Add pretrained/nepali/history.json
Browse files- pretrained/nepali/history.json +177 -0
pretrained/nepali/history.json
ADDED
|
@@ -0,0 +1,177 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"step": 250,
|
| 4 |
+
"tokens": 16384000,
|
| 5 |
+
"train_loss": 7.154766367912292,
|
| 6 |
+
"val_loss": 6.089403772354126,
|
| 7 |
+
"lr": 0.000375
|
| 8 |
+
},
|
| 9 |
+
{
|
| 10 |
+
"step": 500,
|
| 11 |
+
"tokens": 32768000,
|
| 12 |
+
"train_loss": 5.719772349834442,
|
| 13 |
+
"val_loss": 5.261052489280701,
|
| 14 |
+
"lr": 0.0005995865005315875
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"step": 750,
|
| 18 |
+
"tokens": 49152000,
|
| 19 |
+
"train_loss": 5.116953732967377,
|
| 20 |
+
"val_loss": 4.854724955558777,
|
| 21 |
+
"lr": 0.0005948762502592904
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"step": 1000,
|
| 25 |
+
"tokens": 65536000,
|
| 26 |
+
"train_loss": 4.817941832065582,
|
| 27 |
+
"val_loss": 4.607416772842408,
|
| 28 |
+
"lr": 0.0005849994869546038
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"step": 1250,
|
| 32 |
+
"tokens": 81920000,
|
| 33 |
+
"train_loss": 4.613338257312774,
|
| 34 |
+
"val_loss": 4.4259114265441895,
|
| 35 |
+
"lr": 0.0005701488606806481
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"step": 1500,
|
| 39 |
+
"tokens": 98304000,
|
| 40 |
+
"train_loss": 4.473004261016846,
|
| 41 |
+
"val_loss": 4.303855562210083,
|
| 42 |
+
"lr": 0.0005506140386088227
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"step": 1750,
|
| 46 |
+
"tokens": 114688000,
|
| 47 |
+
"train_loss": 4.3708488221168516,
|
| 48 |
+
"val_loss": 4.22306125164032,
|
| 49 |
+
"lr": 0.0005267760549493322
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"step": 2000,
|
| 53 |
+
"tokens": 131072000,
|
| 54 |
+
"train_loss": 4.304188202381134,
|
| 55 |
+
"val_loss": 4.150909876823425,
|
| 56 |
+
"lr": 0.0004990998787325145
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"step": 2250,
|
| 60 |
+
"tokens": 147456000,
|
| 61 |
+
"train_loss": 4.241593008518219,
|
| 62 |
+
"val_loss": 4.10257363319397,
|
| 63 |
+
"lr": 0.00046812534440925373
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"step": 2500,
|
| 67 |
+
"tokens": 163840000,
|
| 68 |
+
"train_loss": 4.197152866363526,
|
| 69 |
+
"val_loss": 4.0591519474983215,
|
| 70 |
+
"lr": 0.000434456622172599
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"step": 2750,
|
| 74 |
+
"tokens": 180224000,
|
| 75 |
+
"train_loss": 4.154413634300232,
|
| 76 |
+
"val_loss": 4.0196410655975345,
|
| 77 |
+
"lr": 0.0003987504333860145
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"step": 3000,
|
| 81 |
+
"tokens": 196608000,
|
| 82 |
+
"train_loss": 4.120177652835846,
|
| 83 |
+
"val_loss": 3.9881181001663206,
|
| 84 |
+
"lr": 0.0003617032409808589
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"step": 3250,
|
| 88 |
+
"tokens": 212992000,
|
| 89 |
+
"train_loss": 4.087694558620453,
|
| 90 |
+
"val_loss": 3.9599430918693543,
|
| 91 |
+
"lr": 0.00032403766467931306
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"step": 3500,
|
| 95 |
+
"tokens": 229376000,
|
| 96 |
+
"train_loss": 4.058541301012039,
|
| 97 |
+
"val_loss": 3.935786688327789,
|
| 98 |
+
"lr": 0.0002864883860190531
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"step": 3750,
|
| 102 |
+
"tokens": 245760000,
|
| 103 |
+
"train_loss": 4.040331825971603,
|
| 104 |
+
"val_loss": 3.912290930747986,
|
| 105 |
+
"lr": 0.0002497878181075778
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"step": 4000,
|
| 109 |
+
"tokens": 262144000,
|
| 110 |
+
"train_loss": 4.01366329908371,
|
| 111 |
+
"val_loss": 3.8883316278457642,
|
| 112 |
+
"lr": 0.00021465181962313854
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"step": 4250,
|
| 116 |
+
"tokens": 278528000,
|
| 117 |
+
"train_loss": 3.9925370585918425,
|
| 118 |
+
"val_loss": 3.8704324841499327,
|
| 119 |
+
"lr": 0.00018176573171617096
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"step": 4500,
|
| 123 |
+
"tokens": 294912000,
|
| 124 |
+
"train_loss": 3.9749942247867582,
|
| 125 |
+
"val_loss": 3.855111062526703,
|
| 126 |
+
"lr": 0.00015177101016683431
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"step": 4750,
|
| 130 |
+
"tokens": 311296000,
|
| 131 |
+
"train_loss": 3.9607299628257753,
|
| 132 |
+
"val_loss": 3.8400635838508608,
|
| 133 |
+
"lr": 0.00012525271354355745
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"step": 5000,
|
| 137 |
+
"tokens": 327680000,
|
| 138 |
+
"train_loss": 3.9453326466083527,
|
| 139 |
+
"val_loss": 3.827047574520111,
|
| 140 |
+
"lr": 0.00010272809141087024
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"step": 5250,
|
| 144 |
+
"tokens": 344064000,
|
| 145 |
+
"train_loss": 3.93602179646492,
|
| 146 |
+
"val_loss": 3.8175846219062803,
|
| 147 |
+
"lr": 8.46364951779092e-05
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"step": 5500,
|
| 151 |
+
"tokens": 360448000,
|
| 152 |
+
"train_loss": 3.924561816692352,
|
| 153 |
+
"val_loss": 3.8100233554840086,
|
| 154 |
+
"lr": 7.133080838037369e-05
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"step": 5750,
|
| 158 |
+
"tokens": 376832000,
|
| 159 |
+
"train_loss": 3.917643396139145,
|
| 160 |
+
"val_loss": 3.803208017349243,
|
| 161 |
+
"lr": 6.307056355157717e-05
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"step": 6000,
|
| 165 |
+
"tokens": 393216000,
|
| 166 |
+
"train_loss": 3.9141684684753417,
|
| 167 |
+
"val_loss": 3.797338342666626,
|
| 168 |
+
"lr": 6.001687994066689e-05
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"step": 6019,
|
| 172 |
+
"tokens": 394461184,
|
| 173 |
+
"train_loss": 3.913020513534546,
|
| 174 |
+
"val_loss": 3.7969972014427187,
|
| 175 |
+
"lr": 6.0000042200290274e-05
|
| 176 |
+
}
|
| 177 |
+
]
|