k8s-yaml

Kubernetes resource manifests in YAML across many resource kinds. One example per manifest file; the filename prefix encodes the resource kind (e.g. deployment_001750.yaml).

A corpus of 246,439 configuration files, packaged as a single parquet split for convenient loading. Part of a collection of schema/config corpora.

Format

One row per source file. Columns:

column	type	description
filename	string	original file name (e.g. deployment_001750.yaml)
text	string	full YAML content of the file

Resource kinds (top 10 by count)

kind	count
deployment	59,090
service	37,741
pod	27,117
configmap	25,283
clusterrole	12,150
secret	8,479
namespace	8,273
persistentvolumeclaim	8,161
serviceaccount	7,823
job	6,673

...and ~30 more kinds (daemonset, statefulset, ingress, crd, role, etc.).

Usage

from datasets import load_dataset

ds = load_dataset("anonymous-tct-authors/k8s-yaml", split="train")
print(ds[0]["filename"])
print(ds[0]["text"])

Reconstruct the original per-file tree:

import os, pyarrow.parquet as pq
t = pq.read_table("data/train-00000-of-00001.parquet")
os.makedirs("out", exist_ok=True)
for name, text in zip(t["filename"].to_pylist(), t["text"].to_pylist()):
    open(os.path.join("out", name), "w").write(text)