README.md

metadata

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

The source code used to generate the dataset can be found on GitHub

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

This dataset is similar to 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.