DuoLingo-AutoDiff : In-Database Automatic Differentiation with MLIR
Gutjahr, Kevin; Ruck, Clemens; Schüle, Maximilian E. (2025): DuoLingo-AutoDiff : In-Database Automatic Differentiation with MLIR, in: DEEM ’25: Proceedings of the Workshop on Data Management for End-to-End Machine Learning, New York: ACM, S. 1–8, doi: 10.1145/3735654.3735943.
Details
- Faculty/Chair
- Faculty/Professorship
- Author
- Gutjahr, Kevin
- Title of the compilation
- DEEM '25: Proceedings of the Workshop on Data Management for End-to-End Machine Learning
- Conference
- SIGMOD/PODS '25: International Conference on Management of Data, June 22 - 27, 2025, Berlin
- Publisher Information
- ACM : New York
- Year of publication
- 2025
- ISBN
- 979-8-4007-1924-0
- Language
- English
- DOI
- Abstract
Forward and reverse mode automatic differentiation evaluate the gradient of a model function efficiently by caching the results of partial derivatives. Just-in-time compilation improves the runtime of automatic differentiation by eliminating function calls and storing partial derivatives in virtual registers. This paper discusses the first open-source implementation of automatic differentiation with MLIR and LingoDB. The evaluation compares optimizations applied to forward and reverse modes. It showed that sub-expressions, that appear frequently within the calculation, will be reused after MLIR performs its optimization. Additionally, reverse mode outperforms forward mode due to less generated code.
- Keywords
- Automatic DiferentiationIn-Database Machine LearningQuery CompilationMLIR
- Type
- Conferenceobject
- URI
- Activation date
- 2025-11-26
- Project(s)