Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,133 @@
+---
+license: mit
+task_categories:
+  - text-generation
+language:
+  - en
+  - code
+tags:
+  - devops
+  - docker
+  - ci-cd
+  - github-actions
+  - build-systems
+  - configuration
+size_categories:
+  - 100K<n<1M
+---
+
+# Build/CI Configuration Corpus
+
+A curated dataset of build, CI/CD, and project configuration files from top GitHub repositories. This fills an underserved niche in LLM training data — models are notoriously weak at DevOps configuration tasks.
+
+## Dataset Description
+
+This corpus contains real-world configuration files extracted from the most popular GitHub repositories (by star count). Each row contains the full file content along with metadata about the source repository and file classification.
+
+### Source
+
+Repositories are sourced from [ronantakizawa/github-top-projects](https://huggingface.co/datasets/ronantakizawa/github-top-projects), which tracks GitHub's trending repositories from 2013–2025.
+
+### Categories
+
+| Category | Description | Example Files |
+|----------|-------------|---------------|
+| `dockerfile` | Docker configuration | `Dockerfile`, `docker-compose.yml` |
+| `github_actions` | GitHub Actions workflows | `.github/workflows/*.yml` |
+| `ci` | Other CI/CD systems | `.travis.yml`, `.gitlab-ci.yml`, `Jenkinsfile` |
+| `makefile` | Make-based build systems | `Makefile`, `CMakeLists.txt` |
+| `build_jvm` | JVM build tools | `build.gradle`, `pom.xml`, `build.sbt` |
+| `config_js` | JavaScript/TypeScript config | `tsconfig.json`, `webpack.config.js`, `vite.config.ts` |
+| `config_python` | Python project config | `pyproject.toml`, `setup.py`, `tox.ini` |
+| `config_rust` | Rust config | `Cargo.toml`, `build.rs` |
+| `config_go` | Go config | `go.mod` |
+| `config_ruby` | Ruby config | `Gemfile`, `Rakefile` |
+| `linting` | Linters and formatters | `.eslintrc.json`, `.prettierrc`, `.editorconfig` |
+| `packaging` | Package managers | `package.json`, `.npmrc` |
+| `infra` | Infrastructure as code | `*.tf`, `flake.nix` |
+| `k8s` | Kubernetes manifests | Detected via `apiVersion`/`kind` fields |
+| `pre_commit` | Pre-commit and dependency bots | `.pre-commit-config.yaml`, `dependabot.yml` |
+| `lockfile` | Dependency lock files | `package-lock.json`, `yarn.lock`, `Cargo.lock` |
+
+### Schema
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `content` | string | Full file content |
+| `file_path` | string | Path within repository |
+| `file_name` | string | Filename only |
+| `category` | string | High-level category (see above) |
+| `config_type` | string | Specific config type (e.g., "docker-compose", "tsconfig") |
+| `repo_name` | string | Repository (owner/name) |
+| `repo_stars` | int64 | Star count |
+| `repo_language` | string | Primary language of repository |
+| `license` | string | SPDX license identifier |
+| `line_count` | int32 | Number of lines |
+| `file_size_bytes` | int32 | File size in bytes |
+| `has_comments` | bool | Whether file contains comments |
+| `quality_score` | float32 | Quality score (0.0–1.0), see below |
+
+### Quality Filtering
+
+The dataset undergoes three quality filtering stages:
+
+1. **Minimum size**: Files with fewer than 5 lines or 50 characters are removed (trivial configs like 2-line `.nvmrc` files add no training signal).
+
+2. **Near-deduplication**: MinHash LSH (128 permutations, Jaccard threshold 0.85) removes near-duplicate files. Within each duplicate cluster, the version from the highest-starred repository is kept. This eliminates hundreds of copies of common starter templates (e.g., default `tsconfig.json`, boilerplate `Dockerfile`).
+
+3. **Makefile scoping**: Makefiles are restricted to root-level and 1 directory deep, preventing large C/C++ repos from flooding the dataset with subdirectory Makefiles.
+
+### Quality Score
+
+Each file receives a quality score (0.0–1.0) based on four equally-weighted factors:
+
+- **Comment density** (0–0.25): Files with comments/annotations teach intent, not just syntax
+- **Content length** (0–0.25): Longer files are more substantive (log-scaled, capped at 500 lines)
+- **Repository quality** (0–0.25): Higher-starred repos signal better engineering practices (log-scaled)
+- **Non-trivial ratio** (0–0.25): Ratio of meaningful lines vs blank/bracket-only lines
+
+Use `quality_score` to filter for higher-quality examples during training:
+```python
+high_quality = ds["train"].filter(lambda x: x["quality_score"] >= 0.5)
+```
+
+### Splits
+
+- **train** (90%): For training
+- **test** (5%): For evaluation
+- **validation** (5%): For validation
+
+Splits are deterministic by repository (all files from a repo go to the same split).
+
+## Usage
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("ronantakizawa/codeconfig")
+
+# Filter by category
+dockerfiles = ds["train"].filter(lambda x: x["category"] == "dockerfile")
+github_actions = ds["train"].filter(lambda x: x["category"] == "github_actions")
+
+# Filter by specific config type
+tsconfigs = ds["train"].filter(lambda x: x["config_type"] == "tsconfig")
+```
+
+## Intended Uses
+
+- Fine-tuning LLMs for DevOps/infrastructure code generation
+- Training code completion models for configuration files
+- Benchmarking LLM performance on build/CI tasks
+- Studying configuration patterns across popular open-source projects
+
+## Limitations
+
+- Only includes repositories with permissive licenses (MIT, Apache-2.0, BSD, etc.)
+- Biased toward popular/trending repositories
+- K8s manifests detected heuristically (may have false positives/negatives)
+- Lock files may be large and low-signal
+
+## License
+
+MIT — the dataset only includes files from permissively-licensed repositories.