Walle

暗物智能 Guangzhou i robot

Results 24 issues of


Walle

kbqa_test.py文件的kb_fuzzy_classify_test函数中的非语义匹配部分我认为应该break跳出循环，为什么没有跳出来呢？

![image](https://user-images.githubusercontent.com/9403745/67384206-4f566280-f5c3-11e9-8681-55b2dfbc6386.png)

Can you add context functionality like tensorflow?

snet-0.1.0.pb and cnet-0.1.0.pb ，I wonder how these two documents were trained? Github above only provides the source code of inferencing, i would like to ask the training part of the source code? Can you provide this part?

hello,can you provide UNET and Deepfill v2 training model source code?thanks!

Hello, I would like to ask why the text mask does not directly use text area detection technology, such as the CTPN model? This mask is a regular rectangle, is it simpler than the U-net model? Then does subsequent Deepfill v2 have an advantage over irregular masks in dealing with rectangular boxes? Looking forward to your answer, thank you!

麻烦你的qabot的文件上传下好吧

7

comment

我debug rasa_nlu的ner_duckling，发现进入train里面之后是空的，是不是意味着这个ner_duckling是无效的啊

4

comment

![image](https://user-images.githubusercontent.com/9403745/54512383-ece31700-498e-11e9-802d-26a840dca337.png)

tracker_store

1

comment

tracker_store: store_type: mongod url: mongodb://192.168.47.185:27017 db: rasa username: password: 这个里面存的是啥？

好有心机啊，跟_rasa_chatbot就差一个字符，哈哈

2

comment

想问下训练集里的develop.json有用到吗？还有测试集eval722.json有标准答案吗？我想做下评估