Multi_Label_Classifier_finetune
Multi_Label_Classifier_finetune copied to clipboard
微调预训练语言模型,解决多标签分类任务(可加载BERT、Roberta、Bert-wwm以及albert等知名开源tf格式的模型)
FAILED: Failed 20210520082028730gvn0uw8a_1b9034b2_0afc_41f2_8580_17231ee747f7:ODPS-1202005:Algo Job Failed-User Error-Tensorflow script runs failed with exit code: 123, please see the details in logview. The tail contents of the stderr file: eval_metric_ops must be (metric_value,...
您好,感谢贡献代码,对我有很大的帮助! 我在使用bert多多标签分类时,看到您的代码的eval中只计算了auc,代码在这里 `current_auc, update_op_auc = tf.metrics.auc(label_id_, probabilities)` 可以看到这里传入是label_id 和probabilities,比如在标签概率中我想设置阈值为0.5的话,该如何设置呢?多谢解答!