This PR upgrades TF1 imports -> TF2 and updates documentation accordingly. Unfortunately, since the keras_exp package depends on TF1, I've removed this functionality. I will open a subsequent PR for TF2 distributed GPU training. Also, I've added a hot fix by removing the embedding_constraint variable in the Embedding Layer (see issue 15 for more info), but hopefully that can be fixed soon as well.
This PR upgrades TF1 imports -> TF2 and updates documentation accordingly. Unfortunately, since the
keras_exp
package depends on TF1, I've removed this functionality. I will open a subsequent PR for TF2 distributed GPU training. Also, I've added a hot fix by removing theembedding_constraint
variable in the Embedding Layer (see issue 15 for more info), but hopefully that can be fixed soon as well.