<![CDATA[nandehuit的专栏]]>https://blog.csdn.net/nandehuitzh-cnhttps://blog.csdn.net/<![CDATA[从梯度到大模型（005）：彻底理解 PyTorch Tensor、dtype、device 与 shape]]>https://blog.csdn.net/nandehuit/article/details/165890679https://blog.csdn.net/nandehuit/article/details/165890679nandehuitFri, 18 Sep 2026 21:48:20 +0800<![CDATA[从梯度到大模型（004）：数值梯度检查，怎样确认梯度没有写错]]>https://blog.csdn.net/nandehuit/article/details/165890146https://blog.csdn.net/nandehuit/article/details/165890146nandehuitFri, 18 Sep 2026 21:16:42 +0800<![CDATA[从梯度到大模型（003）：Loss、梯度与学习率，模型究竟怎样学习]]>https://blog.csdn.net/nandehuit/article/details/165363920https://blog.csdn.net/nandehuit/article/details/165363920nandehuitMon, 14 Sep 2026 22:20:56 +0800<![CDATA[从梯度到大模型（002）：矩阵乘法与线性模型]]>https://blog.csdn.net/nandehuit/article/details/165363679https://blog.csdn.net/nandehuit/article/details/165363679nandehuitMon, 14 Sep 2026 22:05:53 +0800<![CDATA[从梯度到大模型（001）：NumPy 数组、形状与广播]]>https://blog.csdn.net/nandehuit/article/details/165219307https://blog.csdn.net/nandehuit/article/details/165219307nandehuitSun, 13 Sep 2026 12:19:20 +0800<![CDATA[从梯度到大模型（000）：一条面向实践的 LLM 系统学习路线]]>https://blog.csdn.net/nandehuit/article/details/165217771https://blog.csdn.net/nandehuit/article/details/165217771nandehuitSun, 13 Sep 2026 12:06:55 +0800