-
文章给出的lattcie-lstm的结构与我看到的[Chinese NER Using Lattice LSTM](https://arxiv.org/pdf/1805.02023.pdf)有点不一样, 有个问题想要请教一下, 按照[Chinese NER Using Lattice LSTM](https://arxiv.org/pdf/1805.02023.pdf)中的构建方法, 重庆人和药店…
STHSF updated
1 month ago
-
Thank you for your work to modify the lattice LSTM code! I have tried your lattice LSTM code, but my f1-value just achieve 0.82, much lower than the other models with the same dataset. I want to know …
-
Hello, thank you very much for sharing the source code of lattice-lstm. I have some questions about lattice-LSTM, that is, how do we get word_length_tensor and lexicon_word_embedding_inputs. In other …
-
Hi, thank you so much for you elegant reproduction of batched version of Lattice LSTM.
But I can't get the same results on Weibo Datasets, using the same emb\params\data. I just got F1 0.44 (normal…
-
您好，我现在参考的这个代码，但是我的地址识别精度比较细，所以想把召回率提高，这个怎么改？
-
I would like to propose an extension to the API (and CLI) for producing a **character lattice** (directed acyclic character hypotheses graph) as recognition result with LSTM models.
#### _Rational…
-
在找字符c的B、M、E、S的集合时候，以B为例，就需要找到以c开头的属于L的所有的词，L是从哪里来的呢。
-
-
您好，您用的代码中，gaz是怎么用的，因为我遍历了所有文件,gaz_dropout并没有用到在latticeLSTM的model中，只是声明了一下，是我遗漏了什么么，，同时两个latticeLSTm带不带V1有什么具体区别么？谢谢
-