Go backend for MLflow
In order to increase the performance of the tracking server and the various stores, we propose to rewrite the server and store implementation in Go.
Getting started
Installation
pip install mlflow-go-backend
CLI Usage
mlflow-go-backend is meant to be a drop-in-replacement for mlflow.
You can update your existing mlflow command with mlflow-go:
- mlflow server --backend-store-uri postgresql://postgres:postgres@localhost:5432/postgres
+ mlflow-go server --backend-store-uri postgresql://postgres:postgres@localhost:5432/postgres
Every existing setting of mlflow server can be passed to mlflow-go.
Python Usage
import mlflow
import mlflow_go_backend
# Enable the Go client implementation (disabled by default)
mlflow_go_backend.enable_go()
# Set the tracking URI (you can also set it via the environment variable MLFLOW_TRACKING_URI)
# Currently only database URIs are supported
mlflow.set_tracking_uri("sqlite:///mlflow.db")
# Use MLflow as usual
mlflow.set_experiment("my-experiment")
with mlflow.start_run():
mlflow.log_param("param", 1)
mlflow.log_metric("metric", 2)
Python Store Usage
import logging
import mlflow
import mlflow_go_backend
# Enable debug logging
logging.basicConfig()
logging.getLogger('mlflow_go_backend').setLevel(logging.DEBUG)
# Enable the Go client implementation (disabled by default)
mlflow_go_backend.enable_go()
# Instantiate the tracking store with a database URI
tracking_store = mlflow.tracking._tracking_service.utils._get_store('sqlite:///mlflow.db')
# Call any tracking store method
tracking_store.get_experiment(0)
# Instantiate the model registry store with a database URI
model_registry_store = mlflow.tracking._model_registry.utils._get_store('sqlite:///mlflow.db')
# Call any model registry store method
model_registry_store.get_latest_versions("model")
Why bother?
For that sweet performance my friend!
Initial benchmarks show us that critical API calls are faster and Go can handle more concurrent requests.
See benchmarks for more information.
Contribution
Try it out!
We advocate this project as a drop-in replacement for mlflow.
Please give it a try, and let us know if anything isn't working for you!
We look forward to your feedback!
Missing Endpoints
Hey there! We need your help! This project is not yet a full port of the current tracking server. Not every endpoint in the REST API is implemented in Go. Those missing are redirected to the existing Python implementation. Please consider helping us out and implement a missing endpoint.
Not yet a Go wizard? We've collected some resources to learn Go. Trust us, it will be an enriching and fun experience!
Community
There is a dedicated Slack channel called #mlflow-go on the official mlflow Slack.
Shoot us a message if you have any questions!
Metadata
Release files for mlflow-go-backend 0.2.2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Built distributions (wheels)
Total release size: 49.3 MB
Release files / mlflow_go_backend-0.2.2-py3-none-win_arm64.whl
| Download URL | mlflow_go_backend-0.2.2-py3-none-win_arm64.whl |
|---|---|
| Size | 7.1 MB |
| Tags | Python 3 Windows ARM64 |
|
SHA-256 checksum How to use checksums |
c7c4c469229864bf041dc446fcf1c878ebcf73712392762ef74181fd7bfa2078
|
|
BLAKE2b-256 checksum How to use checksums |
a0cf7214f1989916479c728176cf69f382d58c9a519b7ccc95ee6183ad1a203e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Release files / mlflow_go_backend-0.2.2-py3-none-win_amd64.whl
| Download URL | mlflow_go_backend-0.2.2-py3-none-win_amd64.whl |
|---|---|
| Size | 7.8 MB |
| Tags | Python 3 Windows x86-64 |
|
SHA-256 checksum How to use checksums |
cda07bf723554f13508e2042ffb3965d4fcbf8202f74b216245d1c43e61b9f58
|
|
BLAKE2b-256 checksum How to use checksums |
96b75ae0ab1606bde891037e3a780f0462d527c6378eacca74b2837a226cb67d
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Release files / mlflow_go_backend-0.2.2-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
| Download URL | mlflow_go_backend-0.2.2-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl |
|---|---|
| Size | 9.9 MB |
| Tags | Linux glibc 2.17+ x86-64 Python 3 |
|
SHA-256 checksum How to use checksums |
49b014275d6310b926212e5c00f2e4a8f52d1b6e174ead7ffa2b57bf929f15bf
|
|
BLAKE2b-256 checksum How to use checksums |
08b05022e418b733a11992dc86b2338c0d495c5f9a4d04f9ef8cb0d9d47ba134
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Release files / mlflow_go_backend-0.2.2-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
| Download URL | mlflow_go_backend-0.2.2-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl |
|---|---|
| Size | 9.6 MB |
| Tags | Linux glibc 2.17+ ARM64 Python 3 |
|
SHA-256 checksum How to use checksums |
d0131a8b2efcd41174042be1ef2b7149598352a1de7ea3c303810c064ad87b24
|
|
BLAKE2b-256 checksum How to use checksums |
f59d842040a9efe4c4afe5dd3a76a71866eb3e2d8b32f06d56b079d10390ce02
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Release files / mlflow_go_backend-0.2.2-py3-none-macosx_11_0_arm64.whl
| Download URL | mlflow_go_backend-0.2.2-py3-none-macosx_11_0_arm64.whl |
|---|---|
| Size | 7.1 MB |
| Tags | Python 3 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
b4a0dbc780e70532c1966865e5359f1e782e3a459fb2e14330bb3dc4aeb635e7
|
|
BLAKE2b-256 checksum How to use checksums |
29edf152c400c27be8d71599859612eb1f96ef41e1727adeea2391bf41c0e9fc
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Release files / mlflow_go_backend-0.2.2-py3-none-macosx_10_13_x86_64.whl
| Download URL | mlflow_go_backend-0.2.2-py3-none-macosx_10_13_x86_64.whl |
|---|---|
| Size | 7.6 MB |
| Tags | Python 3 macOS 10.13+ x86-64 |
|
SHA-256 checksum How to use checksums |
7536a686a43b3c6e87ffd85de53d71e5d5512bff890965afccaea94aeccfdc2e
|
|
BLAKE2b-256 checksum How to use checksums |
fc5b6b507b4d63d8bd3aa3038367eb286bd0874d3f9f970a6fb2df693d71e6c1
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Security for mlflow-go-backend 0.2.2
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.