Failed to load the model. Python stream closed unexpectedly, exit code: 9009

[GerardoLoffredo]

Hi,
I'm quite new to the use of DJL.

I'm trying to load my model.py from path, following these guidelines:
https://github.com/deepjavalibrary/djl-demo/blob/master/development/python/README.md

When I run the code I get this error:

Exception in thread "main" ai.djl.engine.EngineException: Failed to loaded model.
Caused by: java.lang.IllegalThreadStateException: Python stream closed unexpectedly, exit code: 9009

I added Python in the System Environment Variables, but it still doesn't work. I'm actually using the version 28.0 of DJL.

Is it still possible to use the Python engine? Or should I consider converting the model to PyTorch and then loading it with PyTorch engine?

Thank you in advance!

[Poovarasan46]

Hey @GerardoLoffredo,

DJL's Python engine is picky
pip install numpy pandas torch torchvision djl_python
because DJL won’t work unless it has all its favorite snacks.

• if you're using windows, check if Python is really in your PATH:

  where python

• or mac/Linux:

  which python
  If it points to some mysterious location, we have a plot twist.

Plan B: PyTorch engine 🏃‍♂️

• If DJL keeps ghosting Python, try converting your model to PyTorch and use the PyTorch engine instead. Less drama, more AI magic.

[GerardoLoffredo]

Thank you!

I tested it on Linux and it actually works!

[frankfliu]

@GerardoLoffredo

The best way to use python engine is using DJLServing container, see: https://github.com/deepjavalibrary/djl-serving?tab=readme-ov-file#docker