five

Experimental data LLMProphet - AIST 2025 (co-located with ICST 2025)

收藏
NIAID Data Ecosystem2026-05-02 收录
下载链接:
https://zenodo.org/record/14889624
下载链接
链接失效反馈
官方服务:
资源简介:
Reproducing Results for LLMProphet - AIST 2025 Paper Information Title: From Implemented to Expected Behaviors: Leveraging Regression Oracles for Non-Regression Fault Detection Using LLMs Authors: Stefano Ruberto, Judith Perera, Gunel Jahangirova, Valerio Terragni Affiliations: Stefano Ruberto, JRC European Commission, Ispra, Italy (stefano.ruberto@ec.europa.eu) Judith Perera, University of Auckland, Auckland, New Zealand (jper120@aucklanduni.ac.nz) Gunel Jahangirova, King’s College London, London, United Kingdom (gunel.jahangirova@kcl.ac.uk) Valerio Terragni, University of Auckland, Auckland, New Zealand (v.terragni@auckland.ac.nz) Published in: at the 5th International Workshop on Artificial Intelligence in Software Testing (AIST 2025) co-located with 18th IEEE International Conference on Software Testing, Verification and Validation (ICST) 2025 Reproduction Instructions Requirements To reproduce the results, you will need access to OpenAI's GPT-4o model. Steps Access GPT-4o: Open the OpenAI Playground: GPT-4o Playground Set the following parameters: Temperature: 0.0 Top-p: 1.0 Run the prompts: Copy and paste the prompts from the respective folders into the "User" prompt field, one at a time. Make sure to clear the input before inserting a new prompt. Prompt Folders: Prompts_with_test_case_combinations_evaluating_19/05/22 Prompts_with_test_case_combinations_evaluating_15/05/22 File Naming Convention Inside the folders, files are named as follows: 15_10_wrong_0_correct.java This indicates: - 15/05/2022: The test case being evaluated. - 10_wrong: Number of test cases with faulty assertions. - 0_correct: Number of test cases with correct assertions. - .java: The files are stored in .java format to maintain formatting. Notes Ensure you follow the reproduction steps precisely to achieve the same results as presented in the paper. The files inside the folders contain the necessary test cases for evaluation. Contact For any questions regarding this material, please reach out to the authors via the provided email addresses.
创建时间:
2025-02-18
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作