Error with environment to run tensorflow model.

[hungpham3112]

Hi, I'm new with saturncloud. Sorry for creating an issue in here, I know this is not suitable platform to ask user question but I didn't find anyway to ask, I check the repo saturncloud/images but there is no issues section for reporting.

Firstly, I created jupyterlab server to train model in tensorflow. I used public.ecr.aws/saturncloud/saturn-python-tensorflow:2023.05.01 as docker image for that environment. Expecting to run code smoothly with this environment but then when I run model.fit() some errors appeared:

2023-10-11 09:29:01.001362: W tensorflow/core/framework/op_kernel.cc:1745] OP_REQUIRES failed at xla_ops.cc:296 : INTERNAL: libdevice not found at ./libdevice.10.bc
---------------------------------------------------------------------------
InternalError                             Traceback (most recent call last)
Cell In[36], line 1
----> 1 history = model.fit(X_train, Y_train, epochs=5, batch_size=8, validation_split=0.2, callbacks=[early_stopping], )
      2 # history = model.fit(X_train, Y_train, epochs=5, batch_size=8, validation_split=0.2)

File /opt/saturncloud/envs/saturn/lib/python3.9/site-packages/keras/utils/traceback_utils.py:67, in filter_traceback.<locals>.error_handler(*args, **kwargs)
     65 except Exception as e:  # pylint: disable=broad-except
     66   filtered_tb = _process_traceback_frames(e.__traceback__)
---> 67   raise e.with_traceback(filtered_tb) from None
     68 finally:
     69   del filtered_tb

I guess there are some issues with environment. I searched and tried to debug environment with conda install -c conda-forge cudatoolkit=11.2 cudnn=8.1.0, see here