Process dumped core

[LogExE]

Issue

Hello. I have a problem running my Jupyter Notebook code with XPU. Part of the notebook code works fine and device itself is visible, but then fails when trying to fit model, last visible output is:

2026-04-12 14:50:50.935921: I tensorflow[/core/grappler/optimizers/custom_graph_optimizer_registry.cc:117](http://localhost:8888/core/grappler/optimizers/custom_graph_optimizer_registry.cc#line=116)] Plugin optimizer for device_type XPU is enabled.

Full logs are attached here logs-tensorflow-intel.txt.

Could someone please help me to troubleshoot this issue?

Info

Python and kernel version:

$ python -V
Python 3.11.15
$ uname -r
6.19.11-200.fc43.x86_64

(I'm running Fedora 43 with up-to-date official Intel oneAPI repo packages.)

lspci

$ lspci
...
28:00.0 VGA compatible controller: Intel Corporation Battlemage G21 [Arc B580]
...

Pyproject file with packages:

[project]
name = "tf-oneapi"
version = "0.1.0"
requires-python = ">=3.11"
dependencies = [
    "intel-extension-for-tensorflow[xpu]>=2.15.0.3",
    "jupyterlab>=4.5.6",
    "keras-tuner>=1.4.8",
    "matplotlib>=3.10.8",
    "tensorflow==2.15.1",
    "wget>=3.2",
]

Steps before executing the notebook:

. .venv/bin/activate # activate uv venv
. /opt/intel/oneapi/setvars.sh 
jupyter lab

[alexschroeter]

Hi,

as far as I can tell this repo is dead (last edit to the code is from last year). I believe Intel has stopped the development of this to save company resources in favor of pytorch and jax. If you want you can have a look at my issue #91 and see if you have the same error but I wouldn't expect any help here.

Best of Luck