I have been trying to get Keras working on my laptop running El Capitan but when I attempt to import it I get the following message
Using TensorFlow backend.
Illegal instruction: 4
I've looked for solutions, and have tried updating theano, installing mxnet-mkl, and running an older version of numpy (1.13) to avoid a FutureWarning error. None of these has seemed to fix the issue, though. I feel like I must be missing something somewhere.
Try installing an older version of TensorFlow (1.5 seems to work). AVX2 instructions are now enabled by default in latest TF pip wheels, meaning if your processor doesn't support these instructions, then you will get an illegal instruction error.
Another solution is to install TensorFlow from source, which should detect your processor and configure it accordingly.
Related
Platform: Precision 5820, 32G, rtx4000; Win 10 Pro, Arcgis Pro 2.6 concurrent license;
Issue:
I installed the deep learning tools following the guidelines provided here:
deeplearninginstallation
tersorflow was not found after installation so I manually installed the 2.1.0 version. I now have arcgis 1.8.2, pro 2.6, fastai 1.0.60, python 3.6.12, pytorch 1.4.0, tensorflow-gpu 2.1.0; environment check in arcgis pro python seemed fine.
However, after I select toolbox-image analyst-deeplearning-traindeeplearningmodel, the program seems to go into a hang, with most buttons disabled/unresponsive, this would continue until I force terminate the program. I also ran into "tool not licensed" twice, which was gone after I restarted the program; and a "name 'CallBackHandler' is not defined" once, which was also gone after I restarted.
I tried runing the command from the arcgis pro python prompt:
TrainDeepLearningModel(r"**", r"**", 40, "RETINANET", 16, "# #", None, "RESNET50", None, 10, "STOP_TRAINING", "FREEZE_MODEL")
executing the command would also send the program into a hang similar to the previous one. Monitor shows that ram and GPU usage haven't changed much, so I left the program running for an hour before forcibly terminating it.
I'd greatly appreciate it if anyone can tell me what the issues are here. I'll post any other env parameters if anyone requires. Cheers.
I got the tool up and running now by running conda install -c pytorch -c fastai fastai=1.0.54 pytorch=1.1.0 torchvision scikit-image and removing all the conflicting specifications in the cloned arcgispro-py3 env that I had. Now I still don't understand what went wrong. Presumably one or more packages in the env was conflicting. But seeing as I'm not a python expert, I couldn't identify the exact issue.
Before this I tried the versions stated here deeplearning install guide, but wasn't able to get pass tensorflow-gpu because python kept checking conflications. Now I actually don't have tensorflow-gpu in the env. I have tensorflow 2.1.0, keras-applications 1.0.8/base 2.3.1/preprocesing 1.1.0 (no keras-gpu), scikit-image 0.17.2, pillow 6.2.1, fastai 1.0.54, pytorch 1.1.0, libtiff 4.0.10. Some are different from what the guideline provided.
Thing is when I ran the process, CPU usage was up and GPU wasn't despite the fact that I specified GPU as the processing core. But I have much more pressing things to do right now like getting the analysis finished. So I'll probably tweek the env around a little after I'm done with this bit and see what happens. Meanwhile, anyone's input is still welcome.
I am doing a Differential Privacy Course and have to work with syft. When running the below command, I encountered a error
I have already installed syft package required for the analysis. And I am working in Anaconda on Windows 10
from syft.frameworks.torch.differential_privacy import pate
WARNING:tf_encrypted:Falling back to insecure randomness since the required custom op could not be found for the installed version of TensorFlow (1.13.1). Fix this by compiling custom ops.
It is not an error it is a warning. It comes from tf_encrypted which is used in tensorflow. The problem is that on Windows secure randomness is not supported (according to https://github.com/tf-encrypted/tf-encrypted/issues/513). And as you use PyTorch you can just ignore it.
I'm having an error running simple code using cv2 module.
It's just:
import cv2
img = cv2.imread('sudoku.png',0)
cv2.imshow('image',img)
And it fails with the following error:
QObject::moveToThread: Current thread (0x1b74720) is not the object's thread (0x1e57d70).
Cannot move to target thread (0x1b74720)
I googled this error and tried a lot of things but it doesn't help. I tried installing without pip, I tried using step-by-step installation (from official OpenCV) but nothing helps.
When I run:
cv2.__version__
It returns 3.4.3
As noted already, the basis for this problem is discussed in opencv-python issue 46, and results from the duplication of the following libraries both on the host and the opencv-python distro libQtDBus libQtCore and libQtGui.
It has been lately addressed in the newest release of opencv-python. It is not a fix to the source code, rather the fix is to force pip to compile the newly available source via
pip install --no-binary opencv-python opencv-python
This will cause opencv-python to use the same libraries as the host, so the conflict no longer exists.
According to this issue posted on the OpenCV GitHub, this is a known issue that the developer states is damn near impossible to fix. It is apparently caused by a conflict in any Qt installations on the system with the Qt that is shipped with OpenCV. There are some suggestions floating around to remove the libqt5x11extras5 package from the system. This may fix it for some but anyone running these libraries on a Linux distribution that uses a window manager based on Qt will render their desktop environment unusable by removing this package (having tried it myself).
You can try building OpenCV from source using the WITH_GTK=ON option when running cmake which will use GTK instead of Qt, circumventing the conflict. However, this is hard to make use of in Python when using virtual environments.
I haven't spent the time to fully appreciate this problem, but as I understand this is caused by multiple conflicting versions of some plugin in the environment. I tried installing building opencv-python but there were errors with that approach. Another suggestion is to change your import order, but I've had mixed success with that, and I couldn't get it to work on a project today.
But I found a workaround that worked for me. Install opencv-python-headless instead of opencv-python. This will avoid installing the conflicting plugins. It may not work for you depending on what features of opencv you need.
$ pip uninstall opencv-python
$ pip install opencv-python-headless
The error was fixed on my system, by simply updating one library.
To find out where it is coming from, assuming Linux, try the following,
LD_DEBUG=files python -c "import cv2"
or,
LD_DEBUG=files python -c "import cv2 ; img = cv2.imread('myimage.png',0) ; cv2.imshow('image',img) ; cv2.waitKey(0)"
On my machine, it failed in one of the blas libraries. I updated that library and the code now runs without error.
This is an old bug you can find it discussed in a number of online communities.
My test code is as follows. Notably the error has not reoccurred, and apparently has nothing to do with Qt.
import cv2
img = cv2.imread('sudoku.png',0)
cv2.imshow('image',img)
cv2.waitKey(0)
To me, the solution to this problem was removing Anaconda, then installing pip followed by installing OpenCV with a simple pip install command.
I have following environment in my windows 10 machine
Python : 3.6.0
Anaconda:4.3.1
Tensorflow:1.1.0
Screen Shots
OS:Windows 10-64bit
Now when I am trying to install keras into my system I am getting a huge list of errros.
Detailed Error Log
Now I have two questions here.
Can I install keras into my system when I already have tensorflow GPU version which was really hard to install?
If keras can be installed into my this system configuration then will my tensorflow GPU version work properly afterwards?
I have installed Anaconda, theano , GPU Toolkit ver 8. I am getting this error.
ERROR: refusing to load cuda driver library because the version is blacklisted. Versions 373.06 and below are known to be ok.
If you want to bypass this check and force the driver load define GPUARRAY_FORCE_CUDA_DRIVER_LOAD in your environement.
ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
See discussion here.
I reinstalled libgpuarray and it worked for me.
Your cuda driver is faulty. Install another one, preferably 373.06. Using your current driver will result in wrong computations. DO NOT force the driver load