Pytorch serving practice

TODO

What and Why MLflow?
MLflow Installation
MLflow vs. BentoML
MLflow Simple Usage
1. MLflow tutorial
  1. https://www.mlflow.org/docs/latest/tutorials-and-examples/index.html

Open-source tool to organize your entire ML lifecycle
Non-cloud service solution
MLflow provides solutions for ...
1. Managing ML process and deployment
2. experimentation
3. reproducibility
4. deployment
5. central model registry
MLflow pros
1. model tracking mechanism is easy to set up
2. Offers very intuitive APIs for serving
3. logging is practical and simplified, easy to run experiments
4. Code-first approach
MLflow cons
1. The addition of extra workings to the models is not automatic
2. Not quite easy and ideal for deploying models to different platforms

pip install mlflow bentoml

MLflow provices components that work great for experimentation management, ML project management
BentoML only focuses on serving and deploying trained models
1. Can serve models logged in MLFlow experimentation with BentoML

Name		Name	Last commit message	Last commit date
Latest commit History 8 Commits
.vscode		.vscode
mlruns/0		mlruns/0
notebook		notebook
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
bentoml_quick_start_guide.ipynb		bentoml_quick_start_guide.ipynb
mlflow.py		mlflow.py
mlflow_classifier.py		mlflow_classifier.py
pytorch_fashion_mnist.ipynb		pytorch_fashion_mnist.ipynb