MLEM helps you package and deploy machine learning models. It saves ML models in a standard format that can be used in a variety of production scenarios such as real-time REST serving or batch processing.
-
Run your ML models anywhere: Wrap models as a Python package or Docker Image, or deploy them to Heroku (SageMaker, Kubernetes, and more platforms coming soon). Switch between platforms transparently, with a single command.
-
Model metadata into YAML automatically: Automatically include Python requirements and input data needs into a human-readable, deployment-ready format. Use the same metafile on any ML framework.
-
Stick to your training workflow: MLEM doesn't ask you to rewrite model training code. Add just two lines around your Python code: one to import the library and one to save the model.
-
Developer-first experience: Use the CLI when you feel like DevOps, or the API if you feel like a developer.
The main reason to use MLEM instead of other tools is to adopt a GitOps approach to manage model lifecycles.
-
Git as a single source of truth: MLEM writes model metadata to a plain text file that can be versioned in Git along with code. This enables GitFlow and other software engineering best practices.
-
Unify model and software deployment: Release models using the same processes used for software updates (branching, pull requests, etc.).
-
Reuse existing Git infrastructure: Use familiar hosting like Github or Gitlab for model management, instead of having separate services.
-
UNIX philosophy: MLEM is a modular tool that solves one problem very well. It integrates well into a larger toolset from Iterative.ai, such as DVC and CML.
This a quick walkthrough showcasing deployment and export functionality of MLEM.
Please read Get Started guide for a full version.
MLEM requires Python 3.
$ python -m pip install mlemTo install the pre-release version:
$ python -m pip install git+https://github.com/iterative/mlem
# train.py
from mlem.api import save
from sklearn.ensemble import RandomForestClassifier
from sklearn.datasets import load_iris
def main():
data, y = load_iris(return_X_y=True, as_frame=True)
rf = RandomForestClassifier(
n_jobs=2,
random_state=42,
)
rf.fit(data, y)
save(
rf,
"rf",
sample_data=data,
)
if __name__ == "__main__":
main()Check out what we have:
$ ls
rf
rf.mlem
$ cat rf.mlemClick to show `cat` output
artifacts:
data:
hash: ea4f1bf769414fdacc2075ef9de73be5
size: 163651
uri: rf
model_type:
methods:
predict:
args:
- name: data
type_:
columns:
- sepal length (cm)
- sepal width (cm)
- petal length (cm)
- petal width (cm)
dtypes:
- float64
- float64
- float64
- float64
index_cols: []
type: dataframe
name: predict
returns:
dtype: int64
shape:
- null
type: ndarray
predict_proba:
args:
- name: data
type_:
columns:
- sepal length (cm)
- sepal width (cm)
- petal length (cm)
- petal width (cm)
dtypes:
- float64
- float64
- float64
- float64
index_cols: []
type: dataframe
name: predict_proba
returns:
dtype: float64
shape:
- null
- 3
type: ndarray
sklearn_predict:
args:
- name: X
type_:
columns:
- sepal length (cm)
- sepal width (cm)
- petal length (cm)
- petal width (cm)
dtypes:
- float64
- float64
- float64
- float64
index_cols: []
type: dataframe
name: predict
returns:
dtype: int64
shape:
- null
type: ndarray
sklearn_predict_proba:
args:
- name: X
type_:
columns:
- sepal length (cm)
- sepal width (cm)
- petal length (cm)
- petal width (cm)
dtypes:
- float64
- float64
- float64
- float64
index_cols: []
type: dataframe
name: predict_proba
returns:
dtype: float64
shape:
- null
- 3
type: ndarray
type: sklearn
object_type: model
requirements:
- module: sklearn
version: 1.0.2
- module: pandas
version: 1.4.1
- module: numpy
version: 1.22.3If you want to follow this Quick Start, you'll need to sign up on https://heroku.com,
create an API_KEY and populate HEROKU_API_KEY env var.
First, create an environment to deploy your model:
$ mlem declare env heroku staging
💾 Saving env to staging.mlemNow we can deploy the model with mlem deploy
(you need to use different app_name, since it's going to be published on https://herokuapp.com):
$ mlem deployment run mydeploy -m rf -t staging -c app_name=mlem-quick-start
⏳️ Loading deployment from .mlem/deployment/myservice.mlem
🔗 Loading link to .mlem/env/staging.mlem
🔗 Loading link to .mlem/model/rf.mlem
💾 Updating deployment at .mlem/deployment/myservice.mlem
🏛 Creating Heroku App example-mlem-get-started
💾 Updating deployment at .mlem/deployment/myservice.mlem
🛠 Creating docker image for heroku
💼 Adding model files...
🛠 Generating dockerfile...
💼 Adding sources...
💼 Generating requirements file...
🛠 Building docker image registry.heroku.com/example-mlem-get-started/web...
✅ Built docker image registry.heroku.com/example-mlem-get-started/web
🔼 Pushed image registry.heroku.com/example-mlem-get-started/web to remote registry at host registry.heroku.com
💾 Updating deployment at .mlem/deployment/myservice.mlem
🛠 Releasing app my-mlem-service formation
💾 Updating deployment at .mlem/deployment/myservice.mlem
✅ Service example-mlem-get-started is up. You can check it out at https://mlem-quick-start.herokuapp.com/You could easily export the model to a different format using mlem build:
$ mlem build rf docker -c server.type=fastapi -c image.name=sklearn-model
⏳️ Loading model from rf.mlem
🛠 Building MLEM wheel file...
💼 Adding model files...
🛠 Generating dockerfile...
💼 Adding sources...
💼 Generating requirements file...
🛠 Building docker image sklearn-model:latest...
✅ Built docker image sklearn-model:latest
Contributions are welcome! Please see our Contributing Guide for more details.
Check out the MLEM weekly board to learn about what we do, and about the exciting new functionality that is going to be added soon.
Thanks to all our contributors!
This project is distributed under the Apache license version 2.0 (see the LICENSE file in the project root).
By submitting a pull request to this project, you agree to license your contribution under the Apache license version 2.0 to this project.