stephen-v / zh-ner-keras Goto Github PK
View Code? Open in Web Editor NEWdetails
Home Page: https://www.cnblogs.com/vipyoumay/p/ner-chinese-keras.html
details
Home Page: https://www.cnblogs.com/vipyoumay/p/ner-chinese-keras.html
Line 52 in 78ab9b3
这里处理Y的mask_value是-1,上面X的是0,不一致?
我看keras_contrib中计算crf_accuracy的源代码里,处理y_pred时使用的mask是input_masks[0],也就是其实跟处理X时是同一个。
请问一下 _process_data(data, vocab, chunk_tags, maxlen=None, onehot=False)这个定义里面,x = [[word2idx.get(w[0].lower(), 1) for w in s] for s in data],为什么要把不在vocab里面的词设置为1?原本的1不是应该有个字对应的吗?
我试过几个数据集,训练都是很快就收敛了,感觉都没怎么训练到
如题
loss: 9.3887 - crf_accuracy: 0.8581
我的loss 从14 降到9 就降不动了 但是 crf_accuracy 已经0.85了 请问应该怎么改进呢?
楼主您好,这种写法在预测的时候支持viterbi解码吗
AttributeError: 'Tensor' object has no attribute '_keras_history'
是版本问题吗
关于后期epoch增加对训练效果基本没什么提升 或提升很慢 是因为学习率没设置吗还是别的问题
想问一下数据集的来源哈
目前的程序相当于使用vocab对输入句子进行了编码和padding得到统一长度,那可以使用每个词的词向量作为输入吗?
你能说一下测试集原始数据是什么样的格式吗?就是没有做预测之前的格式是什么样子的?
请问方便提供一下config.pkl吗?
自己换了训练集,按照CRF++训练的格式进行数据处理,结果训练的时候显示,train on 1 sample,validate on 1 samples,我数据集里面标注的很多
tensorflow__version__
2.3.0
tensorflow_addons.version
0.11.2
Raise error at model.add(crf)
"All layers in a Sequential model should have a single output tensor. For multi-output layers, use the functional API."
TypeError: Tensors in list passed to 'values' of 'ConcatV2' Op have types [bool, float32] that don't all match.
when run this line:
model.add(crf)
when I use keras' version 2.2.0 ,the issue is You are trying to load a weight file containing 16 layers into a model with 0 layers.when I load model. when I downgrade keras' version to 2.1.6, the new question is ImportError: cannot import name 'normalize_data_format', I want to know your version for keras, thank you
您好!请问您的数据是怎样进行预处理的呢?我使用的标注工具是doccano,导出来的数据的格式不是我所需要的,但是用代码处理的话,又有点难,请问您是自己写代码预处理的数据呢还是有专门的工具来规范化呢?
BLSTM可以自动标注,但我看你测试数据集和训练数据集都是标注好的,最后用一段话来测试。不应该是用测试集来测试吗?我有点没有看懂
这个model/crf.h5文件是自动生成的吗?
看不太懂...
遇到这么个问题AttributeError: module 'keras.backend' has no attribute 'slice',有人碰到吗?
请问怎么计算precision、recall、以及F1-score呢?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.