blogging_template

A blogging template to make it faster to share project code for blog posts

Table of Contents

1. Usage
2. Requirements
3. Development
   1. Installing Dependencies
   2. Download & Saving a Model
   3. Serving in Development
   4. Testing

Usage

The purpose of this template is to quickly create projects for devs to experiment with code from my blog posts

Requirements

• Python 3.8
• Poetry 1.1
• Docker

Development

Installing Dependencies

From within the root directory:

poetry install

Download & Save a Trained Tensorflow Model

Before serving in development using docker compose, you must follow the setup steps documented here:

Downloading & Saving a Trained Tensorflow Model README

Serving in Development

Included a minimal example from fastapi to get started with a production server.

To rebuild the api run

docker-compose build

To serve the api locally run

docker-compose up api

To serve the full stack locally run

docker-compose up

Once that runs, you can use the api by going to the following endpoints:

1. Main Api: http://localhost:8000/
2. Api Docs (Auto Generated from FastAPI): http://localhost:8000/docs
3. Tensorflow Serving: http://localhost:8501/v1/models/sentiment_model/metadata
4. Main Api endpoint returning model inference: http://localhost:8000/text/

Testing

From within the root directory:

poetry run pytest tests