Viro-BERT / config.json
LatentSpaceResearch's picture
Upload config.json
92ff1cd verified
Raw History Blame Contribute Delete
2.66 kB
{
"model_type": "distilbert",
"architecture": "BiologicalDistilBERT",
"k_mer": 3,
"stride": 3,
"chunk_len": 512,
"overlap": 50,
"num_classes": 14,
"label_to_id": {
"Dengue": 0,
"EnteroA": 1,
"EnteroB": 2,
"HBV": 3,
"HCV": 4,
"HPV": 5,
"HSV-1": 6,
"IAV": 7,
"IBV": 8,
"ICV": 9,
"MPV": 10,
"SFTS": 11,
"WNV": 12,
"Zika": 13
},
"id_to_label": {
"0": "Dengue",
"1": "EnteroA",
"2": "EnteroB",
"3": "HBV",
"4": "HCV",
"5": "HPV",
"6": "HSV-1",
"7": "IAV",
"8": "IBV",
"9": "ICV",
"10": "MPV",
"11": "SFTS",
"12": "WNV",
"13": "Zika"
},
"vocab": {
"[PAD]": 0,
"[UNK]": 1,
"[CLS]": 2,
"[SEP]": 3,
"[MASK]": 4,
"AAA": 5,
"AAC": 6,
"AAG": 7,
"AAN": 8,
"AAT": 9,
"ACA": 10,
"ACC": 11,
"ACG": 12,
"ACN": 13,
"ACT": 14,
"AGA": 15,
"AGC": 16,
"AGG": 17,
"AGN": 18,
"AGT": 19,
"ANA": 20,
"ANC": 21,
"ANG": 22,
"ANN": 23,
"ANT": 24,
"ATA": 25,
"ATC": 26,
"ATG": 27,
"ATN": 28,
"ATT": 29,
"CAA": 30,
"CAC": 31,
"CAG": 32,
"CAN": 33,
"CAT": 34,
"CCA": 35,
"CCC": 36,
"CCG": 37,
"CCN": 38,
"CCT": 39,
"CGA": 40,
"CGC": 41,
"CGG": 42,
"CGN": 43,
"CGT": 44,
"CNA": 45,
"CNC": 46,
"CNG": 47,
"CNN": 48,
"CNT": 49,
"CTA": 50,
"CTC": 51,
"CTG": 52,
"CTN": 53,
"CTT": 54,
"GAA": 55,
"GAC": 56,
"GAG": 57,
"GAN": 58,
"GAT": 59,
"GCA": 60,
"GCC": 61,
"GCG": 62,
"GCN": 63,
"GCT": 64,
"GGA": 65,
"GGC": 66,
"GGG": 67,
"GGN": 68,
"GGT": 69,
"GNA": 70,
"GNC": 71,
"GNG": 72,
"GNN": 73,
"GNT": 74,
"GTA": 75,
"GTC": 76,
"GTG": 77,
"GTN": 78,
"GTT": 79,
"NAA": 80,
"NAC": 81,
"NAG": 82,
"NAN": 83,
"NAT": 84,
"NCA": 85,
"NCC": 86,
"NCG": 87,
"NCN": 88,
"NCT": 89,
"NGA": 90,
"NGC": 91,
"NGG": 92,
"NGN": 93,
"NGT": 94,
"NNA": 95,
"NNC": 96,
"NNG": 97,
"NNN": 98,
"NNT": 99,
"NTA": 100,
"NTC": 101,
"NTG": 102,
"NTN": 103,
"NTT": 104,
"TAA": 105,
"TAC": 106,
"TAG": 107,
"TAN": 108,
"TAT": 109,
"TCA": 110,
"TCC": 111,
"TCG": 112,
"TCN": 113,
"TCT": 114,
"TGA": 115,
"TGC": 116,
"TGG": 117,
"TGN": 118,
"TGT": 119,
"TNA": 120,
"TNC": 121,
"TNG": 122,
"TNN": 123,
"TNT": 124,
"TTA": 125,
"TTC": 126,
"TTG": 127,
"TTN": 128,
"TTT": 129
}
}