PeterPanUnderhill

这段计算eval performance的代码是不是有错误

1

我看了一下代码逻辑，由split后的probabilities和labels来计算accuracy, 下面这个代码块中加粗部分是否应该改成label_ids_split=tf.split(label_ids,FLAGS.num_aspects,axis=-1)？ 这个是否与其他人po的eval_accuracy出错有关？ `def metric_fn(per_example_loss, label_ids, logits): #print("###metric_fn.logits:",logits.shape) # (?,80) #predictions = tf.argmax(logits, axis=-1, output_type=tf.int32) #print("###metric_fn.label_ids:",label_ids.shape,";predictions:",predictions.shape) # label_ids: (?,80);predictions:(?,) logits_split=tf.split(logits,FLAGS.num_aspects,axis=-1) # a list. length is num_aspects **label_ids_split=tf.split(logits,FLAGS.num_aspects,axis=-1)** # a list....