Sovereign Tokenizer Upgrade: H100 Tensor-Core Aligned 40,960 BPE (special_tokens_map.json)
Browse files- special_tokens_map.json +64 -0
special_tokens_map.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"eos_token": "</s>",
|
| 4 |
+
"unk_token": "<unk>",
|
| 5 |
+
"pad_token": "<pad>",
|
| 6 |
+
"additional_special_tokens": [
|
| 7 |
+
"<image>",
|
| 8 |
+
"</image>",
|
| 9 |
+
"<audio>",
|
| 10 |
+
"</audio>",
|
| 11 |
+
"<video>",
|
| 12 |
+
"</video>",
|
| 13 |
+
"<click>",
|
| 14 |
+
"</click>",
|
| 15 |
+
"<touch>",
|
| 16 |
+
"</touch>",
|
| 17 |
+
"<box3d>",
|
| 18 |
+
"</box3d>",
|
| 19 |
+
"<frame>",
|
| 20 |
+
"</frame>",
|
| 21 |
+
"<document>",
|
| 22 |
+
"</document>",
|
| 23 |
+
"<page>",
|
| 24 |
+
"</page>",
|
| 25 |
+
"<table>",
|
| 26 |
+
"</table>",
|
| 27 |
+
"<sheet>",
|
| 28 |
+
"</sheet>",
|
| 29 |
+
"<chart>",
|
| 30 |
+
"</chart>",
|
| 31 |
+
"<layout>",
|
| 32 |
+
"</layout>",
|
| 33 |
+
"<ocr>",
|
| 34 |
+
"</ocr>",
|
| 35 |
+
"<system>",
|
| 36 |
+
"</system>",
|
| 37 |
+
"<user>",
|
| 38 |
+
"</user>",
|
| 39 |
+
"<assistant>",
|
| 40 |
+
"</assistant>",
|
| 41 |
+
"<|sovereign_reserved_0|>",
|
| 42 |
+
"<|sovereign_reserved_1|>",
|
| 43 |
+
"<|sovereign_reserved_2|>",
|
| 44 |
+
"<|sovereign_reserved_3|>",
|
| 45 |
+
"<|sovereign_reserved_4|>",
|
| 46 |
+
"<|sovereign_reserved_5|>",
|
| 47 |
+
"<|sovereign_reserved_6|>",
|
| 48 |
+
"<|sovereign_reserved_7|>",
|
| 49 |
+
"<|sovereign_reserved_8|>",
|
| 50 |
+
"<|sovereign_reserved_9|>",
|
| 51 |
+
"<|sovereign_reserved_10|>",
|
| 52 |
+
"<|sovereign_reserved_11|>",
|
| 53 |
+
"<|sovereign_reserved_12|>",
|
| 54 |
+
"<|sovereign_reserved_13|>",
|
| 55 |
+
"<|sovereign_reserved_14|>",
|
| 56 |
+
"<|sovereign_reserved_15|>",
|
| 57 |
+
"<|sovereign_reserved_16|>",
|
| 58 |
+
"<|sovereign_reserved_17|>",
|
| 59 |
+
"<|sovereign_reserved_18|>",
|
| 60 |
+
"<|sovereign_reserved_19|>",
|
| 61 |
+
"<|sovereign_reserved_20|>",
|
| 62 |
+
"<|sovereign_reserved_21|>"
|
| 63 |
+
]
|
| 64 |
+
}
|