{ "version": "1.0", "truncation": null, "padding": null, "added_tokens": [ { "id": 0, "content": "[START]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 1, "content": "[END]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 2, "content": "[PAD]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 3, "content": "[UNK]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 8, "content": "[None]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 9, "content": "[PF00580]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 10, "content": "[PF00580, PF06114, PF13361]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 11, "content": "[PF00580, PF13361]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 12, "content": "[PF01051]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 13, "content": "[PF01051, PF01051]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 14, "content": "[PF01051, PF06430]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 15, "content": "[PF01446]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 16, "content": "[PF01695]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 17, "content": "[PF01719]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 18, "content": "[PF01719, PF01719]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 19, "content": "[PF01719, PF01719, PF02796]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 20, "content": "[PF02387]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 21, "content": "[PF02387, PF02387]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 22, "content": "[PF02486]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 23, "content": "[PF02486, PF02486]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 24, "content": "[PF02486, PF18106]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 25, "content": "[PF03090]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 26, "content": "[PF03428, PF11800]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 27, "content": "[PF04796]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 28, "content": "[PF05472]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 29, "content": "[PF05732]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 30, "content": "[PF06114, PF13361, PF00580]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 31, "content": "[PF06430]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 32, "content": "[PF06504]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 33, "content": "[PF06970]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 34, "content": "[PF06970, PF06970, PF18008]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 35, "content": "[PF06970, PF18008]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 36, "content": "[PF07042]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 37, "content": "[PF10134]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 38, "content": "[PF10723]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 39, "content": "[PF13175]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 40, "content": "[PF13245]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 41, "content": "[PF13245, PF13361]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 42, "content": "[PF13361]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 43, "content": "[PF13361, PF00580]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 44, "content": "[PF13361, PF13245]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 45, "content": "[PF13481]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 46, "content": "[PF18008]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 47, "content": "[UNKNOWN_REP]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 48, "content": "[UNANNOTATED_REP]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 49, "content": "[Acetobacter pasteurianus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 50, "content": "[Achromobacter xylosoxidans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 51, "content": "[Acidithiobacillus ferridurans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 52, "content": "[Acinetobacter]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 53, "content": "[Acinetobacter baumannii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 54, "content": "[Acinetobacter calcoaceticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 55, "content": "[Acinetobacter defluvii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 56, "content": "[Acinetobacter haemolyticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 57, "content": "[Acinetobacter johnsonii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 58, "content": "[Acinetobacter lwoffii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 59, "content": "[Acinetobacter nosocomialis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 60, "content": "[Acinetobacter pittii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 61, "content": "[Acinetobacter schindleri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 62, "content": "[Acinetobacter soli]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 63, "content": "[Aeromonas]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 64, "content": "[Aeromonas bestiarum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 65, "content": "[Aeromonas hydrophila]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 66, "content": "[Aeromonas salmonicida]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 67, "content": "[Aggregatibacter actinomycetemcomitans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 68, "content": "[Agrobacterium rhizogenes]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 69, "content": "[Agrobacterium tumefaciens]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 70, "content": "[Alicycliphilus denitrificans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 71, "content": "[Aliivibrio wodanis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 72, "content": "[Azotobacter chroococcum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 73, "content": "[Bacillus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 74, "content": "[Bacillus aryabhattai]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 75, "content": "[Bacillus cereus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 76, "content": "[Bacillus cytotoxicus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 77, "content": "[Bacillus endophyticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 78, "content": "[Bacillus licheniformis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 79, "content": "[Bacillus methanolicus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 80, "content": "[Bacillus mycoides]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 81, "content": "[Bacillus pseudofirmus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 82, "content": "[Bacillus pumilus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 83, "content": "[Bacillus stearothermophilus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 84, "content": "[Bacillus subtilis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 85, "content": "[Bacillus thuringiensis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 86, "content": "[Bacteroides fragilis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 87, "content": "[Beet leafhopper]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 88, "content": "[Bhargavaea]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 89, "content": "[Bhargavaea cecembensis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 90, "content": "[Bibersteinia trehalosi]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 91, "content": "[Bifidobacterium asteroides]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 92, "content": "[Bifidobacterium breve]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 93, "content": "[Bifidobacterium catenulatum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 94, "content": "[Bifidobacterium longum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 95, "content": "[Bifidobacterium pseudocatenulatum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 96, "content": "[Buchnera aphidicola]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 97, "content": "[Burkholderia]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 98, "content": "[Butyrivibrio fibrisolvens]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 99, "content": "[Campylobacter coli]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 100, "content": "[Campylobacter fetus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 101, "content": "[Campylobacter jejuni]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 102, "content": "[Campylobacter lari]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 103, "content": "[Candidatus Azobacteroides]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 104, "content": "[Celeribacter manganoxidans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 105, "content": "[Chinaberry witches'-broom]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 106, "content": "[Chlamydia trachomatis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 107, "content": "[Citrobacter]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 108, "content": "[Citrobacter amalonaticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 109, "content": "[Citrobacter farmeri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 110, "content": "[Citrobacter freundii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 111, "content": "[Citrobacter rodentium]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 112, "content": "[Clostridium difficile]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 113, "content": "[Clostridium estertheticum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 114, "content": "[Clostridium pasteurianum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 115, "content": "[Clostridium perfringens]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 116, "content": "[Clostridium taeniosporum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 117, "content": "[Clostridium thermocellum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 118, "content": "[Collimonas fungivorans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 119, "content": "[Corynebacterium glutamicum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 120, "content": "[Deinococcus deserti]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 121, "content": "[Delftia]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 122, "content": "[Delftia acidovorans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 123, "content": "[Desulfurella acetivorans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 124, "content": "[Dichelobacter nodosus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 125, "content": "[Edwardsiella ictaluri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 126, "content": "[Elizabethkingia endophytica]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 127, "content": "[Enterobacter]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 128, "content": "[Enterobacter asburiae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 129, "content": "[Enterobacter cloacae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 130, "content": "[Enterobacter hormaechei]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 131, "content": "[Enterobacteriaceae bacterium]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 132, "content": "[Enterococcus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 133, "content": "[Enterococcus faecalis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 134, "content": "[Enterococcus faecium]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 135, "content": "[Enterococcus hirae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 136, "content": "[Erwinia amylovora]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 137, "content": "[Erysipelothrix rhusiopathiae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 138, "content": "[Escherichia]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 139, "content": "[Escherichia coli]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 140, "content": "[Fibrisoma limi]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 141, "content": "[Fusobacterium nucleatum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 142, "content": "[Geobacillus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 143, "content": "[Haemophilus parasuis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 144, "content": "[Helicobacter cetorum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 145, "content": "[Helicobacter pylori]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 146, "content": "[Klebsiella]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 147, "content": "[Klebsiella michiganensis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 148, "content": "[Klebsiella oxytoca]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 149, "content": "[Klebsiella pneumoniae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 150, "content": "[Klebsiella quasipneumoniae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 151, "content": "[Klebsiella variicola]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 152, "content": "[Komagataeibacter nataicola]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 153, "content": "[Lactobacillus acidipiscis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 154, "content": "[Lactobacillus acidophilus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 155, "content": "[Lactobacillus amylovorus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 156, "content": "[Lactobacillus backii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 157, "content": "[Lactobacillus brevis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 158, "content": "[Lactobacillus casei]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 159, "content": "[Lactobacillus coryniformis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 160, "content": "[Lactobacillus crustorum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 161, "content": "[Lactobacillus curvatus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 162, "content": "[Lactobacillus delbrueckii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 163, "content": "[Lactobacillus farciminis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 164, "content": "[Lactobacillus fermentum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 165, "content": "[Lactobacillus paracasei]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 166, "content": "[Lactobacillus paracollinoides]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 167, "content": "[Lactobacillus pentosus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 168, "content": "[Lactobacillus plantarum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 169, "content": "[Lactobacillus reuteri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 170, "content": "[Lactobacillus rhamnosus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 171, "content": "[Lactobacillus sakei]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 172, "content": "[Lactobacillus salivarius]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 173, "content": "[Lactobacillus sanfranciscensis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 174, "content": "[Lactococcus garvieae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 175, "content": "[Lactococcus lactis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 176, "content": "[Lactococcus piscium]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 177, "content": "[Lactococcus raffinolactis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 178, "content": "[Laribacter hongkongensis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 179, "content": "[Leclercia]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 180, "content": "[Lelliottia]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 181, "content": "[Leuconostoc citreum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 182, "content": "[Leuconostoc lactis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 183, "content": "[Listeria grayi]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 184, "content": "[Listeria monocytogenes]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 185, "content": "[Macrococcus caseolyticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 186, "content": "[Melissococcus plutonius]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 187, "content": "[Mesorhizobium ciceri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 188, "content": "[Moraxella bovis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 189, "content": "[Moraxella catarrhalis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 190, "content": "[Moraxella osloensis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 191, "content": "[Moraxella ovis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 192, "content": "[Paenibacillus larvae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 193, "content": "[Paeniclostridium sordellii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 194, "content": "[Pantoea]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 195, "content": "[Pantoea ananatis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 196, "content": "[Pantoea stewartii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 197, "content": "[Paracoccus marcusii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 198, "content": "[Paracoccus pantotrophus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 199, "content": "[Pectobacterium atrosepticum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 200, "content": "[Pediococcus damnosus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 201, "content": "[Pediococcus inopinatus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 202, "content": "[Pediococcus pentosaceus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 203, "content": "[Persicobacter]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 204, "content": "[Photobacterium damselae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 205, "content": "[Photorhabdus asymbiotica]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 206, "content": "[Planococcus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 207, "content": "[Planococcus antarcticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 208, "content": "[Planococcus citreus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 209, "content": "[Plasmid pRJF]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 210, "content": "[Plesiomonas shigelloides]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 211, "content": "[Proteus vulgaris]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 212, "content": "[Providencia rettgeri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 213, "content": "[Pseudoalteromonas arctica]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 214, "content": "[Pseudoalteromonas nigrifaciens]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 215, "content": "[Pseudomonas]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 216, "content": "[Pseudomonas aeruginosa]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 217, "content": "[Pseudomonas putida]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 218, "content": "[Pseudomonas syringae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 219, "content": "[Psychrobacter]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 220, "content": "[Psychrobacter cryohalolentis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 221, "content": "[Psychrobacter urativorans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 222, "content": "[Rahnella]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 223, "content": "[Raoultella ornithinolytica]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 224, "content": "[Raoultella planticola]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 225, "content": "[Rhodococcus erythropolis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 226, "content": "[Rhodothermus marinus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 227, "content": "[Rickettsiales bacterium]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 228, "content": "[Ruminococcus albus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 229, "content": "[Runella slithyformis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 230, "content": "[Salmonella enterica]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 231, "content": "[Saprospira grandis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 232, "content": "[Selenomonas ruminantium]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 233, "content": "[Serratia marcescens]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 234, "content": "[Shewanella baltica]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 235, "content": "[Shigella flexneri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 236, "content": "[Sinorhizobium meliloti]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 237, "content": "[Sphingobium baderi]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 238, "content": "[Sphingobium yanoikuyae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 239, "content": "[Sporosarcina ureae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 240, "content": "[Staphylococcus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 241, "content": "[Staphylococcus aureus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 242, "content": "[Staphylococcus chromogenes]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 243, "content": "[Staphylococcus cohnii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 244, "content": "[Staphylococcus epidermidis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 245, "content": "[Staphylococcus haemolyticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 246, "content": "[Staphylococcus kloosii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 247, "content": "[Staphylococcus lugdunensis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 248, "content": "[Staphylococcus pasteuri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 249, "content": "[Staphylococcus saprophyticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 250, "content": "[Staphylococcus sciuri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 251, "content": "[Staphylococcus simulans]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 252, "content": "[Staphylococcus warneri]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 253, "content": "[Streptococcus agalactiae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 254, "content": "[Streptococcus dysgalactiae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 255, "content": "[Streptococcus gallolyticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 256, "content": "[Streptococcus parasanguinis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 257, "content": "[Streptococcus pneumoniae]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 258, "content": "[Streptococcus pyogenes]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 259, "content": "[Streptococcus thermophilus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 260, "content": "[Tetragenococcus halophilus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 261, "content": "[Tetragenococcus muriaticus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 262, "content": "[Thermoanaerobacterium thermosaccharolyticum]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 263, "content": "[Thermus thermophilus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 264, "content": "[Thiobacillus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 265, "content": "[Trueperella pyogenes]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 266, "content": "[Uncultured marine]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 267, "content": "[Vibrio tubiashii]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 268, "content": "[Vibrio vulnificus]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 269, "content": "[Weissella koreensis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 270, "content": "[Wigglesworthia glossinidia]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 271, "content": "[Xylella fastidiosa]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 272, "content": "[Yersinia enterocolitica]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 273, "content": "[Yersinia pestis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 274, "content": "[Yersinia pseudotuberculosis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 275, "content": "[Zymomonas mobilis]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 276, "content": "[UNKNOWN_SPECIES]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 277, "content": "[UNANNOTATED_SPECIES]", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false } ], "normalizer": null, "pre_tokenizer": { "type": "Whitespace" }, "post_processor": null, "decoder": null, "model": { "type": "BPE", "dropout": null, "unk_token": "[UNK]", "continuing_subword_prefix": null, "end_of_word_suffix": null, "fuse_unk": false, "byte_fallback": false, "ignore_merges": false, "vocab": { "[START]": 0, "[END]": 1, "[PAD]": 2, "[UNK]": 3, "A": 4, "C": 5, "G": 6, "T": 7 }, "merges": [] } }