https://marovi.ai/index.php?action=history&feed=atom&title=Translations%3AStochastic_Gradient_Descent%2F27%2Fzh Translations:Stochastic Gradient Descent/27/zh - Revision history 2026-04-27T22:02:11Z Revision history for this page on the wiki MediaWiki 1.39.1 https://marovi.ai/index.php?title=Translations:Stochastic_Gradient_Descent/27/zh&diff=5474&oldid=prev DeployBot: Batch translate Stochastic Gradient Descent unit 27 → zh 2026-04-27T03:38:16Z <p>Batch translate Stochastic Gradient Descent unit 27 → zh</p> <p><b>New page</b></p><div>* &#039;&#039;&#039;数据洗牌&#039;&#039;&#039; —— 在每个 epoch 重新打乱数据集，避免出现循环模式。<br /> * &#039;&#039;&#039;{{Term|gradient clipping|梯度裁剪}}&#039;&#039;&#039; —— 对梯度范数进行截断，以防止更新爆炸，尤其是在循环神经网络中。<br /> * &#039;&#039;&#039;{{Term|batch normalization|批归一化}}&#039;&#039;&#039; —— 对层输入进行归一化可降低对{{Term|learning rate|学习率}}的敏感度。<br /> * &#039;&#039;&#039;混合精度训练&#039;&#039;&#039; —— 使用半精度浮点数能在现代 GPU 上加速 SGD，同时几乎不损失精度。</div> DeployBot