Upload config.json
Browse files- config.json +173 -0
config.json
ADDED
|
@@ -0,0 +1,173 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_type": "distilbert",
|
| 3 |
+
"architecture": "BiologicalDistilBERT",
|
| 4 |
+
"k_mer": 3,
|
| 5 |
+
"stride": 3,
|
| 6 |
+
"chunk_len": 512,
|
| 7 |
+
"overlap": 50,
|
| 8 |
+
"num_classes": 14,
|
| 9 |
+
"label_to_id": {
|
| 10 |
+
"Dengue": 0,
|
| 11 |
+
"EnteroA": 1,
|
| 12 |
+
"EnteroB": 2,
|
| 13 |
+
"HBV": 3,
|
| 14 |
+
"HCV": 4,
|
| 15 |
+
"HPV": 5,
|
| 16 |
+
"HSV-1": 6,
|
| 17 |
+
"IAV": 7,
|
| 18 |
+
"IBV": 8,
|
| 19 |
+
"ICV": 9,
|
| 20 |
+
"MPV": 10,
|
| 21 |
+
"SFTS": 11,
|
| 22 |
+
"WNV": 12,
|
| 23 |
+
"Zika": 13
|
| 24 |
+
},
|
| 25 |
+
"id_to_label": {
|
| 26 |
+
"0": "Dengue",
|
| 27 |
+
"1": "EnteroA",
|
| 28 |
+
"2": "EnteroB",
|
| 29 |
+
"3": "HBV",
|
| 30 |
+
"4": "HCV",
|
| 31 |
+
"5": "HPV",
|
| 32 |
+
"6": "HSV-1",
|
| 33 |
+
"7": "IAV",
|
| 34 |
+
"8": "IBV",
|
| 35 |
+
"9": "ICV",
|
| 36 |
+
"10": "MPV",
|
| 37 |
+
"11": "SFTS",
|
| 38 |
+
"12": "WNV",
|
| 39 |
+
"13": "Zika"
|
| 40 |
+
},
|
| 41 |
+
"vocab": {
|
| 42 |
+
"[PAD]": 0,
|
| 43 |
+
"[UNK]": 1,
|
| 44 |
+
"[CLS]": 2,
|
| 45 |
+
"[SEP]": 3,
|
| 46 |
+
"[MASK]": 4,
|
| 47 |
+
"AAA": 5,
|
| 48 |
+
"AAC": 6,
|
| 49 |
+
"AAG": 7,
|
| 50 |
+
"AAN": 8,
|
| 51 |
+
"AAT": 9,
|
| 52 |
+
"ACA": 10,
|
| 53 |
+
"ACC": 11,
|
| 54 |
+
"ACG": 12,
|
| 55 |
+
"ACN": 13,
|
| 56 |
+
"ACT": 14,
|
| 57 |
+
"AGA": 15,
|
| 58 |
+
"AGC": 16,
|
| 59 |
+
"AGG": 17,
|
| 60 |
+
"AGN": 18,
|
| 61 |
+
"AGT": 19,
|
| 62 |
+
"ANA": 20,
|
| 63 |
+
"ANC": 21,
|
| 64 |
+
"ANG": 22,
|
| 65 |
+
"ANN": 23,
|
| 66 |
+
"ANT": 24,
|
| 67 |
+
"ATA": 25,
|
| 68 |
+
"ATC": 26,
|
| 69 |
+
"ATG": 27,
|
| 70 |
+
"ATN": 28,
|
| 71 |
+
"ATT": 29,
|
| 72 |
+
"CAA": 30,
|
| 73 |
+
"CAC": 31,
|
| 74 |
+
"CAG": 32,
|
| 75 |
+
"CAN": 33,
|
| 76 |
+
"CAT": 34,
|
| 77 |
+
"CCA": 35,
|
| 78 |
+
"CCC": 36,
|
| 79 |
+
"CCG": 37,
|
| 80 |
+
"CCN": 38,
|
| 81 |
+
"CCT": 39,
|
| 82 |
+
"CGA": 40,
|
| 83 |
+
"CGC": 41,
|
| 84 |
+
"CGG": 42,
|
| 85 |
+
"CGN": 43,
|
| 86 |
+
"CGT": 44,
|
| 87 |
+
"CNA": 45,
|
| 88 |
+
"CNC": 46,
|
| 89 |
+
"CNG": 47,
|
| 90 |
+
"CNN": 48,
|
| 91 |
+
"CNT": 49,
|
| 92 |
+
"CTA": 50,
|
| 93 |
+
"CTC": 51,
|
| 94 |
+
"CTG": 52,
|
| 95 |
+
"CTN": 53,
|
| 96 |
+
"CTT": 54,
|
| 97 |
+
"GAA": 55,
|
| 98 |
+
"GAC": 56,
|
| 99 |
+
"GAG": 57,
|
| 100 |
+
"GAN": 58,
|
| 101 |
+
"GAT": 59,
|
| 102 |
+
"GCA": 60,
|
| 103 |
+
"GCC": 61,
|
| 104 |
+
"GCG": 62,
|
| 105 |
+
"GCN": 63,
|
| 106 |
+
"GCT": 64,
|
| 107 |
+
"GGA": 65,
|
| 108 |
+
"GGC": 66,
|
| 109 |
+
"GGG": 67,
|
| 110 |
+
"GGN": 68,
|
| 111 |
+
"GGT": 69,
|
| 112 |
+
"GNA": 70,
|
| 113 |
+
"GNC": 71,
|
| 114 |
+
"GNG": 72,
|
| 115 |
+
"GNN": 73,
|
| 116 |
+
"GNT": 74,
|
| 117 |
+
"GTA": 75,
|
| 118 |
+
"GTC": 76,
|
| 119 |
+
"GTG": 77,
|
| 120 |
+
"GTN": 78,
|
| 121 |
+
"GTT": 79,
|
| 122 |
+
"NAA": 80,
|
| 123 |
+
"NAC": 81,
|
| 124 |
+
"NAG": 82,
|
| 125 |
+
"NAN": 83,
|
| 126 |
+
"NAT": 84,
|
| 127 |
+
"NCA": 85,
|
| 128 |
+
"NCC": 86,
|
| 129 |
+
"NCG": 87,
|
| 130 |
+
"NCN": 88,
|
| 131 |
+
"NCT": 89,
|
| 132 |
+
"NGA": 90,
|
| 133 |
+
"NGC": 91,
|
| 134 |
+
"NGG": 92,
|
| 135 |
+
"NGN": 93,
|
| 136 |
+
"NGT": 94,
|
| 137 |
+
"NNA": 95,
|
| 138 |
+
"NNC": 96,
|
| 139 |
+
"NNG": 97,
|
| 140 |
+
"NNN": 98,
|
| 141 |
+
"NNT": 99,
|
| 142 |
+
"NTA": 100,
|
| 143 |
+
"NTC": 101,
|
| 144 |
+
"NTG": 102,
|
| 145 |
+
"NTN": 103,
|
| 146 |
+
"NTT": 104,
|
| 147 |
+
"TAA": 105,
|
| 148 |
+
"TAC": 106,
|
| 149 |
+
"TAG": 107,
|
| 150 |
+
"TAN": 108,
|
| 151 |
+
"TAT": 109,
|
| 152 |
+
"TCA": 110,
|
| 153 |
+
"TCC": 111,
|
| 154 |
+
"TCG": 112,
|
| 155 |
+
"TCN": 113,
|
| 156 |
+
"TCT": 114,
|
| 157 |
+
"TGA": 115,
|
| 158 |
+
"TGC": 116,
|
| 159 |
+
"TGG": 117,
|
| 160 |
+
"TGN": 118,
|
| 161 |
+
"TGT": 119,
|
| 162 |
+
"TNA": 120,
|
| 163 |
+
"TNC": 121,
|
| 164 |
+
"TNG": 122,
|
| 165 |
+
"TNN": 123,
|
| 166 |
+
"TNT": 124,
|
| 167 |
+
"TTA": 125,
|
| 168 |
+
"TTC": 126,
|
| 169 |
+
"TTG": 127,
|
| 170 |
+
"TTN": 128,
|
| 171 |
+
"TTT": 129
|
| 172 |
+
}
|
| 173 |
+
}
|