Instead of Rewriting Foreign Code for Machine Learning, Automatically Synthesize Fast Gradients
This solves the problem of integrating non-ML code into machine learning workflows for developers, though it is incremental as it builds on existing AD and compiler techniques.
The paper tackles the problem of applying machine learning to foreign code without rewriting by introducing Enzyme, an automatic differentiation compiler plugin for LLVM that synthesizes gradients from optimized intermediate representation, achieving a geometric mean speedup of 4.5x over pre-optimized AD and enabling integration with frameworks like PyTorch and TensorFlow.
Applying differentiable programming techniques and machine learning algorithms to foreign programs requires developers to either rewrite their code in a machine learning framework, or otherwise provide derivatives of the foreign code. This paper presents Enzyme, a high-performance automatic differentiation (AD) compiler plugin for the LLVM compiler framework capable of synthesizing gradients of statically analyzable programs expressed in the LLVM intermediate representation (IR). Enzyme synthesizes gradients for programs written in any language whose compiler targets LLVM IR including C, C++, Fortran, Julia, Rust, Swift, MLIR, etc., thereby providing native AD capabilities in these languages. Unlike traditional source-to-source and operator-overloading tools, Enzyme performs AD on optimized IR. On a machine-learning focused benchmark suite including Microsoft's ADBench, AD on optimized IR achieves a geometric mean speedup of 4.5x over AD on IR before optimization allowing Enzyme to achieve state-of-the-art performance. Packaging Enzyme for PyTorch and TensorFlow provides convenient access to gradients of foreign code with state-of-the art performance, enabling foreign code to be directly incorporated into existing machine learning workflows.