PiotrSty's picture
Add English Technical 1B corpus (#1)
3b833d3
|
Raw History Blame Contribute Delete
1.59 kB
---
license: other
license_name: mixed-per-source
language:
- en
pretty_name: English Technical 1B
tags:
- english
- technical
- slayer
- pretraining
---
# English Technical 1B
Consolidated English technical text for Slayer1B continued pretraining. It merges the three audited releases listed below without cross-release normalized-text or ID duplicates.
## Contents
| Split | Documents | cl100k proxy tokens | Slayer tokens |
|---|---:|---:|---:|
| train | 188,904 | 630,008,062 | 899,001,323 |
| valid | 10,370 | 34,721,895 | 49,886,904 |
| test | 10,665 | 35,374,977 | 51,226,632 |
**Total:** 209,939 documents and 1,000,114,859 native Slayer tokens.
## Merged releases
- `english_technical_100m` — 28,848 documents, 100,081,972 native Slayer tokens
- `english_technical_100m_v2` — 50,559 documents, 100,000,038 native Slayer tokens
- `english_technical_800m` — 130,532 documents, 800,032,849 native Slayer tokens
## Licensing
Mixed per-source open licences. This repository does not assert a blanket licence. Use the `license` field on each row and the matching notice under `licenses/` before redistribution.
## Data hygiene
The merge removed 0 additional duplicate rows (none). Source-level duplicate removal totals are retained in `manifest.json`. Direct identifiers were scrubbed in the source releases.
## Slayer pack
`slayer_v6/` contains the Slayer1B-1.06 tokenizer, uint16 token binaries and uint64 document indices. Existing source packs were concatenated with adjusted offsets; no retokenization was performed.