stefanocarrera/autophagycode_D_he_train-mercury_Qwen3-8B_strategy_trust_t1.25_g6_run1_metrics
收藏资源简介:
该数据集是一个包含代码分析相关指标的数据集,适用于编程任务评估和代码质量研究。数据集包含164个训练样本,每个样本具有多个特征,如任务ID(task_id)、入口点(entry_point)、可执行性(is_executable)、正确性(is_correct)、测试通过和失败次数(tests_passed、tests_failed)、错误类型(error_type)、代码复杂度指标(如Halstead词汇量、长度、难度、努力值、时间、圈复杂度、可维护性指数)、代码行数(loc、sloc)、注释比例(comment_percentage)、词汇多样性(TTR)、熵值(shannon_entropy、mean_predictive_entropy、max_predictive_entropy)、定义函数数量(n_func_defined)以及入口点重复性(entry_point_repeated)。这些特征可用于分析代码执行结果、错误模式、代码结构和可维护性。
This dataset is a collection of code analysis metrics, suitable for programming task evaluation and code quality research. It contains 164 training examples, each with multiple features such as task ID (task_id), entry point (entry_point), executability (is_executable), correctness (is_correct), test pass and fail counts (tests_passed, tests_failed), error type (error_type), code complexity metrics (e.g., Halstead vocabulary, length, difficulty, effort, time, cyclomatic complexity, maintainability index), lines of code (loc, sloc), comment percentage (comment_percentage), lexical diversity (TTR), entropy values (shannon_entropy, mean_predictive_entropy, max_predictive_entropy), number of defined functions (n_func_defined), and entry point repetition (entry_point_repeated). These features can be used to analyze code execution outcomes, error patterns, code structure, and maintainability.




