The IOL-AI Challenge: An Open Challenge towards Advancing Linguistic Reasoning

Eduardo Sánchez, Rita Berrada, Dan-Mircea Mirea +8

Reasoning in LLMs is overwhelmingly studied in domains that provide a model with rules: mathematics and code. Linguistic puzzles invert this: the solver must first discover the sys…