mlflow-cli

What problem does it solve?

Manually navigating remote MLflow tracking servers to inspect experiments, runs, metrics, and artifacts is time-consuming and inefficient, especially when comparing performance across multiple machine learning experiments.

Core Features & Use Cases

• Read-Only MLflow Inspection: Query experiments, runs, metrics, parameters, tags, and artifacts directly from the CLI without accessing the web UI or writing custom API code.
• Offline Run Comparison: Export full experiment run tables to CSV for sorting, filtering, and side-by-side performance analysis.
• Use Case: A data scientist can quickly identify the best-performing hyperparameter configuration for a computer vision model by comparing metric trends and artifact contents across 20+ training runs in minutes.

Quick Start

Use the mlflow-cli skill to retrieve the top 3 runs by validation accuracy from the remote MLflow experiment named 'resnet50-training' and summarize their key hyperparameters, final metrics, and artifact contents.