2019, Installing TensorFlow, Keras, & Python 3.7 in Windows

Hi this is Jeff, welcome to Applications
of Deep Neural Networks, with Washington University. In this video I’m going to
show you how to set up a Python environment for the course
with Windows. Even if you’re not a member of the class, this still shows you how to
set up Keras and TensorFlow for use with the Windows operating system I also
have a Macintosh specific version of this video as well. Of course this is
just part of my entire course on deep learning so if you’re more interested in
this sort of thing please be sure to subscribe to my channel and also click
the little bell icon so you receive all the notifications. So, this is one of
several iterations of this particular video. Google and TensorFlow and
Anaconda Python always changed something and I have to keep updating this so I’m
going to try to make this as general as possible and let you know some of the
things you might encounter. But, I do update this video each semester that I
run it so definitely check back here if you’re looking at an older version of
this. This is the January 2019 version of this and I expect this this should last
for a bit but there is TensorFlow 2.0 coming out so will will see what this is
entails now sometimes it’s a bit painful to install these machine learning
packages and I do apologize for that. But, welcome to my world!
This is really just how this stuff works. This stuff is all changing at the speed
of light and if you want to be in machine learning, you’re going to have to
be a bit of a operating system and software tinkerer. I am going to show you
how to do this in Microsoft Windows and I’ll probably create a Mac version of
this as well but this just I’ll point out the difference is kind of as we go.
Now for this one I have already installed a Python version. Anaconda… use
Anaconda Python. I recommend you do as well it simplifies some of the
installation processes for these complicated machine learning packages
like TensorFlow and Keras. If you already had a Python installed you may want to erase
your current Python and start , this is this is what I’m going to show
you how to do real quick. There’s two places that Anaconda will typically be
installed it’ll either be installed in your user directory or at the root of
your C Drive like you see here I installed it under my user directory and
I recommend that approach I think that works well it’s also recommended by
Anaconda. I’ll have links in the description of this video for all of the
places that you might want to go also if you want to have an idea of about how
long this install will take notice the time here I will time lapse this
considerably so that when things are installing you’re not. I mean you don’t
want to sit there watching a YouTube video for 10 minutes while something
downloads. Depending on how fast that would be so let’s go ahead I am going to
go into my users directory that is me lots of stuff in my users directory but
if you go down to Anaconda3 that is where it was installed I’m simply going
to nuke that now. Be careful before you do that if you have other software on
your computer that installed that for you just be aware of what you’re doing
there. If you you may want to install it into a different directory if this is
all you’re doing on Python and you’re comfortable with that. I mean like me I’m
I’m glad to rip out my Python throw in a new one I’ll deal with the consequences
from there. But if you’ve already got Python installed just be aware you could
you could run into some version issues. So I’m going to delete that this might
take some time so this is our first time lapse. And now I am Python-less so let’s
go ahead and get Python installed there’s two flavors of Python that you
can deal with: Anaconda and Miniconda recommend Minoconda simply Google
search form Miniconda and click on the link. Miniconda has just the minimum of
Python installed you’ll have to pip install the rest of the things that you
will commonly need. But it’s so much smaller and cleaner I prefer just to
install what I need. The class instructions assume that you’re doing
Miniconda. So here’s the problem that we have for
this semester. This might be completely different at
the time that you’re that you’re viewing this. So look at the description for the
video look at the pinned comment near the top. I’ll always give you a clue as
far as what might have changed with this and if it changes considerably obviously
I just recorded a new video. In which case there’ll be a link to that new
video so here we have Python 3.7 a few weeks before the class started 3.6 was
here so that’s fine, it’s the latest version of Python. But TensorFlow
requires Python 3.6. Again welcome to the world of machine learning.
Welcome to Python packages in general so what we’re going to do here is download
the 64 bit of Python 3.7. I recommend just getting the latest version of mint
Anaconda, we could go into the archive and find 3.6 but let’s just march forward
and I’ll show you how to create a 3.6 environment once we have it installed. So
you would click this to download it. Now, I already did this because that that can
potentially take a little bit depending on how all your connection speed is.
Let’s go ahead and install that. I’m going to go to the Downloads directory,
where I have that. And don’t worry about that other file, let’s go ahead and run
this this is what I downloaded. Okay you can click Next and agree. So this is
where if you pick “just me”, it’s going to install in your user directory if you
choose all users then it is going to install in your root. I recommend doing
“just me”. You can do either one but you do need special privileges if you’re going
to do the second one this is where it’s going to install it now if you are
leaving an old version in here you’ll give this something different. But this
is going to install my onto three now. I deleted Anaconda so I had anaconda
installed previously Anaconda is just the big full-blown everything I perform
more minimalistic so let’s go ahead and do this next. This is important. So you
can add Anaconda to your path environment variable. If you do that
whenever you go to dos basically it’s going to run this
“conda” now this is typically what I do, but again I’m pretty familiar in advance
with Python so I’m willing to deal with the consequences the problem you would
run into is if there’s other software installed in your machine that is using
other versions of Python installed on here that could get you into trouble.
We’re gonna go ahead and do the default installs for this. So I am NOT going to
add it to the path and we are going to register it as the default one the only
thing is when you run your Anaconda, you won’t run it from your DOS prompt
you’ll run it rather from the Anaconda prompt that you’ll get over here, that
I’ll show you shortly. So let’s go ahead and click install. This
is a pretty quick install. This is what’s great about Miniconda if you did
Anaconda you would you would be sitting here considerably longer
although that being said let’s time lapse. Alright that’s done let’s do next.
Okay now it’s installed, so let’s go ahead and take a look at the class notes
so that you can follow my installation instructions. If you just do a Google
search for “Jeff Heaton WUSTL”, that’ll take you to this
class website, or you can go directly to my GitHub repository. I just click on
this one Applications of Deep Learning, this is the course website there’s a
link to the GitHub repo right here. Alright, you’re going to want to download
all of these files. You can clone it if you know what you’re doing with Git if
you don’t I say just download the ZIP. We’ll deal with that file in a moment
for now we’re mainly interested in the installation instructions that I am
going to have you follow from that is in module one now we’re just looking at
this file in GitHub, so you can’t really run anything in Python but this lets you
see the instructions and you scroll through it. And yeah that’s me!
You’ll want to get down here where it says software installation. I’ve already
done this, we’ve downloaded Miniconda this is talking about dealing with this
python 37 issue ability issue. I’ll remove this once this
is no longer an issue probably TensorFlow 2.0 might well
resolve this but for now what we’re going to do is kind of create this, so
this is going to create a new environment for you in your in your
anaconda or Miniconda for Python 3.6 so you’ll have a 3.6 environment in your
master 3.7 environment. Let me show you what that looks like so even though you
downloaded MiniConda it still says Anaconda kind of annoying but that’s how
they do it you’ll want to click on your Anaconda prompt don’t just open a DOS
window this is your anaconda prompt. It’s basically a DOS prompt now. It is important that’s your environment you always have a base
environment. I, in my day job and work as a data scientist, I don’t use my base
environment for anything it’s it’s just sort of a loading area for everything
and I’m going to suggest you do the same. Here you need to create a new
environment that basically has your Python 3.6 so we’re going to just copy
and paste this. Always copy and paste, believe me,
this is how I roll. OK, so you’re gonna run that and that takes a moment and
then you just say yes ok you now have a new environment so notice here it says
to activate it you do Conda activate tensorflow so I’m going to do that now.
If you’re using a Mac its “source activate tensorflow”
I’m going to I usually just do “activate tensorflow” You can use the “conda” one as
well but that has some additional ramifications to it. I’m just going to do
activate or source activate if I’m using a Mac. And notice now this base has
changed to TensorFlow we are now in this new environment and
notice if we type “python version” we are on 3.6 point eight if I had
done that prior to the activate and when I was in base it’d be 3.7. So
let’s go and do the rest of this installation
we’re going to do Conda installed Jupiter. That’s Jupiter notebooks which
is the IDE of sorts that we’ll be using for this semester can also copy and
paste procede okay that was one of the bigger ones but it’s done so what I’m
gonna do is drag the window to the side okay now they’re side-by-side copy
and pasting made easy so now we need to install each of these this is also a
pretty big one this size each of these will prompt you just say yes next is
pandas and pandas data reader matplotlib is important that gives us access to all
of our graphics pillow lets us directly manipulate images and requests lets us
access HTTP web services h5py lets us read H5 files which are how Kira’s
stores neural network files yes you tell gets this information with the operating
system there’s really only one example that makes use of it so if you can’t
install this one it doesn’t matter and PS util if your if you’ve installed
anaconda instead of Miniconda this will fail for reasons that I won’t go into. If
you’re curious as to why I, ask me in the comments. By the way, I tried to
install h5py twice. If you do that it’ll simply tell you that it’s already been
installed. So let’s install the one I meant to PSUIL, okay now we’re down to
the two big ones TensorFlow and Keras and notice I have the version numbers
specify one point one two for tensorflow which should be the last TensorFlow
before going to 2.0. I do this just so that if somebody’s installing this later
and TensorFlow, upgraded in the middle of the semester you’re going to have the
material that the course was updated for. Now I’ll be updating this to TensorFlow
2.0 during the summer of of 2019 and if that is a seismic enough shift to the
installation process and it might well be I will record another video of this.
So you’ll see probably a pinned comment from me at the very bottom letting you
know if there’s there’s something you need to do here. Especially so
can also remove the double equal especially if you’re not in my class so
if you can just do pip install tensorflow and it’ll install the latest.
But I am going to go ahead and do it like this now if a problem in my
previous video that people were mentioning having was it would tell them
that it couldn’t find TensorFlow the reason for that is that we’re trying to
install TensorFlow into 3.7 Python and TensorFlow doesn’t exist for 3.7 at this
point in the time continuum so it’s not found so if you’re seen and I found
error that’s probably what’s happening you have a tensor flow to Python
mismatch I’ll post the latest information on the on the GitHub and
also in the description of this video and probably append comments so look for
those. Let’s go ahead and without further ado let’s install TensorFlow. If
anything goes wrong this is where it will. Cross your fingers! And happiness is
a TensorFlow install command the machine with no read errors at the
bottom. If you had read errors at the bottom then I suggest copying and
pasting those and Googling to see if you can find out what’s going wrong, if not
feel free to post into the comments and I’ll I’ll chime in as I as I know what
to answer for you or tell you. Where you might want to go for more information
now let’s install cures next by the way a TensorFlow 2.0 point Keras
TensorFlow and 2.00 or effectively merging so if I had a if I had to guess
what this script will look like when the that gets released is this will change
to equals two point oh two point something and this curves line will go
away but we’ll see what that actually looks like when it happens now let’s do
Keras that’s a little bit quicker. Okay Keras tends to be less error-prone and
then TensorFlow and by the way this is all for the CPU
if you want to install this on the GPU there are definitely tutorials to do
that, I’m going to do a video to show you how to use CoLab to do
which is Google’s web cloud-based product they give you a really sweet GPU
to to use with this. But, if you think this is complicated this is this is “easy
level” compared to GPU, GPU you can be a weekend of your life that you will just
never ever get back. Several weekend’s of my life. Okay now that we have all of
this let’s go ahead and do a quick test. Now I don’t have this in the Jupiter
notebook but a real quick way that you can test this is just type Python and
now you’re in Python it says python 3.6 that’s a good sign do “import tensorflow
as tf” enter notice it takes a moment if you get an error here this then you have
some debugging to do, but if you just get this back you’re golden. It’s
installed so let’s do also “print (tf.__version__) make sure you do to
under bars their version underbar underbar close parentheses and if you
see that congratulations you have TensorFlow installed. Now type “quit()” and
that okay and now we’re going to run this last line here what this line does
is makes jupiter aware of this new environment, that you’ve created for 3.6
, so i’m gonna copy that and when we’re going to kinda prompt we need
to make sure that we “activate tensorflow” always do that to get out of your base
and into tensorflow and i’m just going to paste that line it and run it and
this basically makes it so that you you have Jupyter
connected to TensorFlow to that new environment that you’ve
created will see what it’s important in just a second let’s i’ve downloaded the
the files earlier for my course website and from GitHub
and I’ve unzipped those into this jeffh directory you can see that duder command
you can see it right there T81-558 so we’ll change directory into
that make sure that you download that that you downloaded and unzip it into
your C:users wherever your your folder is. So we’re in there let’s go
ahead and launch Jupyter notebook and here we are this is Jupyter notebook so
Jupyter notebook always shows you what’s in the current directory from where you
launched it so that’s why it was so important that we see deed change
directory into where we want to be. Let’s open up module 1 and we’re going
to run just something to test that this is available that we can actually work
this correctly now. You might see this kernel not found Python 3 well if you
choose that that’s going to be base but notice this new one that’s what the
command that I just gave you did if you just see Python 3 it’s not gonna work
because that’s going to be Python 37 so choose that make sure you did that
command I gave you earlier set the kernel if you see that it can’t find
your functions like tensorflow or scikit-learn or missing the libraries
you might be in the wrong kernel so you can always see what kernel you’re in by
change colonel there you see the Python 3 and the Python 3.6 TensorFlow
always check your kernel and make sure that you have selected Python 3.6 you
can also see which one up here at the at the top right let’s go ahead and run
helpful functions this is the top of every one of the course modules we’ll
run that and “silence is bliss” if it does not give you any output, usually that
means it worked. so you’ll see a 1 there if you ran it
again that changed you a 2 it’s just the the number of times it’s been or the
kernel number that it’s run or the script number we’ll scroll down further
Python versions notice this is the output from when I ran
just before the semester started and just before they decided to up up
anaconda to 3.7 and cause numerous problems so let’s let’s go ahead and run this see
what happens that’s full back and that is awesome you
should see these this output one point twelve point oh if you’re seeing this
you’re golden it’s it’s installed. This is how powerful is your machine if you
weren’t able to install PS-utils because maybe you installed the full and a
condom this part might not work. But let’s go ahead and run this this will
show you how powerful your machine is. This is the end of this installation
you’ve got this setup if you found this interesting go ahead and subscribe to my
channel if you run into some problems Google’s your best friend or let me know
in the comments section and I will see what I can do for you thank you very

100 thoughts on “2019, Installing TensorFlow, Keras, & Python 3.7 in Windows

  1. I got error in installing jupyter-downloaded byte didn't match content length… please sir help me out

  2. Hi Jeff! Great tutorial as always. Really thankful for your effort in making these tutorials.

    I hope you don't mind me asking, how to save and run python on a server?

  3. How did you launch Jupyter notebooks? I keep entering "j" after changing the directory and it says that it is an unknown command.

  4. i must be the only one that got an error when copying and pasting pip install — upgrade tensorflow==1.12.0

  5. Getting an error while running command : conda install jupyter
    Error occurred while transaction being executed.
    ERROR: conda.core.link:_execute<568> An error occured while installing the package

  6. I was able to install as desired but I am not able to launch jupyter notebook
    Please help me giving me error
    jupyter is not recognized as internal or external command

    I had installed previously multiple python and now i am not sure if they are removed , please suggests how to remove all python verison

  7. print(tf.__version__)

    Traceback (most recent call last):

    File "<stdin>", line 1, in <module>

    AttributeError: module 'tensorflow' has no attribute '__version__'
    >the print statement showed the tensorflow's version earlier but now it shows this error. what might possibly be wrong here?

  8. thanks alot sir ,thank you thank you . Wasted 3 days of my vacation trying to install the tensorflow,i regret not watching in the first place .You are a life saver.i am short of words and i have tears of joy seeing 'using tensorflow backend' instead of those errors,you earned my subscribe….

  9. What are your thoughts on using conda install for tensorflow? I installed it right after I installed miniconda

  10. After running,
    python -m ipykernel install –user –name tensorflow –display-name "Python 3.6 (tensorflow)"
    If you get an error like,
    No module named ipykernel

    then first run this command,
    pip install ipykernel
    and finally run the command,
    python -m ipykernel install –user –name tensorflow –display-name "Python 3.6 (tensorflow)"

  11. If the jupyter notebook command doesn't work inside the tensorflow environment then,
    Install the jupyter notebook and pandas again inside the environment,
    activate tensorflow
    pip install jupyter
    pip install pandas

    then run jupyter notebook again.
    Source from,

  12. thank you so so so much for sharing your knowledge! You saved me so much time to get started on a project I really want to do.

  13. hi, at minute 16:23 you say "if you get an error here then you hve some debugging to do" how do I do that?

  14. After this command i got this error on Windows 10.

    pip install –upgrade pandas

    WARNING: Retrying (Retry(total=4, connect=None, read=None, redirect=None, status=None)) after connection broken by 'ProtocolError('Connection aborted.', ConnectionResetError(10054, 'An existing connection was forcibly closed by the remote host', None, 10054, None))': /packages/d0/4e/9db3468e504ac9aeadb37eb32bcf0a74d063d24ad1471104bd8a7ba20c97/pandas-0.24.2-cp36-cp36m-win_amd64.whl

  15. ERROR: Could not install packages due to an EnvironmentError: HTTPSConnectionPool(host='files.pythonhosted.org', port=443): Max retries exceeded with url: /packages/d0/4e/9db3468e504ac9aeadb37eb32bcf0a74d063d24ad1471104bd8a7ba20c97/pandas-0.24.2-cp36-cp36m-win_amd64.whl (Caused by ProtocolError('Connection aborted.', ConnectionResetError(10054, 'An existing connection was forcibly closed by the remote host', None, 10054, None)))

  16. Thanks for sharing this tutorial..Sir i am not able to install tensorflow.I have installed Anconda and python 3.7.3 and in Anconda i am trying todowgrade python using "conda downgrade python=3.6" Even this is also not helping to install tensor flow Can you help me sir

  17. LinkError: post-link script failed for package defaults::qt-5.9.7-vc14h73c81de_0
    running your command again with `-v` will provide additional information
    location of failed script: C:UsersjhaniMiniconda3envstensorflowScripts.qt-post-link.bat
    ==> script messages <==


    i get this error while installing jupyter!!

    what should i do now??

  18. I LOVE YOU!!! My first attempt to install tensorflow took me like 4 weekends, then I had to format the hard drive because I've screwed up my computer too much. I've been already mentally prepared to spend another 2 or 3 weeks on this, but your tutorial saved me so much time!!!

  19. importError: DLL load failed: Error en una rutina de inicializacion de biblioteca de vinculos dinamicos (DLL)

    Failed to load the native TensorFlow runtime.

  20. When I try to create the new channel in miniconda, I run into "UnavailableInvalidChannel: The channel is not accessible or is invalid."

  21. i managed to install psutil with anaconda but more on a tinker way: "conda install -c anaconda psutil" which actually failed to install it so ithad to be followed by "pip install psutil"…. but works for me

  22. Thank you for sharing, I found another solution who still has trouble on tensorflow. On Anaconda promp, try for "pip install tensorflow==1.13.1" Good Luck.

  23. Hi, I really appreciate your tutorial, which is informative. Thank you very much. Just a question, how can we access the private videos?

  24. Hi Jeff, Thank you for being the person who bridges theory and applied for deep learning! It is a very important place! CUDOS
    Also, thank you for making this learning available. I am an Australian, who is currently in the UK and wanting to implement (a planned) something radical with AI+ML. For me its a to hard to get to class. LOL

  25. Hi.thanks, but i think i don't get sth. each time that i enter the miniconda i should type ((((((conda create -y –name tensorflow python=3.6. ))))) then typing (((((activate tensorflow))))) but after this when i type import tensorflow it says that you dont have that package. it's mean that each time should i install all that packages?????? Thanks

  26. As an Optical Engineer turned High School teacher. I am learning PIL in Colab and have OpenCV2 on Windows 10 IDLEX . I surveyed TensorFlow, VS Code and StackOverflow for an answer but found none.In WIndows 10, Visual Studio Code, may you recommended a way to install openCV ?
    Is it under TensorFlow or Keras ??

    Is it in Python 3.7

    $ conda update conda
    $ pip list (no OpenCV found under piplist)
    Python3 pip –m install tensorflow

    Python3 pip –m install Keras


    Python pip –m install tensorflow

    Python pip –m install Keras

    Thank you for your consideration.

  27. Unfortunately this instruction is not working for me either: I had all Python installations removed before starting and I have previously tried many different types of instructions to get tensorflow running in windows with and without gpu. When following your instruction, after starting python and running "import tensorflow as tf" I get "ImportError: Could not find 'cudnn64_7.dll". This is strange as the conda package for tf should be the non-gpu version. I am confused and very frustrated. Any ideas of what is wrong?

  28. Thanks alot sir…Really this serves as life saver.
    i was trying from last 10 days but know i just found what i want.
    once again love ,respect from Pakistan <3

  29. GOT this error please let me know how to fix it. thank you 🙂

    File "<stdin>", line 1, in <module>

    File "C:UsersZain UllahMiniconda3envstensorflowlibsite-packagestensorflow__init__.py", line 24, in <module>

    from tensorflow.python import pywrap_tensorflow # pylint: disable=unused-import

    File "C:UsersZain UllahMiniconda3envstensorflowlibsite-packagestensorflowpython__init__.py", line 47, in <module>

    import numpy as np

    File "C:UsersZain UllahAppDataRoamingPythonPython36site-packagesnumpy__init__.py", line 140, in <module>

    from . import _distributor_init

    File "C:UsersZain UllahAppDataRoamingPythonPython36site-packagesnumpy_distributor_init.py", line 26, in <module>


    File "C:UsersZain UllahMiniconda3envstensorflowlibctypes_init__.py", line 348, in __init_

    self._handle = _dlopen(self._name, mode)

    OSError: [WinError 193] %1 is not a valid Win32 application

  30. pip install –exists-action i –upgrade tf-nightly-2.0-preview

    pip install –exists-action i –upgrade keras-rl2 –user
    they are not working
    ERROR: Could not find a version that satisfies the requirement tf-nightly-2.0 (from versions: none)

    ERROR: No matching distribution found for tf-nightly-2.0

    ERROR: Could not find a version that satisfies the requirement tf-nightly-2.0-preview (from keras-rl2) (from versions: none)

    ERROR: No matching distribution found for tf-nightly-2.0-preview (from keras-rl2)

  31. Please note, I've updated the install script a bit to include TensorFlow 2.0. The video is still correct, but just copy/paste the UPDATED pip installs, which are found here: https://github.com/jeffheaton/t81_558_deep_learning/blob/master/t81_558_class_01_1_overview.ipynb

  32. Hi, professor. Hope all are doing well. After days and days, I could fix the problems with my mismatched and error with your videos. I will be happy to learn more and get your courses. I am a computer science major and a passion for Data Science. I like to have your courses as well about this field. Could you mind send a link that I can connect to your knowledge and information that you updated about machine-learning and deep-learning as well? Thanks

  33. Hey, professor! could you please guide regarding installation of opencv2. After following these steps i installed jupyter and running one code gives error: ModuleNotFoundError: No module named 'cv2'. Thanks a lot for this!

  34. It gives the error status 1: python setup.py egg_info everytime I try to install one of the packages after jupyter

  35. I don't understand why professor needed to install keras separately since with tensorflow version 1.12 it is possible to do import tensorflow.keras after installing only tensorflow

  36. Been having long problems getting keras installed (Theano missing C++ compiler, even though I have one and it is in the PATH environment).
    For anyone else having these problems the line:

    conda install mkl-service libpython m2w64-toolchain

    Fixed my problems.
    Hope this helps someone.

  37. Hello sir, I followed everything right but it says "ImportError: DLL load failed with error code -1073741795". Please help me sir. Thank you

  38. Hi prof.
    Such a great video ,you really made life alot easier for alot of people out here But I'm not sure if u will read this or not but after I run "print(tf. version )", instead of getting the version number it's showing me this " <module 'tensorflow._api.v2.version' from ' (path)'>".
    Does this still mean that tensorflow is installed ??
    It would be great if I got a reply from you or anyone else out here in the comments thank you

  39. Please, help me with my problem of Kernel error!

    PermissionError: [Errno 13] Permission denied: 'C:\Users\Lola\AppData\Roaming\jupyter\runtime\kernel-7d3b57f9-c98f-4414-8f7c-39ddba20cd91.json'

  40. Thanks you so much!..while outlining the model i am getting this error..Please help
    AttributeError: module 'tensorflow' has no attribute 'placeholder'

  41. (tensorflow) C:UsershimanshuAppDataLocalProgramsPythonPython36>pip install –upgrade tensorflow==2.0

    Collecting tensorflow==2.0

    ERROR: Could not find a version that satisfies the requirement tensorflow==2.0 (from versions: none)

    ERROR: No matching distribution found for tensorflow==2.0

    (tensorflow) C:UsershimanshuAppDataLocalProgramsPythonPython36>python –version

    Python 3.6.5

    still unable to install tf, even though the version of Python is 3.6.5… pls answer it to rectify this problem..

  42. Hi Jeff I'm having some issues with the opening jupyter notebook part, I did everything right and I'm in t81_558_deep_learning-master folder but I don't know what to do next, thank you!

  43. Just want to share with the community a problem that I ran into(and it's solution). In the last step to create a kernel for the tensorflow environment, i got an error saying "no module named ipykernel". If you get this too, it just means that the ipykernel package is not installed in the newly created tensorflow environment. The simple solution to this problem is just to run the command: conda install ipykernel , and then run the command given in jeff's notes again.

  44. Hello!
    Firstly THANK YOU! That course is pure gold:)

    Secondly, I have a problem with the installation. in the 20:17 minutes it gives me the error :
    ModuleNotFoundError: No module named 'tensorflow.core'
    I search through the internet but it didn't help.
    I installed it once again and the error is still there.
    Does anyone have the same?

    Kind regards:)

  45. Does anyone know how to solve this error: module 'tensorflow' has no attribute 'InteractiveSession'. I will really appreciate it.

  46. Hai Jeff, when I import tensorflow as tf. This is what happened. What should I do?

    >>> import tensorflow as tf

    C:UsersuserAnaconda3envstensorflow2libsite-packagestensorflowpythonframeworkdtypes.py:523: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.

    _np_qint8 = np.dtype([("qint8", np.int8, 1)])

    C:UsersuserAnaconda3envstensorflow2libsite-packagestensorflowpythonframeworkdtypes.py:524: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.

    _np_quint8 = np.dtype([("quint8", np.uint8, 1)])

    C:UsersuserAnaconda3envstensorflow2libsite-packagestensorflowpythonframeworkdtypes.py:525: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.

    _np_qint16 = np.dtype([("qint16", np.int16, 1)])

    C:UsersuserAnaconda3envstensorflow2libsite-packagestensorflowpythonframeworkdtypes.py:526: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.

    _np_quint16 = np.dtype([("quint16", np.uint16, 1)])

    C:UsersuserAnaconda3envstensorflow2libsite-packagestensorflowpythonframeworkdtypes.py:527: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.

    _np_qint32 = np.dtype([("qint32", np.int32, 1)])

    C:UsersuserAnaconda3envstensorflow2libsite-packagestensorflowpythonframeworkdtypes.py:532: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.

    np_resource = np.dtype([("resource", np.ubyte, 1)])

  47. Hi Jeff…..
    Firstly, I want to thank you for the great stuff that you've provided on youtube…

    Could you please make a video on building tensorflow from the source optimized for AVX AVX2 instructions…

    I'm getting an error saying "your cpu uses instructions that this tensorflow binary is not compiled to use: AVX AVX2" and I am unable to resolve it just by going through stackoverflow…!

    Currently I am using anaconda 3.7 on Windows OS, CPU.

    Could you please provide some solution to this. It'll be a great help for me….


  48. Thank you very much sir for providing such a detailed step by step tutorial for installation,very precise and up to date video.

  49. Now that I have installed it. where should I run and edit the python script?
    on jupyter in a browser? tried it, but I didn't see any run command there. Thanks!
    hope you could help me out.

  50. I've got Python 3.7.5 and i want to use tensorflow but I've got this: "ModuleNotFoundError: No module named 'tensorflow.python.platform". How to solve this?

  51. Sir lock what I get as result
    ERROR: Could not find a version that satisfies the requirement tensorflow==1.12.

    0 (from versions: none)

    ERROR: No matching distribution found for tensorflow==1.12.0

  52. Found this video AFTER just watching the original version. Going back and pip installing all the needed upgrades is tedious and confusing at times. 🙁

  53. I really thank you Dr.Jeff Heaton for your effort, your time, and the detailed explanation of installation.
    Also may I know, why installing "psutil" will be failed with Anaconda3 ?


  54. Hello, can you help me? please, I am facing python TensorFlow mismatch error while I am trying to install TensorFlow I actually changed the python environment into 3.6 and followed each of the mentioned steps

  55. Your video helped me to figure out the installation problem of Tensorflow with Python 3.7 after trying a few days. Thanks a lot respected professor.

Leave a Reply

Your email address will not be published. Required fields are marked *