:exclamation::exclamation: This repo will no longer be maintained, please visit https://github.com/milvus-io/bootcamp :exclamation: :exclamation:

README

This project combines Milvus and the model provided by Bert to realize a question and answer system.This project aims to provide a solution to achieve semantic similarity matching with Milvus combined with various AI models.

Data description

The question-and-answer data set needed for this project is a csv file includes questions and answers.

The data set in the data directory is a sample data.

config description

QA/config.py：The script is a configuration file and needs to be modified for the specific environment.

Steps to build a project

1.Install Milvus 0.10.4

2.Install PostgreSQL

3.Install the Python packages you need

pip install -r requriment.txt

4.Start the Bert services (more Bert related)

#Download model
$ cd model
$ wget https://storage.googleapis.com/bert_models/2018_11_03/english_L-12_H-768_A-12.zip
#start service
$ bert-serving-start -model_dir /model/english_L-12_H-768_A-12/ -num_worker=2 -max_seq_len=40

If you want to build a Q&A system in other languages, you can download the corresponding language model.

5. Start the query service

uvicorn main:app --host 127.0.0.1 --port 8000

6. Enter 127.0.0.1:8000/docs in the web page to view the interface provided by this project.