alexjercan/AoC

---
license: apache-2.0
dataset_info:
features:
- name: year
dtype: string
- name: day
dtype: string
- name: part
dtype: string
- name: pass
dtype: string
- name: fail
dtype: string
- name: test
dtype: string
- name: change
dtype: string
- name: i1
dtype: uint32
- name: i2
dtype: uint32
- name: j1
dtype: uint32
- name: j2
dtype: uint32
splits:
- name: train
num_bytes: 21469
num_examples: 15
download_size: 23847
dataset_size: 21469
---

# About the Dataset

This dataset is inspired by [HumanEval](https://github.com/openai/human-eval)

The source code used to generate the dataset can be found on [GitHub](https://github.com/alexjercan/bug-detection/tree/master/aoc-dataset)

A collection of submissions for the Advent of Code challenge.
This repository contains both passing and failing submissions.

This dataset is similar to [BugNet](https://huggingface.co/datasets/alexjercan/bugnet),
however it is meant to be used as an evaluation dataset.

The resulting dataset file will be a csv with the following columns:
- `year`: Used to identify the submission
- `day`: Used to identify the submission
- `part`: Used to identify the submission
- `fail`: The initial (buggy) source code formatted (`black`)
- `pass`: The modified (accepted) source code formatted (`black`)
- `change`: The change that was made (`replace`, `insert`, `delete`)
- `i1`: Start of the change in the buggy source (the line; starting with 1)
- `i2`: End of the change in the buggy source (not inclusive; for insert we have i1 == i2)
- `j1`: Start of the change in the accepted source (the line; starting with 1)
- `j2`: End of the change in the accepted source (not inclusive; for delete we have j1 == j2)
- `test`: The test case that can be used to evaluate the submission.

原始信息汇总

数据集概述

数据集信息

• 特征列表：
  • year：字符串类型，用于标识提交。
  • day：字符串类型，用于标识提交。
  • part：字符串类型，用于标识提交。
  • pass：字符串类型，格式化的修改后（接受）源代码。
  • fail：字符串类型，格式化的初始（错误）源代码。
  • test：字符串类型，可用于评估提交的测试用例。
  • change：字符串类型，所做的更改（replace、insert、delete）。
  • i1：无符号32位整数类型，错误源代码中更改的开始行（从1开始）。
  • i2：无符号32位整数类型，错误源代码中更改的结束行（不包括）。
  • j1：无符号32位整数类型，接受源代码中更改的开始行（从1开始）。
  • j2：无符号32位整数类型，接受源代码中更改的结束行（不包括）。

数据集划分

• 训练集：
  • 名称：train
  • 字节数：21469
  • 样本数：15

数据集大小

• 下载大小：23847字节
• 数据集大小：21469字节