AINeutralarXiv – CS AI · Feb 277/107
🧠
Compositional-ARC: Assessing Systematic Generalization in Abstract Spatial Reasoning
Researchers developed Compositional-ARC, a dataset to test AI models' ability to systematically generalize abstract spatial reasoning tasks. A small 5.7M parameter transformer model trained with meta-learning outperformed large language models like GPT-4o and Gemini 2.0 Flash on novel geometric transformation combinations.