The lost epic
Repair the tokenizer's damaged mapping using its original BERT model and the text of "Pan Tadeusz".
AI task evaluation platform
SprawdzAI combines a broad task collection with an evaluation system designed to recreate real contest conditions while making learning enjoyable.
A constantly growing collection of original tasks and regular contests with prizes.
Tasks from OAI, IOAI, and international olympiads-all in one place and one consistent format.
Fast, automatic evaluation on GPU machines available to everyone completely free of charge.
Compete, compare results, and improve alongside other AI enthusiasts.
A simple path from task to solution.
We publish olympiad-style tasks with clear limits and metrics in a consistent format while preserving the conditions of the original challenge.
We publish olympiad-style tasks with clear limits and metrics in a consistent format while preserving the conditions of the original challenge.
Solve tasks locally or in Google Colab. Every task opens in Colab with one click, so you can focus on the solution instead of environment setup.
Solve tasks locally or in Google Colab. Every task opens in Colab with one click, so you can focus on the solution instead of environment setup.
Submit your solution with one click and receive validation and test-set results without searching for datasets or configuring evaluation.
Submit your solution with one click and receive validation and test-set results without searching for datasets or configuring evaluation.
Track your progress, compare results with other participants, and take part in our contests for a chance to win prizes.
Track your progress, compare results with other participants, and take part in our contests for a chance to win prizes.