Asserts bridge outputs are correct. Must be implemented by child classes

Subclasses should implement their logic here.

Computes attention scores and outputs. Args: query: The query used to calculate attention scores. In seq2seq this is typically the

Clips gradients by global norm.

In addition to standard gradient clipping, also clips embedding gradients to a specified value.

Creates example data for a test

Model-specific preprocessing for features and labels: - Creates vocabulary lookup tables for source and target vocab - Converts tokens

Model-specific preprocessing for features and labels: - Creates vocabulary lookup tables for source and target vocab - Converts tokens

Model-specific preprocessing for features and labels: - Creates vocabulary lookup tables for source and target vocab - Converts tokens

Model-specific preprocessing for features and labels: - Creates vocabulary lookup tables for target vocab - Converts tokens into vocabulary i

Wrapper function that converts tensors to unicode and slices them until the EOS token is found.

Replaces UNK tokens with tokens from the source or a provided mapping based on the attention scores. Args: source_tokens: A numpy array of st

Adds a dictionary to a graph collection. Args: dict_: A dictionary of string keys to tensor values collection_name: The name of the collect

Runs SyncReplicasOptimizer initialization ops.

return an ASTNode with type information only

Wraps the original decoder helper function to append the attention context.

Wraps the original decoder helper function to append the attention context.

Wraps the original decoder helper function to append the attention context.

Wraps the original decoder helper function to append the attention context.

The batch size of the inputs returned by `sample`.

Returns a scalar int32 tensor.