Annotatsiya
Mazkur maqolada generativ sun’iy intellekt texnologiyalaridan foydalanib dasturiy kodni avtomatik shakllantirish, uning funksional to‘g‘riligini tekshirish va hisoblash samaradorligini optimallashtirish usullari tadqiq etilgan. Dasturiy ta’minotni ishlab chiqishda katta til modellarining imkoniyatlari, tabiiy tilda berilgan masalalarni dasturga aylantirish mexanizmlari hamda kod sifatiga ta’sir qiluvchi omillar tahlil qilindi. Qiyosiy tahlil, algoritmik modellashtirish, funksional dekompozitsiya va ko‘p mezonli baholash metodlariga tayangan holda generatsiya, statik tahlil, testlash, optimallashtirish va qayta verifikatsiyalash jarayonlarini birlashtiruvchi bosqichli yondashuv taklif qilindi. Kodning funksional to‘g‘riligi, bajarilish vaqti, xotira sarfi, strukturaviy murakkabligi va xavfsizlik ko‘rsatkichlari asosida baholash mezonlari shakllantirildi. Model istiqbolli metodologik yechim sifatida asoslandi; uning amaliy samaradorligi keyingi eksperimental sinovlarda tasdiqlanishi lozim.
Foydalanilgan adabiyotlar
1. Chen M., Tworek J., Jun H. et al. Evaluating Large Language Models Trained on Code. arXiv:2107.03374. 2021. DOI: 10.48550/arXiv.2107.03374. https://arxiv.org/abs/2107.03374
2. Rozière B., Gehring J., Gloeckle F. et al. Code Llama: Open Foundation Models for Code. arXiv:2308.12950. 2023. https://arxiv.org/abs/2308.12950
3. Waghjale S., Veerendranath V., Wang Z., Fried D. ECCO: Can We Improve Model-Generated Code Efficiency Without Sacrificing Functional Correctness? Proceedings of EMNLP. 2024. P. 15362–15376. https://aclanthology.org/2024.emnlp-main.859/
4. Madaan A., Tandon N., Gupta P. et al. Self-Refine: Iterative Refinement with Self-Feedback. Advances in Neural Information Processing Systems. 2023. https://arxiv.org/abs/2303.17651
5. Jimenez C. E., Yang J., Wettig A. et al. SWE-bench: Can Language Models Resolve Real-world GitHub Issues? International Conference on Learning Representations. 2024. https://openreview.net/forum?id=VTF8yNQM66