-
Thank you very much for your project, it has helped me a lot.
However, I had a problem during the running process that how to train CNN RCNN DPCNN without bert pretrained in your code?
I hope you c…
-
Hi, in the DPCNN code, I find that the function "tf.layers.batch_normalization" is used when training the model. But the parameter "training" is not set True! e.g conv3 = tf.layers.batch_normalizati…
-
你好,看了你的代码受益匪浅,对于Bert的变种我有一个问题想问一下,就是Bert_CNN、Bert_RNN、Bert_RCNN以及Bert_DPCNN模型都是出自于某一篇论文吗?如果是的话可否将论文名称或者论文链接分享一下?谢谢。
ghost updated
2 years ago
-
I think it is wrong that the same conv3 layer is used for every convolution in the network.
You implemented is as if they would share weights.
However, in the paper, the convolutions don't share wei…
-
x = x.squeeze()
#x: [batch_size, filter_num]
change:
x = x.view(batch_size, 2*filter_num)
#x: [batch_size,2* filter_num]
-
`自己写的预测类,输入文本即可得出结果,但是要用tensorflow才行,怎么把tensorflow pad那个删除?
下面写的预测类代码在TextCNN测试通过,但是在 kr.preprocessing.sequence.pad_sequences这个用到了kears代码,怎么把这个换成pyotorch的或者numpy的?请大神指教,大家一起交流下
from tensorflow…
-
初步写了个FastText的配置类:https://github.com/maoding1/sinaCrawler/blob/master/eval.py 其他模型的预测可以改改__init__方法里的配置试试
-
用词来训练准确率只有20%多?
-
> Another area I started looking into (but haven't deeply explored yet) for both figuring out how to map variable names to sections of code in a 'smart' way, and potentially also for module identifica…
-