Upload logs/train_log_T5Gemma-2B_seed456.json with huggingface_hub
Browse files
logs/train_log_T5Gemma-2B_seed456.json
ADDED
|
@@ -0,0 +1,557 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"loss": 1.6751,
|
| 4 |
+
"grad_norm": 35.5,
|
| 5 |
+
"learning_rate": 1.8e-05,
|
| 6 |
+
"epoch": 0.06369426751592357,
|
| 7 |
+
"step": 10
|
| 8 |
+
},
|
| 9 |
+
{
|
| 10 |
+
"loss": 0.753,
|
| 11 |
+
"grad_norm": 52.5,
|
| 12 |
+
"learning_rate": 1.9993345687502566e-05,
|
| 13 |
+
"epoch": 0.12738853503184713,
|
| 14 |
+
"step": 20
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"loss": 0.5992,
|
| 18 |
+
"grad_norm": 30.25,
|
| 19 |
+
"learning_rate": 1.9970354495287275e-05,
|
| 20 |
+
"epoch": 0.1910828025477707,
|
| 21 |
+
"step": 30
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"loss": 0.5004,
|
| 25 |
+
"grad_norm": 28.625,
|
| 26 |
+
"learning_rate": 1.993098203681623e-05,
|
| 27 |
+
"epoch": 0.25477707006369427,
|
| 28 |
+
"step": 40
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"loss": 0.4852,
|
| 32 |
+
"grad_norm": 30.0,
|
| 33 |
+
"learning_rate": 1.987529300093532e-05,
|
| 34 |
+
"epoch": 0.3184713375796178,
|
| 35 |
+
"step": 50
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"loss": 0.5795,
|
| 39 |
+
"grad_norm": 18.875,
|
| 40 |
+
"learning_rate": 1.9803378884584266e-05,
|
| 41 |
+
"epoch": 0.3821656050955414,
|
| 42 |
+
"step": 60
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"loss": 0.5235,
|
| 46 |
+
"grad_norm": 30.375,
|
| 47 |
+
"learning_rate": 1.97153578424674e-05,
|
| 48 |
+
"epoch": 0.445859872611465,
|
| 49 |
+
"step": 70
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"loss": 0.4305,
|
| 53 |
+
"grad_norm": 21.375,
|
| 54 |
+
"learning_rate": 1.96113744929258e-05,
|
| 55 |
+
"epoch": 0.5095541401273885,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"loss": 0.5519,
|
| 60 |
+
"grad_norm": 29.375,
|
| 61 |
+
"learning_rate": 1.949159968032972e-05,
|
| 62 |
+
"epoch": 0.5732484076433121,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"loss": 0.4587,
|
| 67 |
+
"grad_norm": 24.0,
|
| 68 |
+
"learning_rate": 1.935623019438176e-05,
|
| 69 |
+
"epoch": 0.6369426751592356,
|
| 70 |
+
"step": 100
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"loss": 0.5376,
|
| 74 |
+
"grad_norm": 116.5,
|
| 75 |
+
"learning_rate": 1.9205488446791873e-05,
|
| 76 |
+
"epoch": 0.7006369426751592,
|
| 77 |
+
"step": 110
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"loss": 0.476,
|
| 81 |
+
"grad_norm": 16.875,
|
| 82 |
+
"learning_rate": 1.903962210585554e-05,
|
| 83 |
+
"epoch": 0.7643312101910829,
|
| 84 |
+
"step": 120
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"loss": 0.4523,
|
| 88 |
+
"grad_norm": 12.8125,
|
| 89 |
+
"learning_rate": 1.885890368953539e-05,
|
| 90 |
+
"epoch": 0.8280254777070064,
|
| 91 |
+
"step": 130
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"loss": 0.4212,
|
| 95 |
+
"grad_norm": 32.5,
|
| 96 |
+
"learning_rate": 1.8663630117714904e-05,
|
| 97 |
+
"epoch": 0.89171974522293,
|
| 98 |
+
"step": 140
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"loss": 0.4637,
|
| 102 |
+
"grad_norm": 14.625,
|
| 103 |
+
"learning_rate": 1.845412222435984e-05,
|
| 104 |
+
"epoch": 0.9554140127388535,
|
| 105 |
+
"step": 150
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"loss": 0.3916,
|
| 109 |
+
"grad_norm": 53.25,
|
| 110 |
+
"learning_rate": 1.823072423038886e-05,
|
| 111 |
+
"epoch": 1.019108280254777,
|
| 112 |
+
"step": 160
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"loss": 0.3953,
|
| 116 |
+
"grad_norm": 14.875,
|
| 117 |
+
"learning_rate": 1.7993803178119457e-05,
|
| 118 |
+
"epoch": 1.0828025477707006,
|
| 119 |
+
"step": 170
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"loss": 0.3769,
|
| 123 |
+
"grad_norm": 17.125,
|
| 124 |
+
"learning_rate": 1.7743748328218412e-05,
|
| 125 |
+
"epoch": 1.1464968152866242,
|
| 126 |
+
"step": 180
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"loss": 0.3165,
|
| 130 |
+
"grad_norm": 12.5,
|
| 131 |
+
"learning_rate": 1.7480970520147522e-05,
|
| 132 |
+
"epoch": 1.2101910828025477,
|
| 133 |
+
"step": 190
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"loss": 0.3242,
|
| 137 |
+
"grad_norm": 110.0,
|
| 138 |
+
"learning_rate": 1.7205901497155445e-05,
|
| 139 |
+
"epoch": 1.2738853503184713,
|
| 140 |
+
"step": 200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"loss": 0.387,
|
| 144 |
+
"grad_norm": 45.0,
|
| 145 |
+
"learning_rate": 1.691899319692469e-05,
|
| 146 |
+
"epoch": 1.3375796178343948,
|
| 147 |
+
"step": 210
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"loss": 0.3033,
|
| 151 |
+
"grad_norm": 5.0625,
|
| 152 |
+
"learning_rate": 1.6620717009039175e-05,
|
| 153 |
+
"epoch": 1.4012738853503186,
|
| 154 |
+
"step": 220
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"loss": 0.3157,
|
| 158 |
+
"grad_norm": 29.0,
|
| 159 |
+
"learning_rate": 1.631156300049238e-05,
|
| 160 |
+
"epoch": 1.4649681528662422,
|
| 161 |
+
"step": 230
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"loss": 0.254,
|
| 165 |
+
"grad_norm": 13.4375,
|
| 166 |
+
"learning_rate": 1.5992039110508586e-05,
|
| 167 |
+
"epoch": 1.5286624203821657,
|
| 168 |
+
"step": 240
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"loss": 0.3212,
|
| 172 |
+
"grad_norm": 25.625,
|
| 173 |
+
"learning_rate": 1.566267031600003e-05,
|
| 174 |
+
"epoch": 1.5923566878980893,
|
| 175 |
+
"step": 250
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"loss": 0.3061,
|
| 179 |
+
"grad_norm": 9.6875,
|
| 180 |
+
"learning_rate": 1.532399776903124e-05,
|
| 181 |
+
"epoch": 1.6560509554140128,
|
| 182 |
+
"step": 260
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"loss": 0.2912,
|
| 186 |
+
"grad_norm": 28.125,
|
| 187 |
+
"learning_rate": 1.4976577907707603e-05,
|
| 188 |
+
"epoch": 1.7197452229299364,
|
| 189 |
+
"step": 270
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"loss": 0.3149,
|
| 193 |
+
"grad_norm": 21.75,
|
| 194 |
+
"learning_rate": 1.4620981541949024e-05,
|
| 195 |
+
"epoch": 1.78343949044586,
|
| 196 |
+
"step": 280
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"loss": 0.2663,
|
| 200 |
+
"grad_norm": 8.625,
|
| 201 |
+
"learning_rate": 1.4257792915650728e-05,
|
| 202 |
+
"epoch": 1.8471337579617835,
|
| 203 |
+
"step": 290
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"loss": 0.272,
|
| 207 |
+
"grad_norm": 40.25,
|
| 208 |
+
"learning_rate": 1.3887608746772064e-05,
|
| 209 |
+
"epoch": 1.910828025477707,
|
| 210 |
+
"step": 300
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"loss": 0.2492,
|
| 214 |
+
"grad_norm": 12.6875,
|
| 215 |
+
"learning_rate": 1.3511037246930476e-05,
|
| 216 |
+
"epoch": 1.9745222929936306,
|
| 217 |
+
"step": 310
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"loss": 0.2056,
|
| 221 |
+
"grad_norm": 11.1875,
|
| 222 |
+
"learning_rate": 1.312869712211141e-05,
|
| 223 |
+
"epoch": 2.038216560509554,
|
| 224 |
+
"step": 320
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"loss": 0.1938,
|
| 228 |
+
"grad_norm": 18.25,
|
| 229 |
+
"learning_rate": 1.2741216556136e-05,
|
| 230 |
+
"epoch": 2.1019108280254777,
|
| 231 |
+
"step": 330
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"loss": 0.1736,
|
| 235 |
+
"grad_norm": 8.8125,
|
| 236 |
+
"learning_rate": 1.2349232178556703e-05,
|
| 237 |
+
"epoch": 2.1656050955414012,
|
| 238 |
+
"step": 340
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"loss": 0.1291,
|
| 242 |
+
"grad_norm": 9.875,
|
| 243 |
+
"learning_rate": 1.1953388018676618e-05,
|
| 244 |
+
"epoch": 2.229299363057325,
|
| 245 |
+
"step": 350
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"loss": 0.2206,
|
| 249 |
+
"grad_norm": 7.8125,
|
| 250 |
+
"learning_rate": 1.1554334447411019e-05,
|
| 251 |
+
"epoch": 2.2929936305732483,
|
| 252 |
+
"step": 360
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"loss": 0.1709,
|
| 256 |
+
"grad_norm": 5.0,
|
| 257 |
+
"learning_rate": 1.1152727108729675e-05,
|
| 258 |
+
"epoch": 2.356687898089172,
|
| 259 |
+
"step": 370
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"loss": 0.1411,
|
| 263 |
+
"grad_norm": 8.25,
|
| 264 |
+
"learning_rate": 1.0749225842435498e-05,
|
| 265 |
+
"epoch": 2.4203821656050954,
|
| 266 |
+
"step": 380
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"loss": 0.1655,
|
| 270 |
+
"grad_norm": 4.8125,
|
| 271 |
+
"learning_rate": 1.0344493600049509e-05,
|
| 272 |
+
"epoch": 2.484076433121019,
|
| 273 |
+
"step": 390
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"loss": 0.1753,
|
| 277 |
+
"grad_norm": 8.4375,
|
| 278 |
+
"learning_rate": 9.939195355583199e-06,
|
| 279 |
+
"epoch": 2.5477707006369426,
|
| 280 |
+
"step": 400
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"loss": 0.113,
|
| 284 |
+
"grad_norm": 3.15625,
|
| 285 |
+
"learning_rate": 9.53399701298801e-06,
|
| 286 |
+
"epoch": 2.611464968152866,
|
| 287 |
+
"step": 410
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"loss": 0.1698,
|
| 291 |
+
"grad_norm": 8.3125,
|
| 292 |
+
"learning_rate": 9.129564312076861e-06,
|
| 293 |
+
"epoch": 2.6751592356687897,
|
| 294 |
+
"step": 420
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"loss": 0.1449,
|
| 298 |
+
"grad_norm": 8.125,
|
| 299 |
+
"learning_rate": 8.726561734715388e-06,
|
| 300 |
+
"epoch": 2.738853503184713,
|
| 301 |
+
"step": 430
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"loss": 0.1286,
|
| 305 |
+
"grad_norm": 5.9375,
|
| 306 |
+
"learning_rate": 8.325651413080003e-06,
|
| 307 |
+
"epoch": 2.802547770700637,
|
| 308 |
+
"step": 440
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"loss": 0.1644,
|
| 312 |
+
"grad_norm": 18.75,
|
| 313 |
+
"learning_rate": 7.927492041776452e-06,
|
| 314 |
+
"epoch": 2.8662420382165603,
|
| 315 |
+
"step": 450
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"loss": 0.1375,
|
| 319 |
+
"grad_norm": 12.5,
|
| 320 |
+
"learning_rate": 7.532737795606299e-06,
|
| 321 |
+
"epoch": 2.9299363057324843,
|
| 322 |
+
"step": 460
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"loss": 0.1238,
|
| 326 |
+
"grad_norm": 5.9375,
|
| 327 |
+
"learning_rate": 7.142037254759469e-06,
|
| 328 |
+
"epoch": 2.9936305732484074,
|
| 329 |
+
"step": 470
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"loss": 0.0463,
|
| 333 |
+
"grad_norm": 1.390625,
|
| 334 |
+
"learning_rate": 6.756032339198675e-06,
|
| 335 |
+
"epoch": 3.0573248407643314,
|
| 336 |
+
"step": 480
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"loss": 0.0913,
|
| 340 |
+
"grad_norm": 9.75,
|
| 341 |
+
"learning_rate": 6.375357253986614e-06,
|
| 342 |
+
"epoch": 3.121019108280255,
|
| 343 |
+
"step": 490
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"loss": 0.0608,
|
| 347 |
+
"grad_norm": 4.375,
|
| 348 |
+
"learning_rate": 6.000637447288719e-06,
|
| 349 |
+
"epoch": 3.1847133757961785,
|
| 350 |
+
"step": 500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"loss": 0.0731,
|
| 354 |
+
"grad_norm": 3.203125,
|
| 355 |
+
"learning_rate": 5.6324885827634665e-06,
|
| 356 |
+
"epoch": 3.248407643312102,
|
| 357 |
+
"step": 510
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"loss": 0.0545,
|
| 361 |
+
"grad_norm": 2.140625,
|
| 362 |
+
"learning_rate": 5.271515528028592e-06,
|
| 363 |
+
"epoch": 3.3121019108280256,
|
| 364 |
+
"step": 520
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"loss": 0.0846,
|
| 368 |
+
"grad_norm": 2.765625,
|
| 369 |
+
"learning_rate": 4.91831136086518e-06,
|
| 370 |
+
"epoch": 3.375796178343949,
|
| 371 |
+
"step": 530
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"loss": 0.0621,
|
| 375 |
+
"grad_norm": 3.296875,
|
| 376 |
+
"learning_rate": 4.573456394792446e-06,
|
| 377 |
+
"epoch": 3.4394904458598727,
|
| 378 |
+
"step": 540
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"loss": 0.0646,
|
| 382 |
+
"grad_norm": 6.75,
|
| 383 |
+
"learning_rate": 4.237517225614138e-06,
|
| 384 |
+
"epoch": 3.5031847133757963,
|
| 385 |
+
"step": 550
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"loss": 0.0716,
|
| 389 |
+
"grad_norm": 8.625,
|
| 390 |
+
"learning_rate": 3.911045800503101e-06,
|
| 391 |
+
"epoch": 3.56687898089172,
|
| 392 |
+
"step": 560
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"loss": 0.0643,
|
| 396 |
+
"grad_norm": 7.40625,
|
| 397 |
+
"learning_rate": 3.5945785111535547e-06,
|
| 398 |
+
"epoch": 3.6305732484076434,
|
| 399 |
+
"step": 570
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"loss": 0.0596,
|
| 403 |
+
"grad_norm": 2.953125,
|
| 404 |
+
"learning_rate": 3.2886353124909267e-06,
|
| 405 |
+
"epoch": 3.694267515923567,
|
| 406 |
+
"step": 580
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"loss": 0.0705,
|
| 410 |
+
"grad_norm": 9.375,
|
| 411 |
+
"learning_rate": 2.9937188683872796e-06,
|
| 412 |
+
"epoch": 3.7579617834394905,
|
| 413 |
+
"step": 590
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"loss": 0.0563,
|
| 417 |
+
"grad_norm": 3.359375,
|
| 418 |
+
"learning_rate": 2.7103137257858867e-06,
|
| 419 |
+
"epoch": 3.821656050955414,
|
| 420 |
+
"step": 600
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"loss": 0.0605,
|
| 424 |
+
"grad_norm": 2.390625,
|
| 425 |
+
"learning_rate": 2.438885518591878e-06,
|
| 426 |
+
"epoch": 3.8853503184713376,
|
| 427 |
+
"step": 610
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"loss": 0.053,
|
| 431 |
+
"grad_norm": 1.7421875,
|
| 432 |
+
"learning_rate": 2.179880202636966e-06,
|
| 433 |
+
"epoch": 3.949044585987261,
|
| 434 |
+
"step": 620
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"loss": 0.0379,
|
| 438 |
+
"grad_norm": 0.462890625,
|
| 439 |
+
"learning_rate": 1.9337233229751686e-06,
|
| 440 |
+
"epoch": 4.012738853503185,
|
| 441 |
+
"step": 630
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"loss": 0.0183,
|
| 445 |
+
"grad_norm": 1.46875,
|
| 446 |
+
"learning_rate": 1.7008193147134167e-06,
|
| 447 |
+
"epoch": 4.076433121019108,
|
| 448 |
+
"step": 640
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"loss": 0.0349,
|
| 452 |
+
"grad_norm": 4.9375,
|
| 453 |
+
"learning_rate": 1.4815508385257316e-06,
|
| 454 |
+
"epoch": 4.140127388535032,
|
| 455 |
+
"step": 650
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"loss": 0.0266,
|
| 459 |
+
"grad_norm": 0.6640625,
|
| 460 |
+
"learning_rate": 1.2762781519427324e-06,
|
| 461 |
+
"epoch": 4.203821656050955,
|
| 462 |
+
"step": 660
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"loss": 0.0173,
|
| 466 |
+
"grad_norm": 1.9765625,
|
| 467 |
+
"learning_rate": 1.0853385174494758e-06,
|
| 468 |
+
"epoch": 4.267515923566879,
|
| 469 |
+
"step": 670
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"loss": 0.022,
|
| 473 |
+
"grad_norm": 5.21875,
|
| 474 |
+
"learning_rate": 9.090456483640686e-07,
|
| 475 |
+
"epoch": 4.3312101910828025,
|
| 476 |
+
"step": 680
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"loss": 0.0218,
|
| 480 |
+
"grad_norm": 0.291015625,
|
| 481 |
+
"learning_rate": 7.476891934075126e-07,
|
| 482 |
+
"epoch": 4.3949044585987265,
|
| 483 |
+
"step": 690
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"loss": 0.0392,
|
| 487 |
+
"grad_norm": 0.2236328125,
|
| 488 |
+
"learning_rate": 6.015342608116092e-07,
|
| 489 |
+
"epoch": 4.45859872611465,
|
| 490 |
+
"step": 700
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"loss": 0.0322,
|
| 494 |
+
"grad_norm": 3.90625,
|
| 495 |
+
"learning_rate": 4.708209827468457e-07,
|
| 496 |
+
"epoch": 4.522292993630574,
|
| 497 |
+
"step": 710
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"loss": 0.0242,
|
| 501 |
+
"grad_norm": 0.130859375,
|
| 502 |
+
"learning_rate": 3.5576412078585755e-07,
|
| 503 |
+
"epoch": 4.585987261146497,
|
| 504 |
+
"step": 720
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"loss": 0.0169,
|
| 508 |
+
"grad_norm": 2.515625,
|
| 509 |
+
"learning_rate": 2.565527130507295e-07,
|
| 510 |
+
"epoch": 4.649681528662421,
|
| 511 |
+
"step": 730
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"loss": 0.0183,
|
| 515 |
+
"grad_norm": 5.9375,
|
| 516 |
+
"learning_rate": 1.7334976362386458e-07,
|
| 517 |
+
"epoch": 4.713375796178344,
|
| 518 |
+
"step": 740
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"loss": 0.0237,
|
| 522 |
+
"grad_norm": 11.4375,
|
| 523 |
+
"learning_rate": 1.0629197473269736e-07,
|
| 524 |
+
"epoch": 4.777070063694268,
|
| 525 |
+
"step": 750
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"loss": 0.03,
|
| 529 |
+
"grad_norm": 6.1875,
|
| 530 |
+
"learning_rate": 5.548952214829762e-08,
|
| 531 |
+
"epoch": 4.840764331210191,
|
| 532 |
+
"step": 760
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"loss": 0.0222,
|
| 536 |
+
"grad_norm": 1.15625,
|
| 537 |
+
"learning_rate": 2.1025874166864523e-08,
|
| 538 |
+
"epoch": 4.904458598726115,
|
| 539 |
+
"step": 770
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"loss": 0.0305,
|
| 543 |
+
"grad_norm": 1.0859375,
|
| 544 |
+
"learning_rate": 2.9576544715370636e-09,
|
| 545 |
+
"epoch": 4.968152866242038,
|
| 546 |
+
"step": 780
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"train_runtime": 1208.3302,
|
| 550 |
+
"train_samples_per_second": 5.197,
|
| 551 |
+
"train_steps_per_second": 0.65,
|
| 552 |
+
"total_flos": 5599444378429440.0,
|
| 553 |
+
"train_loss": 0.22870425369329514,
|
| 554 |
+
"epoch": 5.0,
|
| 555 |
+
"step": 785
|
| 556 |
+
}
|
| 557 |
+
]
|