Training in progress, epoch 1
Browse files
config.json
CHANGED
|
@@ -10,23 +10,195 @@
|
|
| 10 |
"hidden_size": 768,
|
| 11 |
"id2label": {
|
| 12 |
"0": "0",
|
| 13 |
-
"1": "
|
| 14 |
-
"
|
| 15 |
-
"
|
| 16 |
-
"
|
| 17 |
-
"
|
| 18 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
},
|
| 20 |
"initializer_range": 0.02,
|
| 21 |
"intermediate_size": 3072,
|
| 22 |
"label2id": {
|
| 23 |
"0": "0",
|
| 24 |
-
"
|
| 25 |
-
"
|
| 26 |
-
"
|
| 27 |
-
"
|
| 28 |
-
"
|
| 29 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
},
|
| 31 |
"layer_norm_eps": 1e-12,
|
| 32 |
"max_length": 1024,
|
|
|
|
| 10 |
"hidden_size": 768,
|
| 11 |
"id2label": {
|
| 12 |
"0": "0",
|
| 13 |
+
"1": "101",
|
| 14 |
+
"10": "151",
|
| 15 |
+
"11": "156",
|
| 16 |
+
"12": "166",
|
| 17 |
+
"13": "169",
|
| 18 |
+
"14": "171",
|
| 19 |
+
"15": "172",
|
| 20 |
+
"16": "18",
|
| 21 |
+
"17": "182",
|
| 22 |
+
"18": "187",
|
| 23 |
+
"19": "189",
|
| 24 |
+
"2": "106",
|
| 25 |
+
"20": "190",
|
| 26 |
+
"21": "192",
|
| 27 |
+
"22": "200",
|
| 28 |
+
"23": "205",
|
| 29 |
+
"24": "207",
|
| 30 |
+
"25": "209",
|
| 31 |
+
"26": "211",
|
| 32 |
+
"27": "218",
|
| 33 |
+
"28": "219",
|
| 34 |
+
"29": "221",
|
| 35 |
+
"3": "112",
|
| 36 |
+
"30": "224",
|
| 37 |
+
"31": "226",
|
| 38 |
+
"32": "227",
|
| 39 |
+
"33": "229",
|
| 40 |
+
"34": "237",
|
| 41 |
+
"35": "239",
|
| 42 |
+
"36": "242",
|
| 43 |
+
"37": "244",
|
| 44 |
+
"38": "257",
|
| 45 |
+
"39": "26",
|
| 46 |
+
"4": "117",
|
| 47 |
+
"40": "260",
|
| 48 |
+
"41": "262",
|
| 49 |
+
"42": "265",
|
| 50 |
+
"43": "278",
|
| 51 |
+
"44": "281",
|
| 52 |
+
"45": "3",
|
| 53 |
+
"46": "312",
|
| 54 |
+
"47": "317",
|
| 55 |
+
"48": "328",
|
| 56 |
+
"49": "343",
|
| 57 |
+
"5": "122",
|
| 58 |
+
"50": "351",
|
| 59 |
+
"51": "354",
|
| 60 |
+
"52": "356",
|
| 61 |
+
"53": "358",
|
| 62 |
+
"54": "359",
|
| 63 |
+
"55": "368",
|
| 64 |
+
"56": "369",
|
| 65 |
+
"57": "371",
|
| 66 |
+
"58": "372",
|
| 67 |
+
"59": "373",
|
| 68 |
+
"6": "129",
|
| 69 |
+
"60": "378",
|
| 70 |
+
"61": "380",
|
| 71 |
+
"62": "383",
|
| 72 |
+
"63": "385",
|
| 73 |
+
"64": "386",
|
| 74 |
+
"65": "391",
|
| 75 |
+
"66": "394",
|
| 76 |
+
"67": "397",
|
| 77 |
+
"68": "4",
|
| 78 |
+
"69": "422",
|
| 79 |
+
"7": "134",
|
| 80 |
+
"70": "423",
|
| 81 |
+
"71": "424",
|
| 82 |
+
"72": "426",
|
| 83 |
+
"73": "427",
|
| 84 |
+
"74": "428",
|
| 85 |
+
"75": "46",
|
| 86 |
+
"76": "49",
|
| 87 |
+
"77": "5",
|
| 88 |
+
"78": "50",
|
| 89 |
+
"79": "58",
|
| 90 |
+
"8": "137",
|
| 91 |
+
"80": "6",
|
| 92 |
+
"81": "66",
|
| 93 |
+
"82": "67",
|
| 94 |
+
"83": "69",
|
| 95 |
+
"84": "7",
|
| 96 |
+
"85": "71",
|
| 97 |
+
"86": "73",
|
| 98 |
+
"87": "82",
|
| 99 |
+
"88": "84",
|
| 100 |
+
"89": "86",
|
| 101 |
+
"9": "139",
|
| 102 |
+
"90": "87",
|
| 103 |
+
"91": "89",
|
| 104 |
+
"92": "96"
|
| 105 |
},
|
| 106 |
"initializer_range": 0.02,
|
| 107 |
"intermediate_size": 3072,
|
| 108 |
"label2id": {
|
| 109 |
"0": "0",
|
| 110 |
+
"101": "1",
|
| 111 |
+
"106": "2",
|
| 112 |
+
"112": "3",
|
| 113 |
+
"117": "4",
|
| 114 |
+
"122": "5",
|
| 115 |
+
"129": "6",
|
| 116 |
+
"134": "7",
|
| 117 |
+
"137": "8",
|
| 118 |
+
"139": "9",
|
| 119 |
+
"151": "10",
|
| 120 |
+
"156": "11",
|
| 121 |
+
"166": "12",
|
| 122 |
+
"169": "13",
|
| 123 |
+
"171": "14",
|
| 124 |
+
"172": "15",
|
| 125 |
+
"18": "16",
|
| 126 |
+
"182": "17",
|
| 127 |
+
"187": "18",
|
| 128 |
+
"189": "19",
|
| 129 |
+
"190": "20",
|
| 130 |
+
"192": "21",
|
| 131 |
+
"200": "22",
|
| 132 |
+
"205": "23",
|
| 133 |
+
"207": "24",
|
| 134 |
+
"209": "25",
|
| 135 |
+
"211": "26",
|
| 136 |
+
"218": "27",
|
| 137 |
+
"219": "28",
|
| 138 |
+
"221": "29",
|
| 139 |
+
"224": "30",
|
| 140 |
+
"226": "31",
|
| 141 |
+
"227": "32",
|
| 142 |
+
"229": "33",
|
| 143 |
+
"237": "34",
|
| 144 |
+
"239": "35",
|
| 145 |
+
"242": "36",
|
| 146 |
+
"244": "37",
|
| 147 |
+
"257": "38",
|
| 148 |
+
"26": "39",
|
| 149 |
+
"260": "40",
|
| 150 |
+
"262": "41",
|
| 151 |
+
"265": "42",
|
| 152 |
+
"278": "43",
|
| 153 |
+
"281": "44",
|
| 154 |
+
"3": "45",
|
| 155 |
+
"312": "46",
|
| 156 |
+
"317": "47",
|
| 157 |
+
"328": "48",
|
| 158 |
+
"343": "49",
|
| 159 |
+
"351": "50",
|
| 160 |
+
"354": "51",
|
| 161 |
+
"356": "52",
|
| 162 |
+
"358": "53",
|
| 163 |
+
"359": "54",
|
| 164 |
+
"368": "55",
|
| 165 |
+
"369": "56",
|
| 166 |
+
"371": "57",
|
| 167 |
+
"372": "58",
|
| 168 |
+
"373": "59",
|
| 169 |
+
"378": "60",
|
| 170 |
+
"380": "61",
|
| 171 |
+
"383": "62",
|
| 172 |
+
"385": "63",
|
| 173 |
+
"386": "64",
|
| 174 |
+
"391": "65",
|
| 175 |
+
"394": "66",
|
| 176 |
+
"397": "67",
|
| 177 |
+
"4": "68",
|
| 178 |
+
"422": "69",
|
| 179 |
+
"423": "70",
|
| 180 |
+
"424": "71",
|
| 181 |
+
"426": "72",
|
| 182 |
+
"427": "73",
|
| 183 |
+
"428": "74",
|
| 184 |
+
"46": "75",
|
| 185 |
+
"49": "76",
|
| 186 |
+
"5": "77",
|
| 187 |
+
"50": "78",
|
| 188 |
+
"58": "79",
|
| 189 |
+
"6": "80",
|
| 190 |
+
"66": "81",
|
| 191 |
+
"67": "82",
|
| 192 |
+
"69": "83",
|
| 193 |
+
"7": "84",
|
| 194 |
+
"71": "85",
|
| 195 |
+
"73": "86",
|
| 196 |
+
"82": "87",
|
| 197 |
+
"84": "88",
|
| 198 |
+
"86": "89",
|
| 199 |
+
"87": "90",
|
| 200 |
+
"89": "91",
|
| 201 |
+
"96": "92"
|
| 202 |
},
|
| 203 |
"layer_norm_eps": 1e-12,
|
| 204 |
"max_length": 1024,
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c818b58a6061cffaca8f9b3f909814059ce0feaef646bfc0fe6276edbaf7df3e
|
| 3 |
+
size 345115321
|
runs/Jun03_18-48-59_5f08d837f8f4/1685818562.333823/events.out.tfevents.1685818562.5f08d837f8f4.623.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9037393e67353ad0388fc83922ed3d301b4435649a5ffe1d19ade71ec44ec44a
|
| 3 |
+
size 5859
|
runs/Jun03_18-48-59_5f08d837f8f4/events.out.tfevents.1685818562.5f08d837f8f4.623.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bb7f9ff33c45e085819715c5e3353cccb106cd72a2565c25d34a8e412bf55b4b
|
| 3 |
+
size 19137
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3579
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:46f7fd9ddc1935acdc1bb714b730ff212802c0bdf9c0702500d3774437c5bb19
|
| 3 |
size 3579
|