Upload folder using huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
language:
|
| 4 |
+
- fw
|
| 5 |
+
tags:
|
| 6 |
+
- tokenizer
|
| 7 |
+
- bpe
|
| 8 |
+
- flexitok
|
| 9 |
+
- fineweb2
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Byte-Level BPE Tokenizer: fw_edu (32K)
|
| 13 |
+
|
| 14 |
+
A **Byte-Level BPE** tokenizer trained on **fw_edu** data from Fineweb-2-HQ.
|
| 15 |
+
|
| 16 |
+
## Training Details
|
| 17 |
+
|
| 18 |
+
| Parameter | Value |
|
| 19 |
+
|-----------|-------|
|
| 20 |
+
| Algorithm | Byte-Level BPE |
|
| 21 |
+
| Language | `fw_edu` |
|
| 22 |
+
| Target Vocab Size | 32,000 |
|
| 23 |
+
| Final Vocab Size | 0 |
|
| 24 |
+
| Pre-tokenizer | ByteLevel |
|
| 25 |
+
| Normalizer | NFC |
|
| 26 |
+
| Special Tokens | `<s>`, `</s>`, `<pad>`, `<unk>` |
|
| 27 |
+
| Training Shards | 2 |
|
| 28 |
+
| Data Source | `/scratch/gsa/data/flexitok//fw_edu/` |
|
| 29 |
+
|
| 30 |
+
## Usage
|
| 31 |
+
|
| 32 |
+
```python
|
| 33 |
+
from transformers import AutoTokenizer
|
| 34 |
+
|
| 35 |
+
tokenizer = AutoTokenizer.from_pretrained("<repo_id>")
|
| 36 |
+
tokens = tokenizer.encode("Hello, world!")
|
| 37 |
+
```
|
| 38 |
+
|
| 39 |
+
## Files
|
| 40 |
+
|
| 41 |
+
- `tokenizer.json` — Full HuggingFace tokenizer
|
| 42 |
+
- `vocab.json` — Vocabulary mapping
|
| 43 |
+
- `merges.txt` — BPE merge rules
|