Add pretrained/hindi/history.json
Browse files- pretrained/hindi/history.json +198 -0
pretrained/hindi/history.json
ADDED
|
@@ -0,0 +1,198 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"step": 250,
|
| 4 |
+
"tokens": 16384000,
|
| 5 |
+
"train_loss": 6.636879325866699,
|
| 6 |
+
"val_loss": 5.579191946983338,
|
| 7 |
+
"lr": 0.000375
|
| 8 |
+
},
|
| 9 |
+
{
|
| 10 |
+
"step": 500,
|
| 11 |
+
"tokens": 32768000,
|
| 12 |
+
"train_loss": 5.284028044223786,
|
| 13 |
+
"val_loss": 4.918073010444641,
|
| 14 |
+
"lr": 0.0005997000845339562
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"step": 750,
|
| 18 |
+
"tokens": 49152000,
|
| 19 |
+
"train_loss": 4.739607352256775,
|
| 20 |
+
"val_loss": 4.501560378074646,
|
| 21 |
+
"lr": 0.000596280709172829
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"step": 1000,
|
| 25 |
+
"tokens": 65536000,
|
| 26 |
+
"train_loss": 4.449326189517975,
|
| 27 |
+
"val_loss": 4.252658152580262,
|
| 28 |
+
"lr": 0.000589092717074061
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"step": 1250,
|
| 32 |
+
"tokens": 81920000,
|
| 33 |
+
"train_loss": 4.2429333193302154,
|
| 34 |
+
"val_loss": 4.071026241779327,
|
| 35 |
+
"lr": 0.000578237838428388
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"step": 1500,
|
| 39 |
+
"tokens": 98304000,
|
| 40 |
+
"train_loss": 4.0969563086032865,
|
| 41 |
+
"val_loss": 3.949920582771301,
|
| 42 |
+
"lr": 0.0005638697001263056
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"step": 1750,
|
| 46 |
+
"tokens": 114688000,
|
| 47 |
+
"train_loss": 4.008148776769638,
|
| 48 |
+
"val_loss": 3.8674664974212645,
|
| 49 |
+
"lr": 0.0005461916515079224
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"step": 2000,
|
| 53 |
+
"tokens": 131072000,
|
| 54 |
+
"train_loss": 3.9327919991016387,
|
| 55 |
+
"val_loss": 3.8020901441574098,
|
| 56 |
+
"lr": 0.0005254538864010942
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"step": 2250,
|
| 60 |
+
"tokens": 147456000,
|
| 61 |
+
"train_loss": 3.8751951913833618,
|
| 62 |
+
"val_loss": 3.7550268173217773,
|
| 63 |
+
"lr": 0.0005019499021790462
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"step": 2500,
|
| 67 |
+
"tokens": 163840000,
|
| 68 |
+
"train_loss": 3.832032141685486,
|
| 69 |
+
"val_loss": 3.711101257801056,
|
| 70 |
+
"lr": 0.00047601234595172375
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"step": 2750,
|
| 74 |
+
"tokens": 180224000,
|
| 75 |
+
"train_loss": 3.7913031642436983,
|
| 76 |
+
"val_loss": 3.678738808631897,
|
| 77 |
+
"lr": 0.0004480083066788768
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"step": 3000,
|
| 81 |
+
"tokens": 196608000,
|
| 82 |
+
"train_loss": 3.7590393693447113,
|
| 83 |
+
"val_loss": 3.6446539163589478,
|
| 84 |
+
"lr": 0.00041833411983464874
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"step": 3250,
|
| 88 |
+
"tokens": 212992000,
|
| 89 |
+
"train_loss": 3.7263370764255526,
|
| 90 |
+
"val_loss": 3.6173880100250244,
|
| 91 |
+
"lr": 0.00038740975815219176
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"step": 3500,
|
| 95 |
+
"tokens": 229376000,
|
| 96 |
+
"train_loss": 3.700405584335327,
|
| 97 |
+
"val_loss": 3.592095470428467,
|
| 98 |
+
"lr": 0.00035567288783496225
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"step": 3750,
|
| 102 |
+
"tokens": 245760000,
|
| 103 |
+
"train_loss": 3.674201312303543,
|
| 104 |
+
"val_loss": 3.5669969797134398,
|
| 105 |
+
"lr": 0.00032357267435592814
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"step": 4000,
|
| 109 |
+
"tokens": 262144000,
|
| 110 |
+
"train_loss": 3.6556076962947843,
|
| 111 |
+
"val_loss": 3.549611473083496,
|
| 112 |
+
"lr": 0.0002915634255099528
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"step": 4250,
|
| 116 |
+
"tokens": 278528000,
|
| 117 |
+
"train_loss": 3.6357395310401914,
|
| 118 |
+
"val_loss": 3.530433976650238,
|
| 119 |
+
"lr": 0.0002600981616879423
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"step": 4500,
|
| 123 |
+
"tokens": 294912000,
|
| 124 |
+
"train_loss": 3.617664345741272,
|
| 125 |
+
"val_loss": 3.51442095041275,
|
| 126 |
+
"lr": 0.00022962220437135994
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"step": 4750,
|
| 130 |
+
"tokens": 311296000,
|
| 131 |
+
"train_loss": 3.6010146491527557,
|
| 132 |
+
"val_loss": 3.497687542438507,
|
| 133 |
+
"lr": 0.00020056687358783996
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"step": 5000,
|
| 137 |
+
"tokens": 327680000,
|
| 138 |
+
"train_loss": 3.5828054022789,
|
| 139 |
+
"val_loss": 3.484951174259186,
|
| 140 |
+
"lr": 0.000173343383526531
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"step": 5250,
|
| 144 |
+
"tokens": 344064000,
|
| 145 |
+
"train_loss": 3.5735193407535553,
|
| 146 |
+
"val_loss": 3.471825587749481,
|
| 147 |
+
"lr": 0.00014833702270728415
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"step": 5500,
|
| 151 |
+
"tokens": 360448000,
|
| 152 |
+
"train_loss": 3.557482803583145,
|
| 153 |
+
"val_loss": 3.4588160037994387,
|
| 154 |
+
"lr": 0.00012590170107056644
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"step": 5750,
|
| 158 |
+
"tokens": 376832000,
|
| 159 |
+
"train_loss": 3.547771404504776,
|
| 160 |
+
"val_loss": 3.449340450763702,
|
| 161 |
+
"lr": 0.00010635494116202957
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"step": 6000,
|
| 165 |
+
"tokens": 393216000,
|
| 166 |
+
"train_loss": 3.5455921549797056,
|
| 167 |
+
"val_loss": 3.441956579685211,
|
| 168 |
+
"lr": 8.997338430047869e-05
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"step": 6250,
|
| 172 |
+
"tokens": 409600000,
|
| 173 |
+
"train_loss": 3.5336987044811248,
|
| 174 |
+
"val_loss": 3.4355835199356077,
|
| 175 |
+
"lr": 7.698887532953336e-05
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"step": 6500,
|
| 179 |
+
"tokens": 425984000,
|
| 180 |
+
"train_loss": 3.528030238866806,
|
| 181 |
+
"val_loss": 3.428183174133301,
|
| 182 |
+
"lr": 6.758518136469335e-05
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"step": 6750,
|
| 186 |
+
"tokens": 442368000,
|
| 187 |
+
"train_loss": 3.519493240594864,
|
| 188 |
+
"val_loss": 3.4235813856124877,
|
| 189 |
+
"lr": 6.189539097471628e-05
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"step": 6998,
|
| 193 |
+
"tokens": 458620928,
|
| 194 |
+
"train_loss": 3.5168786826133727,
|
| 195 |
+
"val_loss": 3.419961929321289,
|
| 196 |
+
"lr": 6.0000030606162754e-05
|
| 197 |
+
}
|
| 198 |
+
]
|