Conference
Optimizing code generation with mutational prompts and automated testing in LLMs.
| Τίτλος: | Optimizing code generation with mutational prompts and automated testing in LLMs. |
|---|---|
| Συγγραφείς: | Jayaram1 (AUTHOR) jayaram_258m@yahoo.com, Bhutkar, Yogesh1 (AUTHOR) yogeshbhutkar3@gmail.com, Chaithanya, Dasari1 (AUTHOR) chaithanyareddybi@gmail.com, Narsamma1 (AUTHOR) narsammag443@gmail.com |
| Πηγή: | AIP Conference Proceedings. 2026, Vol. 3341 Issue 1, p1-7. 7p. |
| Θεματικοί όροι: | *Code generators, *Computer software testing, *Gemini (Chatbot), *Language models |
| Εταιρία/Οντότητα: | OpenAI Inc. |
| Περίληψη: | Recent advancements in Large Language Models (LLMs) have opened doors to code generation using user-defined prompts. However, achieving optimal code quality remains a challenge. This paper proposes a novel web application built with Next.js that leverages the strengths of OpenAI and Gemini LLMs to address this limitation. Our approach utilizes prompt mutation and continuous testing to iteratively refine code generation. The application takes the initial prompt and user-specified number of code generations as input. It then creates a set of mutated prompts through strategic modifications. Each mutated prompt generates code, which is evaluated based on user-defined criteria and feedback. This feedback loop guides the mutation process, leading to prompts that consistently generate higher-quality code that aligns with user preferences. [ABSTRACT FROM AUTHOR] |
| Βάση Δεδομένων: | Academic Search Index |
καταχωρήστε σχόλιο πρώτοι!