Trying to backward through the graph a second time

原因是把创建loss的语句loss_aux = torch.tensor(0.)放在循环体外了，可能的解释是第一次backward后把计算图删除，第二次backward就会找不到父节点，也就无法反向传播。参考：https://stackoverflow.com/questions/55268726/pytorch-why-does-preallocating-memory-cause-trying-to-backward-through-the-gr

本栏目推荐文章
• Eloquent 模型使用详解 Has One Through 远程一对一
• 基于正则化的图自编码器在推荐算法中的应用 Application of graph auto-encoders based on regularization in recommendation algorithms
• 使用 Microsoft Graph oFFICE365 sendmail C#
• If you are running inside a VM, you may need to enable SSE4.2 pass-through. 报错问题解析
• 【略读论文|大模型相关】Zero-Shot Relational Learning on Temporal Knowledge Graphs with Large Language Models
• P7830 [CCO2021] Through Another Maze Darkly
• [ARC105E] Keep Graph Disconnected 题解
• pytorch反向传播错误解决：RuntimeError: Trying to backward through the graph a second time, but the buffers have already been freed. Specify retain_graph=True when calling backward the first time.
• CF295B Greg and Graph 题解 floyd性质题
• Unity3D Shader Graph 使用 DDXY 节点达到抗锯齿的原理详解