Pratik Worah, Randall Bateman · 2026 · arXiv
DOI: 10.48550/arXiv.2610.28214 · open-access PDF
La pérdida neuronal progresiva es la característica de la enfermedad neurodegenerativa, sin embargo, cómo la pérdida de neuronas forma el patrón de declive cognitivo sigue siendo mal comprendido. Aquí sistemáticamente debilitamos las neuronas (unidades computacionales) en modelos de lenguaje que abarcan las arquitecturas de memoria a corto plazo recurrente (LSTM) y transformador y aplicamos ajustes finos limitados como un simple proxy para la plasticidad compensatoria. Con la ablación de la unidad entera, las LSTMs se degradan gradualmente con facultades sintácticas y semánticas disminuyendo en paralelo a través de daños leves a moderados.
Machine translation (Meta M2M100); the English original follows.
Progressive neuronal loss is the hallmark of neurodegenerative disease, yet how loss of neurons shapes the pattern of cognitive decline remains poorly understood. Here we systematically ablate neurons (computational units) in language models spanning recurrent long short-term memory (LSTM) and Transformer architectures and apply limited fine-tuning as a simple proxy for compensatory plasticity. With whole-unit ablation, LSTMs degrade gradually with syntactic and semantic faculties declining in parallel across mild-to-moderate damage. By contrast, Transformers exhibit late threshold-like collap
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28214, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange