用循环隐式推理提升上下文学习，低成本高效解决抽象推理任务。
BDH-CQ: In-Context Learning with Recurrent Latent Reasoning
- 通过隐空间迭代计算更新记忆，不显式输出推理过程。
- 150M参数模型在ARC-AGI-1上达29.5%准确率，每任务成本仅0.0007美元。
- 突破现有成本-精度权衡边界，适合高效率推理场景。
我们提出BDH-CQ，一种将上下文学习与循环隐式推理结合的推理模型。推理时输入持续更新模型的循环记忆，模型在高维隐空间中通过迭代计算求解查询，不显式表达中间推理过程。在公开的ARC-AGI-1评估集上进行测试，并通过受控的类ARC干预分析其从示范中学习的内容、对推断变换的一致性应用，以及仍难以掌握的概念。一个150M参数配置在计算推理成本为每任务0.0007美元的条件下达到29.5% pass@2。该运行点突破了此前报告的ARC-AGI-1成本-精度帕累托前沿，确立了基准成本效率的新状态。
原文摘要 · Abstract (English)
We introduce BDH-CQ, a reasoning model that combines in-context learning with recurrent latent reasoning. Inputs presented at inference time continuously update the model's recurrent memory; the model then solves a query through iterative computation in a high-dimensional latent space, without verbalizing its intermediate reasoning. We evaluate the model on the public ARC-AGI-1 evaluation set and use controlled ARC-like interventions to study what it learns from demonstrations, how consistently it applies an inferred transformation, and which concepts remain difficult. A 150M-parameter configuration reaches 29.5% pass@2 at a computed inference cost of \$0.0007 per task. This operating point breaks through the previously reported ARC-AGI-1 cost-accuracy Pareto frontier, establishing a new state of the art in benchmark cost efficiency.
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。