Sentence_paraphrasing_t5_model

Project Structure -

Dataset: This model was trained on T5 model which uses C4 dataset(Common Crawl's web crawl corpus).

Codes:

Files used:

1. app.py: runs on 6000 port, given a sentence the system tries to find the different variations of the question for the user input
2. logs: The file stores the logging of the service. The projects uses python standard logging to log date, time with information and error type.
3. dockerfile: The file consits of various commands and libraries to run the file in the docker

Dependencies: The project is built using python 3.X. Open source packges:

• tensorflow(gpu)
• tensorflow-hub
• pytorch(CUDA 10.1)
• flask
• numpy
• transformers

Installing libraries:

requirements.txt: contains all the packages required to run the project. Run the below command to install the required libraries for the project pip install -r requirements.txt Pytorch version needs to be installed depending upon the CUDA Version from pytorch.org

Build Docker Image:

1. Build docker using the command docker build -t myname/myapp
2. Run the docker using docker run -p 6000:6000 myname/myapp

Response from API:

Request Example: {"comment": query string}

Response Type: {"text" : answer string1 answer string2 answer string3}