Datasetpublic

akaruineko/git-commits

Dataset: dataset.jsonl Auto-labeled commit dataset scraped from GitHub repositories. Each line is a JSON object representing one commit with extracted features and an inferred label. Features Field Type Description Stats text string Commit message first line, conventional prefix stripped — files_count int Number of files changed mean 4.3, median 1, max 300 additions int Lines added mean 88, median 6, max 187K deletions int Lines deleted mean 172… See the full description on the dataset page: https://huggingface.co/datasets/akaruineko/git-commits.

sourceHugging Facemitupdated 2mo agoView on Hugging Face

0likes73downloads

Dataset Card

Dataset: `dataset.jsonl`

Auto-labeled commit dataset scraped from GitHub repositories. Each line is a JSON object representing one commit with extracted features and an inferred label.

Features

Field	Type	Description	Stats
`text`	string	Commit message first line, conventional prefix stripped	—
`files_count`	int	Number of files changed	mean 4.3, median 1, max 300
`additions`	int	Lines added	mean 88, median 6, max 187K
`deletions`	int	Lines deleted	mean 172, median 3, max 2.3M
`changed_tests`	int	Files matching test/spec patterns	mean 0.4, median 0, max 286
`changed_docs`	int	Files matching doc/documentation patterns	mean 0.6, median 0, max 300
`changed_source`	int	Files with known source extensions	mean 2.3, median 1, max 300
`has_tests`	bool	Any file in test/spec directory	—
`has_docs`	bool	Any file in docs directory	—
`extensions`	[string]	Unique file extensions in the diff	592 unique

Top extensions: md, go, cc, js, h, json, ts, yml, rc, mm

Top directories: docs, pkg, shell, atom, spec, lib, test, script

Labeling heuristics

Labels are inferred from the commit message and file changes using a cascade of signals:

Conventional commit prefix (highest confidence)

fix(scope):     → bugfix
feat(scope):    → feature
chore(scope):   → chore
refactor(scope):→ refactor
docs(scope):    → docs
test(scope):    → test
perf(scope):    → perf
ci(scope):      → ci
build(scope):   → build
style(scope):   → style
revert(scope):  → revert

Gitmoji prefix

:bug:           → bugfix
:sparkles:      → feature
:books:         → docs
:hammer:        → refactor
...             → (28 emoji mapped)

Keyword start (first word)

fix/correct     → bugfix   (unless "lint"→chore, "typo"→docs, docs-only→docs)
doc/readme      → docs
add/new/support → feature
remove/cleanup  → refactor
update/bump     → chore
test/spec       → test
revert/rollback → revert

Free-text keyword (within the whole first line)

crash/corrupt/leak/deadlock       → bugfix
prevent/avoid/guard               → bugfix
cleanup/tidy/polish               → refactor
renam/move                        → refactor
refactor/restruct/redesign/rewrite→ refactor

File-only fallback

—All files in docs directories → docs
—All files in test/spec directories → test

Source repos

Data was scraped from the commits endpoint of these public repositories:

—electron/electron
—yargs/yargs
—istanbuljs/istanbuljs
—fastify/fastify
—jenkins-x/jx