Browse the use cases below to find tutorials that match your model, framework, and optimization goal. Each use case walks through a real workflow you can follow directly.
Transformer training acceleration
| Use case | What you'll learn |
|---|---|
| Accelerate Transformer model training | How to use PAI-Rapidformer to speed up PyTorch Transformer model training |
Inference acceleration
The following use cases cover model inference optimization with PAI-Blade. Each entry notes the framework and model type so you can jump straight to the one that fits your setup.
| Use case | Framework | Model type | What you'll learn |
|---|---|---|---|
| Optimize a RetinaNet model in Detectron2 | PyTorch (Detectron2) | Object detection | How to use PAI-Blade to optimize a RetinaNet model in the Detectron2 framework |
| Optimize RetinaNet with custom C++ operators | PyTorch (TorchScript) | Object detection | How to use TorchScript custom C++ operators to build the post-processing network and then optimize the model with PAI-Blade |
| Optimize RetinaNet with TensorRT plug-ins | PyTorch (TensorRT) | Object detection | How to use PAI-Blade to optimize a detection model whose post-processing network is built with TensorRT plug-ins |
| Optimize a TensorFlow BERT model | TensorFlow | NLP | How to use PAI-Blade to optimize a TensorFlow BERT model |
| Optimize TensorFlow ResNet50 | TensorFlow | Image classification | How to use PAI-Blade to optimize a TensorFlow ResNet50 model |
| Optimize ResNet50 with dynamic input shapes | — | Image classification | How to use PAI-Blade to optimize a ResNet50 model with dynamic input shapes |