ml-experiment-tracking

A lightweight, developer-friendly ML experiment tracking library that automatically records model runs, parameters, metrics, training time, and dataset information using local storage. It provides a CLI and web-based dashboard for organizing, analyzing, and managing experiments without requiring external tracking services or cloud infrastructure.

Framework-agnostic ML experiment tracking. Log metrics from any training loop to local JSON, inspect runs from the CLI.

AI companion for ML training. Track experiments, analyze patterns, get intelligent suggestions from Claude.

Reproducible deep learning experiments on the IQ-OTH/NCCD lung CT dataset examining data leakage, CT-slice partitioning, and performance overestimation in AI-based lung cancer classification.