follow the steps or example but crash

[maochaokuo]

Hi there:

I try to clone this repository, or follow the steps from readme.md. They both crash with the following messages:

C:\Program Files\dotnet\dotnet.exe (process 8912) exited with code -1073741795.
To automatically close the console when debugging stops, enable Tools->Options->Debugging->Automatically close the console when debugging stops.
Press any key to close this window . . .

Is there any further hint for me to resolve it? Thanks

[deepakkumar1984]

Which example project you running? Can you please take the latest and run the basic example which should work fine.

[maochaokuo]

hi there: It does not matter which example (including Basic, Image, Text, and a complete empty new project with XOR sample). They all crash at the line: var model = new Sequential(); I guess the problem might concern my environment. I did: 1. download and install dotnet core 2.2 2. install anaconda (default with python 3.7) 3. install python 3.6.8 and update path 4. install numpy If possible, please give me recommended process to setup the developing environment. A few environment verification steps will be appreciated. Regards, Thomas Mao

…

On Wed, Jul 17, 2019 at 12:31 PM Deepak Kumar Battini < ***@***.***> wrote: Which example project you running? Can you please take the latest and run the basic example which should work fine. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#6?email_source=notifications&email_token=ABUKGQL4Y6AQOFSRCGUF5S3P72OA5A5CNFSM4IEI3ZR2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD2C73AA#issuecomment-512097664>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABUKGQO7KPW3XVX2WPRBOQTP72OA5ANCNFSM4IEI3ZRQ> .

[henon]

We know that Pythonnet (which we use to access Python) does not find Python with an Anaconda install. It works with a normal Python installation, that can be said.
If you could clone the Pythonnet repo from source and execute the example on their main readme you should get the same error. Then if you will, please try setting the path right until it works with Anaconda. We then have to discuss how to support Anaconda (automatically, or via manual steps).

[maochaokuo]

No, I just clean my environment and reinstall Python 3.6.8. The problem is still the same.

[henon]

does the sample from the pythonnet readme work for you? if not when you single step, where does it fail? what message?

[maochaokuo]

Hi How do I specify which virtual environment which I wish to use? in Keras.net? Thanks Regards, Thomas Mao

…

On Wed, Jul 17, 2019 at 9:57 PM Meinrad Recheis ***@***.***> wrote: does the sample from the pythonnet readme work for you? if not when you single step, where does it fail? what message? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#6?email_source=notifications&email_token=ABUKGQOFNA6IKB6L6S6PV4TP74QLRA5CNFSM4IEI3ZR2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD2EI5DY#issuecomment-512265871>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABUKGQLUMVKEUEZFMXQEELLP74QLRANCNFSM4IEI3ZRQ> .

[deepakkumar1984]

Can you try setting the Python.Runtime.PythonEngine.PythonPath and Python.Runtime.PythonEngine.PythonHome to the virtual env directory?

[maochaokuo]

When I tried to set either one of the two variables, I got the following error:

Unable to load DLL 'python36' or one of its dependencies; The specified module could not be found.

[deepakkumar1984]

I never tested with virtual environment, will give a try and let you know if its supported or not.

[adnanlanewla]

When I tried to set either one of the two variables, I got the following error:

Unable to load DLL 'python36' or one of its dependencies; The specified module could not be found.

I am getting the exact same error. Spent two days on trying to fix this error.

I set my PythonHome and PythonPath same as my anaconda virtual env (C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64)

[itzukah]

Same problem. I have already installed python3.6.8 win64 and python 3.7

[philvanzu]

Trying to load a trained keras model file to make predictions here.

I think I found a solution to make my miniconda environment work, but then I crash on the next line because Keras module is not found, I imagine that's because I installed tensorflow 2.0 rc1 in that environment and keras is no longer its own thing, it is tensorflow.keras now. Anyway the "Python36.dll not found" crash is resolved for me after this.

            string envPythonHome = "C:\\Users\\username\\Miniconda3\\envs\\tensorflow\\";
            string envPythonLib = envPythonHome + "Lib;" + envPythonHome + "Lib\\site-packages";
            Environment.SetEnvironmentVariable("PYTHONHOME", envPythonHome, EnvironmentVariableTarget.Process);
            Environment.SetEnvironmentVariable("PATH", envPythonHome + ";" + Environment.GetEnvironmentVariable("PATH", EnvironmentVariableTarget.Machine), EnvironmentVariableTarget.Process);
            Environment.SetEnvironmentVariable("PYTHONPATH", envPythonLib, EnvironmentVariableTarget.Process);

            PythonEngine.PythonHome = envPythonHome;
            PythonEngine.PythonPath = Environment.GetEnvironmentVariable("PYTHONPATH");`

I wonder if I should try to bypass keras.net and try my luck with pythonnet embedding. I would just have to execute a few python lines and the necessary imports.

[deepakkumar1984]

Can I close this issue?

[philvanzu]

Not my issue originally, I was just providing a possible fix for it.

…

On Fri, 27 Sep 2019 at 09:34, Deepak Kumar Battini ***@***.***> wrote: Can I close this issue? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#6?email_source=notifications&email_token=AHQQHTIPH52VENSRBEFGSY3QLWZQPA5CNFSM4IEI3ZR2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD7YBIVQ#issuecomment-535827542>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHQQHTN7FAG76YU2JPWYHTLQLWZQPANCNFSM4IEI3ZRQ> .

[deepakkumar1984]

I know dear :), I was asking the original requester

[deepakkumar1984]

Closing this issue, please reopen if still having the problem