The kernel appears to have died. It will restart automatically.

[silverswingkiller]

After modifying the name of the installation package, I successfully created a virtual environment with the official script and installed tensorflow, but when I use jupyter to run the code, it still fails to run。The kernel appears to have died when i training

[anna-tikhonova]

Thank you for reporting this issue. We will experiment with Jupyter notebooks.

[icenando]

After modifying the name of the installation package, I successfully created a virtual environment with the official script and installed tensorflow, but when I use jupyter to run the code, it still fails to run。The kernel appears to have died when i training

I was having the same issue. Apparently it's an architecture (chip) issue. If you run jupyter from terminal and observe the terminal, it will show you what the kernel is doing. Mine was giving me a "wrong architecture" error (or something like that) before the kernel died. I was using a Tensorflow 2.0 though, so I assumed that it didn't work with my Mac M1. I since then haven't been able to install the arm version of Tensorflow (see issue 48)

[jmbernabotto]

I have the same issue @icenando. It's a disappoint ! I was waiting before to buy an MacBook Pro or Air M1. So, if I can't work with it, I will wait again...

[jmbernabotto]

@anna-tikhonova : some news about Jupiter notebook experiment ?

[dmmajithia]

@silverswingkiller can you post your model summary/architecture?
Also, have you tried training the model from #25 ?

[seongilp]

please refer to below links

• https://towardsdatascience.com/how-to-run-jupyter-notebooks-on-an-apple-m1-mac-ac3a8bf39c6c

[jmbernabotto]

Thanks. I tried, but nothing better !

…

Le 9 déc. 2020 à 03:09, Seongil Park ***@***.***> a écrit :  please refer to below links https://towardsdatascience.com/how-to-run-jupyter-notebooks-on-an-apple-m1-mac-ac3a8bf39c6c — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

[anna-tikhonova]

@silverswingkiller Could you let me know if the discussion in #45 helps?

[maeverst]

I am receiving the same error (Kernel appears to have died...) when I launch jupyter notebook from the terminal. This just started happening.

Does anyone know why please?

[vitdegtyarev]

I'm getting the same error ("The kernel appears to have died. It will restart automatically") on Windows when I run the "import keras" or "import tensorflow" command. The commands worked fine about a week ago. Not sure what has changed.

[icenando]

Same here. It was fine until about 2 weeks ago, when I last tested.

[vitdegtyarev]

I resolved this by updating hdf5 to 1.10.5.
conda install -c conda-forge hdf5=1.10.5

[asima-azmat]

I am having the same issue when I run the cell where I am training my network.

[asima-azmat]

Adding following lines in my notebook helped me:
import os
os.environ['KMP_DUPLICATE_LIB_OK']='True'

[asima-azmat]

Installing nomkl also works.

[borinmin]

My experience on Linux, I killed and reopened Jupyter was fine.

[kartikeypro]

I have tried everything mentioned in this thread to resolve this issue but no success. If anyone has resolved this problem please share.

[kartikeypro]

I have solved this issue by creating the environment after deactivating the base, which can be done using "Conda Deactivate".

[kartikeypro]

I started to face this issue again when I downloaded Keras over the top of tensorflow. So, I deleted the Keras package and downloaded tensorflow again, in virtual environment. tensorflow model worked fine after that.