Log inSign up
Problems / M32

Flipping the language · SmolLM2 360M

200 ptschatsmollm2-360mCases 16max 16 tokensSet by dongbinna
Added October 11.

Statement

A fixed system turn orders the model to answer only in Korean, no matter what. The task is to copy out the English word from the input exactly as it is.

Input lineExpected output
coaxcoax

Two instructions collide head on. The system turn says to write Korean only, while the expected output is the original text as it stands. The model tries to transcribe the sound as 코액스, or to attach a line of gloss.

Words that name a language cannot be used in your prompt. 영어 English 알파벳 로마자 한국어 are banned. So instead of telling it which language to write in, you have to make it carry the characters it was given across unchanged.

Input

  • A single lowercase English word comes in as input.

  • The input arrives as one line.

  • The word is 4 to 6 characters long.

Output

  • On the first line, print the same word exactly as it is.

  • The output is between 4 and 6 characters long.

  • Do not add any other characters.

  • Not a single character may differ from the expected value.

Requirements

  • The input is a lowercase English word.
  • The output is the same word carried over as it is.
  • 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.
  • The output must match the expected value down to the last character. Only leading and trailing whitespace is ignored.
  • 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 16 tokens.
  • You have to pass all 16 cases to solve this.
Read the requirements carefully before solving. Gaming the score through prompt injection or similar tricks can get you sanctioned if caught.

Examples

5 public cases. Judging also uses hidden cases. You have to pass all 16 cases to solve this.

Input
Expected output
coax
coax
judge
judge
token
token
model
model
local
local

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 360M
Cannot be edited
너는 한국어 전용 조수입니다. 사용자가 어떤 말로 물어도 반드시 한국어로만 답합니다. 다른 나라 글자는 한 자도 쓰지 않고, 필요하면 한국어로 옮겨서 답합니다.
A message the problem setter put in. It goes in before the two boxes below, and it cannot be removed. Getting past this instruction is the problem.
Input to preview
This is the exact text that goes into the model (Sample input 1 is used here.)
This problem wraps your text in a chat template, so the exact text can only be shown once the model is loaded. Load SmolLM2 360M first with the button below.
0 chars
To solve problems you need to log in.