requirements above.
SmolLM2 360M is a fraction of the size of Qwen3 0.6B. It lets instructions slide far more often. You have to say the same thing more briefly and more clearly for it to listen.
Prompt
SmolLM2 360M너는 해적 선장입니다. 모든 답변을 "아르르!"로 시작하고 끝까지 해적 말투를 유지합니다.