＜Rasa实战＞第八章实例运行

1. 指令执行步骤

1).训练数据

rasa train

2).启动Rasa动作服务器

rasa run actions

3).启动Rasa服务器和客户端(在新建的命令行里)

rasa shell

2. 问题追踪

Q1:训练数据时报警告"constrain_similarities is set to False. It is recommended to set it to True when using cross-entropy loss."

参考:
https://stackoverflow.com/questions/71361366/userwarning-constrain-similarities-is-set-to-false-it-is-recommended-to-set
https://serveanswer.com/questions/userwarning-constrain_similarities-is-set-to-false-it-is-recommended-to-set-it-to-true-when-using-cross-entropy-loss-nlu-model-loaded
https://github.com/RasaHQ/rasa/pull/6753

Q2:训练数据时报警告"Misaligned entity annotation…the tokenizer gives an unexpected result, due to languages such as Chinese that don’t use whitespace for word separation"

与JiebaTokenizer有关,第9章使用自定义分词组件,可以完美解决这个难题.

3. 参考:

我正在「AI爱好者社区」服务器中聊天，来和我一起畅聊吧 ~ 点击加入：https://fanbook.mobi/3H6D5FVN