Researchers introduce the idea of compiler-generated feedback for Large Language Models (LLMs) to optimize code size of LLVM assembly. The proposed model outperforms the original model by 0.53% and improves code optimization.
31 Impressions
Researchers introduce the idea of compiler-generated feedback for Large Language Models (LLMs) to optimize code size of LLVM assembly. The proposed model outperforms the original model by 0.53% and improves code optimization.