Closed GuangyanZhang closed 4 years ago
我该怎么操作才能解决这个问题呢?希望您能指导一下,谢谢
You can try this docker: kesun/pose:pytorch-0.4.1-bn-cudnn-off If possible, you adopt the official sync-bn provided by pytorch.
这个docker "kesun/pose:pytorch-0.4.1-bn-cudnn-off" 可以提供一个链接吗?谢谢!
https://cloud.docker.com/u/kesun/repository/docker/kesun/pose
I think you can pull this docker without any link.
好的,非常感谢
hi your website 'https://cloud.docker.com/u/kesun/repository/docker/kesun/pose' is not working now. Could you tell me what is wrong?
This site maybe requires over the wall,I think
Hi I can visit google,but I cannot visit this website,Therefore, it is not the reason of wall
I think website is limitted,you can wait a moment.......
@GuangyanZhang @sde123 Sorry about that. https://hub.docker.com/r/kesun/pose/.
However, I have no idea why you need to visit this website. You can just run "docker pull kesun/pose:pytorch-0.4.1-bn-cudnn-off" to pull this image.
Ok,Do you solve the above problem of BN and pytorch that you said?
my website is limitted,so I want download by other computer.but I find that my environment cann't use docker.....
@sde123 I don't,I should adopt the official sync-bn provided by pytorch.that's bad.
@sde123 你好,之前那个搞定了,现在又出现这样的问题,我是不是哪里路径不对或者参数不对呢?,能指导下吗?
File "tools/train.py", line 245, in
I am confused by this problem(ninjia),too. Could you teach me to solve it,please?
you should install pytorch 0.4.0 and ninja~
@GuangyanZhang Did you check your config (yaml) file? Does it contain STAGE4?
@GuangyanZhang 你好,我安装pytorch0.4.1和您出现了同样的问题,又安装了pytorch0.4.0和ninja,出现了下面这样的问题,请问您出现过吗,怎么解决呢?
File "tools/train.py", line 27, in
@Frostinassiky you should change the default.py~
@YanYunYun ,没有出现过,这个应该还是ninja的问题
@YanYunYun
Seg/lib/models/sync_bn/inplace_abn/src/inplace_abn.cpp -o inplace_abn.o
ninja: build stopped: subcommand failed.
did you solve this problem?
@sxhsunday
I solved this issue by installing pytorch 1.1 and using pytorch-v1.1 branch.
@YanYunYun Thank you very much. it is work now.
@sxhsunday @YanYunYun
I install pytorch 1.1 using conda and checkout pytorch v1.1 branch. However, I still got the same error. Can you provide instruction how you install pytorch 1.1 and checkout pytorch v1.1 branch? ALso which version ninja fo you use?
Thanks,
@sxhsunday @YanYunYun
I install pytorch 1.1 using conda and checkout pytorch v1.1 branch. However, I still got the same error. Can you provide instruction how you install pytorch 1.1 and checkout pytorch v1.1 branch? ALso which version ninja fo you use?
Thanks,
I have the same error as you have you solve it? thanks,
I found that this error relates to version matching problem, specifying between Torch and ninja version. Therefore I solved this problem by finding ninja version which matching with Torch version. My environment as show as below: Ubuntu: 16.04 Gcc: 6.5.0 Python: 3.5.2 Torch: 0.4.1 ninja: 1.8.2 Hopefully that it can help you guys!
@sxhsunday I solved this issue by installing pytorch 1.1 and using pytorch-v1.1 branch.
It works!
您好,首先我出现这样的问题: RuntimeError: Ninja is required to load C++ extensions 然后我pip install ninja成功以后 又出现这样的问题: /usr/local/lib/python3.5/dist-packages/torch/utils/cpp_extension.py:118: UserWarning:
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Your compiler (c++) may be ABI-incompatible with PyTorch! Please use a compiler that is ABI-compatible with GCC 4.9 and above. See https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html.
See https://gist.github.com/goldsborough/d466f43e8ffc948ff92de7486c5216d6 for instructions on how to install GCC 4.9 or higher. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
warnings.warn(ABI_INCOMPATIBILITY_WARNING.format(compiler)) Traceback (most recent call last): File "tools/train.py", line 27, in
import models
File "/data/HRNet-Semantic-Segmentation-master/tools/../lib/models/init.py", line 11, in
import models.seg_hrnet
File "/data/HRNet-Semantic-Segmentation-master/tools/../lib/models/seg_hrnet.py", line 22, in
from .sync_bn.inplace_abn.bn import InPlaceABNSync
File "/data/HRNet-Semantic-Segmentation-master/tools/../lib/models/sync_bn/init.py", line 1, in
from .inplace_abn import bn
File "/data/HRNet-Semantic-Segmentation-master/tools/../lib/models/sync_bn/inplace_abn/init.py", line 1, in
from .bn import ABN, InPlaceABN, InPlaceABNSync
File "/data/HRNet-Semantic-Segmentation-master/tools/../lib/models/sync_bn/inplace_abn/bn.py", line 14, in
from functions import *
File "/data/HRNet-Semantic-Segmentation-master/lib/models/sync_bn/inplace_abn/functions.py", line 16, in
extra_cuda_cflags=["--expt-extended-lambda"])
File "/usr/local/lib/python3.5/dist-packages/torch/utils/cpp_extension.py", line 514, in load
with_cuda=with_cuda)
File "/usr/local/lib/python3.5/dist-packages/torch/utils/cpp_extension.py", line 690, in _jit_compile
return _import_module_from_library(name, build_directory)
File "/usr/local/lib/python3.5/dist-packages/torch/utils/cpp_extension.py", line 773, in _import_module_from_library
return imp.load_module(module_name, file, path, description)
File "/usr/lib/python3.5/imp.py", line 242, in load_module
return load_dynamic(name, filename, file)
File "/usr/lib/python3.5/imp.py", line 342, in load_dynamic
return _load(spec)
ImportError: /tmp/torch_extensions/inplace_abn/inplace_abn.so: undefined symbol: _ZN2at5ErrorC1ENS_14SourceLocationESs
请问这个BN和pytorch是要同步编译吗?我的pytorch==0.4.1