Add dataset card

README.md

@@ -0,0 +1,35 @@
+---
+license: cc-by-4.0
+pretty_name: "IOAI 2026 – Robot Chasing"
+tags:
+- ioai
+- olympiad
+- grid-world
+- instruction-following
+viewer: false
+---
+
+# IOAI 2026 — Robot Chasing
+
+Official contest data for **Robot Chasing**, task 2 (Day 1) of the [IOAI 2026](https://ioai-official.org) Individual Contest, held in Astana, Kazakhstan.
+
+Six robots operate in `6×6` grid rooms (an `8×8` array including walls). Each robot receives an English instruction describing a behaviour; contestants must predict the robots' actions from observations.
+
+The task statement, translations, baseline and grader live in the [IOAI-2026 GitHub repository](https://github.com/IOAI-official/IOAI-2026/tree/main/Individual-Contest/2_Robot_Chasing). This repository holds **data only**.
+
+## Subsets
+
+| Subset | Path | Contestants saw it | Role |
+|---|---|---|---|
+| Public – train | `public/train/` + `public/train_answers.json` | yes | labelled training observations |
+| Public – dry-run | `public/pretrain/`, `public/pretest/` + answers | yes | small slices for pipeline validation |
+| Private – Leaderboard A | `private/test_leaderboard_a/` + `private/test_leaderboard_a_answers.json` | no | **live leaderboard** during the contest |
+| Private – Leaderboard B | `private/test_leaderboard_b/` + `private/test_leaderboard_b_answers.json` | no | **final ranking** |
+
+Answers files sit at the subset root (`*_answers.json`), not inside the split folders — the layout the contest grader expects.
+
+> ⚠️ **The `private/` subset contains the graded splits and their answer keys.** Published for post-contest study; the task cannot be reused for competition.
+
+## Credits
+
+Proposed by **Salem Lahlou**; developed by **Anuar Aimoldin**, **Kamalkhan Artykbayev** and **Nurdaulet Akhanov**. Published by the IOAI 2026 Scientific Committee under CC BY 4.0.