search

megvii-research / PETR

[ECCV2022] PETR: Position Embedding Transformation for Multi-View 3D Object Detection & [ICCV2023] PETRv2: A Unified Framework for 3D Perception from Multi-Camera Images
Other
879 stars 132 forks source link

No loss displayed while training #34

Closed GuuuuuG closed 2 years ago

GuuuuuG commented 2 years ago

just this. No loss displayed while training 2022-08-03 17:16:37,279 - mmdet - INFO - workflow: [('train', 1)], max: 24 epochs 2022-08-03 17:16:37,281 - mmdet - INFO - Checkpoints will be saved to /root/paddlejob/workspace/wangguangjie/PETR/work_dirs/petr_r50dcn_gridmask_p4 by HardDiskBackend. 2022-08-03 17:19:34,002 - mmdet - INFO - Saving checkpoint at 1 epochs 2022-08-03 17:22:33,118 - mmdet - INFO - Saving checkpoint at 2 epochs 2022-08-03 17:25:32,982 - mmdet - INFO - Saving checkpoint at 3 epochs

yingfei1016 commented 2 years ago

Do you modified some codes, such as nuscenes-mini dataset?

GuuuuuG commented 2 years ago

是否修改了一些代码，例如 nuscenes-mini 数据集？

Yes, I changed the data to v1.0-mini, do you mean to change it to full data?

yingfei1016 commented 2 years ago

Yes, the log will be printed every 50 iters,. When sue the nuscenes-mini dataset the iters may not be enough for printing. https://github.com/open-mmlab/mmdetection3d/blob/v0.17.1/configs/_base_/default_runtime.py#L7

GuuuuuG commented 2 years ago

Yes, the log will be printed every 50 iters,. When sue the nuscenes-mini dataset the iters may not be enough for printing. https://github.com/open-mmlab/mmdetection3d/blob/v0.17.1/configs/_base_/default_runtime.py#L7

Thanks, that's the problem, it's solved