To prepare the data, You need to create a directory under the path data/, which contains train.txt, valid.txt, and test.txt.
The functions that process datasets are located in the file data/data_process.py. After the processing, each line in the files should be a dialogue sample. Different utterances are separated with \t and the last utterance is the response that the model needs to generate.
你好，知道北京市规划展览馆在什么地方吗？ 嗯呢，地址在北京市东城区前门东大街20号（老北京火车站东侧），电话你要不？ 不用了，我已经知道了，电话是010-67017074，帮我查一下这里可以玩多久好了？ 大概可以我1小时 - 2小时，知道这里啥时间开放吗？ 周一闭馆，周二-周日9:00-17:00，16:00停止入馆，门票贵不？ 免费开放，凭有效证件领票入场。 挺好呀，该景点周边有没有其他好玩的地方啊？ 好多呢，比如故宫，天安门广场，恭王府等，都是历史遗留的产物。 我想去恭王府看看，能把详细地址发给我吗？ 可以，地址在北京市西城区柳荫街甲14号，电话你有没？ 有的，电话是010-83288149，这里能够玩多久呀？ 差不多能玩2小时 - 4小时吧，门票贵不？
All the python scripts to run is in src/.
generate.py: Invoke the API of the model and generate the response.metric.py: Calculate metrics according to the generated responses.generation_metrics.py: Include utility functions used to calculate metrics.
Before running the code, please change model in the script according to the URL. The results can be found in results/{{model}}/{{testset}}.
The code for creating database is located in human evaluation/database.
init.py: Create the mongodb database and collections.
You need to first modify line 4 and input your own ip address.
Then, run init.py to initialize mongodb.
The code for generating self-chat dialogue is located in human evaluation/self-chat generation.
db_op.py: Methods for manipulating mongodb.self_chat_creator.py: Generate self-chat dialogue.
Run self_chat_creator to generate self-chat dialogue.
The code for deploying backend is located in human evaluation/backend.
db_op.py: Methods for manipulating mongodb.chatbot_api: Getting response from available chatbot.app.py: Deploy the flask backend.
You should first modify line 4 in db_op.py and input your own mongodb information.
Then, run app.py to deploy the backend.
The code for deploying frontend is located in human evaluation/frontend
run npm install to setup the project.
run npm run dev to compile and hot-reload for development
run npm run build to compile and minify for production
The code for creating database is located in implicit human evaluation/database.
init.py: Create the mongodb database and collections.
You need to first modify line 4 and input your own ip address.
Then, run init.py to initialize mongodb.
The code for deploying backend is located in implicit human evaluation/backend.
db_op.py: Methods for manipulating mongodb.chatbot_api: Getting response from available chatbot.app.py: Deploy the flask backend.
You should first modify line 4 in db_op.py and input your own mongodb information.
Then, run app.py to deploy the backend.
The code for deploying frontend is located in implicit human evaluation/frontend
run npm install to setup the project.
run npm run dev to compile and hot-reload for development
run npm run build to compile and minify for production