Upload histories/C2_BytePairTokenizer_gelu_history.json with huggingface_hub
Browse files
histories/C2_BytePairTokenizer_gelu_history.json
ADDED
|
@@ -0,0 +1,182 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"epoch": 1,
|
| 4 |
+
"train_loss": 1.7624475087902762,
|
| 5 |
+
"val_loss": 1.1151122884842957,
|
| 6 |
+
"train_acc": 0.4803972833806818,
|
| 7 |
+
"val_acc": 0.6656418554476807,
|
| 8 |
+
"train_sec": 58.75240993499756,
|
| 9 |
+
"peak_gb": 0.944777728
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"epoch": 2,
|
| 13 |
+
"train_loss": 0.5108237018097531,
|
| 14 |
+
"val_loss": 0.1522014513595719,
|
| 15 |
+
"train_acc": 0.8382568359375,
|
| 16 |
+
"val_acc": 0.9471514293419633,
|
| 17 |
+
"train_sec": 58.77429413795471,
|
| 18 |
+
"peak_gb": 0.944777728
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 3,
|
| 22 |
+
"train_loss": 0.11274288888343356,
|
| 23 |
+
"val_loss": 0.051911772089932987,
|
| 24 |
+
"train_acc": 0.962689541903409,
|
| 25 |
+
"val_acc": 0.9846615426105717,
|
| 26 |
+
"train_sec": 57.8922758102417,
|
| 27 |
+
"peak_gb": 0.944777728
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 4,
|
| 31 |
+
"train_loss": 0.0385091987201436,
|
| 32 |
+
"val_loss": 0.013644262950991993,
|
| 33 |
+
"train_acc": 0.9884419389204545,
|
| 34 |
+
"val_acc": 0.9961771844660194,
|
| 35 |
+
"train_sec": 58.206791400909424,
|
| 36 |
+
"peak_gb": 0.944777728
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 5,
|
| 40 |
+
"train_loss": 0.020106018228583378,
|
| 41 |
+
"val_loss": 0.009553470273884947,
|
| 42 |
+
"train_acc": 0.994150390625,
|
| 43 |
+
"val_acc": 0.9972761596548004,
|
| 44 |
+
"train_sec": 58.04266905784607,
|
| 45 |
+
"peak_gb": 0.944777728
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 6,
|
| 49 |
+
"train_loss": 0.013522688390221447,
|
| 50 |
+
"val_loss": 0.006664294412735339,
|
| 51 |
+
"train_acc": 0.9960284978693181,
|
| 52 |
+
"val_acc": 0.9981425296655879,
|
| 53 |
+
"train_sec": 57.91779971122742,
|
| 54 |
+
"peak_gb": 0.944777728
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 7,
|
| 58 |
+
"train_loss": 0.010302849772005258,
|
| 59 |
+
"val_loss": 0.005835580717603462,
|
| 60 |
+
"train_acc": 0.9970348011363637,
|
| 61 |
+
"val_acc": 0.9984155879180151,
|
| 62 |
+
"train_sec": 59.05021333694458,
|
| 63 |
+
"peak_gb": 0.944777728
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 8,
|
| 67 |
+
"train_loss": 0.008762956866253675,
|
| 68 |
+
"val_loss": 0.005082864396474018,
|
| 69 |
+
"train_acc": 0.997529296875,
|
| 70 |
+
"val_acc": 0.9986212243797196,
|
| 71 |
+
"train_sec": 59.73148488998413,
|
| 72 |
+
"peak_gb": 0.944777728
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 9,
|
| 76 |
+
"train_loss": 0.007837512882595713,
|
| 77 |
+
"val_loss": 0.002842013011339895,
|
| 78 |
+
"train_acc": 0.9978173828125,
|
| 79 |
+
"val_acc": 0.9993055555555556,
|
| 80 |
+
"train_sec": 57.72091603279114,
|
| 81 |
+
"peak_gb": 0.944777728
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 10,
|
| 85 |
+
"train_loss": 0.005858481948380358,
|
| 86 |
+
"val_loss": 0.012970390834844598,
|
| 87 |
+
"train_acc": 0.998343838778409,
|
| 88 |
+
"val_acc": 0.9970974919093851,
|
| 89 |
+
"train_sec": 57.91335153579712,
|
| 90 |
+
"peak_gb": 0.944777728
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 11,
|
| 94 |
+
"train_loss": 0.0055208700603212825,
|
| 95 |
+
"val_loss": 0.0028324383224070295,
|
| 96 |
+
"train_acc": 0.9984592507102272,
|
| 97 |
+
"val_acc": 0.9993358953613808,
|
| 98 |
+
"train_sec": 58.04611110687256,
|
| 99 |
+
"peak_gb": 0.944777728
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 12,
|
| 103 |
+
"train_loss": 0.005396796792470427,
|
| 104 |
+
"val_loss": 0.0024209556266757888,
|
| 105 |
+
"train_acc": 0.9985258345170455,
|
| 106 |
+
"val_acc": 0.9994403991370011,
|
| 107 |
+
"train_sec": 58.18194007873535,
|
| 108 |
+
"peak_gb": 0.944777728
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 13,
|
| 112 |
+
"train_loss": 0.0042736622006420725,
|
| 113 |
+
"val_loss": 0.0029658884202866556,
|
| 114 |
+
"train_acc": 0.9988432173295455,
|
| 115 |
+
"val_acc": 0.9992482470334412,
|
| 116 |
+
"train_sec": 59.30978345870972,
|
| 117 |
+
"peak_gb": 0.944777728
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 14,
|
| 121 |
+
"train_loss": 0.003965111205735328,
|
| 122 |
+
"val_loss": 0.0027619202999097214,
|
| 123 |
+
"train_acc": 0.9989000355113636,
|
| 124 |
+
"val_acc": 0.9993864617044229,
|
| 125 |
+
"train_sec": 58.864203691482544,
|
| 126 |
+
"peak_gb": 0.944777728
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 15,
|
| 130 |
+
"train_loss": 0.0035830637618015646,
|
| 131 |
+
"val_loss": 0.001652343539179701,
|
| 132 |
+
"train_acc": 0.9990287642045454,
|
| 133 |
+
"val_acc": 0.9995685005393743,
|
| 134 |
+
"train_sec": 58.31713509559631,
|
| 135 |
+
"peak_gb": 0.944777728
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 16,
|
| 139 |
+
"train_loss": 0.003869527549885044,
|
| 140 |
+
"val_loss": 0.0025518046446875975,
|
| 141 |
+
"train_acc": 0.9989417613636363,
|
| 142 |
+
"val_acc": 0.9994977076591154,
|
| 143 |
+
"train_sec": 59.23481750488281,
|
| 144 |
+
"peak_gb": 0.944777728
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 17,
|
| 148 |
+
"train_loss": 0.003034974182453218,
|
| 149 |
+
"val_loss": 0.002608602105912641,
|
| 150 |
+
"train_acc": 0.99919921875,
|
| 151 |
+
"val_acc": 0.9993561218985976,
|
| 152 |
+
"train_sec": 59.45095705986023,
|
| 153 |
+
"peak_gb": 0.944777728
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 18,
|
| 157 |
+
"train_loss": 0.0029995176076508044,
|
| 158 |
+
"val_loss": 0.003107864049538576,
|
| 159 |
+
"train_acc": 0.9991916725852272,
|
| 160 |
+
"val_acc": 0.9993628640776699,
|
| 161 |
+
"train_sec": 59.402247190475464,
|
| 162 |
+
"peak_gb": 0.944777728
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 19,
|
| 166 |
+
"train_loss": 0.0031176625257119833,
|
| 167 |
+
"val_loss": 0.0019467750582050334,
|
| 168 |
+
"train_acc": 0.9991366299715909,
|
| 169 |
+
"val_acc": 0.999558387270766,
|
| 170 |
+
"train_sec": 59.30274057388306,
|
| 171 |
+
"peak_gb": 0.944777728
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 20,
|
| 175 |
+
"train_loss": 0.002729102304438129,
|
| 176 |
+
"val_loss": 0.001676761161761023,
|
| 177 |
+
"train_acc": 0.9992857776988636,
|
| 178 |
+
"val_acc": 0.9995887270765912,
|
| 179 |
+
"train_sec": 59.69143867492676,
|
| 180 |
+
"peak_gb": 0.944777728
|
| 181 |
+
}
|
| 182 |
+
]
|