macanv / BERT-BiLSTM-CRF-NER

Tensorflow solution of NER task Using BiLSTM-CRF model with Google BERT Fine-tuning And private Server services
https://github.com/macanv/BERT-BiLSMT-CRF-NER
4.68k stars 1.25k forks source link

修改优化器为Lookahead+RAdam,无效 #321

Open currenttime opened 4 years ago

currenttime commented 4 years ago

修改优化器为Lookahead+RAdam 参考https://github.com/Clarence-wen/Ranger-tensorflow里的Ranger实现 将optimizer.py改为optimizer = Ranger(learning_rate=learning_rate) 结果不生效accuracy: 85.49%; precision: 0.00%; recall: 0.00%; FB1: 0.00

另外也尝试过Radam和单纯Lookahead+AdamW也没有效果 请问是什么原因,是只能用AdamW这一中优化器吗?还是有其他地方需要改动?