Bert embeddings sentiment analysis

Part I. Install requirements & train model

• Docker (https://docs.docker.com/get-docker)
• Git (https://git-scm.com/downloads)

1. Go to project dir

git clone https://github.com/Jason-Oleana/bert-embeddings-sentiment-analysis.git
cd bert-embeddings-sentiment-analysis

2. Docker build image

docker build -t notebook .

wait...

3. Docker run container

For Windows Users

docker run -p 8888:8888 -v %cd%/models:/work/models notebook

For MacOS & Linux Users

step 1

sudo chown -R 1000 models/

step 2

docker run -p 8888:8888 -v "$(pwd)"/models:/work/models notebook

6. Run machine learning notebook

• Copy and paste http://127.0.0.1:8888/lab?token=[insert-token]
• Open notebooks
• Open BERT_mlp_model.ipynb
• Run notebook before proceeding to part 2
• Make sure the model is succesfully saved

Part 2. Serve model with FastAPI

1. Docker build image

docker build -t mlapp ./fastapi

wait...

2. Docker run container

For Windows Users

docker run -p 8000:8000 -v %cd%/models:/app/models -v %cd%/fastapi:/app mlapp

For MacOS Users

docker run -p 8000:8000 -v "$(pwd)"/models:/app/models -v "$(pwd)"/fastapi:/app mlapp

3. Test the model

• Open a browser
• Go to localhost:8000/docs

Handy Docker commands

1. Show running containers

docker ps

2. Show Docker images

docker images

3. stop container

docker stop [container-id]

4. Remove image

docker rmi [image-id]

or

docker rmi -f [image-id]