error: "Instruction non permise" when I executed "python 02_example_region_from_model.py"

liu-bioinfo-lab / caesar

An accurate and interpretable computational model for predicting nucleosome resolution chromatin contact maps

MIT License

10 stars 1 forks source link

error: "Instruction non permise" when I executed "python 02_example_region_from_model.py" #4

Open hanshanmengqi opened 2 years ago

hanshanmengqi commented 2 years ago

Dear developer,

Thank you for so great tools.

But now I met a problem as the title saying. What should I do?

Best, Xiaoning

hanshanmengqi commented 2 years ago

First step looks good and I got 'example_outputs', 'input_epi_HFF', 'input_hic' 3 folders in temp.

dmcbffeng commented 2 years ago

Hi Xiaoning,

Thank you for your interest in our work!

It seems like a bug at the system level rather than from Python. Could you provide more information like:

The complete error message
Your Python and tensorflow version (tensorflow 1 or 2, whether tensorflow-gpu)
Your CPU version

Best, Fan

hanshanmengqi commented 2 years ago

Dear Fan,

Thank you a lot for your swift reply.

This is screenshot of error message. CAESAR

My python version is Python 3.6.8 and tensorflow is 2.4.1.

cpuinfo.txt

Thank you again.

Best, Xiaoning

dmcbffeng commented 2 years ago

Hi Xiaoning,

I think the error comes from the conflict between tensorflow 2.0 and certain CPU versions. Could you check whether this happens when you run "import tensorflow as tf" only?

I found some references here: https://stackoverflow.com/questions/61745843/how-to-install-tensorflow-on-intel-xeon-processor-based-pc. They mentioned that downgrading to tensorflow 1 might solve the issue.

However, the CAESAR_model.py is written with tf.keras, which is only supported by tensorflow 2. To make it work with tf 1.x, I think you need to install keras as well and replace all "tf.keras" with "keras".

Best, Fan