tfa.text.crf_forward

Stay organized with collections Save and categorize content based on your preferences.

Computes the alpha values in a linear-chain CRF.

tfa.text.crf_forward(
    inputs: tfa.types.TensorLike,
    state: tfa.types.TensorLike,
    transition_params: tfa.types.TensorLike,
    sequence_lengths: tfa.types.TensorLike
) -> tf.Tensor

See http://www.cs.columbia.edu/~mcollins/fb.pdf for reference.

inputs A [batch_size, num_tags] matrix of unary potentials.
state A [batch_size, num_tags] matrix containing the previous alpha values.
transition_params A [num_tags, num_tags] matrix of binary potentials. This matrix is expanded into a [1, num_tags, num_tags] in preparation for the broadcast summation occurring within the cell.
sequence_lengths A [batch_size] vector of true sequence lengths.

new_alphas A [batch_size, num_tags] matrix containing the new alpha values.