https://terms-en.ai-term-hub.com/zh/tags/empirical/Recent content in Empirical on 中文AI术语词典Hugozh-cnSat, 18 Jul 2026 11:44:45 +0000https://terms-en.ai-term-hub.com/zh/terms/neural_scaling_law/Sat, 18 Jul 2026 11:28:12 +0000https://terms-en.ai-term-hub.com/zh/terms/neural_scaling_law/<h2 id="definition">Definition</h2> <p>神经缩放定律描述了模型性能与其规模（包括数据集大小、参数量和计算预算）之间的可预测的幂律关系。这些定律表明，随着规模的扩大，模型性能会按特定规律提升。</p> <h3 id="summary">Summary</h3> <p>基于数据、参数或计算量增加来预测模型性能提升的经验关系。</p> <h2 id="key-concepts">Key Concepts</h2> <ul> <li>幂律关系</li> <li>参数缩放</li> <li>数据效率</li> <li>计算最优训练</li> </ul> <h2 id="use-cases">Use Cases</h2> <ul> <li>大语言模型的资源规划</li> <li>模型架构选择</li> <li>性能预测</li> </ul> <h2 id="related-terms">Related Terms</h2> <ul> <li><a href="https://terms-en.ai-term-hub.com/en/terms/chinchilla-optimization-chinchilla%E4%BC%98%E5%8C%96/">Chinchilla optimization (Chinchilla优化)</a></li> <li><a href="https://terms-en.ai-term-hub.com/en/terms/loss-scaling-%E6%8D%9F%E5%A4%B1%E7%BC%A9%E6%94%BE/">Loss scaling (损失缩放)</a></li> <li><a href="https://terms-en.ai-term-hub.com/en/terms/emergent-abilities-%E6%B6%8C%E7%8E%B0%E8%83%BD%E5%8A%9B/">Emergent abilities (涌现能力)</a></li> <li><a href="https://terms-en.ai-term-hub.com/en/terms/compute-budget-%E8%AE%A1%E7%AE%97%E9%A2%84%E7%AE%97/">Compute budget (计算预算)</a></li> </ul>