不靠预训练，用压缩最小化解出20%的视觉谜题。
ARC-AGI Without Pretraining
- 仅在推理时优化目标谜题的描述长度，无需任何预训练。
- 76K参数模型在无训练数据下解出20%的ARC-AGI测试题。
- 适合研究通用智能与极小数据学习的学者参考。
当前主流观点认为，解决类似于智力测验的视觉谜题（如ARC-AGI-1基准）需依赖大规模预训练模型。为挑战这一观点，我们提出CompressARC，一个仅76K参数、完全无预训练的模型，通过在推理阶段最小化目标谜题的描述长度（MDL），成功解决了20%的评估谜题。该方法使模型具备极强泛化能力，远超传统深度学习模型。据我们所知，CompressARC是首个在单一样本（目标推理谜题本身）上训练且不使用预提供的ARC-AGI“训练集”的深度学习方法。在极端数据受限条件下，通常预期无法解决任何谜题，但该模型仍能应对多样且富有创造性的测试题，表明描述长度最小化可作为替代预训练的可行智能生成路径。
原文摘要 · Abstract (English)
Conventional wisdom in the age of LLMs dictates that solving IQ-test-like visual puzzles from the ARC-AGI-1 benchmark requires capabilities derived from massive pretraining. To counter this, we introduce CompressARC, a 76K parameter model without any pretraining that solves 20% of evaluation puzzles by minimizing the description length (MDL) of the target puzzle purely during inference time. The MDL endows CompressARC with extreme generalization abilities typically unheard of in deep learning. To our knowledge, CompressARC is the only deep learning method for ARC-AGI where training happens only on a single sample: the target inference puzzle itself, with the final solution information removed. Moreover, CompressARC does not train on the pre-provided ARC-AGI "training set". Under these extremely data-limited conditions, we do not ordinarily expect any puzzles to be solvable at all. Yet CompressARC still solves a diverse distribution of creative ARC-AGI puzzles, suggesting MDL to be an alternative feasible way to produce intelligence, besides conventional pretraining.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。