Upload tokenizer_config.json with huggingface_hub
Browse files- tokenizer_config.json +82 -0
tokenizer_config.json
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"backend": "tokenizers",
|
| 3 |
+
"bos_token": "<s>",
|
| 4 |
+
"clean_up_tokenization_spaces": true,
|
| 5 |
+
"eos_token": "</s>",
|
| 6 |
+
"extra_special_tokens": [
|
| 7 |
+
"<user>",
|
| 8 |
+
"<assistant>",
|
| 9 |
+
"<system>",
|
| 10 |
+
"<sudan>",
|
| 11 |
+
"<url>",
|
| 12 |
+
"<mention>",
|
| 13 |
+
"<hashtag>",
|
| 14 |
+
"<elongation>",
|
| 15 |
+
"<emoji_purple_heart>",
|
| 16 |
+
"<emoji_sleep>",
|
| 17 |
+
"<emoji_money>",
|
| 18 |
+
"<emoji_confetti>",
|
| 19 |
+
"<emoji_trophy>",
|
| 20 |
+
"<emoji_heart>",
|
| 21 |
+
"<emoji_broken_heart>",
|
| 22 |
+
"<emoji_angry>",
|
| 23 |
+
"<emoji_soccer>",
|
| 24 |
+
"<emoji_camera>",
|
| 25 |
+
"<emoji_arrow_right>",
|
| 26 |
+
"<emoji_hearts>",
|
| 27 |
+
"<emoji_tea>",
|
| 28 |
+
"<emoji_green_circle>",
|
| 29 |
+
"<emoji_ok>",
|
| 30 |
+
"<emoji_sparkles>",
|
| 31 |
+
"<emoji_party>",
|
| 32 |
+
"<emoji_star>",
|
| 33 |
+
"<emoji_green_heart>",
|
| 34 |
+
"<emoji_notes>",
|
| 35 |
+
"<emoji_joy>",
|
| 36 |
+
"<emoji_sad>",
|
| 37 |
+
"<emoji_wow>",
|
| 38 |
+
"<emoji_rofl>",
|
| 39 |
+
"<emoji_sweat>",
|
| 40 |
+
"<emoji_check>",
|
| 41 |
+
"<emoji_100>",
|
| 42 |
+
"<emoji_love>",
|
| 43 |
+
"<emoji_mask>",
|
| 44 |
+
"<emoji_thumbsdown>",
|
| 45 |
+
"<emoji_bell>",
|
| 46 |
+
"<emoji_blue_heart>",
|
| 47 |
+
"<emoji_rage>",
|
| 48 |
+
"<emoji_grin>",
|
| 49 |
+
"<emoji_black_heart>",
|
| 50 |
+
"<emoji_sun>",
|
| 51 |
+
"<emoji_white_heart>",
|
| 52 |
+
"<emoji_cross>",
|
| 53 |
+
"<emoji_rainbow>",
|
| 54 |
+
"<emoji_red_circle>",
|
| 55 |
+
"<emoji_arrow_left>",
|
| 56 |
+
"<emoji_shock>",
|
| 57 |
+
"<emoji_music>",
|
| 58 |
+
"<emoji_phone>",
|
| 59 |
+
"<emoji_star2>",
|
| 60 |
+
"<emoji_neutral>",
|
| 61 |
+
"<emoji_coffee>",
|
| 62 |
+
"<emoji_fire>",
|
| 63 |
+
"<emoji_clapper>",
|
| 64 |
+
"<emoji_moon>",
|
| 65 |
+
"<emoji_clap>",
|
| 66 |
+
"<emoji_rose>",
|
| 67 |
+
"<emoji_gift_heart>",
|
| 68 |
+
"<emoji_mic>",
|
| 69 |
+
"<emoji_pray>",
|
| 70 |
+
"<emoji_yellow_heart>",
|
| 71 |
+
"<emoji_computer>",
|
| 72 |
+
"<emoji_think>",
|
| 73 |
+
"<emoji_cry>",
|
| 74 |
+
"<emoji_thumbsup>",
|
| 75 |
+
"<emoji_smile>"
|
| 76 |
+
],
|
| 77 |
+
"mask_token": "<mask>",
|
| 78 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 79 |
+
"pad_token": "<pad>",
|
| 80 |
+
"tokenizer_class": "TokenizersBackend",
|
| 81 |
+
"unk_token": "<unk>"
|
| 82 |
+
}
|