AWhat changed
Traced annotated pure-and-statically-composed Python/NumPy subroutines into XLA computations while preserving dynamic Python orchestration and compatibility with Autograd transformations.
Scientific scope
Traced annotated pure-and-statically-composed Python/NumPy subroutines into XLA computations while preserving dynamic Python orchestration and compatibility with Autograd transformations. OriginalSysML2018three-person report, not current project-maintainer citation. Inherits Autograd/NumPy/XLA; not invention of automatic differentiation or every later JAX transformation. Shape/type changes can recompile; staged subroutines obey purity/static-composition assumptions. Warmed-up speedups exclude potentially slow compilation; no arbitrary Python acceleration guarantee.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. OriginalSysML2018three-person report, not current project-maintainer citation. Inherits Autograd/NumPy/XLA; not invention of automatic differentiation or every later JAX transformation. Shape/type changes can recompile; staged subroutines obey purity/static-composition assumptions. Warmed-up speedups exclude potentially slow compilation; no arbitrary Python acceleration guarantee.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
- Compiling machine learning programs via high-level tracing
- Google Research: high-level tracing / JAX (SysML2018)
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction