tf.nn.ctc_loss_v2(
labels,
logits,
label_length,
logit_length,
logits_time_major=True,
unique=None,
blank_index=None,
name=None
)
Defined in tensorflow/python/ops/ctc_ops.py
.
Computes CTC (Connectionist Temporal Classification) loss.
This op implements the CTC loss as presented in the article:
Notes: - Same as the "Classic CTC" in TensorFlow 1.x's tf.nn.ctc_loss setting of preprocess_collapse_repeated=False, ctc_merge_repeated=True - Labels may be supplied as either a dense, zero-padded tensor with a vector of label sequence lengths OR as a SparseTensor. - On TPU and GPU: - Only dense padded labels are supported. - On CPU: - Caller may use SparseTensor or dense padded labels but calling with a SparseTensor will be significantly faster. - Default blank label is 0 rather num_classes - 1, unless overridden by blank_index.
Args:
labels
: tensor of shape [batch_size, max_label_seq_length] or SparseTensorlogits
: tensor of shape [frames, batch_size, num_labels], if logits_time_major == False, shape is [batch_size, frames, num_labels].label_length
: tensor of shape [batch_size], None if labels is SparseTensor Length of reference label sequence in labels.logit_length
: tensor of shape [batch_size] Length of input sequence in logits.logits_time_major
: (optional) If True (default), logits is shaped [time, batch, logits]. If False, shape is [batch, time, logits]unique
: (optional) Unique label indices as computed by ctc_unique_labels(labels). If supplied, enable a faster, memory efficient implementation on TPU.blank_index
: (optional) Set the class index to use for the blank label. Negative values will start from num_classes, ie, -1 will reproduce the ctc_loss behavior of using num_classes - 1 for the blank symbol. There is some memory/performance overhead to switching from the default of 0 as an additional shifted copy of the logits may be created.name
: A name for thisOp
. Defaults to "ctc_loss_dense".
Returns:
loss
: tensor of shape [batch_size], negative log probabilities.