Shy

> radius() radius_graph() can be calculated in data processing stage. You are right. The edge radius can be calculated offline. However, during online inference, this procedure cannot be accelerated by...

@HLSS-Hen Hi, I've come up with a great idea. I experimented with using PyTorch 2's features and found that invoking `model = torch.compile(model)` in the `train_qcnet.py` significantly reduces memory usage....

For example, you can have a try to use different lr or use the default lr. ``` model = QCNet.load_from_checkpoint(checkpoint_path=ckpt_path) model = QCNet.load_from_checkpoint(checkpoint_path=ckpt_path, lr=lr) ```

[ m = self.mode_emb.weight.repeat(scene_enc['x_a'].size(0), 1)](https://github.com/ZikangZhou/QCNet/blob/55cacb418cbbce3753119c1f157360e66993d0d0/modules/qcnet_decoder.py#L151)

你可以参考test函数中的代码，将模型的轨迹映射回全局坐标系，然后将字典里的map_point可视化点，map_polygon可视化线的连接关系。

收到，非常感谢您！我到时候试一试------------------ 原始邮件 ------------------ 发件人: "ybko"