hsingyu-chou

Results 4 comments of hsingyu-chou

ImportError: cannot import name 'feature_decorator_ext' from partially initialized module 'mmdet3d.ops.feature_decorator' (most likely due to a circular import) (/data/bevfusion/mmdet3d/ops/feature_decorator/__init__.py)

I found this solution from other website. It works but I don't know why. You could try this. 1. mmdet3d/ops/feature_decorator/feature_decorator.py : Change “from mmdet3d.ops.feature_decorator import feature_decorator_ext” to “from . import...

compilation terminated. error: command '/usr/bin/gcc' failed with exit code 1

Try this. 1. \bevfusion\mmdet3d\ops\ball_query\src\ball_query.cpp 2. \bevfusion\mmdet3d\ops\knn\src\knn.cpp 3. \bevfusion\mmdet3d\ops\group_points\src\group_points.cpp 4. \bevfusion\mmdet3d\ops\interpolate\src\interpolate.cpp 5. \bevfusion\mmdet3d\ops\furthest_point_sample\src\furthest_point_sample.cpp 6. \bevfusion\mmdet3d\ops\gather_points\src\gather_points.cpp #include extern THCState *state; cudaStream_t stream = THCState_getCurrentStream(state); Replace with → #include #include cudaStream_t stream =...

Conda Environment：ModuleNotFoundError: No module named 'flash_attn'

Try this. /bevfusion/mmdet3d/models/backbones/radar_encoder.py : from flash_attn.flash_attention import FlashMHA → from flash_attn.modules.mha import MHA I think this error is caused by the difference between the version 1.0.9 and 2.0.0 of flash-attention.

I am running the official training and have encountered the following error, which has caused me to be unable to train properly. Can anyone help me?

Try this. !pip install yapf==0.40.1