Upload distilled_mobilellm_v2_360M/tokenizer_config.json with huggingface_hub
Browse files
distilled_mobilellm_v2_360M/tokenizer_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"backend": "tokenizers",
|
| 3 |
+
"bos_token": "<|begin_of_text|>",
|
| 4 |
+
"clean_up_tokenization_spaces": true,
|
| 5 |
+
"eos_token": "<|eot_id|>",
|
| 6 |
+
"extra_special_tokens": [
|
| 7 |
+
"<|im_start|>",
|
| 8 |
+
"<|im_end|>",
|
| 9 |
+
"<think>",
|
| 10 |
+
"</think>"
|
| 11 |
+
],
|
| 12 |
+
"is_local": false,
|
| 13 |
+
"model_input_names": [
|
| 14 |
+
"input_ids",
|
| 15 |
+
"attention_mask"
|
| 16 |
+
],
|
| 17 |
+
"model_max_length": 32768,
|
| 18 |
+
"model_specific_special_tokens": {},
|
| 19 |
+
"pad_token": "<|eot_id|>",
|
| 20 |
+
"padding_side": "right",
|
| 21 |
+
"tokenizer_class": "TokenizersBackend"
|
| 22 |
+
}
|