Metraipolis › archive › dc_c493ccf71feb

LLM-Based Embeddings for Program Analysis and Optimization — arXiv:2608.07894

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.07894
License (as archived)
CC0-1.0
Archived
2026-10-19T09:19:38.432Z
SHA-256 of the archived text
76f3cab78fa9f3f9ea8512b532cee209b60b7c58ef79e90354f5676c7cee97d5
Stable URL of this page
https://metraipolis.com/documento/dc_c493ccf71feb

How to cite

LLM-Based Embeddings for Program Analysis and Optimization — arXiv:2608.07894 — arXiv (en), https://arxiv.org/abs/2608.07894 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c493ccf71feb (Metraipolis archive dc_c493ccf71feb, sha256 76f3cab78fa9f3f9ea8512b532cee209b60b7c58ef79e90354f5676c7cee97d5)
BibTeX
@misc{dcc493ccf71feb,
  title = {LLM-Based Embeddings for Program Analysis and Optimization — arXiv:2608.07894},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.07894},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c493ccf71feb; sha256 76f3cab78fa9f3f9ea8512b532cee209b60b7c58ef79e90354f5676c7cee97d5},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.07894. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recent advances have highlighted the potential of machine learning, particularly Large Language Models (LLMs), for analyzing and optimizing programs. We present the first application of program embeddings from LLMCompiler---an LLM massively pretrained on intermediate representation (IR) code---to representative program analysis and optimization tasks. We generate program embeddings directly from source and IR code using a simple approach: split programs into chunks, independently embed each chunk with pretrained LLMs, and then aggregate the chunk embeddings into a single program embedding. Our experiments show that combining source and IR code embeddings achieves an error rate of 1.54\% in algorithm classification, a 12\% improvement over the current state-of-the-art, and a competitive accuracy on heterogeneous device mapping. These findings suggest that training a performance-aware LLM for embedding IR code might yield state-of-the-art results in code optimization tasks.

Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-08-11 in cs.PL).
Categorie arXiv: Machine Learning (cs.LG) ; Programming Languages (cs.PL).
Pagina del lavoro: https://arxiv.org/abs/2608.07894 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-19T09:15:58.327Z dalla pagina https://arxiv.org/catchup/cs.PL/2026-08-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).