Quick End-to-end Example

Get introduced to the Ray AI Libraries in this overview. These libraries are built on top of Ray Core and provide a unified framework for machine learning tasks with a focus on performance and scalability.

Plain Hugging Face Image Classification

While this notebook focuses on using Hugging Face without Ray, it sets the foundation for subsequent notebooks, where Ray Train, Ray Data, and Ray Serve will be introduced for scalable machine learning. Explore the basics of fine-tuning ViTs for image classification, building a strong starting point for more advanced tasks in the following notebooks.

Adding Ray Train for Distributed Fine-tuning

Leverage the power of Ray Train for distributed machine learning training in this notebook focused on ViT image classification. Explore the capabilities of Ray Train, which enables distributed training and efficient resource utilization.

Adding Ray Data for Distributed Data Processing

Optimize your data pipeline for ViT image classification using Ray Data in this notebook. The notebook covers topics like distributed data processing, creating a Ray Dataset, image preprocessing, setting up training logic, launching distributed fine-tuning, and performing batch inference with Ray Data.

Adding Ray Serve for Scalable Deployments

Bridge the gap between development and production by creating a fully deployed, scalable, and efficient serving solution. Deploy your fine-tuned ViT model efficiently with Ray Serve in this notebook.