Update tokenizer_config.json
#6
by AyaGL - opened
- tokenizer_config.json +16 -0
tokenizer_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"tokenizer_class": "SteerlingTokenizer",
|
| 3 |
+
"auto_map": {
|
| 4 |
+
"AutoTokenizer": ["tokenization_steerling.SteerlingTokenizer", null]
|
| 5 |
+
},
|
| 6 |
+
"pad_token": "<|pad|>",
|
| 7 |
+
"bos_token": "<|bos|>",
|
| 8 |
+
"eos_token": "<|endoftext|>",
|
| 9 |
+
"additional_special_tokens": ["<|endofchunk|>", "<|mask|>"],
|
| 10 |
+
"encoding_name": "cl100k_base",
|
| 11 |
+
"pad_token_id": 100277,
|
| 12 |
+
"bos_token_id": 100278,
|
| 13 |
+
"eos_token_id": 100257,
|
| 14 |
+
"endofchunk_token_id": 100279,
|
| 15 |
+
"mask_token_id": 100280
|
| 16 |
+
}
|