Torchy: A Tracing JIT Compiler for PyTorch
Nuno P. Lopes
Abstract
Machine learning (ML) models keep getting larger and more complex. Whereas before models used to be represented by static data-flow graphs, they are now implemented via arbitrary Python code. Eager-mode frameworks, such as PyTorch, are now the standard for developing new ML models. The semantics of eager-mode frameworks is that operations are computed straight away. This greatly simplifies the development process, and it enables more dynamic ML models.
Topics & Concepts
Computer sciencePython (programming language)CompilerProgramming languageTracingJust-in-time compilationParallel computingRuntime systemParallel Computing and Optimization TechniquesComputational Physics and Python ApplicationsScientific Computing and Data Management