LINTO-PLATFORM-PUNCTUATION

LinTO-platform-punctuation is a LinTO service for punctuation prediction. It predicts punctuation from raw text or raw transcription.

LinTO-platform-punctuation can either be used as a standalone punctuation service or deployed as a micro-service.

Table of content

• Prerequisites

  • Models

• Deploy

  • HTTP
  • MicroService

• Usage

  • HTTP API
    • /healthcheck
    • /punctuation
    • /docs
  • Using celery

• License

---

Pre-requisites

Models

The punctuation service relies on a trained punctuation prediction model.

We provide homebrew models on dl.linto.ai.

Docker

The punctuation service requires docker up and running.

(micro-service) Service broker

The punctuation only entry point in job mode are tasks posted on a REDIS message broker using Celery.

Deploy

linto-platform-punctuation can be deployed two different ways:

• As a standalone punctuation service through an HTTP API.
• As a micro-service connected to a task queue.

1- First step is to build the image:

git clone https://github.com/linto-ai/linto-platform-punctuation.git
cd linto-platform-punctuation
docker build . -t linto-platform-punctuation:latest

or

docker pull registry.linto.ai/lintoai/linto-platform-punctuation:latest

2- Download the models

Have the punctuation model (.mar) ready at MODEL_PATH.

HTTP

1- Fill the .env

cp .env_default_http .env

Fill the .env with your values.

Parameters:

Variables	Description	Example
SERVICE_NAME	The service's name	my_punctuation_service
CONCURRENCY	Number of worker	> 1

2- Run with docker

docker run --rm \
-v MODEL_PATH:/usr/src/app/model-store/punctuation.mar \
-p HOST_SERVING_PORT:80 \
--env-file .env \
linto-platform-punctuation:latest

This will run a container providing an http API binded on the host HOST_SERVING_PORT port.

Micro-service

LinTO-platform-punctuation can be deployed as a microservice. Used this way, the container spawn celery workers waiting for punctuation tasks on a dedicated task queue. LinTO-platform-punctuation in task mode requires a configured REDIS broker.

You need a message broker up and running at MY_SERVICE_BROKER. Instance are typically deployed as services in a docker swarm using the docker compose command:

1- Fill the .env

cp .env_default_task .env

Fill the .env with your values.

Parameters:

Variables	Description	Example
SERVICES_BROKER	Service broker uri	redis://my_redis_broker:6379
BROKER_PASS	Service broker password (Leave empty if there is no password)	my_password
QUEUE_NAME	(Optionnal) overide the generated queue's name (See Queue name bellow)	my_queue
SERVICE_NAME	Service's name	punctuation-ml
LANGUAGE	Language code as a BCP-47 code	en-US or * or languages separated by "|"
MODEL_INFO	Human readable description of the model	"Bert based model for french punctuation prediction"
CONCURRENCY	Number of worker (1 worker = 1 cpu)	>1

Do not use spaces or character "_" for SERVICE_NAME or language.

2- Fill the docker-compose.yml

#docker-compose.yml

version: '3.7'

services:
  punctuation-service:
    image: linto-platform-punctuation:latest
    volumes:
      - /my/path/to/models/punctuation.mar:/usr/src/app/model-store/punctuation.mar
    env_file: .env
    deploy:
      replicas: 1
    networks:
      - your-net

networks:
  your-net:
    external: true

3- Run with docker compose

docker stack deploy --resolve-image always --compose-file docker-compose.yml your_stack

Queue name:

By default the service queue name is generated using SERVICE_NAME and LANGUAGE: punctuation_{LANGUAGE}_{SERVICE_NAME}.

The queue name can be overided using the QUEUE_NAME env variable.

Service discovery:

As a micro-service, the instance will register itself in the service registry for discovery. The service information are stored as a JSON object in redis's db0 under the id service:{HOST_NAME}.

The following information are registered:

{
  "service_name": $SERVICE_NAME,
  "host_name": $HOST_NAME,
  "service_type": "punctuation",
  "service_language": $LANGUAGE,
  "queue_name": $QUEUE_NAME,
  "version": "1.2.0", # This repository's version
  "info": "Bert Based Punctuation model for french punctuation prediction",
  "last_alive": 65478213,
  "concurrency": 1
}

Usages

HTTP API

/healthcheck

Returns the state of the API

Method: GET

Returns "1" if healthcheck passes.

/punctuation

Punctuation API

• Method: POST
• Response content: text/plain or application/json
• Body: A json object structured as follows:

{
  "sentences": [
    "this is sentence 1", "is that a second sentence", "yet an other sentence"
  ]
}

Return the punctuated text as a json object structured as follows:

{
  "punctuated_sentences": [
    "This is sentence 1",
    "Is that a second sentence ?",
    "Yet an other sentence"
  ]
}

/docs

The /docs route offers a OpenAPI/swagger interface.

Using Celery

Punctuation-Worker accepts celery tasks with the following arguments: text: Union[str, List[str]]

• text: (str or list) A sentence or a list of sentences.

Return format

Returns a string or a list of string depending on the input parameter.

Test

Curl

You can test you http API using curl:

curl -X POST "http://YOUR_SERVICE:YOUR_PORT/punctuation" -H  "accept: application/json" -H  "Content-Type: application/json" -d "{  \"sentences\": [    \"this is sentence 1\", \"is that a second sentence\", \"yet an other sentence\"  ]}"

License

This project is developped under the AGPLv3 License (see LICENSE).