MLXP

Spec sheet

What it does

MLXP is an open-source Python framework for managing machine learning experiments from launch and logging through querying results.Read 9 Oct 2026 · github.com

Purpose

MLXP is an open-source Python framework for managing experiments from launching and logging through querying results.Read 4 Oct 2026 · github.com

License

The project says MLXP is distributed under the MIT license.Read 9 Oct 2026 · github.com

Cluster jobs

It can submit jobs to a cluster using a job scheduler.Read 4 Oct 2026 · github.com

Code versioning

Its Git based version manager can generate a deployment version of code based on the latest Git commit.Read 9 Oct 2026 · github.com

Configuration

Logger, version manager, scheduler, and interactive mode settings can be adjusted in an automatically created mlxp.yaml file.Read 9 Oct 2026 · inria-thoth.github.io

Dependencies

Documented requirements include Hydra Core, OmegaConf, TinyDB, PyYAML, pandas, dill, and GitPython.Read 9 Oct 2026 · inria-thoth.github.io

Documentation

The repository links to a quick start guide, a tutorial, and detailed documentation on the project website.Read 4 Oct 2026 · github.com

Experiment launching

A decorator and Hydra hierarchical configurations let users launch several jobs with different option values.Read 9 Oct 2026 · github.com

Experiment records

It logs metrics, artifacts, checkpoints, metadata, and configuration options in a uniquely assigned job directory.Read 4 Oct 2026 · github.com

Installation

The project documents installation with pip from PyPI or by cloning the GitHub repository.Read 9 Oct 2026 · inria-thoth.github.io

Job launching

It can launch several jobs automatically using Hydra and hierarchical configs, with a decorator on the main task function.Read 4 Oct 2026 · github.com

Lazy loading

The MLXP DataFrame can defer loading metric files and artifacts until their fields are accessed.Read 9 Oct 2026 · inria-thoth.github.io

Logging

Runs can log metrics, artifacts, checkpoints, metadata, and configuration options in uniquely assigned directories.Read 9 Oct 2026 · github.com

Result analysis

The Reader supports querying, grouping, and aggregating results across runs.Read 9 Oct 2026 · inria-thoth.github.io

Results

It supports reading, querying, grouping, and aggregating outputs from multiple jobs.Read 4 Oct 2026 · github.com

Scheduler integrations

The documentation lists SLURM, OAR, TORQUE, SGE, MWM, and LSF as supported job schedulers.Read 9 Oct 2026 · inria-thoth.github.io

Where Specifiction places it

• Experiment Tracking#17 of 226.6
• ML Experiment Tracking Software#32 of 396.5