采用这篇博客里的断点训练https://blog.csdn.net/txpp520/article/details/104406880?utm_medium=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-2.edu_weight&depth_1-utm_source=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-2.edu_weight]
报错![](https://i-blog.csdnimg.cn/blog_migrate/05e0c47a6bb00771171570b5da915d3c.png)
这个原因是说 原先的参数值是29858,现在是31668,那我就将29858对应的参数的值改成31668.
修改完后又报错:
报错: Dst tensor is not initialized.
原因: GPU不够用
报错: ResourceExhaustedError (see above for traceback): OOM when allocating tensor with shape
原因: GPU不够用