-
Notifications
You must be signed in to change notification settings - Fork 74k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
TensorFlow upgrade to 1.0.0 breaking import #7980
Comments
Same issue here. List of package versions: |
Yeah, I wouldn't expect it to break like that if 0.9.0 works fine. Any necessary updates of packages should be handled at installation. |
We test our upgrade script from 0.12 to 1.0 |
Gunan, I tried all the upgrades in between 0.10.0, 0.11.0, and 0.12.0 and they all have the same problem. 0.9.0 still works when I pip install it though. |
Ah, wait now I see. Could you try running |
I'm in macOS 10.12.3 and I have the same error. When I try to install mock it says it is already installed. Forgot to mention that I'm installing it directly from PIP and not using Anaconda. |
Sergio could you copy paste your full error message?
The last error can look similar, but the real error may be hidden in the
big stack trace preceeding the last import error.
…On Mar 3, 2017 12:38 AM, "Sergio Díaz" ***@***.***> wrote:
I'm in macOS 10.12.3 and I have the same error. When I try to install mock
it says it is already installed.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#7980 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AHlCOWQP1VOrSCHHBsuyAl_CsXe9Zrogks5rh9GhgaJpZM4MQC1z>
.
|
No gunan, I have mock. I have used mock before in my own projects. That is not the issue. This appears to be a tensorflow issue and it seems to have appeared when tensorflow went from one version to two versions - one for the CPU and one for the GPU. |
The first stack trace you have in the report is tied to mock. Tensorflow also had one CPU and one GPU version in 0.9.0, too. I am suspecting pip tool itself may be struggling to update everything correctly. Could you try first uninstalling then reinstalling tensorflow?
|
pip install --upgrade --ignore_installed tensorflow "--ignore_installed" solved the issue in my case. Otherwise some of the packages are using system installed ones instead of in virtual environment. Thanks! |
pip install --upgrade --ignore_installed tensorflow should be: pip install --upgrade --ignore-installed tensorflow and it did not solve my issue... same error as before. But I am not using a virtual environment. |
@xiaoliangbai I'm glad the issue is resolved for you. @MikeTam1021 This is interesting. What is your Anaconda installation located? |
My anaconda installation is in my home directory.. I use pip without sudo all the time, so I would imagine we do not need to use sudo. I think I am going to try to start developing in a container. I would really like to get this working though as it would speed my development time. |
Sorry for all the inconvenience @MikeTam1021 Have you been using the GPU version, or the CPU version with 0.9? |
I believe it is the CPU version |
Got it!!! So, I tried the virtual environment conda install. It worked (but was buggy, as noted). Then I thought "maybe this will work on my native system". It did. Now I can access tensorflow 1.0.0 from my system. Many thanks! conda install -c conda-forge tensorflow |
I encountered the same problem. I tried all the method mentioned but they did not work. How about you? @MikeTam1021 |
I did a conda install ningyuma. That solved it for me. |
@MikeTam1021 the problem is that such solution don't install the GPU version, as it is documented here: https://www.tensorflow.org/install/install_linux |
I was not installing the GPU version hfranco. Conda installed worked for me. pip install was breaking on the import statement in the REPL. |
@MikeTam1021 Good approach if you are not using CUDA. The other side has the nightmare of installing tensorflow and anaconda alongside |
Still facing the same problem..i did try the all the above mentioned solution. Nothing worked for me.. |
I am also facing the issue. Tensorflow was working fine in my system, but suddenly this problem appears. I am using ubuntu 16.04 and installation is running on virtualenv. My tensorflow verson is tensorflow==1.1.0
|
try this |
@imanabbasnejad Thanks, it solved my problem. |
export PBR_VERSION='your pbr version', it's will be worked, but i don't know why VersionInfo('mock') is wrong, maybe setup.cfg in mock has some problem. |
Imported from GitHub PR openxla/xla#7980 This PR is originally submited as part of openxla/xla#7940. It aims to make load/store logic more general so that it can be optimized to vector load/store pattern. It changes IR from ```llvm %linear_index_plus_base = add nuw nsw i32 %linear_index_base, %loop.indvar %linear_index1 = add nuw nsw i32 %linear_index_plus_base, 1 %linear_index2 = add nuw nsw i32 %linear_index_plus_base, 2 %linear_index3 = add nuw nsw i32 %linear_index_plus_base, 3 %21 = getelementptr inbounds float, ptr %0, i32 %linear_index_plus_base %22 = load float, ptr %21, align 4, !invariant.load !4 %26 = getelementptr inbounds float, ptr %0, i32 %linear_index1 %27 = load float, ptr %26, align 4, !invariant.load !4 %31 = getelementptr inbounds float, ptr %0, i32 %linear_index2 %32 = load float, ptr %31, align 4, !invariant.load !4 %36 = getelementptr inbounds float, ptr %0, i32 %linear_index3 %37 = load float, ptr %36, align 4, !invariant.load !4 ``` to ```llvm %linear_index_plus_base = add nuw nsw i32 %linear_index_base, %loop.indvar %21 = getelementptr float, ptr %0, i32 %linear_index_plus_base %22 = getelementptr inbounds float, ptr %21, i32 0 %23 = load float, ptr %22, align 4, !invariant.load !4 %29 = getelementptr float, ptr %0, i32 %linear_index_plus_base %30 = getelementptr inbounds float, ptr %29, i32 1 %31 = load float, ptr %30, align 4, !invariant.load !4 %37 = getelementptr float, ptr %0, i32 %linear_index_plus_base %38 = getelementptr inbounds float, ptr %37, i32 2 %39 = load float, ptr %38, align 4, !invariant.load !4 %45 = getelementptr float, ptr %0, i32 %linear_index_plus_base %46 = getelementptr inbounds float, ptr %45, i32 3 %47 = load float, ptr %46, align 4, !invariant.load !4 ``` The former one does not always work for different backends since it needs additional pass to handle GEP pattern. There are only ~20 lines of core code changes, the others are all UTs changes. Copybara import of the project: -- df1286a48461dd6337c841d4a353b694ce60bf86 by Sheng, Yang <yang.sheng@intel.com>: Make vector load/store logic more general -- 66f0782ad62e90e01fd8de7b41d20d607f974844 by Sheng, Yang <yang.sheng@intel.com>: fix IR in UTs -- b944ee015c177d4180f9cba6b564cf72e1c80bbc by Sheng, Yang <yang.sheng@intel.com>: Add comments Merging this change closes #7980 PiperOrigin-RevId: 595953368
On the Mac OS X El Capitan V 10.11.6 using python 2.7.11 with anaconda
when I run this code:
pip install https://storage.googleapis.com/tensorflow/mac/tensorflow-0.9.0-py2-none-any.whl
I get a working version of tensorflow 0.9.0. import tensorflow in python throws no errors.
I type in the command:
pip install --upgrade tensorflow
I get a successful install
and I get a broken import statement in python (.. modify dir names)
when I import again in the same interface, I get a different error which repeats if I do it anymore
The text was updated successfully, but these errors were encountered: