接入crf层后，首先不太清楚 log_likelihood输出究竟是什么，那么怎么获取精准度这个指标呢？@Slyne if self.config.crf: log_likelihood, self.transition_params = tf.contrib.crf.crf_log_likelihood( self.logits, self.labels, self.sequence_lengths) self.loss = tf.reduce_mean(-log_likelihood)
接入crf层后，首先不太清楚 log_likelihood输出究竟是什么，那么怎么获取精准度这个指标呢？@Slyne
if self.config.crf:
log_likelihood, self.transition_params = tf.contrib.crf.crf_log_likelihood(
self.logits, self.labels, self.sequence_lengths)
self.loss = tf.reduce_mean(-log_likelihood)