Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/martinkalema/bentoml-test
Learnt bentoML
https://github.com/martinkalema/bentoml-test
bentoml localstorage mlops rest-api
Last synced: about 1 month ago
JSON representation
Learnt bentoML
- Host: GitHub
- URL: https://github.com/martinkalema/bentoml-test
- Owner: MartinKalema
- License: apache-2.0
- Created: 2024-04-03T16:24:29.000Z (10 months ago)
- Default Branch: main
- Last Pushed: 2024-04-03T18:31:33.000Z (10 months ago)
- Last Synced: 2024-11-12T11:09:00.676Z (3 months ago)
- Topics: bentoml, localstorage, mlops, rest-api
- Language: Python
- Homepage:
- Size: 11.7 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
### BentoML
BentoML is an open-source platform for serving, managing, and deploying machine learning models. It allows you to package trained models with their dependencies into a format that can be easily deployed and managed in various production environments.
BentoML provides a unified interface for serving machine learning models via REST API endpoints and web applications. It supports popular ML frameworks like TensorFlow, PyTorch, Scikit-learn, XGBoost, and others.
Its used during model training, and saves different model versions to your local storage which can be retrieved and used by creating a service.
### CLI
- Create a virtual environment
```bash
conda create -p venv python==3.9 -y
```
- Activate the environment
```bash
conda activate venv/
```
- Install dependencies
```bash
pip install bentoml scikit-learn pandas
```
- Create a `download_model.py` file as shown below.
```python
import bentoml
from sklearn import svm
from sklearn import datasets
# Load training data set
iris = datasets.load_iris()
X, y = iris.data, iris.target
# Train the model
clf = svm.SVC(gamma='scale')
clf.fit(X, y)
# Save model to the BentoML local Model Store
saved_model = bentoml.sklearn.save_model("iris_clf", clf)
```
- Run the script to download the model
```bash
python download_model.py
```
- The model is now saved in the Model Store with the name iris_clf and an automatically generated version. You can retrieve this model later by using the name and version to create a BentoML Service. Run the command below to view all the available models in the Model Store.
```bash
bentoml models list
```
- Create a bentoML service and a model Runner by creating the file below named `service.py`.
```python
import numpy as np
import bentoml
from bentoml.io import NumpyNdarray
iris_clf_runner = bentoml.sklearn.get("iris_clf:latest").to_runner()
svc = bentoml.Service("iris_classifier", runners=[iris_clf_runner])
@svc.api(input=NumpyNdarray(), output=NumpyNdarray())
def classify(input_series: np.ndarray) -> np.ndarray:
result = iris_clf_runner.predict.run(input_series)
return result
```
- Run the service
```bash
bentoml serve service:svc
```
- Build a bento
After the Service is ready, you can package it into a Bento by specifying a configuration YAML file (`bentofile.yaml`) that defines the build options.
```bash
service: "service:svc" # Same as the argument passed to `bentoml serve`
labels:
owner: bentoml-team
stage: dev
include:
- "*.py" # A pattern for matching which files to include in the Bento
python:
packages: # Additional pip packages required by the Service
- scikit-learn
- pandas
models: # The model to be used for building the Bento.
- iris_clf:latest
```
- Run `bentoml build` in your project directory to build the Bento.
```bash
bentoml build
```
- View all available bentos
```bash
bentoml list
```
- Deploy a bento
```bash
bentoml containerize iris_classifier:latest
```
Next steps:
- Deploy to BentoCloud:
```bash
bentoml deploy iris_classifier:ojx7ofxr46mocxha -n ${DEPLOYMENT_NAME}
```
- Update an existing deployment on BentoCloud:
```bash
bentoml deployment update --bento iris_classifier:ojx7ofxr46mocxha ${DEPLOYMENT_NAME}
```
- Containerize your Bento with `bentoml containerize`:
```bash
bentoml containerize iris_classifier:ojx7ofxr46mocxha [or bentoml build --containerize]
```
- Push to BentoCloud with `bentoml push`:
```bash
bentoml push iris_classifier:ojx7ofxr46mocxha [or bentoml build --push]
```