Issue Description:
I followed the instructions to run the timesfm model on the cpu and set up the conda environment based on the environment_cpu.yml file. The setup of the conda environment ran through without any errors or warnings. However, when attempting to import the timesfm module and create the model, I encountered an ImportError related to TensorFlow's native runtime.
Error Message:
File "pywrap_tensorflow.py", line 77, in <module>
raise ImportError(ImportError: Traceback (most recent call last):
File "pywrap_tensorflow.py", line 62, in <module>
from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: /.../_pywrap_tensorflow_internal.so: undefined symbol: _ZNK10tensorflow4data11DatasetBase8FinalizeEPNS_15OpKernelContextESt8functionIFNS_8StatusOrISt10unique_ptrIS1_NS_4core15RefCountDeleterEEEEvEE
Failed to load the native TensorFlow runtime.
Additional Information:
I am seeking help to understand this error and find a possible fix. The detailed ImportError suggests an issue with _pywrap_tensorflow_internal.so, which could be related to an improper TensorFlow build or configuration specific to CPU environments.
Environment Details:
Issue Description: I followed the instructions to run the timesfm model on the cpu and set up the conda environment based on the environment_cpu.yml file. The setup of the conda environment ran through without any errors or warnings. However, when attempting to import the timesfm module and create the model, I encountered an ImportError related to TensorFlow's native runtime.
Error Message:
Additional Information: I am seeking help to understand this error and find a possible fix. The detailed ImportError suggests an issue with
_pywrap_tensorflow_internal.so, which could be related to an improper TensorFlow build or configuration specific to CPU environments.