tfa.text.crf_filtered_inputs

Constrains the inputs to filter out certain tags at each time step.

tfa.text.crf_filtered_inputs(
    inputs: tfa.types.TensorLike,
    tag_bitmap: tfa.types.TensorLike
) -> tf.Tensor

tag_bitmap limits the allowed tags at each input time step. This is useful when an observed output at a given time step needs to be constrained to a selected set of tags.

inputs A [batch_size, max_seq_len, num_tags] tensor of unary potentials to use as input to the CRF layer.
tag_bitmap A [batch_size, max_seq_len, num_tags] boolean tensor representing all active tags at each index for which to calculate the unnormalized score.

filtered_inputs A [batch_size] vector of unnormalized sequence scores.