BROWSE
Detail
Conference Paper
Backward Graph Construction and Lowering in DL Compiler for Model Training on AI Accelerators
Cited 2 time in 
Share 
Share
- Authors
- Hyunjeong Kwon, Youngsu Kwon, Jinho Han
- Issue Date
- 2022-10
- Citation
- International SoC Design Conference (ISOCC) 2022, pp.91-92
- Publisher
- 반도체공학회
- Language
- English
- Type
- Conference Paper
- Abstract
- A deep learning (DL) compiler is required to acceler ate model inference and training on AI accelerators. In this work, we propose a novel approach to constructing a backward graph from a PyTorch model, and lowering it to machine codes. The backward graph is constructed using information from PyTorch's autograd engine. The newly proposed lexer and parser convert the generated graph into an abstract syntax tree (AST). The AST is converted to GIR, an intermediate representation within the MLIR framework. IR lowering is then applied to the GIR, producing an LLVM IR for the LLVM backend. Among operators, those that can be accelerated using a DL accelerator are processed by the accelerator. This is achieved through the LLVM IR call function, which calls the accelerator's backend. In the experiment, the proposed compiler estimated the training loss with an average error of 1.46% within 6.7 seconds while processing 100 epochs.
- KSP Keywords
- Average error, Intermediate representation, LLVM IR, Machine code, Model Inference, Novel approach, Using information, abstract syntax tree, deep learning(DL), graph construction