README.md

metadata

configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*
content: ''
dataset_info:
  features:
    - name: problem_id
      dtype: int64
    - name: problem_name
      dtype: string
    - name: language
      dtype: string
    - name: interactive
      dtype: bool
    - name: time_limit_s
      dtype: float64
    - name: memory_limit_mb
      dtype: float64
    - name: console_input
      dtype: bool
    - name: input_filename
      dtype: string
    - name: output_filename
      dtype: string
    - name: multiple_solutions
      dtype: bool
    - name: original_source
      dtype: string
    - name: challenge
      dtype: string
    - name: examples_in
      sequence: string
    - name: examples_out
      sequence: string
    - name: tags
      sequence: string
    - name: year
      dtype: int64
    - name: grade
      dtype: string
  splits:
    - name: test
      num_bytes: 820507
      num_examples: 304
  download_size: 398279
  dataset_size: 820507

RO OJI

This dataset contains Romanian county level Computer Science challenges (OJI - Olimpiada Județeană de Informatică) from grade V to grade XII, spanning 2002 - 2023.

Only toy test examples are made public. Real tests can be performed on kilonova.

Reference

@misc{dumitran2024evaluatingperformancelargelanguage,
      title={Evaluating the Performance of Large Language Models in Competitive Programming: A Multi-Year, Multi-Grade Analysis}, 
      author={Adrian Marius Dumitran and Adrian Catalin Badea and Stefan-Gabriel Muscalu},
      year={2024},
      eprint={2409.09054},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      url={https://arxiv.org/abs/2409.09054}, 
}