deep-learning-mlflow 0.0.10

Local MLflow integration layer for dl-core.

deep-learning-mlflow

Local MLflow integration layer for deep-learning-core.

deep-learning-mlflow adds local MLflow tracking on top of deep-learning-core without Azure dependencies. It is the public MLflow variant behind deep-learning-core[mlflow].

Install

Install from PyPI through the core extra:

pip install "deep-learning-core[mlflow]"

Install the package directly:

pip install deep-learning-mlflow

Install in a uv project:

uv add "deep-learning-core[mlflow]" deep-learning-mlflow

Quick Start

uv init
uv add deep-learning-mlflow
uv run dl-init --root-dir . --with-mlflow
uv run dl-run --config configs/base.yaml
uv run dl-sweep experiments/lr_sweep.yaml

The scaffold points MLflow at a local ./mlruns directory by default. Tracker experiment naming defaults to the repository root name unless tracking.experiment_name overrides it. Run artifacts are uploaded from config.yaml, per-epoch epoch_<n>/ directories after checkpointing, and final/ at the end of training.

Concrete local tracking flow:

uv run dl-init --root-dir . --with-mlflow
uv run dl-run --config configs/base.yaml
uv run dl-analyze --sweep experiments/lr_sweep.yaml

What You Get

• the mlflow callback for local training runs
• dl-init --with-mlflow scaffold support
• local ./mlruns tracking defaults for generated experiment repositories
• automatic upload of epoch_<n>/, final/, and config.yaml artifacts

Azure-backed MLflow wiring remains part of dl-azure.

Companion Packages

• dl-core
• dl-azure
• dl-wandb

Docs

• Documentation Index
• GitHub Repository

License

MIT. See LICENSE.