Text Analytics Microservice

Getting Started

# 1. Create venv
python3 -m venv text-analytics-env

# 2. Activate venv
source text-analytics-env/bin/activate

# 3. Install Dependencies
pip install -r requirements.txt

Installing Models

Source: https://public.ukp.informatik.tu-darmstadt.de/reimers/sentence-transformers/v0.2/
We are using paraphrase-distilroberta-base-v2 so lets download it.

1. cd serverless-bert
2. wget https://public.ukp.informatik.tu-darmstadt.de/reimers/sentence-transformers/v0.2/paraphrase-distilroberta-base-v2.zip

API Format

Request Body | Expected Input (Data Format From Rails Backend)

Request_body: {
        "data": [model_answer, student_res_1, student_res_2, student_res_3] #len = n
    }

Response Payload

{
    "data": [
        {"lexical": 0.5, "semantic": 0.7, "overall": 0.6}, ... , {"lexical": 0.5, "semantic": 0.7, "overall": 0.6}] #len = n-1
}

Note:

Do take note to uncomment the lines accordingly on depending whether you're running docker locally or pushing to aws!

---

Build Custom docker Image

docker build -t bert-lambda .

Test our function locally

Start docker locally:

docker run -p 8080:8080 bert-lambda

Subsequently, we can then locally invoke the function using curl, a REST-Client or run python req_script.py and we should expect a 200 status code.

Deploy a custom docker image to ECR

Since we now have a local docker image, we can deploy this to ECR. Therefore we need to create an ECR repository with the name bert-lambda.

aws ecr create-repository --repository-name bert-lambda > /dev/null

To be able to push our images we need to login to ECR. We need to define some environment variables to make deploying easier.

aws_region=ap-southeast-1
aws_account_id=<insert aws_account_id>

aws ecr get-login-password \
    --region $aws_region \
| docker login \
    --username AWS \
    --password-stdin $aws_account_id.dkr.ecr.$aws_region.amazonaws.com

Next we need to tag / rename our previously created image to an ECR format. The format for this is {AccountID}.dkr.ecr.{region}.amazonaws.com/{repository-name}

docker tag bert-lambda $aws_account_id.dkr.ecr.$aws_region.amazonaws.com/bert-lambda

To check if it worked we can run docker images and should see an image with our tag as name

Finally, we push the image to ECR Registry.

 docker push $aws_account_id.dkr.ecr.$aws_region.amazonaws.com/bert-lambda

---

Deploy AWS Lambda function with a custom docker image

service: serverless-bert-lambda-docker

provider:
  name: aws # provider
  region: ap-southeast-1 # aws region
  memorySize: 5120 # optional, in MB, default is 1024
  timeout: 60 # optional, in seconds, default is 6

functions:
  lexicalSemanticScoring:
    image: 029104380498.dkr.ecr.ap-southeast-1.amazonaws.com/bert-lambda:latest #ecr url
    events:
      - http:
          path: qa # http path
          method: post # http method

For an ECR image, the URL should look like this {AccountID}.dkr.ecr.{region}.amazonaws.com/{repository-name}@{digest}

To deploy the function, we run:

serverless deploy

---

Test our Serverless BERT API

To test our Lambda function, uncomment the "request aws" section and run python req_script.py

---