search

mit-han-lab / bevfusion

[ICRA'23] BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representation
https://bevfusion.mit.edu
Apache License 2.0
2.26k stars 409 forks source link

Segmentation fault(core dumped) #157

Closed jiapeng789 closed 1 year ago

jiapeng789 commented 2 years ago

Hello, when I run: "torchpack dist-run -np 2 python tools/train.py configs/nuscenes/det/transfusion/secfpn/camera+lidar/swint_v0p075/convfuser.yaml --model.encoders.camera.backbone.init_cfg.checkpoint pretrained/swint-nuimages-pretrained.pth" to train the model, I get an error:

Process received signal Signal: Segmentation fault (11) Signal code: Address not mapped (1) Failing at address: 0xfffffee8

This error occurs when the model starts training or after training for a while, I tried to run: python setup.py develop to recompile the code, but the problem is still not solved, I suspect it is the version of spconv. Please how can I solve this problem, thank you.

kentang-mit commented 2 years ago

What about trying out the docker? We have provided some instructions in the README.md.

jiapeng789 commented 2 years ago

Thanks for your reply, I will try to use docker to configure the environment, when will the training method of the fusion part be released. Thanks again for your great work.

kentang-mit commented 1 year ago

Temporarily closed since the issue is resolved.