ntedgi / albert-fine-tuning-squad-2.0

Fine-tuning a Transformer albert model for Question Answering on Squad 2.0

the Albert Model for question answering fine tuning squad 2.0

links:

• [Squad 2.0] (https://rajpurkar.github.io/SQuAD-explorer/)
• [Albert] (https://ai.googleblog.com/2019/12/albert-lite-bert-for-self-supervised.html)

How I Run This Project Locally ?

• clone this repository

• Just Want to Play ?

  • Run : bash install.sh
  • Run : bash run.sh

• request:

  • End-Point : GET http://localhost:8000/predict
  • Params : { question : String , paragraphs : Array}

• result:

  • Type : Json
  • Result : {result : [paragraphIndex:Int , Answer:String]

• example:

curl -X GET \  http://localhost:8000/predict \ -H 'Content-Type: text/plain' \ -d '{"question":"Who was Jim Henson?", "paragraphs":[ {"id": 1, "text": "has a nice car. Jim Henson was a nice puppet."} , {"id":2,"text": "All the 2023(GoF) design patterns implemented inJavascript.Jim Henson was a monkey king"}]}'

{ "result": [ [ 1, "a nice puppet" ], [ 2, "a monkey king" ] ] }