metadata
dataset_info:
- config_name: clone-detection
features:
- name: label
dtype:
class_label:
names:
'0': '0'
'1': '1'
- name: source_code
dtype: string
splits:
- name: train
num_bytes: 2118123393
num_examples: 901028
- name: validation
num_bytes: 2009829980
num_examples: 830832
- name: test
num_bytes: 900449709
num_examples: 415416
download_size: 1986648636
dataset_size: 5028403082
- config_name: complexity-prediction
features:
- name: source_code
dtype: string
- name: label
dtype:
class_label:
names:
'0': constant
'1': cubic
'2': linear
'3': logn
'4': nlogn
'5': np
'6': quadratic
splits:
- name: train
num_bytes: 10643844.587115342
num_examples: 3800
- name: validation
num_bytes: 560202.3466902812
num_examples: 200
- name: test
num_bytes: 1448123.066194377
num_examples: 517
download_size: 4536582
dataset_size: 12652170
- config_name: defect-detection
features:
- name: label
dtype:
class_label:
names:
'0': '0'
'1': '1'
- name: source_code
dtype: string
splits:
- name: train
num_bytes: 44656083
num_examples: 21854
- name: validation
num_bytes: 5449113
num_examples: 2732
- name: test
num_bytes: 5513310
num_examples: 2732
download_size: 21139308
dataset_size: 55618506
configs:
- config_name: clone-detection
data_files:
- split: train
path: clone-detection/train-*
- split: validation
path: clone-detection/validation-*
- split: test
path: clone-detection/test-*
- config_name: complexity-prediction
data_files:
- split: train
path: complexity-prediction/train-*
- split: validation
path: complexity-prediction/validation-*
- split: test
path: complexity-prediction/test-*
- config_name: defect-detection
data_files:
- split: train
path: defect-detection/train-*
- split: validation
path: defect-detection/validation-*
- split: test
path: defect-detection/test-*
This dataset collates three class-balanced code classification datasets in the wild, where splits have been stratified by label. The sources are:
- CodeXGLUE defect detection (sourced from: semeru/code-code-DefectDetection)
- BigCloneBench clone detection (sourced from: nchen909/bigclonebench-processed)
- CodeComplex code runtime complexity prediction (sourced from: codeparrot/codecomplex)