Everything needed to develop a solution to this task and to score it the way the contest did.
code/
baseline/ unofficial educational baseline — for use outside the contest environment
grading/ unofficial educational evaluation — replays the grading outside the contest
baseline-original/ official: the baseline exactly as kept during the contest
grading-original/ official: the grader exactly as used to score submissions
Run it outside the contest (unofficial, educational)
The *-original folders are the official contest artifacts, preserved verbatim — they are what actually ran during IOAI 2026, and they remain the authoritative reference. baseline/ and grading/ are unofficial educational versions, provided purely for convenience: they let anyone use the task code outside the contest environment and read the data directly from the Hugging Face dataset, with one click on Google Colab. Where the two disagree, the originals win.
| Run the baseline — fetches the dataset from Hugging Face and runs the untouched baseline | |
To evaluate your own attempt: open the evaluation notebook, upload your edited solution.ipynb via the Files pane, Run all. |
The data is not in this repository — it is published as a standalone Hugging Face dataset:
The dataset has public/ (what contestants received) and private/ (the graded leaderboard A/B splits) — its dataset card documents every subset.
Download it and place it so the notebook can find it (see the note below), then open the baseline.
Running the baseline
The baseline is the longest of the six and walks through loading both classifiers, comparing their predictions, and assembling a submission.
cd code/baseline-original
jupyter lab solution.ipynb
Needs torch, torchvision and timm. Model R is torchvision.models.resnet18 and Model V is timm/vit_tiny_patch16_224, both loaded directly from their hubs. Paths are set by the DATA_DIR, MODELS_DIR, TRAIN_SPLIT and TEST_SPLIT environment variables.
Scoring a solution
code/grading-original/ holds the grader used during the contest. It reads a solution plus the private data and prints a score. The grader's own README documents how it is invoked and what it expects; it was built to run inside a prepared container image, so running it locally takes some setup.
A note on the baselines
These are deliberately simple — several score close to zero. They are starting points that show the input and output format correctly, not strong solutions. Improving on them is the exercise.