Team Ai
Datasetpublic

OpenScientificCodeRegistry/Database

Open Scientific Code Registry (OSCR): the authors' scripts The code published by the authors of open-access neuroscience papers, as found and verified by Open Scientific Code Registry (OSCR). Each file is here exactly as it is at the source, at the verified commit, under the license of its repository. 421,275 unique files (3,782 MB of text) from 8,849 repositories, in 24 Parquet block(s). Only files whose repository's license allows redistribution, confirmed by the repository's… See the full description on the dataset page: https://huggingface.co/datasets/OpenScientificCodeRegistry/Database.

sourceHugging Faceotherupdated 2d agoView on Hugging Face
0likes5.6kdownloads
README.md76 linesDownload Raw Back to root
1---2license: other3license_name: per-file4license_link: LICENSE.md5pretty_name: Open Scientific Code Registry (OSCR) scripts6tags:7- neuroscience8- research-software9- code10- reproducibility11configs:12- config_name: scripts13  data_files: blocks/*.parquet14---15 16# Open Scientific Code Registry (OSCR): the authors' scripts17 18The code published by the authors of open-access neuroscience papers, as found and19verified by Open Scientific Code Registry (OSCR). Each file is here exactly as it is at the source, at the verified20commit, **under the license of its repository**.21 22- **421,275 unique files** (3,782 MB of text) from **8,849 repositories**,23  in 24 Parquet block(s).24- **Only files whose repository's license allows redistribution**, confirmed by the25  repository's own license file (or, for an archive without one, by its record). Files26  without such a license are never copied here: Open Scientific Code Registry (OSCR) links to them at the source.27- No text of any paper.28 29## Layout30 31- `blocks/NNNNN.parquet`: one row per unique file (SHA-256 of its text):32  `sha256`, `language`, `size`, `lines`, `content`. zstd page compression, 64 rows per33  row group: a single file can be read with HTTP range requests (for example with34  hyparquet in a browser). A published block never changes; new files go into new35  blocks.36- `manifests/<xx>/<repository>.json`: per repository, its URL, commit and license, and37  for each file its path, digest, and the (block, row) holding its text.38 39## Licenses40 41Each file keeps the license of its repository, given in the repository's manifest42(`license`, `license_confirmed_by`). Repositories by license:43 44| license | repositories |45|---|---|46| MIT | 3596 |47| CC-BY-4.0 | 2018 |48| GPL-3.0 | 1339 |49| Apache-2.0 | 676 |50| BSD-3-Clause | 407 |51| CC0-1.0 | 150 |52| GPL-2.0 | 115 |53| BSD-2-Clause | 104 |54| AGPL-3.0 | 89 |55| CC-BY-NC-4.0 | 75 |56| CC-BY-NC-SA-4.0 | 60 |57| other-open | 52 |58| LGPL-3.0 | 33 |59| CC-BY-NC-ND-4.0 | 30 |60| CC-BY-SA-4.0 | 29 |61| MPL-2.0 | 22 |62| LGPL-2.1 | 19 |63| Artistic-2.0 | 11 |64| Unlicense | 7 |65| CECILL-2.1 | 7 |66| EUPL-1.2 | 6 |67| CC-BY-ND-4.0 | 3 |68| BSL-1.0 | 1 |69 70Files under a non-commercial license (CC BY-NC…) may only be reused non-commercially.71 72## Takedown73 74An author who wants a file removed: open an issue at75https://github.com/NeuraCrypt/OSCR/issues.76