Delete tokenizer_config.json
Browse files- tokenizer_config.json +0 -57
tokenizer_config.json
DELETED
|
@@ -1,57 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"add_prefix_space": false,
|
| 3 |
-
"auto_map": {
|
| 4 |
-
"AutoTokenizer": [
|
| 5 |
-
null,
|
| 6 |
-
"tokenization_minicpmo_fast.MiniCPMOTokenizerFast"
|
| 7 |
-
]
|
| 8 |
-
},
|
| 9 |
-
"backend": "tokenizers",
|
| 10 |
-
"bos_token": "<|im_start|>",
|
| 11 |
-
"clean_up_tokenization_spaces": false,
|
| 12 |
-
"eos_token": "<|im_end|>",
|
| 13 |
-
"errors": "replace",
|
| 14 |
-
"extra_special_tokens": [
|
| 15 |
-
"<image>",
|
| 16 |
-
"</image>",
|
| 17 |
-
"<ref>",
|
| 18 |
-
"</ref>",
|
| 19 |
-
"<box>",
|
| 20 |
-
"</box>",
|
| 21 |
-
"<quad>",
|
| 22 |
-
"</quad>",
|
| 23 |
-
"<point>",
|
| 24 |
-
"</point>",
|
| 25 |
-
"<slice>",
|
| 26 |
-
"</slice>",
|
| 27 |
-
"<image_id>",
|
| 28 |
-
"</image_id>",
|
| 29 |
-
"<unit>",
|
| 30 |
-
"</unit>",
|
| 31 |
-
"<asr>",
|
| 32 |
-
"</asr>",
|
| 33 |
-
"<query>",
|
| 34 |
-
"</query>",
|
| 35 |
-
"<|audio_start|>",
|
| 36 |
-
"<|audio|>",
|
| 37 |
-
"<|audio_end|>",
|
| 38 |
-
"<|spk_bos|>",
|
| 39 |
-
"<|spk|>",
|
| 40 |
-
"<|spk_eos|>",
|
| 41 |
-
"<|tts_bos|>",
|
| 42 |
-
"<|tts_eos|>",
|
| 43 |
-
"<|listen|>",
|
| 44 |
-
"<|speak|>",
|
| 45 |
-
"<|interrupt|>",
|
| 46 |
-
"<|vad_start|>",
|
| 47 |
-
"<|vad_end|>",
|
| 48 |
-
"<reserved_43>",
|
| 49 |
-
"<reserved_53>"
|
| 50 |
-
],
|
| 51 |
-
"is_local": false,
|
| 52 |
-
"model_max_length": 131072,
|
| 53 |
-
"pad_token": "<|endoftext|>",
|
| 54 |
-
"split_special_tokens": false,
|
| 55 |
-
"tokenizer_class": "MiniCPMOTokenizer",
|
| 56 |
-
"unk_token": "<unk>"
|
| 57 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|