Datasets:
v2: add 1,750 files from newly accepted artifacts (4.58M tokens); v1 rows unchanged; add release column
2114d01 verified | license: cc-by-4.0 | |
| language: | |
| - en | |
| pretty_name: ALEA synthetic code files | |
| task_categories: | |
| - text-generation | |
| size_categories: | |
| - 10K<n<100K | |
| tags: | |
| - code | |
| - c | |
| - rust | |
| - typescript | |
| - python | |
| - synthetic | |
| - alea | |
| - alea-r4 | |
| configs: | |
| - config_name: default | |
| data_files: | |
| - split: train | |
| path: data/train-*.parquet | |
| # ALEA synthetic code files | |
| Raw files from synthetic code artifacts in C, Rust, TypeScript and Python: focused samples and textbook-style walkthroughs on topics from mathematics and data structures to file formats, quantitative finance and application foundations. One document per file, with a path header in the file's own comment syntax; files include source, tests, text fixtures and data, READMEs, walkthroughs and package manifests. | |
| **12,680 documents, 33,200,905 tokens** (`alea-v2b-sw32k`, the ALEA R4 tokenizer, counting each document's start and end tokens). | |
| ## Releases | |
| The `release` column says which release a row came from. | |
| - `v1` (2026-10-20): 10,930 documents, 28,621,267 tokens. These are the rows ALEA's R4 model trained on in its stable pretraining mixture (second data stage, one pass each), except 2 files omitted for internal-path hygiene (R4 trained on 10,932 documents, 28,629,787 tokens). They are unchanged from the first release; every row is byte-identical to the trained text. | |
| - `v2` (2026-10-26): 1,750 documents, 4,579,638 tokens. New files from artifacts completed since v1, and revised versions of v1 files. 151 file paths therefore appear twice, once per release (`id` is the path; `text_sha256` tells the versions apart). v2 files come only from artifacts that passed the campaign's validation and independent review; a revised file that is a near-duplicate of its v1 version is not repeated. These rows are prepared for a later R4 training stage. | |
| `text_sha256` is the SHA-256 of `text`. Held-out evaluation items are not part of this dataset. | |
| ## Composition | |
| | task | documents | tokens | | |
| |---|---:|---:| | |
| | `source` | 3,725 | 10,970,666 | | |
| | `test` | 2,482 | 11,204,317 | | |
| | `readme` | 2,297 | 2,256,781 | | |
| | `manifest` | 1,428 | 140,668 | | |
| | `walkthrough` | 1,125 | 4,692,240 | | |
| | `data` | 1,012 | 2,861,833 | | |
| | `prose` | 611 | 1,074,400 | | |
| | language | documents | tokens | | |
| |---|---:|---:| | |
| | `rust` | 3,950 | 9,055,813 | | |
| | `typescript` | 3,238 | 8,065,402 | | |
| | `c` | 2,859 | 9,354,107 | | |
| | `python` | 2,633 | 6,725,583 | | |
| | generator | documents | tokens | | |
| |---|---:|---:| | |
| | `codex` | 8,627 | 21,170,055 | | |
| | `unknown` | 2,466 | 7,263,907 | | |
| | `claude-sonnet` | 1,185 | 3,528,834 | | |
| | `codex-gpt-6.1-sol` | 392 | 1,216,173 | | |
| | `claude-sonnet+codex-draft` | 10 | 21,936 | | |
| New in v2: | |
| | task | documents | tokens | | |
| |---|---:|---:| | |
| | `source` | 731 | 1,536,232 | | |
| | `test` | 358 | 1,324,743 | | |
| | `prose` | 203 | 408,568 | | |
| | `manifest` | 167 | 22,399 | | |
| | `data` | 144 | 460,639 | | |
| | `walkthrough` | 114 | 758,161 | | |
| | `readme` | 33 | 68,896 | | |
| | language | documents | tokens | | |
| |---|---:|---:| | |
| | `rust` | 726 | 1,688,571 | | |
| | `c` | 388 | 1,228,808 | | |
| | `python` | 326 | 907,366 | | |
| | `typescript` | 310 | 754,893 | | |
| ## How it was made | |
| The artifacts were written by coding agents in an ALEA code-synthesis campaign; each artifact is checked with its language's native toolchain and reviewed before it is accepted. The `generator` column gives the generator recorded in each artifact's provenance file: `codex` (OpenAI Codex), `codex-gpt-6.1-sol` (OpenAI Codex with GPT-6.1-sol), `claude-sonnet` (Anthropic Claude Sonnet), `claude-sonnet+codex-draft` (a Claude Sonnet artifact from a Codex draft), and `unknown` where the provenance file does not record a model. | |
| ## Sources | |
| - Synthetic code artifacts (no third-party repositories) | |
| ## Verification | |
| - File-level drops (detection only, a file is dropped whole and never edited): infrastructure strings and absolute host paths, secrets, binaries, captured logs and stdout, provenance, review and history records, exact duplicates, walkthrough code near-duplicating the same topic's sample, and Markdown dominated by process text. v2 also drops generated data tables (files over 256 KB and large generated include tables) and files already released in v1 verbatim or as near-duplicates. | |
| - A blind review judged sampled files of each release against the bar for a curated natural-code corpus. | |
| - Decontamination: v1 rows that shared a word 13-gram with a held-out evaluation item of a sibling family were removed; v2 rows were screened the same way against all of ALEA's R4 held-out and evaluation items, and none matched. | |
| - Every file in this release was scanned for API keys, access tokens, private keys and host paths; none were found. | |
| ## Columns | |
| | column | content | | |
| |---|---| | |
| | `id` | the file's path (`<language>/<topic>/<artifact kind>/<file>`); revised files appear once per release | | |
| | `text` | the document exactly as prepared for training: a path header line, then the file verbatim | | |
| | `task` | file kind: `source`, `test`, `walkthrough`, `readme`, `prose` (other Markdown), `data`, `manifest` | | |
| | `phase` | `stable` (R4's stable pretraining mixture) | | |
| | `generator` | the generator recorded for the file's artifact | | |
| | `group_id` | topic and language group; files of one topic in one language share it | | |
| | `prompt_chars` | 0 (plain documents) | | |
| | `n_tokens` | tokens under `alea-v2b-sw32k`, including the document start and end tokens | | |
| | `text_sha256` | SHA-256 of `text` | | |
| | `meta` | JSON: `lang`, `topic`, `artifact_kind` (`focused-sample` or `textbook-walkthrough`) | | |
| | `release` | `v1` (trained in R4's second data stage) or `v2` (added 2026-10-26) | | |
| ## Known limitations | |
| - Files keep natural-repository quirks: some docstrings and READMEs mention tiers, contracts, fixtures or review steps, and some identifiers are seed words. | |
| - Tests may reference fixtures that are not included (binaries, large generated tables). | |
| - Code was written by models; review it before reuse. | |
| ## License | |
| CC-BY-4.0 for this compilation and the generated text. OpenGloss is CC-BY-4.0. | |
| Published by the [ALEA Institute](https://aleainstitute.ai). | |