A scratchpad for machine-learning experiments — notebooks, training scripts, and evaluation utilities used while iterating on small models and dataset pipelines.
This repository is intentionally lightweight: it collects reusable snippets (data loading, metrics, plotting) that graduate into their own projects once they stabilize. For polished, production-ready work see the other repositories on this profile.