stefanocarrera/autophagycode_D_he_train-mercury_Qwen3-4B_strategy_trust_t1.25_g9_run2_metrics
收藏资源简介:
该数据集包含164个训练示例,专注于代码分析和评估任务。每个示例由多个特征组成,包括任务标识符(task_id)、入口点函数(entry_point)、代码可执行性(is_executable)和正确性(is_correct)的布尔标志、测试通过和失败的数量(tests_passed、tests_failed)、测试运行时间(test_run_time_ms,当前为null值)以及错误类型(error_type)。此外,数据集还提供了广泛的代码复杂度度量,如Halstead词汇量、长度、体积、难度、努力度和时间(halstead_vocabulary等)、圈复杂度(cyclomatic_complexity)、可维护性指数(maintainability_index)、代码行数(loc和sloc)、注释百分比(comment_percentage)、词汇多样性(TTR)、Shannon熵和预测熵(shannon_entropy、mean_predictive_entropy、max_predictive_entropy)、定义函数数量(n_func_defined)以及入口点重复标志(entry_point_repeated)。数据以字典形式存储(token_dict),可能用于进一步分析。整体上,该数据集适用于代码质量评估、程序理解和机器学习任务,强调代码的静态和动态特性。
This dataset contains 164 training examples focused on code analysis and evaluation tasks. Each example comprises multiple features, including a task identifier (task_id), an entry point function (entry_point), boolean flags for code executability (is_executable) and correctness (is_correct), counts of tests passed and failed (tests_passed, tests_failed), test run time (test_run_time_ms, currently null), and error type (error_type). Additionally, the dataset provides extensive code complexity metrics, such as Halstead vocabulary, length, volume, difficulty, effort, and time (halstead_vocabulary, etc.), cyclomatic complexity (cyclomatic_complexity), maintainability index (maintainability_index), lines of code (loc and sloc), comment percentage (comment_percentage), lexical diversity (TTR), Shannon entropy and predictive entropy (shannon_entropy, mean_predictive_entropy, max_predictive_entropy), number of defined functions (n_func_defined), and an entry point repetition flag (entry_point_repeated). Data is stored in dictionary format (token_dict) for potential further analysis. Overall, this dataset is suitable for code quality assessment, program comprehension, and machine learning tasks, emphasizing both static and dynamic code characteristics.




