JupyterLab notebook crashes consitently when using huggingface's transformers

[Raffinate]

Product Version

• [ ] Amazon SageMaker Studio Classic
• [X] Amazon SageMaker Studio
• [ ] Issue is not related to SageMaker Studio

Issue Description

Sagemaker notebooks always crash and restart when using transformers processors regarless of memory machine spec.

Tried following machine specs:

• 4, 8 and 16 GB of RAM
• 5, 10, 15 GB of disk space

Python Kernel: Python3 ipykernel

Minimum example I was able to reproduce the issues 100% of the time.

from transformers import CLIPProcessor

processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32") # or literally any other string

"DONE"

Expected Behavior

1. Create empty notebook and launch any machine
2. Create following cell

   
   from transformers import CLIPProcessor

processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32") # or any literally any other string

print("DONE")

3. Run the code

Expected: cell to succeed and "DONE" is printed

### Observed Behavior

1. Create empty notebook and launch any machine
2. Create following cell

from transformers import CLIPProcessor

processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32") # or any literally any other string

print("DONE")

3. Run the code

Observed: Kerner Restarted

### Product Category

JupyterLab

### Feedback Category

Reliability and Stability

### Other Details

_No response_

[ComputerVisionFans]

same problem appeals here, when loading model from transformer using SageMaker Studio Classic , those kernel and notebook wont have any problem, but using this new sagemaker studio loading model from transformers from Huggingface. the kernel often died automatically, no matter how many Storage (GB) you set, or what instance you set, sometime i saw the instance cpu only used 10%...no sure what leads to notebook often cash...