发表机构
National Yang Ming Chiao Tung University（国立阳明交通大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
URCHIN提出一种基于水平脉冲神经网络的语言模型，采用Dale定律连接组和SSM/RSNN双实现，在BabyLM挑战中实现生物合理、高效且可直接部署的预训练。
AI 中文摘要
BabyLM挑战衡量的是模型从发展上合理、儿童规模的数据而非互联网规模语料库中能学习多少语言，然而以往的语言模型忽视了获取人类语言的神经回路的生物学约束：分为兴奋性和抑制性群体的脉冲神经元，通过循环侧向连接组进行连接。本文提出URCHIN（统一循环连接组与水平整合-发放神经元），将并行分层连接组脉冲状态空间模型（PHCSSM）应用于语言建模：由Dale定律侧向连接组耦合的泄漏整合-发放神经元通过多传输循环解析每个词元，将活动循环至不动点。该实例刻意保持最小化：单层128个神经元，无注意力机制，参数量为423万。两种实现共享同一组权重并产生相同的基准分数，因此URCHIN只需训练一次，无需转换步骤即可任意部署：一种适用于GPU高效训练的并行状态空间模型（SSM）扫描，或一种适用于CPU或神经形态边缘部署的具有恒定成本推理的事件驱动循环脉冲神经网络（RSNN）。在所有三个BabyLM赛道（Strict-100M、Strict-Small和Multilingual）中，URCHIN提供了一个生物学上合理、高效且可直接部署的参考点。
英文摘要
The BabyLM challenge measures how much language a model can learn from developmentally-plausible, child-scale data rather than internet-scale corpora, yet prior language models forgo the biological constraints of the neural circuitry that acquires human language: spiking neurons separated into excitatory and inhibitory populations wired by a recurrent lateral connectome. This paper presents URCHIN (Unified Recurrent Connectome with Horizontal Integrate-and-fire Neurons), which applies the Parallelized Hierarchical Connectome Spiking State-space Model (PHCSSM) to language modeling: leaky integrate-and-fire neurons coupled by a Dale's-law lateral connectome resolve each token through a multi-transmission loop that recirculates activity to a fixed point. The instantiation is deliberately minimal: a single horizontal layer of 128 neurons, no attention, and 4.23M parameters. Two implementations share one set of weights and produce identical benchmark scores, so URCHIN is trained once and deployed either way with no conversion step: a parallel state-space model (SSM) scan that is GPU-efficient for training, or an event-driven recurrent spiking neural network (RSNN) with constant-cost inference for CPU or neuromorphic edge deployment. Across all three BabyLM tracks (Strict-100M, Strict-Small, and Multilingual), URCHIN offers a biologically plausible, efficient, and directly deployable reference point.
Comments12 pages, 2 figures, 6 tables. Accepted to the BabyLM Challenge 2026 (BabyLM Workshop, EMNLP 2026)
