ludwig

Why we included this project

Ludwig lets you describe a model in a YAML file instead of wiring up PyTorch code. You define inputs, outputs, and training settings in one config and run a single command, whether you are fine-tuning an LLM with LoRA or fitting a tabular classifier. That declarative setup is easy to version and reuse, which helps data scientists and ML engineers keep experiments reproducible and iterate quickly without hand-writing training loops. Ludwig also covers vision and timeseries forecasting, so it works as a general starting point rather than a single-task tool. If you want a low-code route to custom models while keeping the training logic visible in the config, it is worth a try.