ImportError: cannot import name '_get_trace_graph'

[123mw123]

** fvcore version of PathManager will be deprecated soon. **
** Please migrate to the version in iopath repo. **
https://github.com/facebookresearch/iopath

Traceback (most recent call last):
File "object_det_pipeline.py", line 31, in
from modules.deep_learning.MaskRCNN_detectron2.api import RunModel as RunModel_Object
File "/home/Saiteja/pipeline/modules/object_detection/source/../../../modules/deep_learning/MaskRCNN_detectron2/api.py", line 3, in
from detectron2.engine import DefaultPredictor
File "/home/Saiteja/detectron2/detectron2/engine/init.py", line 11, in
from .hooks import *
File "/home/Saiteja/detectron2/detectron2/engine/hooks.py", line 15, in
from fvcore.nn.precise_bn import get_bn_modules, update_bn_stats
File "/home/Saiteja/venv_Pipeline/lib/python3.6/site-packages/fvcore-0.1.3.post20210223-py3.6.egg/fvcore/nn/init.py", line 2, in
File "/home/Saiteja/venv_Pipeline/lib/python3.6/site-packages/fvcore-0.1.3.post20210223-py3.6.egg/fvcore/nn/activation_count.py", line 9, in
File "/home/Saiteja/venv_Pipeline/lib/python3.6/site-packages/fvcore-0.1.3.post20210223-py3.6.egg/fvcore/nn/jit_analysis.py", line 14, in
ImportError: cannot import name '_get_trace_graph'

How to resolve this? Can somebody help me

Environment:

Distributor ID: Ubuntu
Description: Ubuntu 18.04.5 LTS
Release: 18.04
Codename: bionic

GPU: GeForce RTX 3070
Cuda: 10.0

torch installation: pip3.6 install torch==1.3.1+cu100 torchvision==0.4.2+cu100 -f https://download.pytorch.org/whl/torch_stable.html

git commit ID: 94d0f13

[github-actions]

You've chosen to report an unexpected problem or bug. Unless you already know the root cause of it, please include details about it by filling the issue template.
The following information is missing: "Instructions To Reproduce the Issue and Full Logs"; "Your Environment";

[ppwwyyxx]

We do not support pytorch 1.3 as https://detectron2.readthedocs.io/en/latest/tutorials/install.html says

[123mw123]

We do not support pytorch 1.3 as https://detectron2.readthedocs.io/en/latest/tutorials/install.html says

But I am able to train and test on Ubuntu 16.04, NVIDIA TITAN RTX, CUDA 10.0 using torch 1.3

[voldemortX]

@ramadevsaiteja I met a similar issue with torch1.3 since the lastest fvcore does not support torch1.3 anymore. Rolling fvcore back to earlier versions can solve this.
For example, I use this version. You can do this by:

pip uninstall fvcore
git clone https://github.com/facebookresearch/fvcore.git
cd fvcore
git checkout c0ba80ac330ac08dd27ef1ce2e69c455d6f48e56
cd ..
pip install -e fvcore

[123mw123]

Found the solution. Just do pip install fvcore=='0.1.dev200114'