Datasets:
Add dataset card
Browse files
README.md
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
task_categories:
|
| 4 |
+
- token-classification
|
| 5 |
+
language:
|
| 6 |
+
- zh
|
| 7 |
+
- ja
|
| 8 |
+
- en
|
| 9 |
+
tags:
|
| 10 |
+
- tokenizer
|
| 11 |
+
- bpe
|
| 12 |
+
- sentencepiece
|
| 13 |
+
- wordpiece
|
| 14 |
+
- chinese
|
| 15 |
+
- japanese
|
| 16 |
+
- cjk
|
| 17 |
+
pretty_name: CJK Tokenizer From Scratch (papers + code dump)
|
| 18 |
+
---
|
| 19 |
+
|
| 20 |
+
# CJK Tokenizer From Scratch
|
| 21 |
+
|
| 22 |
+
Offline research dump for **building Chinese / Japanese tokenizers from scratch**.
|
| 23 |
+
|
| 24 |
+
Download `tokenizer.zip` (≈464 MB), then:
|
| 25 |
+
|
| 26 |
+
```bash
|
| 27 |
+
unzip tokenizer.zip
|
| 28 |
+
```
|
| 29 |
+
|
| 30 |
+
Contents:
|
| 31 |
+
|
| 32 |
+
- `papers/` — 41 free arXiv / ACL PDFs (BPE, WordPiece, Unigram, SentencePiece, Chinese SubChar, Japanese MeCab/Sudachi bake-off, …)
|
| 33 |
+
- `github/` — 23 shallow-cloned repos (`minbpe`, `sentencepiece`, `tokenizers`, `jieba`, `mecab`, `Sudachi`, …)
|
| 34 |
+
- `docs/` — Hugging Face NLP course chapter 6 (train BPE / WordPiece / Unigram)
|
| 35 |
+
- `INDEX.md` — catalog and reading order
|
| 36 |
+
|
| 37 |
+
This is a snapshot of public papers and GitHub sources, not a trained tokenizer. Licenses follow the original papers and repos.
|