guillaumegenthial / tf_ner

Simple and Efficient Tensorflow implementations of NER models with tf.estimator and tf.data
Apache License 2.0
923 stars 275 forks source link

same result on F1, Accuracy, precision #68

Open wentinghome opened 4 years ago

wentinghome commented 4 years ago

Hi,

I have applied a customized dataset in the few shot setting, the training contains 500, valid contains 1k, and the test contains 10k examples, Then, the result on F1, Accuracy, precision I got is exactly the same in the following. Is anyone has a similar problem and any suggestions on this? Thanks so much!


Saving dict for global step 1879: acc = 0.9462044, f1 = 0.9462044, global_step = 1879, loss = 14.065788, precision = 0.9462044, recall = 0.9462044