invertible condition

```
w_hat = self.w / torch.norm(self.w, p=2)
```

==>

```
w_hat = self.w / torch.norm(self.w, p=2)**2
```

Look at the planar flow implementation in the repository, likely in a file like planar_flow.py or a Jupyter notebook. Find the line where w_hat is computed and change the denominator from torch.norm(self.w, p=2) to torch.norm(self.w, p=2)**2. Verify the change by running any existing tests or examples to ensure the flow remains invertible.

由索引模型根據 Issue 內容生成。