validation loss="nan"
vanikanjirangat opened this issue · comments
While I am trying to run the model, the training loss decreases, while the validation loss is always "nan".
I am trying to run the code in google colab with python 3.8.15 and tensorflow 2.9.2. I solved the compatibility issues in tensorflow using tf.compat.v1. and had to do tf.compat.v1.disable_eager_execution().
Can you please point out the possible reasons?
Thanks.