lic2020_baselines icon indicating copy to clipboard operation
lic2020_baselines copied to clipboard

some baselines for lic2020 (http://lic2020.cipsc.org.cn/)

Results 5 lic2020_baselines issues
Sort by recently updated
recently updated
newest added

大佬,问个问题,这句代码 nodes = model.predict([[token_ids], [segment_ids]])[0]出现了这个bug Layer "model_21" expects 2 input(s), but it received 1 input tensors. Inputs received: [],我感觉没问题呀,输入的数据无论是数量还是格式都正确呀,可否解答一下,怎么修改。

苏神您好,感谢baseline! 想请教您一个问题,为什么[bert.model.layers[-2].get_output_at(-1)](https://github.com/bojone/lic2020_baselines/blob/78ab5e13721472e3be41cbd3ae596f9109096fa8/ie.py#L178) 获取的输出,是 bert.model.layers[-2], 而不是 bert.model.output呢? 希望可以得到您的回复,感谢!

748/748 [==============================] - 1023s 1s/step - loss: 32.8033 - sparse_accuracy: 0.8512 45%|####################2 | 673/1498 [01:36

![image](https://user-images.githubusercontent.com/16755610/98643263-bdee9580-2369-11eb-95e6-dbcb65c0377a.png) 如果存在多个主体,那对主体的随机抽取,是否应该保持该主体的start和end一致的呢?

File "rc.py", line 99, in call return K.softmax(inputs, 1) TypeError: softmax() takes 1 positional argument but 2 were given 请问这个要怎么解决呀