1714 lines
33 KiB
JSON
1714 lines
33 KiB
JSON
{
|
|
"version": "1.0",
|
|
"truncation": null,
|
|
"padding": null,
|
|
"added_tokens": [
|
|
{
|
|
"id": 0,
|
|
"content": "<s>",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": true
|
|
},
|
|
{
|
|
"id": 1,
|
|
"content": "<pad>",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": false,
|
|
"special": true
|
|
},
|
|
{
|
|
"id": 2,
|
|
"content": "</s>",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": true
|
|
},
|
|
{
|
|
"id": 3,
|
|
"content": "<unk>",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": true
|
|
},
|
|
{
|
|
"id": 4,
|
|
"content": "[CH1-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 5,
|
|
"content": "[=S@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 6,
|
|
"content": "[/F]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 7,
|
|
"content": "[/123I]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 8,
|
|
"content": "[C-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 9,
|
|
"content": "[C@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 10,
|
|
"content": "[/CH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 11,
|
|
"content": "[Cl]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 12,
|
|
"content": "[#Branch1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 13,
|
|
"content": "[/C@@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 14,
|
|
"content": "[/C@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 15,
|
|
"content": "[C@@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 16,
|
|
"content": "[/P@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 17,
|
|
"content": "[B@@-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 18,
|
|
"content": "[\\O]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 19,
|
|
"content": "[N]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 20,
|
|
"content": "[O]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 21,
|
|
"content": "[/S@@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 22,
|
|
"content": "[\\C@@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 23,
|
|
"content": "[#N]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 24,
|
|
"content": "[/NH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 25,
|
|
"content": "[=B]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 26,
|
|
"content": "[\\F]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 27,
|
|
"content": "[=NH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 28,
|
|
"content": "[CH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 29,
|
|
"content": "[C@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 30,
|
|
"content": "[=O]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 31,
|
|
"content": "[/C@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 32,
|
|
"content": "[=P@@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 33,
|
|
"content": "[=Branch1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 34,
|
|
"content": "[-\\Ring2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 35,
|
|
"content": "[=C]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 36,
|
|
"content": "[127I]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 37,
|
|
"content": "[\\Cl]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 38,
|
|
"content": "[/Si]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 39,
|
|
"content": "[125I]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 40,
|
|
"content": "[#S]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 41,
|
|
"content": "[CH1+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 42,
|
|
"content": "[=S@@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 43,
|
|
"content": "[\\C]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 44,
|
|
"content": "[=S@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 45,
|
|
"content": "[S@@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 46,
|
|
"content": "[\\C@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 47,
|
|
"content": "[/S+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 48,
|
|
"content": "[\\S]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 49,
|
|
"content": "[\\C@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 50,
|
|
"content": "[F+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 51,
|
|
"content": "[PH2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 52,
|
|
"content": "[=Ring2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 53,
|
|
"content": "[=P@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 54,
|
|
"content": "[BH1-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 55,
|
|
"content": "[/NH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 56,
|
|
"content": "[B@@H1-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 57,
|
|
"content": "[S@@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 58,
|
|
"content": "[O-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 59,
|
|
"content": "[P@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 60,
|
|
"content": "[S]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 61,
|
|
"content": "[S+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 62,
|
|
"content": "[Br]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 63,
|
|
"content": "[I]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 64,
|
|
"content": "[\\B-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 65,
|
|
"content": "[C+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 66,
|
|
"content": "[F]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 67,
|
|
"content": "[/N]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 68,
|
|
"content": "[/C-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 69,
|
|
"content": "[/O]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 70,
|
|
"content": "[Ring1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 71,
|
|
"content": "[OH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 72,
|
|
"content": "[S@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 73,
|
|
"content": "[=S]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 74,
|
|
"content": "[\\P]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 75,
|
|
"content": "[BH3-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 76,
|
|
"content": "[18OH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 77,
|
|
"content": "[/S@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 78,
|
|
"content": "[=CH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 79,
|
|
"content": "[/P]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 80,
|
|
"content": "[P]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 81,
|
|
"content": "[P@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 82,
|
|
"content": "[P+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 83,
|
|
"content": "[/I]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 84,
|
|
"content": "[N-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 85,
|
|
"content": "[O+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 86,
|
|
"content": "[=P]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 87,
|
|
"content": "[#P]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 88,
|
|
"content": "[/CH1-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 89,
|
|
"content": "[\\123I]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 90,
|
|
"content": "[Ring2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 91,
|
|
"content": "[BH2-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 92,
|
|
"content": "[\\N+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 93,
|
|
"content": "[S@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 94,
|
|
"content": "[P@@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 95,
|
|
"content": "[\\S+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 96,
|
|
"content": "[=O+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 97,
|
|
"content": "[18F]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 98,
|
|
"content": "[=P+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 99,
|
|
"content": "[SnH2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 100,
|
|
"content": "[=Branch2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 101,
|
|
"content": "[SH2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 102,
|
|
"content": "[SH3]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 103,
|
|
"content": "[\\NH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 104,
|
|
"content": "[Branch1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 105,
|
|
"content": "[=N]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 106,
|
|
"content": "[=S@@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 107,
|
|
"content": "[11CH3]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 108,
|
|
"content": "[B]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 109,
|
|
"content": "[SnH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 110,
|
|
"content": "[CH2-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 111,
|
|
"content": "[N@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 112,
|
|
"content": "[/S]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 113,
|
|
"content": "[\\P@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 114,
|
|
"content": "[Sn]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 115,
|
|
"content": "[CH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 116,
|
|
"content": "[\\S@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 117,
|
|
"content": "[=SH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 118,
|
|
"content": "[/C]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 119,
|
|
"content": "[Si]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 120,
|
|
"content": "[#N+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 121,
|
|
"content": "[B@H1-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 122,
|
|
"content": "[B-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 123,
|
|
"content": "[SH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 124,
|
|
"content": "[123I]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 125,
|
|
"content": "[/B]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 126,
|
|
"content": "[/CH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 127,
|
|
"content": "[\\C-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 128,
|
|
"content": "[#Ring1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 129,
|
|
"content": "[-/Ring1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 130,
|
|
"content": "[NH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 131,
|
|
"content": "[N@@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 132,
|
|
"content": "[=PH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 133,
|
|
"content": "[=S+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 134,
|
|
"content": "[PH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 135,
|
|
"content": "[Sn+3]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 136,
|
|
"content": "[124I]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 137,
|
|
"content": "[3H]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 138,
|
|
"content": "[\\N]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 139,
|
|
"content": "[C]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 140,
|
|
"content": "[Branch2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 141,
|
|
"content": "[\\O-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 142,
|
|
"content": "[\\Br]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 143,
|
|
"content": "[17F]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 144,
|
|
"content": "[=N+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 145,
|
|
"content": "[\\C@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 146,
|
|
"content": "[CH2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 147,
|
|
"content": "[=P@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 148,
|
|
"content": "[NH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 149,
|
|
"content": "[/Br]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 150,
|
|
"content": "[N+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 151,
|
|
"content": "[\\SH1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 152,
|
|
"content": "[/N+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 153,
|
|
"content": "[\\CH1-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 154,
|
|
"content": "[S@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 155,
|
|
"content": "[/Cl]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 156,
|
|
"content": "[-/Ring2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 157,
|
|
"content": "[=P@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 158,
|
|
"content": "[P@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 159,
|
|
"content": "[\\B]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 160,
|
|
"content": "[/S@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 161,
|
|
"content": "[=N-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 162,
|
|
"content": "[N@@H1+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 163,
|
|
"content": "[=Ring1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 164,
|
|
"content": "[-\\Ring1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 165,
|
|
"content": "[#C]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 166,
|
|
"content": "[Br+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 167,
|
|
"content": "[=S@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 168,
|
|
"content": "[\\Si]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 169,
|
|
"content": "[\\S@@+1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 170,
|
|
"content": "[P@@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 171,
|
|
"content": "[/O-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 172,
|
|
"content": "[\\I]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 173,
|
|
"content": "[C@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 174,
|
|
"content": "[/OH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 175,
|
|
"content": "[P@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 176,
|
|
"content": "[\\CH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 177,
|
|
"content": "[/C@H1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 178,
|
|
"content": "[#Branch2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 179,
|
|
"content": "[B@-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 180,
|
|
"content": "[\\S@@]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 181,
|
|
"content": "[#C-1]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 182,
|
|
"content": "[SH0]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 183,
|
|
"content": "[Sn+2]",
|
|
"single_word": false,
|
|
"lstrip": false,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": false
|
|
},
|
|
{
|
|
"id": 184,
|
|
"content": "<mask>",
|
|
"single_word": false,
|
|
"lstrip": true,
|
|
"rstrip": false,
|
|
"normalized": true,
|
|
"special": true
|
|
}
|
|
],
|
|
"normalizer": null,
|
|
"pre_tokenizer": {
|
|
"type": "ByteLevel",
|
|
"add_prefix_space": false,
|
|
"trim_offsets": true,
|
|
"use_regex": true
|
|
},
|
|
"post_processor": {
|
|
"type": "RobertaProcessing",
|
|
"sep": [
|
|
"</s>",
|
|
2
|
|
],
|
|
"cls": [
|
|
"<s>",
|
|
0
|
|
],
|
|
"trim_offsets": true,
|
|
"add_prefix_space": false
|
|
},
|
|
"decoder": {
|
|
"type": "ByteLevel",
|
|
"add_prefix_space": true,
|
|
"trim_offsets": true,
|
|
"use_regex": true
|
|
},
|
|
"model": {
|
|
"type": "BPE",
|
|
"dropout": null,
|
|
"unk_token": null,
|
|
"continuing_subword_prefix": "",
|
|
"end_of_word_suffix": "",
|
|
"fuse_unk": false,
|
|
"byte_fallback": false,
|
|
"vocab": {
|
|
"<s>": 0,
|
|
"<pad>": 1,
|
|
"</s>": 2,
|
|
"<unk>": 3
|
|
},
|
|
"merges": []
|
|
}
|
|
} |