Jupyter kernel dies when importing pipeline function from transformers class on Mac OS
Question:
I’m unable to import pipeline function of transformers class as my jupyter kernel keeps dying. Tried on transformers-4.15.0 and 4.16.2. Anyone faced this issue?
I tried importing the class in a new notebook as you can see in the image and it keeps killing the kernel.
Answers:
It works fine for me.
You could try creating a fresh conda environment and reinstalling the app.
You could also try using jupyterlab instead of jupyter-notebook.
Are you on Mac OS? I couldn’t get it to run at first using conda install transformers
my jupyterlab kept hanging as well.
Then I did this, conda install -c huggingface transformers
and here is the result.
It works fine for me on linux and Mac now.
If you are working on a Mac with a M1 silicon chip, you could try reinstalling transformers through huggingface using conda install -c huggingface transformers
. It worked for me with Python 3.9 in a notebook when i had the same issue.
I’m unable to import pipeline function of transformers class as my jupyter kernel keeps dying. Tried on transformers-4.15.0 and 4.16.2. Anyone faced this issue?
I tried importing the class in a new notebook as you can see in the image and it keeps killing the kernel.
It works fine for me.
You could try creating a fresh conda environment and reinstalling the app.
You could also try using jupyterlab instead of jupyter-notebook.
Are you on Mac OS? I couldn’t get it to run at first using conda install transformers
my jupyterlab kept hanging as well.
Then I did this, conda install -c huggingface transformers
and here is the result.
It works fine for me on linux and Mac now.
If you are working on a Mac with a M1 silicon chip, you could try reinstalling transformers through huggingface using conda install -c huggingface transformers
. It worked for me with Python 3.9 in a notebook when i had the same issue.