Publish validated code corpus

README.md

@@ -0,0 +1,130 @@
+---
+pretty_name: Personal Code Corpus (raw-max)
+license: other
+language:
+- code
+annotations_creators:
+- no-annotation
+language_creators:
+- found
+source_datasets:
+- original
+size_categories:
+- 1K<n<10K
+task_categories:
+- text-generation
+tags:
+- code
+- source-code
+- code-completion
+- deduplicated
+- datasets
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data/train-*.parquet
+  - split: valid
+    path: data/valid-*.parquet
+dataset_info:
+  features:
+  - name: text
+    dtype: string
+  - name: repo
+    dtype: string
+  - name: path
+    dtype: string
+  - name: language
+    dtype: string
+  - name: sha
+    dtype: string
+  - name: chunk_index
+    dtype: int32
+  - name: n_tokens
+    dtype: int32
+  splits:
+  - name: train
+    num_examples: 7172
+    num_bytes: 19583125
+  - name: valid
+    num_examples: 1189
+    num_bytes: 3248759
+---
+
+# Personal code corpus: raw-max
+
+This is a repo-walked style/volume corpus, **not a curated accepted-state dataset**. It captures
+files present in local repository checkouts when the builder ran. It does not establish that every
+chunk is correct, reviewed, authored exclusively by one person, or suitable as a preferred answer.
+
+The corpus is intended for code-language modeling, code completion, and controlled personalization
+experiments. It must not be used to infer repository ownership, software quality, security, or the
+identity of an author.
+
+## Dataset summary
+
+- Rows: 8,361
+- Repositories with retained rows: 53
+- Completion field: `text`
+- Provenance fields: `repo`, `path`, `language`, `sha`, `chunk_index`, `n_tokens`
+- Split: deterministic repository-level train/valid assignment; a repository never crosses splits
+- Hash: SHA-256 of the emitted chunk text
+- Token counts: deterministic tokenizer-independent lexical estimates used for approximate chunking
+
+## Load the dataset
+
+```python
+from datasets import load_dataset
+
+dataset = load_dataset("JulianAT/code-corpus")
+```
+
+Each row has the following stable schema:
+
+- `text`: source-code chunk used as the completion field
+- `repo`, `path`: source provenance within the local corpus
+- `language`: language inferred from the file extension
+- `sha`: SHA-256 of the emitted `text`
+- `chunk_index`: zero-based chunk position within the source file
+- `n_tokens`: tokenizer-independent lexical token estimate
+
+`raw-max` is bounded by code that actually exists after filtering. A claim of several million rows
+is only credible when the source repositories contain enough retained code; `--max-rows` is a
+ceiling, not a promised row count. This implementation supports at most 1,000,000 rows per variant.
+
+## Deduplication
+
+Exact duplicate chunks are removed by SHA-256. Near duplicates are removed online with
+`datasketch.MinHashLSH`, 128 permutations, token
+5-grams, and a Jaccard threshold of 0.85.
+
+## Exclusions and limitations
+
+The walk excludes gitignored paths, `.git`, dependency/environment directories, build outputs,
+vendored/generated directories and filename patterns, lockfiles, minified files, symlinks, binaries,
+non-UTF-8 files, unsupported extensions, and files above 1,048,576 bytes.
+The quality variant may additionally exclude tests, fixtures, and low-history files.
+
+Repository licenses and obligations still apply to the source code. The dataset-level `other`
+license value does not replace per-repository licenses. Paths and code can contain sensitive data;
+inspect the artifacts before publishing. MinHash is approximate, lexical token counts are not model
+token counts, current checkouts omit deleted historical code, and repository-level splitting can
+produce an empty validation split when fewer than two repositories contribute rows.
+
+This corpus can contain insecure, incomplete, duplicated, outdated, or otherwise low-quality code.
+It is not a benchmark and has no correctness labels. The secret scanner uses a conservative set of
+high-confidence patterns and is not proof that the corpus is free of private or identifying data.
+
+## Reproducibility and audit
+
+`statistics.json` records row counts, split assignments, language distribution, filter decisions,
+deduplication totals, and build parameters. Splits are deterministic for the recorded seed. Source
+repository checkouts and the builder configuration are deliberately not included in the Hub repo.
+
+## Hub files
+
+- `data/*.parquet`: Hugging Face loader source
+- `dataset_infos.json`, `statistics.json`: schema, counts, filters, and dedup audit
+
+The local build also creates Arrow and JSONL representations for training. They are intentionally
+excluded from Hub publication because they duplicate the Parquet data.

data/train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb7a5af89fbe1147611287b2072b6b5dd3daba2580c184dcd286cfba25f4348f
+size 6548017

data/valid-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e742a19f733d2715d9de717d29f32d3acc358b478f15690fa6e19790fbd22b0f
+size 1177351

dataset_infos.json

@@ -0,0 +1,54 @@
+{
+  "default": {
+    "builder_name": "parquet",
+    "config_name": "default",
+    "dataset_name": "raw-max",
+    "dataset_size": 22831884,
+    "description": "Repository-walked personal code chunks for completion training.",
+    "download_size": 7725368,
+    "features": {
+      "chunk_index": {
+        "_type": "Value",
+        "dtype": "int32"
+      },
+      "language": {
+        "_type": "Value",
+        "dtype": "string"
+      },
+      "n_tokens": {
+        "_type": "Value",
+        "dtype": "int32"
+      },
+      "path": {
+        "_type": "Value",
+        "dtype": "string"
+      },
+      "repo": {
+        "_type": "Value",
+        "dtype": "string"
+      },
+      "sha": {
+        "_type": "Value",
+        "dtype": "string"
+      },
+      "text": {
+        "_type": "Value",
+        "dtype": "string"
+      }
+    },
+    "license": "other",
+    "size_in_bytes": 30557252,
+    "splits": {
+      "train": {
+        "name": "train",
+        "num_bytes": 19583125,
+        "num_examples": 7172
+      },
+      "valid": {
+        "name": "valid",
+        "num_bytes": 3248759,
+        "num_examples": 1189
+      }
+    }
+  }
+}

statistics.json

@@ -0,0 +1,189 @@
+{
+  "by_language": {
+    "CSS": 144,
+    "HTML": 53,
+    "Java": 13,
+    "JavaScript": 214,
+    "Python": 1530,
+    "SQL": 69,
+    "Shell": 29,
+    "Swift": 15,
+    "TOML": 9,
+    "TypeScript": 6170,
+    "XML": 3,
+    "YAML": 112
+  },
+  "by_repo": {
+    "1AHIF-BOT": 1,
+    "Illegale_Strassenrennen-Webprojekt": 8,
+    "Kriegsmarine_Webprojekt": 19,
+    "POS-23-24-Java": 13,
+    "Valorant-Webprojekt": 29,
+    "adapter-lab": 289,
+    "ai-video": 129,
+    "aucta": 262,
+    "brz-chatbot": 129,
+    "catchi": 235,
+    "clawguard": 448,
+    "codex-hackathon": 333,
+    "cursor-austria": 40,
+    "cursor-vienna-discord": 95,
+    "datagvat-mcp": 565,
+    "daytona-generative-ui": 354,
+    "diplomarbeit-ideen": 267,
+    "diplomarbeit-os": 1016,
+    "hacknation": 553,
+    "hacknation-app": 61,
+    "hacknation-landing": 203,
+    "hacknation-venture": 15,
+    "hacknation-web": 35,
+    "hands-on-ai": 70,
+    "htl-donaustadt-chatbot": 40,
+    "html-starter": 2,
+    "ifg-agentic": 250,
+    "interior-3d": 32,
+    "interiorly": 377,
+    "interiorly-dashboard": 64,
+    "interiorly-pitch": 96,
+    "interiorly-preview": 40,
+    "interiorly-promo-fake": 259,
+    "interiorly_changelog": 17,
+    "interiorly_landingpage": 55,
+    "jku-assignments": 83,
+    "jku-current-topics-report": 24,
+    "jku-encyclopedia": 335,
+    "mlpc-2026-task3": 65,
+    "mlpc-exam": 59,
+    "mlpc_task_4": 45,
+    "mystahd": 65,
+    "next-todo": 13,
+    "openai-build-week": 68,
+    "portfolio": 131,
+    "qdrant-hackathon": 172,
+    "start-hackathon": 64,
+    "synth-ui": 163,
+    "synth-ui-training": 15,
+    "tdot_chatbot": 139,
+    "untis-lehrer-extension": 19,
+    "young_ai_leaders_linz": 65,
+    "zero_one_hack_01": 435
+  },
+  "cap_reached": false,
+  "chunks_considered": 9443,
+  "dedup_report": {
+    "exact_duplicates_dropped": 533,
+    "near_duplicates_dropped": 549,
+    "rows_kept": 8361
+  },
+  "files_considered": 5501,
+  "files_selected": 5455,
+  "filter_report": {
+    "empty": 45,
+    "excluded_directory": 63,
+    "extension": 4626,
+    "gitignored": 5,
+    "lockfile": 55,
+    "secret_openai_key": 1,
+    "sensitive_path": 31,
+    "symlink": 2,
+    "symlink_directory": 2,
+    "too_large": 121
+  },
+  "generated_at": "2026-07-16T15:37:20+00:00",
+  "max_rows": 1000000,
+  "parameters": {
+    "chunk_overlap_tokens": 64,
+    "chunk_tokens": 896,
+    "max_file_bytes": 1048576,
+    "max_rows": 1000000,
+    "min_validation_repos": 3,
+    "minhash_permutations": 128,
+    "near_dedup_threshold": 0.85,
+    "seed": 42,
+    "shingle_tokens": 5,
+    "validation_ratio": 0.1
+  },
+  "repo_split": {
+    "1AHIF-BOT": "train",
+    "Illegale_Strassenrennen-Webprojekt": "train",
+    "Kriegsmarine_Webprojekt": "train",
+    "POS-23-24-Java": "train",
+    "Valorant-Webprojekt": "train",
+    "adapter-lab": "train",
+    "ai-video": "train",
+    "aucta": "train",
+    "brz-chatbot": "train",
+    "catchi": "train",
+    "clawguard": "train",
+    "codex-hackathon": "train",
+    "cursor-austria": "train",
+    "cursor-vienna-discord": "train",
+    "datagvat-mcp": "valid",
+    "daytona-generative-ui": "train",
+    "diplomarbeit-ideen": "train",
+    "diplomarbeit-os": "train",
+    "hacknation": "train",
+    "hacknation-app": "valid",
+    "hacknation-landing": "train",
+    "hacknation-venture": "train",
+    "hacknation-web": "train",
+    "hands-on-ai": "train",
+    "htl-donaustadt-chatbot": "train",
+    "html-starter": "train",
+    "ifg-agentic": "train",
+    "interior-3d": "valid",
+    "interiorly": "train",
+    "interiorly-dashboard": "train",
+    "interiorly-pitch": "valid",
+    "interiorly-preview": "train",
+    "interiorly-promo-fake": "train",
+    "interiorly_changelog": "train",
+    "interiorly_landingpage": "train",
+    "jku-assignments": "train",
+    "jku-current-topics-report": "train",
+    "jku-encyclopedia": "train",
+    "mlpc-2026-task3": "train",
+    "mlpc-exam": "train",
+    "mlpc_task_4": "train",
+    "mystahd": "train",
+    "next-todo": "train",
+    "openai-build-week": "train",
+    "portfolio": "train",
+    "qdrant-hackathon": "train",
+    "start-hackathon": "train",
+    "synth-ui": "train",
+    "synth-ui-training": "train",
+    "tdot_chatbot": "train",
+    "untis-lehrer-extension": "train",
+    "young_ai_leaders_linz": "train",
+    "zero_one_hack_01": "valid"
+  },
+  "repositories_discovered": 60,
+  "repositories_with_rows": 53,
+  "rows": {
+    "total": 8361,
+    "train": 7172,
+    "valid": 1189
+  },
+  "self_test": {
+    "loader": "datasets.load_dataset(\"JulianAT/code-corpus\")",
+    "splits": {
+      "train": 7172,
+      "valid": 1189
+    },
+    "status": "passed"
+  },
+  "sources": {
+    "config": "local config.yaml (not published)",
+    "repository_root": "local repository checkouts (not published)"
+  },
+  "tokens": {
+    "histogram": {
+      "0-255": 1891,
+      "256-511": 1495,
+      "512-1023": 4975
+    },
+    "total": 4968790
+  },
+  "variant": "raw-max"
+}