Deep learning based recommender systems

These notebooks build deep learning recommender systems on AI Runtime. They show how to create a two-tower recommendation model using PyTorch Lightning for distributed training.

Example	Interface	Description
Two-tower recommendation model	Notebook	Learn how to create a two-tower recommendation model using PyTorch Lightning for distributed training.
Two-tower recommendation with PyTorch	CLI	Train and evaluate a two-tower movie recommender on MovieLens 1M on one A10 GPU.
Search technical documents with Qwen3 embeddings	CLI	Create Qwen3 text embeddings for technical documents on one A10 GPU and return semantically similar documents for each query.
Deep learning recommendation model with TorchRec	CLI	Train a click-through-rate model on Criteo data across four A10 nodes with TorchRec sharding.
Learning to rank with XGBoost	CLI	Train an XGBoost LambdaMART model on one A10 GPU to rank search results by relevance.

Example	Interface	Description
Two-tower recommendation model	Notebook	Learn how to create a two-tower recommendation model using PyTorch Lightning for distributed training.
Two-tower recommendation with PyTorch	CLI	Train and evaluate a two-tower movie recommender on MovieLens 1M on one A10 GPU.
Search technical documents with Qwen3 embeddings	CLI	Create Qwen3 text embeddings for technical documents on one A10 GPU and return semantically similar documents for each query.
Deep learning recommendation model with TorchRec	CLI	Train a click-through-rate model on Criteo data across four A10 nodes with TorchRec sharding.
Learning to rank with XGBoost	CLI	Train an XGBoost LambdaMART model on one A10 GPU to rank search results by relevance.