training failed on win10 with GPU or CPU

mbrossar / ai-imu-dr

AI-IMU Dead-Reckoning

MIT License

788 stars 222 forks source link

training failed on win10 with GPU or CPU #53

Open xxn59 opened 3 years ago

xxn59 commented 3 years ago

Hi, great job there! I wished to reproduce the results and encountered some issue. I changed the value of train_filter to 1 in main_kitti.py and ran the script, but when first epoch loss was calculated, the back propagation stuck for hours and cannot be trained. (in train_torch_filter.py train_loop func) tried for both GPU and CPU settings and failed the same. my environment: CPU: intel 7700 GPU: nvidia rtx2070 w/ cuda 10.1 OS: win10 2004 python3.7 pytorch1.7_gpu

hqtang-git commented 3 years ago

Hi, Did you find out the reason, I also take about 434 secs for the first epoch in loss_train.backward() by cpu training model.

Akudavale commented 7 months ago

@xxn59 any update on ur issues, did you solve it?