MCPcopy Create free account
hub / github.com/tensorflow/lingvo / _ComputeXentLoss

Method _ComputeXentLoss

lingvo/tasks/mt/decoder.py:81–132  ·  view source on GitHub ↗

Computes cross-entropy loss given the softmax input, labels and weights. Args: theta: A `.NestedMap` object containing weights' values of this layer and its children layers. softmax_input: A tensor of shape [time, batch, p.softmax.input_dim]. target_labels: A matrix of

(self,
                       theta,
                       softmax_input,
                       target_labels,
                       target_weights,
                       target_paddings,
                       target_segment_ids=None,
                       time_axis=0)

Source from the content-addressed store, hash-verified

source not stored for this graph (policy: none)

Callers 1

_FPropSoftmaxMethod · 0.95

Calls 1

FPropMethod · 0.45

Tested by

no test coverage detected