Update log_history.json
Browse files- log_history.json +429 -0
log_history.json
ADDED
|
@@ -0,0 +1,429 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"loss": 5.60179638671875,
|
| 4 |
+
"learning_rate": 4.9189600959512466e-05,
|
| 5 |
+
"epoch": 0.048623942429252165,
|
| 6 |
+
"total_flos": 127437846067872,
|
| 7 |
+
"step": 500
|
| 8 |
+
},
|
| 9 |
+
{
|
| 10 |
+
"loss": 4.56484326171875,
|
| 11 |
+
"learning_rate": 4.837920191902493e-05,
|
| 12 |
+
"epoch": 0.09724788485850433,
|
| 13 |
+
"total_flos": 258224201311392,
|
| 14 |
+
"step": 1000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"loss": 4.172212890625,
|
| 18 |
+
"learning_rate": 4.756880287853739e-05,
|
| 19 |
+
"epoch": 0.1458718272877565,
|
| 20 |
+
"total_flos": 389157743771424,
|
| 21 |
+
"step": 1500
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"loss": 4.0773701171875,
|
| 25 |
+
"learning_rate": 4.6758403838049856e-05,
|
| 26 |
+
"epoch": 0.19449576971700866,
|
| 27 |
+
"total_flos": 515696696481312,
|
| 28 |
+
"step": 2000
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"loss": 3.90251953125,
|
| 32 |
+
"learning_rate": 4.594800479756232e-05,
|
| 33 |
+
"epoch": 0.24311971214626082,
|
| 34 |
+
"total_flos": 643145055921792,
|
| 35 |
+
"step": 2500
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"loss": 3.74609375,
|
| 39 |
+
"learning_rate": 4.513760575707479e-05,
|
| 40 |
+
"epoch": 0.291743654575513,
|
| 41 |
+
"total_flos": 771103313933760,
|
| 42 |
+
"step": 3000
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"loss": 3.63271875,
|
| 46 |
+
"learning_rate": 4.432720671658725e-05,
|
| 47 |
+
"epoch": 0.34036759700476515,
|
| 48 |
+
"total_flos": 899392743182880,
|
| 49 |
+
"step": 3500
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"loss": 3.617484375,
|
| 53 |
+
"learning_rate": 4.3516807676099715e-05,
|
| 54 |
+
"epoch": 0.3889915394340173,
|
| 55 |
+
"total_flos": 1031971628456064,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"loss": 3.59037109375,
|
| 60 |
+
"learning_rate": 4.270640863561218e-05,
|
| 61 |
+
"epoch": 0.4376154818632695,
|
| 62 |
+
"total_flos": 1159745902447392,
|
| 63 |
+
"step": 4500
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"loss": 3.5753984375,
|
| 67 |
+
"learning_rate": 4.189600959512464e-05,
|
| 68 |
+
"epoch": 0.48623942429252165,
|
| 69 |
+
"total_flos": 1288030075010208,
|
| 70 |
+
"step": 5000
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"loss": 3.5130390625,
|
| 74 |
+
"learning_rate": 4.1085610554637105e-05,
|
| 75 |
+
"epoch": 0.5348633667217738,
|
| 76 |
+
"total_flos": 1414432353876192,
|
| 77 |
+
"step": 5500
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"loss": 3.469171875,
|
| 81 |
+
"learning_rate": 4.027521151414957e-05,
|
| 82 |
+
"epoch": 0.583487309151026,
|
| 83 |
+
"total_flos": 1539068386144032,
|
| 84 |
+
"step": 6000
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"loss": 3.4336640625,
|
| 88 |
+
"learning_rate": 3.946481247366203e-05,
|
| 89 |
+
"epoch": 0.6321112515802781,
|
| 90 |
+
"total_flos": 1667946564259200,
|
| 91 |
+
"step": 6500
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"loss": 3.39712109375,
|
| 95 |
+
"learning_rate": 3.8654413433174495e-05,
|
| 96 |
+
"epoch": 0.6807351940095303,
|
| 97 |
+
"total_flos": 1794422436733440,
|
| 98 |
+
"step": 7000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"loss": 3.3770703125,
|
| 102 |
+
"learning_rate": 3.784401439268696e-05,
|
| 103 |
+
"epoch": 0.7293591364387825,
|
| 104 |
+
"total_flos": 1922738149414080,
|
| 105 |
+
"step": 7500
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"loss": 3.3730703125,
|
| 109 |
+
"learning_rate": 3.703361535219942e-05,
|
| 110 |
+
"epoch": 0.7779830788680346,
|
| 111 |
+
"total_flos": 2052888445614816,
|
| 112 |
+
"step": 8000
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"loss": 3.310546875,
|
| 116 |
+
"learning_rate": 3.622321631171189e-05,
|
| 117 |
+
"epoch": 0.8266070212972868,
|
| 118 |
+
"total_flos": 2186350454187072,
|
| 119 |
+
"step": 8500
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"loss": 3.260265625,
|
| 123 |
+
"learning_rate": 3.5412817271224355e-05,
|
| 124 |
+
"epoch": 0.875230963726539,
|
| 125 |
+
"total_flos": 2314024851138624,
|
| 126 |
+
"step": 9000
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"loss": 3.2306796875,
|
| 130 |
+
"learning_rate": 3.460241823073682e-05,
|
| 131 |
+
"epoch": 0.9238549061557911,
|
| 132 |
+
"total_flos": 2442719045233152,
|
| 133 |
+
"step": 9500
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"loss": 3.2746796875,
|
| 137 |
+
"learning_rate": 3.379201919024928e-05,
|
| 138 |
+
"epoch": 0.9724788485850433,
|
| 139 |
+
"total_flos": 2569378901728032,
|
| 140 |
+
"step": 10000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"loss": 3.1525703125,
|
| 144 |
+
"learning_rate": 3.2981620149761745e-05,
|
| 145 |
+
"epoch": 1.0211027910142954,
|
| 146 |
+
"total_flos": 2698954904950056,
|
| 147 |
+
"step": 10500
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"loss": 3.2410859375,
|
| 151 |
+
"learning_rate": 3.217122110927421e-05,
|
| 152 |
+
"epoch": 1.0697267334435476,
|
| 153 |
+
"total_flos": 2828821340090376,
|
| 154 |
+
"step": 11000
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"loss": 3.0903984375,
|
| 158 |
+
"learning_rate": 3.136082206878667e-05,
|
| 159 |
+
"epoch": 1.1183506758727997,
|
| 160 |
+
"total_flos": 2958682518544392,
|
| 161 |
+
"step": 11500
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"loss": 3.08865625,
|
| 165 |
+
"learning_rate": 3.0550423028299135e-05,
|
| 166 |
+
"epoch": 1.166974618302052,
|
| 167 |
+
"total_flos": 3088039055113224,
|
| 168 |
+
"step": 12000
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"loss": 3.1113671875,
|
| 172 |
+
"learning_rate": 2.9740023987811598e-05,
|
| 173 |
+
"epoch": 1.215598560731304,
|
| 174 |
+
"total_flos": 3214535954332680,
|
| 175 |
+
"step": 12500
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"loss": 3.088109375,
|
| 179 |
+
"learning_rate": 2.8929624947324065e-05,
|
| 180 |
+
"epoch": 1.2642225031605563,
|
| 181 |
+
"total_flos": 3340123446821544,
|
| 182 |
+
"step": 13000
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"loss": 3.03825,
|
| 186 |
+
"learning_rate": 2.8119225906836528e-05,
|
| 187 |
+
"epoch": 1.3128464455898083,
|
| 188 |
+
"total_flos": 3469900518294696,
|
| 189 |
+
"step": 13500
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"loss": 3.1165703125,
|
| 193 |
+
"learning_rate": 2.730882686634899e-05,
|
| 194 |
+
"epoch": 1.3614703880190606,
|
| 195 |
+
"total_flos": 3597664278913416,
|
| 196 |
+
"step": 14000
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"loss": 3.0786875,
|
| 200 |
+
"learning_rate": 2.6498427825861454e-05,
|
| 201 |
+
"epoch": 1.4100943304483127,
|
| 202 |
+
"total_flos": 3726390013125768,
|
| 203 |
+
"step": 14500
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"loss": 3.0769453125,
|
| 207 |
+
"learning_rate": 2.5688028785373918e-05,
|
| 208 |
+
"epoch": 1.458718272877565,
|
| 209 |
+
"total_flos": 3854211597293832,
|
| 210 |
+
"step": 15000
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"loss": 2.9751015625,
|
| 214 |
+
"learning_rate": 2.487762974488638e-05,
|
| 215 |
+
"epoch": 1.5073422153068172,
|
| 216 |
+
"total_flos": 3979215597602952,
|
| 217 |
+
"step": 15500
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"loss": 3.004953125,
|
| 221 |
+
"learning_rate": 2.4067230704398848e-05,
|
| 222 |
+
"epoch": 1.5559661577360693,
|
| 223 |
+
"total_flos": 4105360298840040,
|
| 224 |
+
"step": 16000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"loss": 3.0202265625,
|
| 228 |
+
"learning_rate": 2.325683166391131e-05,
|
| 229 |
+
"epoch": 1.6045901001653213,
|
| 230 |
+
"total_flos": 4232114775688392,
|
| 231 |
+
"step": 16500
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"loss": 2.967203125,
|
| 235 |
+
"learning_rate": 2.2446432623423774e-05,
|
| 236 |
+
"epoch": 1.6532140425945736,
|
| 237 |
+
"total_flos": 4357849455393768,
|
| 238 |
+
"step": 17000
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"loss": 2.95409375,
|
| 242 |
+
"learning_rate": 2.1636033582936237e-05,
|
| 243 |
+
"epoch": 1.7018379850238259,
|
| 244 |
+
"total_flos": 4487106114922824,
|
| 245 |
+
"step": 17500
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"loss": 3.068265625,
|
| 249 |
+
"learning_rate": 2.08256345424487e-05,
|
| 250 |
+
"epoch": 1.750461927453078,
|
| 251 |
+
"total_flos": 4614885645600456,
|
| 252 |
+
"step": 18000
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"loss": 2.992953125,
|
| 256 |
+
"learning_rate": 2.0015235501961167e-05,
|
| 257 |
+
"epoch": 1.79908586988233,
|
| 258 |
+
"total_flos": 4744000374599304,
|
| 259 |
+
"step": 18500
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"loss": 2.9246953125,
|
| 263 |
+
"learning_rate": 1.920483646147363e-05,
|
| 264 |
+
"epoch": 1.8477098123115823,
|
| 265 |
+
"total_flos": 4872962659695336,
|
| 266 |
+
"step": 19000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"loss": 2.980765625,
|
| 270 |
+
"learning_rate": 1.8394437420986094e-05,
|
| 271 |
+
"epoch": 1.8963337547408345,
|
| 272 |
+
"total_flos": 5000447815939944,
|
| 273 |
+
"step": 19500
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"loss": 2.95859375,
|
| 277 |
+
"learning_rate": 1.7584038380498557e-05,
|
| 278 |
+
"epoch": 1.9449576971700866,
|
| 279 |
+
"total_flos": 5130419384806344,
|
| 280 |
+
"step": 20000
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"loss": 2.951875,
|
| 284 |
+
"learning_rate": 1.677363934001102e-05,
|
| 285 |
+
"epoch": 1.9935816395993387,
|
| 286 |
+
"total_flos": 5260301590005576,
|
| 287 |
+
"step": 20500
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"loss": 2.90753125,
|
| 291 |
+
"learning_rate": 1.5963240299523484e-05,
|
| 292 |
+
"epoch": 2.0422055820285907,
|
| 293 |
+
"total_flos": 5386174914812220,
|
| 294 |
+
"step": 21000
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"loss": 2.8725,
|
| 298 |
+
"learning_rate": 1.5152841259035949e-05,
|
| 299 |
+
"epoch": 2.090829524457843,
|
| 300 |
+
"total_flos": 5515216050202812,
|
| 301 |
+
"step": 21500
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"loss": 2.813546875,
|
| 305 |
+
"learning_rate": 1.4342442218548414e-05,
|
| 306 |
+
"epoch": 2.1394534668870953,
|
| 307 |
+
"total_flos": 5644814394341628,
|
| 308 |
+
"step": 22000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"loss": 2.867796875,
|
| 312 |
+
"learning_rate": 1.3532043178060877e-05,
|
| 313 |
+
"epoch": 2.1880774093163473,
|
| 314 |
+
"total_flos": 5770811908362204,
|
| 315 |
+
"step": 22500
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"loss": 2.797328125,
|
| 319 |
+
"learning_rate": 1.272164413757334e-05,
|
| 320 |
+
"epoch": 2.2367013517455994,
|
| 321 |
+
"total_flos": 5901487873193340,
|
| 322 |
+
"step": 23000
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"loss": 2.8206875,
|
| 326 |
+
"learning_rate": 1.1911245097085805e-05,
|
| 327 |
+
"epoch": 2.285325294174852,
|
| 328 |
+
"total_flos": 6032090244416220,
|
| 329 |
+
"step": 23500
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"loss": 2.835703125,
|
| 333 |
+
"learning_rate": 1.110084605659827e-05,
|
| 334 |
+
"epoch": 2.333949236604104,
|
| 335 |
+
"total_flos": 6158765870970012,
|
| 336 |
+
"step": 24000
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"loss": 2.888296875,
|
| 340 |
+
"learning_rate": 1.0290447016110734e-05,
|
| 341 |
+
"epoch": 2.382573179033356,
|
| 342 |
+
"total_flos": 6289205284917468,
|
| 343 |
+
"step": 24500
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"loss": 2.82090625,
|
| 347 |
+
"learning_rate": 9.480047975623198e-06,
|
| 348 |
+
"epoch": 2.431197121462608,
|
| 349 |
+
"total_flos": 6419592132001884,
|
| 350 |
+
"step": 25000
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"loss": 2.832359375,
|
| 354 |
+
"learning_rate": 8.669648935135662e-06,
|
| 355 |
+
"epoch": 2.4798210638918605,
|
| 356 |
+
"total_flos": 6547245502208220,
|
| 357 |
+
"step": 25500
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"loss": 2.81575,
|
| 361 |
+
"learning_rate": 7.859249894648125e-06,
|
| 362 |
+
"epoch": 2.5284450063211126,
|
| 363 |
+
"total_flos": 6676738712620956,
|
| 364 |
+
"step": 26000
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"loss": 2.750359375,
|
| 368 |
+
"learning_rate": 7.048850854160589e-06,
|
| 369 |
+
"epoch": 2.5770689487503646,
|
| 370 |
+
"total_flos": 6806862725390172,
|
| 371 |
+
"step": 26500
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"loss": 2.80775,
|
| 375 |
+
"learning_rate": 6.2384518136730525e-06,
|
| 376 |
+
"epoch": 2.6256928911796167,
|
| 377 |
+
"total_flos": 6936345422430300,
|
| 378 |
+
"step": 27000
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"loss": 2.7769375,
|
| 382 |
+
"learning_rate": 5.4280527731855174e-06,
|
| 383 |
+
"epoch": 2.674316833608869,
|
| 384 |
+
"total_flos": 7062795011473020,
|
| 385 |
+
"step": 27500
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"loss": 2.812875,
|
| 389 |
+
"learning_rate": 4.617653732697981e-06,
|
| 390 |
+
"epoch": 2.7229407760381212,
|
| 391 |
+
"total_flos": 7193202885302652,
|
| 392 |
+
"step": 28000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"loss": 2.82128125,
|
| 396 |
+
"learning_rate": 3.807254692210445e-06,
|
| 397 |
+
"epoch": 2.7715647184673733,
|
| 398 |
+
"total_flos": 7325014294375452,
|
| 399 |
+
"step": 28500
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"loss": 2.7645625,
|
| 403 |
+
"learning_rate": 2.9968556517229085e-06,
|
| 404 |
+
"epoch": 2.8201886608966253,
|
| 405 |
+
"total_flos": 7450607043550620,
|
| 406 |
+
"step": 29000
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"loss": 2.867234375,
|
| 410 |
+
"learning_rate": 2.1864566112353722e-06,
|
| 411 |
+
"epoch": 2.8688126033258774,
|
| 412 |
+
"total_flos": 7577203819809852,
|
| 413 |
+
"step": 29500
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"loss": 2.772796875,
|
| 417 |
+
"learning_rate": 1.3760575707478364e-06,
|
| 418 |
+
"epoch": 2.91743654575513,
|
| 419 |
+
"total_flos": 7705477479000060,
|
| 420 |
+
"step": 30000
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"loss": 2.80721875,
|
| 424 |
+
"learning_rate": 5.656585302603003e-07,
|
| 425 |
+
"epoch": 2.966060488184382,
|
| 426 |
+
"total_flos": 7829551045833372,
|
| 427 |
+
"step": 30500
|
| 428 |
+
}
|
| 429 |
+
]
|