please provides one requirements.txt

[wfs123456]

方便提供一个requirements安装环境，不然会有各种报错，
例如：
File "/home/sam_road-main/model.py", line 102, in forward
src_features = point_features[batch_indices, pairs[:, :, 0]]
IndexError: tensors used as indices must be long, byte or bool tensors

python3.8/site-packages/torch/nn/modules/transformer.py", line 437, in forward
return torch._transformer_encoder_layer_fwd(
RuntimeError: expected scalar type BFloat16 but found Float

[wfs123456]

另外咨询一下单卡v100-32G训练速度怎么会这么慢，

[havinggood]

Hello, I have also encountered the above error, may I ask how you solve it?

[Nodemana]

This is a versioning issue. We need the exact versions of all the dependencies used in this project. If @htcr could please provide a requirements.txt or their python environment.

[TCXX]

One of the authors here. Can you try updating all the dependencies to the latest version?

And yes, training can be slow. There is potential room for optimization.

[wfs123456]

这里的作者之一。您可以尝试将所有依赖项更新到最新版本吗？

是的，训练可能会很慢。有潜在的优化空间。

One of the authors here. Can you try updating all the dependencies to the latest version?

And yes, training can be slow. There is potential room for optimization.

I have another problem, I have trained my own dataset (0.2m/one pixels), the following image is an example（contains test_gprah_label_generator output in dataset.py）. This data lane line is slender but there are few intersections node (degree=1/3), so there are few points of interest with a degree != 2, which leads to the training results of RoadMask is very good, and itscMask is very poor, which ultimately affects the prediction results of TOPO. How should it be adjusted in function test_graph_label_generator()?

[wfs123456]

Hello, I have also encountered the above error, may I ask how you solve it?

更新torch版本

[htcr]

方便提供一个requirements安装环境，不然会有各种报错， 例如： File "/home/sam_road-main/model.py", line 102, in forward src_features = point_features[batch_indices, pairs[:, :, 0]] IndexError: tensors used as indices must be long, byte or bool tensors

python3.8/site-packages/torch/nn/modules/transformer.py", line 437, in forward return torch._transformer_encoder_layer_fwd( RuntimeError: expected scalar type BFloat16 but found Float

Hi, could you pls pull the code again? I remember I have fixed this.

[htcr]

另外咨询一下单卡v100-32G训练速度怎么会这么慢，
I was using a 4090. Also were you using the smallest ViT variant as the original paper?

[htcr]

This is a versioning issue. We need the exact versions of all the dependencies used in this project. If @htcr could please provide a requirements.txt or their python environment.

Thanks for looking into this. I can try to sort them out these days.

[wfs123456]

This is a versioning issue. We need the exact versions of all the dependencies used in this project. If @htcr could please provide a requirements.txt or their python environment.

Thanks for looking into this. I can try to sort them out these days.

请教下我6楼私有数据集的问题，下面是预测结果point_mask，road_mask, topo，其中 pointmask和topo效果很差