Examples

Framework-specific Examples

• Convert existing PyTorch code to Ray AIR: Get started with Ray AIR from an existing PyTorch codebase

• Convert existing Tensorflow/Keras code to Ray AIR: Get started with Ray AIR from an existing Tensorflow/Keras codebase.

• Training a model with distributed LightGBM: Distributed training with LightGBM

• Training a model with distributed XGBoost: Distributed training with XGBoost

• Hyperparameter tuning with XGBoostTrainer: Distributed tuning with XGBoost

• Training a model with Sklearn: Integrating with Scikit-Learn (non-distributed)

Simple Machine Learning

• AutoML for time series forecasting with Ray AIR: Build an AutoML system for time-series forecasting with Ray AIR

• Batch training & tuning on Ray Tune: Perform batch tuning on NYC Taxi Dataset with Ray AIR

• Parallel demand forecasting at scale using Ray Tune: Perform batch forecasting on NYC Taxi Dataset with Prophet, ARIMA and Ray AIR

Text/NLP

• Fine-tune a 🤗 Transformers model: How to use Ray AIR to run Hugging Face Transformers fine-tuning on a text classification task.

• GPT-J-6B Fine-Tuning with Ray AIR and DeepSpeed: How to use Ray AIR to run Hugging Face Transformers with DeepSpeed for fine-tuning a large model.

• GPT-J-6B Batch Prediction with Ray AIR: How to use Ray AIR to do batch prediction with the Hugging Face Transformers GPT-J model.

• GPT-J-6B Serving with Ray AIR: How to use Ray AIR to do online serving with the Hugging Face Transformers GPT-J model.

• Fine-tuning DreamBooth with Ray AIR: How to fine-tune a DreamBooth text-to-image model with your own images.

Image/CV

• Training a Torch Image Classifier

• Fine-tuning a Torch object detection model

• Performing GPU Batch Prediction on Images with a PyTorch Model

• Stable Diffusion Batch Prediction with Ray AIR: How to use Ray AIR to do batch prediction with the Stable Diffusion text-to-image model.

Logging & Observability

• Logging results and uploading models to Comet ML: How to log results and upload models to Comet ML.

• Logging results and uploading models to Weights & Biases: How to log results and upload models to Weights and Biases.

RL (RLlib)

• Serving reinforcement learning policy models

• Online reinforcement learning with Ray AIR

• Offline reinforcement learning with Ray AIR

Advanced

• Incremental Learning with Ray AIR: Incrementally train and deploy a PyTorch CV model

• Integrate Ray AIR with Feast feature store: Integrate with Feast feature store in both train and inference