Contributing

Guides for contributing to Torch-TensorRT — writing converters, lowering passes, understanding the system architecture, and managing resources.

Architecture

• System Overview
• Compiler Phases
• Built-in Dynamo Lowering Passes
• Observer / Callback System
• Writing TorchScript Converters
• Useful Links for Torch-TensorRT Development

Feature Design

• Engine Caching
• Weight Refit
• CUDAGraphs
• Dynamic Shape Support
• Complex Number Support
• Dynamic Memory Allocation
• Autocast and Precision Management
• Export and Serialization