autotrain-advanced

AutoTrain Advanced was a no-code tool from Hugging Face that let people train machine learning models without writing training code. Instead of setting up complex training scripts, users could point it at a dataset and choose a task, and the tool would handle the rest. The README prominently warns that this project is no longer maintained, no new features will be added, and bugs will not be fixed. Hugging Face recommends other tools instead.

When it was active, the tool supported a range of tasks: fine-tuning large language models using several training methods, classifying text, tagging individual words in sentences, question answering, image classification, and scoring images. Each task had a corresponding config file format and, for many tasks, a runnable notebook example.

You could run it in two ways: through a visual interface hosted on Hugging Face Spaces or Google Colab, or locally on your own machine using a Python package installed via pip. The local version launches a small web app on a port of your choosing. There was also a command-line path where you wrote a YAML config file describing your model, dataset, and training settings, then ran a single command.

The tool was free to use. When running locally you paid only for your own hardware. When running on Hugging Face Spaces you paid for the cloud compute used during training.

Because the project is archived and no longer receiving fixes, anyone encountering it today should treat it as a historical reference rather than an active option for new projects.