Publish TR-HASH 32K tokenizer
Browse files- README.md +46 -0
- special_tokens_map.json +6 -0
- tokenizer.json +0 -0
- tokenizer_config.json +46 -0
README.md
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: cc-by-nc-4.0
|
| 4 |
+
language:
|
| 5 |
+
- en
|
| 6 |
+
tags:
|
| 7 |
+
- tokenizer
|
| 8 |
+
- bpe
|
| 9 |
+
- byte-level
|
| 10 |
+
- tr-hash
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# TR-HASH Tokenizer 32K
|
| 14 |
+
|
| 15 |
+
The 32,000-token ByteLevel BPE tokenizer used by the TR-HASH language-model
|
| 16 |
+
research line and the 200B-token pretraining mixture.
|
| 17 |
+
|
| 18 |
+
## Properties
|
| 19 |
+
|
| 20 |
+
- Vocabulary size: 32,000
|
| 21 |
+
- Model: ByteLevel BPE
|
| 22 |
+
- Normalization: Unicode NFC
|
| 23 |
+
- EOS: `</s>` (ID 0)
|
| 24 |
+
- PAD: `<pad>` (ID 1)
|
| 25 |
+
- BOS: `<s>` (ID 2)
|
| 26 |
+
- UNK: `<unk>` (ID 3)
|
| 27 |
+
- Declared model context: 2,048 tokens
|
| 28 |
+
|
| 29 |
+
## Usage
|
| 30 |
+
|
| 31 |
+
```python
|
| 32 |
+
from transformers import AutoTokenizer
|
| 33 |
+
|
| 34 |
+
tokenizer = AutoTokenizer.from_pretrained(
|
| 35 |
+
"AETHORIA-AI/TR-HASH-Tokenizer-32K"
|
| 36 |
+
)
|
| 37 |
+
encoded = tokenizer("Token routing starts with tokenization.")
|
| 38 |
+
decoded = tokenizer.decode(encoded["input_ids"])
|
| 39 |
+
```
|
| 40 |
+
|
| 41 |
+
This repository contains the tokenizer only. It does not define a chat
|
| 42 |
+
template or a model architecture.
|
| 43 |
+
|
| 44 |
+
## Associated dataset
|
| 45 |
+
|
| 46 |
+
[AETHORIA-AI/data-32k-200b-tokens](https://huggingface.co/datasets/AETHORIA-AI/data-32k-200b-tokens)
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"eos_token": "</s>",
|
| 4 |
+
"pad_token": "<pad>",
|
| 5 |
+
"unk_token": "<unk>"
|
| 6 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": true,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"0": {
|
| 6 |
+
"content": "</s>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
},
|
| 13 |
+
"1": {
|
| 14 |
+
"content": "<pad>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false,
|
| 19 |
+
"special": true
|
| 20 |
+
},
|
| 21 |
+
"2": {
|
| 22 |
+
"content": "<s>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false,
|
| 27 |
+
"special": true
|
| 28 |
+
},
|
| 29 |
+
"3": {
|
| 30 |
+
"content": "<unk>",
|
| 31 |
+
"lstrip": false,
|
| 32 |
+
"normalized": false,
|
| 33 |
+
"rstrip": false,
|
| 34 |
+
"single_word": false,
|
| 35 |
+
"special": true
|
| 36 |
+
}
|
| 37 |
+
},
|
| 38 |
+
"bos_token": "<s>",
|
| 39 |
+
"clean_up_tokenization_spaces": false,
|
| 40 |
+
"eos_token": "</s>",
|
| 41 |
+
"extra_special_tokens": {},
|
| 42 |
+
"model_max_length": 2048,
|
| 43 |
+
"pad_token": "<pad>",
|
| 44 |
+
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 45 |
+
"unk_token": "<unk>"
|
| 46 |
+
}
|