苏剑林(Jianlin Su)

Results 390 comments of 苏剑林(Jianlin Su)

you need to try model.fit_generator(), see: http://keras-cn.readthedocs.io/en/latest/models/sequential/

我的问题是哪里用了batch_labels,不是哪里创建了batch_labels。创建了不代表用了,它可能只是一个占位符。 你既然没找到哪里用了batch_labels,怎么可以这么自信地断定“相似度矩阵y_pred不应该为1吗”?

https://github.com/bojone/bert4keras/tree/master/examples/datasets

自行写代码设置。总有办法的。

```python train_generator = data_generator(train_token_ids, 64) ``` 里边的64就是batch_size

load_model的时候传入custom_objects={'PtuningEmbedding': PtuningEmbedding}

oh, sorry. it is just a handwriting mistake while I prepare the code to github. I have corrected it now.