| { | |
| "version": "1.0", | |
| "truncation": null, | |
| "padding": null, | |
| "added_tokens": [ | |
| { | |
| "id": 0, | |
| "content": "<UNK>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 1, | |
| "content": "<PAD>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 2, | |
| "content": "<CLS>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 3, | |
| "content": "<SEP>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 4, | |
| "content": "<MASK>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 5, | |
| "content": "<EOS>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 6, | |
| "content": "<MOLECULAR_ENTITY>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 7, | |
| "content": "<GLOBAL_INTERACTION_ATTRIBUTES>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 8, | |
| "content": "<MOLECULAR_ENTITY_ANTIGEN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 9, | |
| "content": "<MOLECULAR_ENTITY_EPITOPE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 10, | |
| "content": "<MOLECULAR_ENTITY_ANTIBODY_HEAVY_CHAIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 11, | |
| "content": "<MOLECULAR_ENTITY_ANTIBODY_LIGHT_CHAIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 12, | |
| "content": "<MOLECULAR_ENTITY_TCR_ALPHA_CHAIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 13, | |
| "content": "<MOLECULAR_ENTITY_TCR_BETA_VDJ>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 14, | |
| "content": "<MOLECULAR_ENTITY_TCR_BETA_CDR3>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 15, | |
| "content": "<BINDING_AFFINITY_CLASS>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 16, | |
| "content": "<DECODER_START>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 17, | |
| "content": "<BINDING>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 18, | |
| "content": "<FILLIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 19, | |
| "content": "<REORDER>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 20, | |
| "content": "<TOAA>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 21, | |
| "content": "<ACTIVE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 22, | |
| "content": "<GENESEQ>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 23, | |
| "content": "<INCREASE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 24, | |
| "content": "<DECREASE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 25, | |
| "content": "<STRUCTURE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 26, | |
| "content": "<DISTANCE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 27, | |
| "content": "<SOLUBILITY>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 28, | |
| "content": "<TOXICITY>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 29, | |
| "content": "<AB>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 30, | |
| "content": "<ISACTIVE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 31, | |
| "content": "<ISSYNTHETIC>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 32, | |
| "content": "<PENETR>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 33, | |
| "content": "<ABSORPTION>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 34, | |
| "content": "<DISTRIBUTION>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 35, | |
| "content": "<METABOLISM>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 36, | |
| "content": "<EXCRETION>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 37, | |
| "content": "<FLUORESCENCE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 38, | |
| "content": "<STABILITY>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 39, | |
| "content": "<DISORDER>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 40, | |
| "content": "<DISEASE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 41, | |
| "content": "<BINARY>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 42, | |
| "content": "<REGRESSION>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 43, | |
| "content": "<ORGANISM>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 44, | |
| "content": "<0>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 45, | |
| "content": "<1>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 46, | |
| "content": "<2>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 47, | |
| "content": "<3>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 48, | |
| "content": "<4>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 49, | |
| "content": "<5>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 50, | |
| "content": "<6>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 51, | |
| "content": "<7>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 52, | |
| "content": "<8>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 53, | |
| "content": "<9>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 54, | |
| "content": "<.>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 55, | |
| "content": "<YES>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 56, | |
| "content": "<NO>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 57, | |
| "content": "<SENTINEL_ID_0>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 58, | |
| "content": "<SENTINEL_ID_1>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 59, | |
| "content": "<SENTINEL_ID_2>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 60, | |
| "content": "<SENTINEL_ID_3>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 61, | |
| "content": "<SENTINEL_ID_4>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 62, | |
| "content": "<SENTINEL_ID_5>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 63, | |
| "content": "<SENTINEL_ID_6>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 64, | |
| "content": "<SENTINEL_ID_7>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 65, | |
| "content": "<SENTINEL_ID_8>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 66, | |
| "content": "<SENTINEL_ID_9>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 67, | |
| "content": "<SENTINEL_ID_10>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 68, | |
| "content": "<SENTINEL_ID_11>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 69, | |
| "content": "<SENTINEL_ID_12>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 70, | |
| "content": "<SENTINEL_ID_13>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 71, | |
| "content": "<SENTINEL_ID_14>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 72, | |
| "content": "<SENTINEL_ID_15>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 73, | |
| "content": "<SENTINEL_ID_16>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 74, | |
| "content": "<SENTINEL_ID_17>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 75, | |
| "content": "<SENTINEL_ID_18>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 76, | |
| "content": "<SENTINEL_ID_19>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 77, | |
| "content": "<SENTINEL_ID_20>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 78, | |
| "content": "<SENTINEL_ID_21>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 79, | |
| "content": "<SENTINEL_ID_22>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 80, | |
| "content": "<SENTINEL_ID_23>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 81, | |
| "content": "<SENTINEL_ID_24>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 82, | |
| "content": "<SENTINEL_ID_25>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 83, | |
| "content": "<SENTINEL_ID_26>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 84, | |
| "content": "<SENTINEL_ID_27>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 85, | |
| "content": "<SENTINEL_ID_28>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 86, | |
| "content": "<SENTINEL_ID_29>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 87, | |
| "content": "<SENTINEL_ID_30>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 88, | |
| "content": "<SENTINEL_ID_31>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 89, | |
| "content": "<SENTINEL_ID_32>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 90, | |
| "content": "<SENTINEL_ID_33>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 91, | |
| "content": "<SENTINEL_ID_34>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 92, | |
| "content": "<SENTINEL_ID_35>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 93, | |
| "content": "<SENTINEL_ID_36>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 94, | |
| "content": "<SENTINEL_ID_37>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 95, | |
| "content": "<SENTINEL_ID_38>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 96, | |
| "content": "<SENTINEL_ID_39>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 97, | |
| "content": "<SENTINEL_ID_40>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 98, | |
| "content": "<SENTINEL_ID_41>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 99, | |
| "content": "<SENTINEL_ID_42>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 100, | |
| "content": "<SENTINEL_ID_43>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 101, | |
| "content": "<SENTINEL_ID_44>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 102, | |
| "content": "<SENTINEL_ID_45>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 103, | |
| "content": "<SENTINEL_ID_46>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 104, | |
| "content": "<SENTINEL_ID_47>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 105, | |
| "content": "<SENTINEL_ID_48>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 106, | |
| "content": "<SENTINEL_ID_49>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 107, | |
| "content": "<SENTINEL_ID_50>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 108, | |
| "content": "<SENTINEL_ID_51>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 109, | |
| "content": "<SENTINEL_ID_52>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 110, | |
| "content": "<SENTINEL_ID_53>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 111, | |
| "content": "<SENTINEL_ID_54>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 112, | |
| "content": "<SENTINEL_ID_55>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 113, | |
| "content": "<SENTINEL_ID_56>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 114, | |
| "content": "<SENTINEL_ID_57>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 115, | |
| "content": "<SENTINEL_ID_58>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 116, | |
| "content": "<SENTINEL_ID_59>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 117, | |
| "content": "<SENTINEL_ID_60>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 118, | |
| "content": "<SENTINEL_ID_61>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 119, | |
| "content": "<SENTINEL_ID_62>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 120, | |
| "content": "<SENTINEL_ID_63>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 121, | |
| "content": "<SENTINEL_ID_64>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 122, | |
| "content": "<SENTINEL_ID_65>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 123, | |
| "content": "<SENTINEL_ID_66>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 124, | |
| "content": "<SENTINEL_ID_67>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 125, | |
| "content": "<SENTINEL_ID_68>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 126, | |
| "content": "<SENTINEL_ID_69>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 127, | |
| "content": "<SENTINEL_ID_70>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 128, | |
| "content": "<SENTINEL_ID_71>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 129, | |
| "content": "<SENTINEL_ID_72>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 130, | |
| "content": "<SENTINEL_ID_73>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 131, | |
| "content": "<SENTINEL_ID_74>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 132, | |
| "content": "<SENTINEL_ID_75>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 133, | |
| "content": "<SENTINEL_ID_76>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 134, | |
| "content": "<SENTINEL_ID_77>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 135, | |
| "content": "<SENTINEL_ID_78>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 136, | |
| "content": "<SENTINEL_ID_79>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 137, | |
| "content": "<SENTINEL_ID_80>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 138, | |
| "content": "<SENTINEL_ID_81>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 139, | |
| "content": "<SENTINEL_ID_82>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 140, | |
| "content": "<SENTINEL_ID_83>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 141, | |
| "content": "<SENTINEL_ID_84>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 142, | |
| "content": "<SENTINEL_ID_85>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 143, | |
| "content": "<SENTINEL_ID_86>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 144, | |
| "content": "<SENTINEL_ID_87>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 145, | |
| "content": "<SENTINEL_ID_88>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 146, | |
| "content": "<SENTINEL_ID_89>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 147, | |
| "content": "<SENTINEL_ID_90>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 148, | |
| "content": "<SENTINEL_ID_91>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 149, | |
| "content": "<SENTINEL_ID_92>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 150, | |
| "content": "<SENTINEL_ID_93>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 151, | |
| "content": "<SENTINEL_ID_94>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 152, | |
| "content": "<SENTINEL_ID_95>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 153, | |
| "content": "<SENTINEL_ID_96>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 154, | |
| "content": "<SENTINEL_ID_97>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 155, | |
| "content": "<SENTINEL_ID_98>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 156, | |
| "content": "<SENTINEL_ID_99>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 157, | |
| "content": "<SENTINEL_ID_100>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 158, | |
| "content": "<SENTINEL_ID_101>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 159, | |
| "content": "<SENTINEL_ID_102>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 160, | |
| "content": "<SENTINEL_ID_103>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 161, | |
| "content": "<SENTINEL_ID_104>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 162, | |
| "content": "<SENTINEL_ID_105>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 163, | |
| "content": "<SENTINEL_ID_106>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 164, | |
| "content": "<SENTINEL_ID_107>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 165, | |
| "content": "<SENTINEL_ID_108>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 166, | |
| "content": "<SENTINEL_ID_109>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 167, | |
| "content": "<SENTINEL_ID_110>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 168, | |
| "content": "<SENTINEL_ID_111>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 169, | |
| "content": "<SENTINEL_ID_112>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 170, | |
| "content": "<SENTINEL_ID_113>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 171, | |
| "content": "<SENTINEL_ID_114>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 172, | |
| "content": "<SENTINEL_ID_115>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 173, | |
| "content": "<SENTINEL_ID_116>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 174, | |
| "content": "<SENTINEL_ID_117>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 175, | |
| "content": "<SENTINEL_ID_118>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 176, | |
| "content": "<SENTINEL_ID_119>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 177, | |
| "content": "<SENTINEL_ID_120>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 178, | |
| "content": "<SENTINEL_ID_121>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 179, | |
| "content": "<SENTINEL_ID_122>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 180, | |
| "content": "<SENTINEL_ID_123>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 181, | |
| "content": "<SENTINEL_ID_124>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 182, | |
| "content": "<SENTINEL_ID_125>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 183, | |
| "content": "<SENTINEL_ID_126>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 184, | |
| "content": "<SENTINEL_ID_127>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 185, | |
| "content": "<SENTINEL_ID_128>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 186, | |
| "content": "<SENTINEL_ID_129>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 187, | |
| "content": "<SENTINEL_ID_130>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 188, | |
| "content": "<SENTINEL_ID_131>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 189, | |
| "content": "<SENTINEL_ID_132>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 190, | |
| "content": "<SENTINEL_ID_133>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 191, | |
| "content": "<SENTINEL_ID_134>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 192, | |
| "content": "<SENTINEL_ID_135>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 193, | |
| "content": "<SENTINEL_ID_136>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 194, | |
| "content": "<SENTINEL_ID_137>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 195, | |
| "content": "<SENTINEL_ID_138>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 196, | |
| "content": "<SENTINEL_ID_139>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 197, | |
| "content": "<SENTINEL_ID_140>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 198, | |
| "content": "<SENTINEL_ID_141>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 199, | |
| "content": "<SENTINEL_ID_142>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 200, | |
| "content": "<SENTINEL_ID_143>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 201, | |
| "content": "<SENTINEL_ID_144>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 202, | |
| "content": "<SENTINEL_ID_145>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 203, | |
| "content": "<SENTINEL_ID_146>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 204, | |
| "content": "<SENTINEL_ID_147>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 205, | |
| "content": "<SENTINEL_ID_148>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 206, | |
| "content": "<SENTINEL_ID_149>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 207, | |
| "content": "<SENTINEL_ID_150>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 208, | |
| "content": "<SENTINEL_ID_151>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 209, | |
| "content": "<SENTINEL_ID_152>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 210, | |
| "content": "<SENTINEL_ID_153>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 211, | |
| "content": "<SENTINEL_ID_154>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 212, | |
| "content": "<SENTINEL_ID_155>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 213, | |
| "content": "<SENTINEL_ID_156>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 214, | |
| "content": "<SENTINEL_ID_157>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 215, | |
| "content": "<SENTINEL_ID_158>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 216, | |
| "content": "<SENTINEL_ID_159>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 217, | |
| "content": "<SENTINEL_ID_160>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 218, | |
| "content": "<SENTINEL_ID_161>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 219, | |
| "content": "<SENTINEL_ID_162>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 220, | |
| "content": "<SENTINEL_ID_163>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 221, | |
| "content": "<SENTINEL_ID_164>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 222, | |
| "content": "<SENTINEL_ID_165>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 223, | |
| "content": "<SENTINEL_ID_166>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 224, | |
| "content": "<SENTINEL_ID_167>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 225, | |
| "content": "<SENTINEL_ID_168>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 226, | |
| "content": "<SENTINEL_ID_169>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 227, | |
| "content": "<SENTINEL_ID_170>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 228, | |
| "content": "<SENTINEL_ID_171>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 229, | |
| "content": "<SENTINEL_ID_172>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 230, | |
| "content": "<SENTINEL_ID_173>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 231, | |
| "content": "<SENTINEL_ID_174>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 232, | |
| "content": "<SENTINEL_ID_175>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 233, | |
| "content": "<SENTINEL_ID_176>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 234, | |
| "content": "<SENTINEL_ID_177>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 235, | |
| "content": "<SENTINEL_ID_178>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 236, | |
| "content": "<SENTINEL_ID_179>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 237, | |
| "content": "<SENTINEL_ID_180>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 238, | |
| "content": "<SENTINEL_ID_181>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 239, | |
| "content": "<SENTINEL_ID_182>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 240, | |
| "content": "<SENTINEL_ID_183>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 241, | |
| "content": "<SENTINEL_ID_184>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 242, | |
| "content": "<SENTINEL_ID_185>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 243, | |
| "content": "<SENTINEL_ID_186>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 244, | |
| "content": "<SENTINEL_ID_187>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 245, | |
| "content": "<SENTINEL_ID_188>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 246, | |
| "content": "<SENTINEL_ID_189>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 247, | |
| "content": "<SENTINEL_ID_190>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 248, | |
| "content": "<SENTINEL_ID_191>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 249, | |
| "content": "<SENTINEL_ID_192>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 250, | |
| "content": "<SENTINEL_ID_193>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 251, | |
| "content": "<SENTINEL_ID_194>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 252, | |
| "content": "<SENTINEL_ID_195>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 253, | |
| "content": "<SENTINEL_ID_196>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 254, | |
| "content": "<SENTINEL_ID_197>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 255, | |
| "content": "<SENTINEL_ID_198>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 256, | |
| "content": "<SENTINEL_ID_199>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 257, | |
| "content": "<MOLECULAR_ENTITY_TYPE_ANTIGEN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 258, | |
| "content": "<MOLECULAR_ENTITY_TYPE_ANTIBODY_LIGHT_CHAIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 259, | |
| "content": "<MOLECULAR_ENTITY_TYPE_ANTIBODY_HEAVY_CHAIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 260, | |
| "content": "<ATTRIBUTE_ORGANISM>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 261, | |
| "content": "<ATTRIBUTE_ORGANISM_HUMAN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 262, | |
| "content": "<ATTRIBUTE_ORGANISM_RABBIT>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 263, | |
| "content": "<ATTRIBUTE_ORGANISM_RAT>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 264, | |
| "content": "<ATTRIBUTE_ORGANISM_MOUSE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 265, | |
| "content": "<ATTRIBUTE_ORGANISM_MONKEY>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 266, | |
| "content": "<ATTRIBUTE_ORGANISM_CAMEL>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 267, | |
| "content": "<EPITOPE_PARATOPE_PREDICTION>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 268, | |
| "content": "<MOLECULAR_ENTITY_ANTIBODY_HEAVY_CHAIN_CDR1>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 269, | |
| "content": "<MOLECULAR_ENTITY_ANTIBODY_LIGHT_CHAIN_CDR3>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 270, | |
| "content": "<MOLECULAR_ENTITY_ANTIBODY_HEAVY_CHAIN_CDR3>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 271, | |
| "content": "<MOLECULAR_ENTITY_ANTIBODY_LIGHT_CHAIN_CDR2>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 272, | |
| "content": "<MOLECULAR_ENTITY_ANTIBODY_HEAVY_CHAIN_CDR2>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 273, | |
| "content": "<MOLECULAR_ENTITY_ANTIBODY_LIGHT_CHAIN_CDR1>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 274, | |
| "content": "<MOLECULAR_ENTITY_GENERAL_PROTEIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 275, | |
| "content": "<TIMESTEP>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 276, | |
| "content": "<DIFFUSION>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 277, | |
| "content": "<SEQUENCE_NATURAL_END>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 278, | |
| "content": "<SMILES_SEQUENCE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 279, | |
| "content": "<SELFIES_SEQUENCE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 280, | |
| "content": "<AMINO_ACID_SEQUENCE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 281, | |
| "content": "<GENERAL_AFFINITY_CLASS>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 282, | |
| "content": "<BACKSPACE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 283, | |
| "content": "<SEQUENCE_NATURAL_START>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 284, | |
| "content": "<NOOP>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 285, | |
| "content": "<TARGETED_ANTIBODY_DESIGN_ENCODER_ONLY_MODE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 286, | |
| "content": "<MOLECULAR_ENTITY_SMALL_MOLECULE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 287, | |
| "content": "<MOLECULAR_ENTITY_CELL_GENE_EXPRESSION_RANKED>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 288, | |
| "content": "<CELL_TYPE_CLASS>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 289, | |
| "content": "<TISSUE_TYPE_CLASS>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 290, | |
| "content": "<CORRUPTED_AREA_START>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 291, | |
| "content": "<CORRUPTED_AREA_END>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 292, | |
| "content": "<MOLECULAR_ENTITY_MUTATED_PROTEIN_CHAIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 293, | |
| "content": "<MOLECULAR_ENTITY_PROTEIN_CHAIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 294, | |
| "content": "<COMPLEX_ENTITY>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 295, | |
| "content": "<ALTERNATIVE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 296, | |
| "content": "<CDR3_REGION>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 297, | |
| "content": "<GENERAL_CHAIN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 298, | |
| "content": "<SUBMOLECULAR_ENTITY>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 299, | |
| "content": "<MUTATED>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 300, | |
| "content": "<MOLECULAR_ENTITY_TCR_ALPHA_CDR3>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 301, | |
| "content": "<MOLECULAR_ENTITY_TCR_DELTA_CDR3>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 302, | |
| "content": "<MOLECULAR_ENTITY_TCR_DELTA_VAR>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 303, | |
| "content": "<MOLECULAR_ENTITY_TCR_GAMMA_CDR3>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 304, | |
| "content": "<MOLECULAR_ENTITY_TCR_GAMMA_VAR>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 305, | |
| "content": "<SCALAR>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 306, | |
| "content": "<VECTOR>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 307, | |
| "content": "<MASKED_SCALAR>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 308, | |
| "content": "<MASKED_VECTOR>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 309, | |
| "content": "<AUTOENCODER_LATENT_LOG_VARIANCE>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 310, | |
| "content": "<AUTOENCODER_LATENT_MEAN>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 311, | |
| "content": "<AUTOENCODER_LATENT_SAMPLED_Z>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 312, | |
| "content": "<AUTOENCODER_TASK>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| }, | |
| { | |
| "id": 313, | |
| "content": "<DECODED_FROM_LATENT>", | |
| "single_word": false, | |
| "lstrip": false, | |
| "rstrip": false, | |
| "normalized": false, | |
| "special": true | |
| } | |
| ], | |
| "normalizer": null, | |
| "pre_tokenizer": { | |
| "type": "Sequence", | |
| "pretokenizers": [ | |
| { | |
| "type": "Split", | |
| "pattern": { | |
| "Regex": "<.*?>|\\S" | |
| }, | |
| "behavior": "Removed", | |
| "invert": true | |
| } | |
| ] | |
| }, | |
| "post_processor": null, | |
| "decoder": null, | |
| "model": { | |
| "type": "WordLevel", | |
| "vocab": { | |
| "<UNK>": 0, | |
| "<PAD>": 1, | |
| "<CLS>": 2, | |
| "<SEP>": 3, | |
| "<MASK>": 4, | |
| "<EOS>": 5, | |
| "<MOLECULAR_ENTITY>": 6, | |
| "<GLOBAL_INTERACTION_ATTRIBUTES>": 7, | |
| "<MOLECULAR_ENTITY_ANTIGEN>": 8, | |
| "<MOLECULAR_ENTITY_EPITOPE>": 9, | |
| "<MOLECULAR_ENTITY_ANTIBODY_HEAVY_CHAIN>": 10, | |
| "<MOLECULAR_ENTITY_ANTIBODY_LIGHT_CHAIN>": 11, | |
| "<MOLECULAR_ENTITY_TCR_ALPHA_CHAIN>": 12, | |
| "<MOLECULAR_ENTITY_TCR_BETA_VDJ>": 13, | |
| "<MOLECULAR_ENTITY_TCR_BETA_CDR3>": 14, | |
| "<BINDING_AFFINITY_CLASS>": 15, | |
| "<DECODER_START>": 16, | |
| "<BINDING>": 17, | |
| "<FILLIN>": 18, | |
| "<REORDER>": 19, | |
| "<TOAA>": 20, | |
| "<ACTIVE>": 21, | |
| "<GENESEQ>": 22, | |
| "<INCREASE>": 23, | |
| "<DECREASE>": 24, | |
| "<STRUCTURE>": 25, | |
| "<DISTANCE>": 26, | |
| "<SOLUBILITY>": 27, | |
| "<TOXICITY>": 28, | |
| "<AB>": 29, | |
| "<ISACTIVE>": 30, | |
| "<ISSYNTHETIC>": 31, | |
| "<PENETR>": 32, | |
| "<ABSORPTION>": 33, | |
| "<DISTRIBUTION>": 34, | |
| "<METABOLISM>": 35, | |
| "<EXCRETION>": 36, | |
| "<FLUORESCENCE>": 37, | |
| "<STABILITY>": 38, | |
| "<DISORDER>": 39, | |
| "<DISEASE>": 40, | |
| "<BINARY>": 41, | |
| "<REGRESSION>": 42, | |
| "<ORGANISM>": 43, | |
| "<0>": 44, | |
| "<1>": 45, | |
| "<2>": 46, | |
| "<3>": 47, | |
| "<4>": 48, | |
| "<5>": 49, | |
| "<6>": 50, | |
| "<7>": 51, | |
| "<8>": 52, | |
| "<9>": 53, | |
| "<.>": 54, | |
| "<YES>": 55, | |
| "<NO>": 56, | |
| "<SENTINEL_ID_0>": 57, | |
| "<SENTINEL_ID_1>": 58, | |
| "<SENTINEL_ID_2>": 59, | |
| "<SENTINEL_ID_3>": 60, | |
| "<SENTINEL_ID_4>": 61, | |
| "<SENTINEL_ID_5>": 62, | |
| "<SENTINEL_ID_6>": 63, | |
| "<SENTINEL_ID_7>": 64, | |
| "<SENTINEL_ID_8>": 65, | |
| "<SENTINEL_ID_9>": 66, | |
| "<SENTINEL_ID_10>": 67, | |
| "<SENTINEL_ID_11>": 68, | |
| "<SENTINEL_ID_12>": 69, | |
| "<SENTINEL_ID_13>": 70, | |
| "<SENTINEL_ID_14>": 71, | |
| "<SENTINEL_ID_15>": 72, | |
| "<SENTINEL_ID_16>": 73, | |
| "<SENTINEL_ID_17>": 74, | |
| "<SENTINEL_ID_18>": 75, | |
| "<SENTINEL_ID_19>": 76, | |
| "<SENTINEL_ID_20>": 77, | |
| "<SENTINEL_ID_21>": 78, | |
| "<SENTINEL_ID_22>": 79, | |
| "<SENTINEL_ID_23>": 80, | |
| "<SENTINEL_ID_24>": 81, | |
| "<SENTINEL_ID_25>": 82, | |
| "<SENTINEL_ID_26>": 83, | |
| "<SENTINEL_ID_27>": 84, | |
| "<SENTINEL_ID_28>": 85, | |
| "<SENTINEL_ID_29>": 86, | |
| "<SENTINEL_ID_30>": 87, | |
| "<SENTINEL_ID_31>": 88, | |
| "<SENTINEL_ID_32>": 89, | |
| "<SENTINEL_ID_33>": 90, | |
| "<SENTINEL_ID_34>": 91, | |
| "<SENTINEL_ID_35>": 92, | |
| "<SENTINEL_ID_36>": 93, | |
| "<SENTINEL_ID_37>": 94, | |
| "<SENTINEL_ID_38>": 95, | |
| "<SENTINEL_ID_39>": 96, | |
| "<SENTINEL_ID_40>": 97, | |
| "<SENTINEL_ID_41>": 98, | |
| "<SENTINEL_ID_42>": 99, | |
| "<SENTINEL_ID_43>": 100, | |
| "<SENTINEL_ID_44>": 101, | |
| "<SENTINEL_ID_45>": 102, | |
| "<SENTINEL_ID_46>": 103, | |
| "<SENTINEL_ID_47>": 104, | |
| "<SENTINEL_ID_48>": 105, | |
| "<SENTINEL_ID_49>": 106, | |
| "<SENTINEL_ID_50>": 107, | |
| "<SENTINEL_ID_51>": 108, | |
| "<SENTINEL_ID_52>": 109, | |
| "<SENTINEL_ID_53>": 110, | |
| "<SENTINEL_ID_54>": 111, | |
| "<SENTINEL_ID_55>": 112, | |
| "<SENTINEL_ID_56>": 113, | |
| "<SENTINEL_ID_57>": 114, | |
| "<SENTINEL_ID_58>": 115, | |
| "<SENTINEL_ID_59>": 116, | |
| "<SENTINEL_ID_60>": 117, | |
| "<SENTINEL_ID_61>": 118, | |
| "<SENTINEL_ID_62>": 119, | |
| "<SENTINEL_ID_63>": 120, | |
| "<SENTINEL_ID_64>": 121, | |
| "<SENTINEL_ID_65>": 122, | |
| "<SENTINEL_ID_66>": 123, | |
| "<SENTINEL_ID_67>": 124, | |
| "<SENTINEL_ID_68>": 125, | |
| "<SENTINEL_ID_69>": 126, | |
| "<SENTINEL_ID_70>": 127, | |
| "<SENTINEL_ID_71>": 128, | |
| "<SENTINEL_ID_72>": 129, | |
| "<SENTINEL_ID_73>": 130, | |
| "<SENTINEL_ID_74>": 131, | |
| "<SENTINEL_ID_75>": 132, | |
| "<SENTINEL_ID_76>": 133, | |
| "<SENTINEL_ID_77>": 134, | |
| "<SENTINEL_ID_78>": 135, | |
| "<SENTINEL_ID_79>": 136, | |
| "<SENTINEL_ID_80>": 137, | |
| "<SENTINEL_ID_81>": 138, | |
| "<SENTINEL_ID_82>": 139, | |
| "<SENTINEL_ID_83>": 140, | |
| "<SENTINEL_ID_84>": 141, | |
| "<SENTINEL_ID_85>": 142, | |
| "<SENTINEL_ID_86>": 143, | |
| "<SENTINEL_ID_87>": 144, | |
| "<SENTINEL_ID_88>": 145, | |
| "<SENTINEL_ID_89>": 146, | |
| "<SENTINEL_ID_90>": 147, | |
| "<SENTINEL_ID_91>": 148, | |
| "<SENTINEL_ID_92>": 149, | |
| "<SENTINEL_ID_93>": 150, | |
| "<SENTINEL_ID_94>": 151, | |
| "<SENTINEL_ID_95>": 152, | |
| "<SENTINEL_ID_96>": 153, | |
| "<SENTINEL_ID_97>": 154, | |
| "<SENTINEL_ID_98>": 155, | |
| "<SENTINEL_ID_99>": 156, | |
| "<SENTINEL_ID_100>": 157, | |
| "<SENTINEL_ID_101>": 158, | |
| "<SENTINEL_ID_102>": 159, | |
| "<SENTINEL_ID_103>": 160, | |
| "<SENTINEL_ID_104>": 161, | |
| "<SENTINEL_ID_105>": 162, | |
| "<SENTINEL_ID_106>": 163, | |
| "<SENTINEL_ID_107>": 164, | |
| "<SENTINEL_ID_108>": 165, | |
| "<SENTINEL_ID_109>": 166, | |
| "<SENTINEL_ID_110>": 167, | |
| "<SENTINEL_ID_111>": 168, | |
| "<SENTINEL_ID_112>": 169, | |
| "<SENTINEL_ID_113>": 170, | |
| "<SENTINEL_ID_114>": 171, | |
| "<SENTINEL_ID_115>": 172, | |
| "<SENTINEL_ID_116>": 173, | |
| "<SENTINEL_ID_117>": 174, | |
| "<SENTINEL_ID_118>": 175, | |
| "<SENTINEL_ID_119>": 176, | |
| "<SENTINEL_ID_120>": 177, | |
| "<SENTINEL_ID_121>": 178, | |
| "<SENTINEL_ID_122>": 179, | |
| "<SENTINEL_ID_123>": 180, | |
| "<SENTINEL_ID_124>": 181, | |
| "<SENTINEL_ID_125>": 182, | |
| "<SENTINEL_ID_126>": 183, | |
| "<SENTINEL_ID_127>": 184, | |
| "<SENTINEL_ID_128>": 185, | |
| "<SENTINEL_ID_129>": 186, | |
| "<SENTINEL_ID_130>": 187, | |
| "<SENTINEL_ID_131>": 188, | |
| "<SENTINEL_ID_132>": 189, | |
| "<SENTINEL_ID_133>": 190, | |
| "<SENTINEL_ID_134>": 191, | |
| "<SENTINEL_ID_135>": 192, | |
| "<SENTINEL_ID_136>": 193, | |
| "<SENTINEL_ID_137>": 194, | |
| "<SENTINEL_ID_138>": 195, | |
| "<SENTINEL_ID_139>": 196, | |
| "<SENTINEL_ID_140>": 197, | |
| "<SENTINEL_ID_141>": 198, | |
| "<SENTINEL_ID_142>": 199, | |
| "<SENTINEL_ID_143>": 200, | |
| "<SENTINEL_ID_144>": 201, | |
| "<SENTINEL_ID_145>": 202, | |
| "<SENTINEL_ID_146>": 203, | |
| "<SENTINEL_ID_147>": 204, | |
| "<SENTINEL_ID_148>": 205, | |
| "<SENTINEL_ID_149>": 206, | |
| "<SENTINEL_ID_150>": 207, | |
| "<SENTINEL_ID_151>": 208, | |
| "<SENTINEL_ID_152>": 209, | |
| "<SENTINEL_ID_153>": 210, | |
| "<SENTINEL_ID_154>": 211, | |
| "<SENTINEL_ID_155>": 212, | |
| "<SENTINEL_ID_156>": 213, | |
| "<SENTINEL_ID_157>": 214, | |
| "<SENTINEL_ID_158>": 215, | |
| "<SENTINEL_ID_159>": 216, | |
| "<SENTINEL_ID_160>": 217, | |
| "<SENTINEL_ID_161>": 218, | |
| "<SENTINEL_ID_162>": 219, | |
| "<SENTINEL_ID_163>": 220, | |
| "<SENTINEL_ID_164>": 221, | |
| "<SENTINEL_ID_165>": 222, | |
| "<SENTINEL_ID_166>": 223, | |
| "<SENTINEL_ID_167>": 224, | |
| "<SENTINEL_ID_168>": 225, | |
| "<SENTINEL_ID_169>": 226, | |
| "<SENTINEL_ID_170>": 227, | |
| "<SENTINEL_ID_171>": 228, | |
| "<SENTINEL_ID_172>": 229, | |
| "<SENTINEL_ID_173>": 230, | |
| "<SENTINEL_ID_174>": 231, | |
| "<SENTINEL_ID_175>": 232, | |
| "<SENTINEL_ID_176>": 233, | |
| "<SENTINEL_ID_177>": 234, | |
| "<SENTINEL_ID_178>": 235, | |
| "<SENTINEL_ID_179>": 236, | |
| "<SENTINEL_ID_180>": 237, | |
| "<SENTINEL_ID_181>": 238, | |
| "<SENTINEL_ID_182>": 239, | |
| "<SENTINEL_ID_183>": 240, | |
| "<SENTINEL_ID_184>": 241, | |
| "<SENTINEL_ID_185>": 242, | |
| "<SENTINEL_ID_186>": 243, | |
| "<SENTINEL_ID_187>": 244, | |
| "<SENTINEL_ID_188>": 245, | |
| "<SENTINEL_ID_189>": 246, | |
| "<SENTINEL_ID_190>": 247, | |
| "<SENTINEL_ID_191>": 248, | |
| "<SENTINEL_ID_192>": 249, | |
| "<SENTINEL_ID_193>": 250, | |
| "<SENTINEL_ID_194>": 251, | |
| "<SENTINEL_ID_195>": 252, | |
| "<SENTINEL_ID_196>": 253, | |
| "<SENTINEL_ID_197>": 254, | |
| "<SENTINEL_ID_198>": 255, | |
| "<SENTINEL_ID_199>": 256, | |
| "<MOLECULAR_ENTITY_TYPE_ANTIGEN>": 257, | |
| "<MOLECULAR_ENTITY_TYPE_ANTIBODY_LIGHT_CHAIN>": 258, | |
| "<MOLECULAR_ENTITY_TYPE_ANTIBODY_HEAVY_CHAIN>": 259, | |
| "<ATTRIBUTE_ORGANISM>": 260, | |
| "<ATTRIBUTE_ORGANISM_HUMAN>": 261, | |
| "<ATTRIBUTE_ORGANISM_RABBIT>": 262, | |
| "<ATTRIBUTE_ORGANISM_RAT>": 263, | |
| "<ATTRIBUTE_ORGANISM_MOUSE>": 264, | |
| "<ATTRIBUTE_ORGANISM_MONKEY>": 265, | |
| "<ATTRIBUTE_ORGANISM_CAMEL>": 266, | |
| "<EPITOPE_PARATOPE_PREDICTION>": 267, | |
| "<MOLECULAR_ENTITY_ANTIBODY_HEAVY_CHAIN_CDR1>": 268, | |
| "<MOLECULAR_ENTITY_ANTIBODY_LIGHT_CHAIN_CDR3>": 269, | |
| "<MOLECULAR_ENTITY_ANTIBODY_HEAVY_CHAIN_CDR3>": 270, | |
| "<MOLECULAR_ENTITY_ANTIBODY_LIGHT_CHAIN_CDR2>": 271, | |
| "<MOLECULAR_ENTITY_ANTIBODY_HEAVY_CHAIN_CDR2>": 272, | |
| "<MOLECULAR_ENTITY_ANTIBODY_LIGHT_CHAIN_CDR1>": 273, | |
| "<MOLECULAR_ENTITY_GENERAL_PROTEIN>": 274, | |
| "<TIMESTEP>": 275, | |
| "<DIFFUSION>": 276, | |
| "<SEQUENCE_NATURAL_END>": 277, | |
| "<SMILES_SEQUENCE>": 278, | |
| "<SELFIES_SEQUENCE>": 279, | |
| "<AMINO_ACID_SEQUENCE>": 280, | |
| "<GENERAL_AFFINITY_CLASS>": 281, | |
| "<BACKSPACE>": 282, | |
| "<SEQUENCE_NATURAL_START>": 283, | |
| "<NOOP>": 284, | |
| "<TARGETED_ANTIBODY_DESIGN_ENCODER_ONLY_MODE>": 285, | |
| "<MOLECULAR_ENTITY_SMALL_MOLECULE>": 286, | |
| "<MOLECULAR_ENTITY_CELL_GENE_EXPRESSION_RANKED>": 287, | |
| "<CELL_TYPE_CLASS>": 288, | |
| "<TISSUE_TYPE_CLASS>": 289, | |
| "<CORRUPTED_AREA_START>": 290, | |
| "<CORRUPTED_AREA_END>": 291, | |
| "<MOLECULAR_ENTITY_MUTATED_PROTEIN_CHAIN>": 292, | |
| "<MOLECULAR_ENTITY_PROTEIN_CHAIN>": 293, | |
| "<COMPLEX_ENTITY>": 294, | |
| "<ALTERNATIVE>": 295, | |
| "<CDR3_REGION>": 296, | |
| "<GENERAL_CHAIN>": 297, | |
| "<SUBMOLECULAR_ENTITY>": 298, | |
| "<MUTATED>": 299, | |
| "<MOLECULAR_ENTITY_TCR_ALPHA_CDR3>": 300, | |
| "<MOLECULAR_ENTITY_TCR_DELTA_CDR3>": 301, | |
| "<MOLECULAR_ENTITY_TCR_DELTA_VAR>": 302, | |
| "<MOLECULAR_ENTITY_TCR_GAMMA_CDR3>": 303, | |
| "<MOLECULAR_ENTITY_TCR_GAMMA_VAR>": 304, | |
| "<SCALAR>": 305, | |
| "<VECTOR>": 306, | |
| "<MASKED_SCALAR>": 307, | |
| "<MASKED_VECTOR>": 308, | |
| "<AUTOENCODER_LATENT_LOG_VARIANCE>": 309, | |
| "<AUTOENCODER_LATENT_MEAN>": 310, | |
| "<AUTOENCODER_LATENT_SAMPLED_Z>": 311, | |
| "<AUTOENCODER_TASK>": 312, | |
| "<DECODED_FROM_LATENT>": 313, | |
| "A": 501, | |
| "B": 502, | |
| "C": 503, | |
| "D": 504, | |
| "E": 505, | |
| "F": 506, | |
| "G": 507, | |
| "H": 508, | |
| "I": 509, | |
| "K": 510, | |
| "L": 511, | |
| "M": 512, | |
| "N": 513, | |
| "O": 514, | |
| "P": 515, | |
| "Q": 516, | |
| "R": 517, | |
| "S": 518, | |
| "T": 519, | |
| "U": 520, | |
| "V": 521, | |
| "W": 522, | |
| "X": 523, | |
| "Y": 524, | |
| "Z": 525, | |
| ":": 526 | |
| }, | |
| "unk_token": "<UNK>" | |
| } | |
| } |