hi @pasquini-dario . Nice work!
I am trying to run the training experiments. However, python train.py configs.UNCM_Custom yields several errors, which IMO result from the package version for tf/keras.
For example,
ValueError: Cannot convert '1' to a shape.ValueError: A KerasTensor cannot be used as input to a TensorFlow function. A KerasTensor is a symbolic placeholder for a shape and dtype, used when constructing Keras Functional models or Keras Functions. You can only use it as input to a Keras layer or a Keras operation
Could you specify the detailed TF, keras package version you used and environment variables like export TF_USE_LEGACY_KERAS=1 to correctly run the training?
hi @pasquini-dario . Nice work!
I am trying to run the training experiments. However,
python train.py configs.UNCM_Customyields several errors, which IMO result from the package version for tf/keras.For example,
ValueError: Cannot convert '1' to a shape.ValueError: A KerasTensor cannot be used as input to a TensorFlow function. A KerasTensor is a symbolic placeholder for a shape and dtype, used when constructing Keras Functional models or Keras Functions. You can only use it as input to a Keras layer or a Keras operationCould you specify the detailed TF, keras package version you used and environment variables like
export TF_USE_LEGACY_KERAS=1to correctly run the training?