Checklist AICC 2026 Round 4 · Task 2
Alchemy
Predict the result of Little Alchemy element combinations by assigning each of 70 test pairs a unique result from a candidate list, using only bert-base-uncased.
The task
In the game Little Alchemy, players combine elements to discover new ones (for example, fire and water make steam; dinosaur and fire make a dragon).
The training set gives 150 combination rules (item1 + item2 → result). The test set gives 70 pairs with hidden results, and a list of 70 candidate results is provided. All items are lowercase English words or short phrases, and the assignment is bijective: each test pair maps to exactly one candidate and each candidate is used once.
Contestants predict the correct result for each test pair.
Abridged by SOTA from the official materials. The official statement has the exact rules, and it wins wherever this summary differs.
At a glance
- You get
train.csv(150 rows: item1, item2, result),test.csv(70 rows: Id, item1, item2) andcandidates.csv(70 unique results).- You submit
submission.csvwith columns Id and result.- Scoring
- Accuracy: correct assignments divided by 70. Baseline about 0.08, reference solution about 0.41.
- Rules
- bert-base-uncased is the only model allowed; no other pretrained models or external data.
- Individual participation (maximum team size 1); at most 30 submissions per day.
- AICC contest rules (stated on each Kaggle rules page, not enforceable): no use of LLMs for writing code or getting task ideas; no internet use other than official library documentation and the contest platform; no communication with anyone during the contest; clarifications only via the #clarification-requests channel on the AICC Discord server.
- Format
- AICC Round 4, online on Kaggle, 20 Feb 2026 18:00 UTC – 22 Feb 2026 18:00 UTC.