Final training metrics
Browse files- training_metrics.json +414 -0
training_metrics.json
ADDED
|
@@ -0,0 +1,414 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metrics_history": [
|
| 3 |
+
{
|
| 4 |
+
"epoch": 0.05819366852886406,
|
| 5 |
+
"global_step": 500,
|
| 6 |
+
"eval_loss": 4.736684322357178,
|
| 7 |
+
"eval_cosine_accuracy": 0.9417916536331177
|
| 8 |
+
},
|
| 9 |
+
{
|
| 10 |
+
"epoch": 0.11638733705772812,
|
| 11 |
+
"global_step": 1000,
|
| 12 |
+
"eval_loss": 4.456507682800293,
|
| 13 |
+
"eval_cosine_accuracy": 0.9421085715293884
|
| 14 |
+
},
|
| 15 |
+
{
|
| 16 |
+
"epoch": 0.17458100558659218,
|
| 17 |
+
"global_step": 1500,
|
| 18 |
+
"eval_loss": 4.417118072509766,
|
| 19 |
+
"eval_cosine_accuracy": 0.9389393329620361
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 0.23277467411545624,
|
| 23 |
+
"global_step": 2000,
|
| 24 |
+
"eval_loss": 4.384617805480957,
|
| 25 |
+
"eval_cosine_accuracy": 0.9376716613769531
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.2909683426443203,
|
| 29 |
+
"global_step": 2500,
|
| 30 |
+
"eval_loss": 4.382342338562012,
|
| 31 |
+
"eval_cosine_accuracy": 0.929537296295166
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.34916201117318435,
|
| 35 |
+
"global_step": 3000,
|
| 36 |
+
"eval_loss": 4.398530006408691,
|
| 37 |
+
"eval_cosine_accuracy": 0.9236213564872742
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.4073556797020484,
|
| 41 |
+
"global_step": 3500,
|
| 42 |
+
"eval_loss": 4.428633689880371,
|
| 43 |
+
"eval_cosine_accuracy": 0.9207690954208374
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 0.4655493482309125,
|
| 47 |
+
"global_step": 4000,
|
| 48 |
+
"eval_loss": 4.413504600524902,
|
| 49 |
+
"eval_cosine_accuracy": 0.9240439534187317
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.5237430167597765,
|
| 53 |
+
"global_step": 4500,
|
| 54 |
+
"eval_loss": 4.584815979003906,
|
| 55 |
+
"eval_cosine_accuracy": 0.9183393120765686
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.5819366852886406,
|
| 59 |
+
"global_step": 5000,
|
| 60 |
+
"eval_loss": 4.731673240661621,
|
| 61 |
+
"eval_cosine_accuracy": 0.9231988191604614
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.6401303538175046,
|
| 65 |
+
"global_step": 5500,
|
| 66 |
+
"eval_loss": 4.590743064880371,
|
| 67 |
+
"eval_cosine_accuracy": 0.9202408790588379
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.6983240223463687,
|
| 71 |
+
"global_step": 6000,
|
| 72 |
+
"eval_loss": 4.662985324859619,
|
| 73 |
+
"eval_cosine_accuracy": 0.9188675284385681
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.7565176908752328,
|
| 77 |
+
"global_step": 6500,
|
| 78 |
+
"eval_loss": 4.784646511077881,
|
| 79 |
+
"eval_cosine_accuracy": 0.9199239611625671
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.8147113594040968,
|
| 83 |
+
"global_step": 7000,
|
| 84 |
+
"eval_loss": 4.925645351409912,
|
| 85 |
+
"eval_cosine_accuracy": 0.9054510593414307
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 0.8729050279329609,
|
| 89 |
+
"global_step": 7500,
|
| 90 |
+
"eval_loss": 4.843690395355225,
|
| 91 |
+
"eval_cosine_accuracy": 0.9056623578071594
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 0.931098696461825,
|
| 95 |
+
"global_step": 8000,
|
| 96 |
+
"eval_loss": 4.833749771118164,
|
| 97 |
+
"eval_cosine_accuracy": 0.907247006893158
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.9892923649906891,
|
| 101 |
+
"global_step": 8500,
|
| 102 |
+
"eval_loss": 4.809013366699219,
|
| 103 |
+
"eval_cosine_accuracy": 0.9006972312927246
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 1.047486033519553,
|
| 107 |
+
"global_step": 9000,
|
| 108 |
+
"eval_loss": 4.823788166046143,
|
| 109 |
+
"eval_cosine_accuracy": 0.8939362168312073
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.105679702048417,
|
| 113 |
+
"global_step": 9500,
|
| 114 |
+
"eval_loss": 4.753988265991211,
|
| 115 |
+
"eval_cosine_accuracy": 0.9032326340675354
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 1.1638733705772812,
|
| 119 |
+
"global_step": 10000,
|
| 120 |
+
"eval_loss": 4.775505065917969,
|
| 121 |
+
"eval_cosine_accuracy": 0.9001690149307251
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.2220670391061452,
|
| 125 |
+
"global_step": 10500,
|
| 126 |
+
"eval_loss": 4.801342010498047,
|
| 127 |
+
"eval_cosine_accuracy": 0.895415186882019
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 1.2802607076350094,
|
| 131 |
+
"global_step": 11000,
|
| 132 |
+
"eval_loss": 4.581014156341553,
|
| 133 |
+
"eval_cosine_accuracy": 0.9016479849815369
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 1.3384543761638734,
|
| 137 |
+
"global_step": 11500,
|
| 138 |
+
"eval_loss": 4.6524434089660645,
|
| 139 |
+
"eval_cosine_accuracy": 0.8949925899505615
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 1.3966480446927374,
|
| 143 |
+
"global_step": 12000,
|
| 144 |
+
"eval_loss": 4.835806369781494,
|
| 145 |
+
"eval_cosine_accuracy": 0.8828438520431519
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 1.4548417132216014,
|
| 149 |
+
"global_step": 12500,
|
| 150 |
+
"eval_loss": 4.768213272094727,
|
| 151 |
+
"eval_cosine_accuracy": 0.8947813510894775
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.5130353817504656,
|
| 155 |
+
"global_step": 13000,
|
| 156 |
+
"eval_loss": 4.828193664550781,
|
| 157 |
+
"eval_cosine_accuracy": 0.8974223732948303
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 1.5712290502793296,
|
| 161 |
+
"global_step": 13500,
|
| 162 |
+
"eval_loss": 4.6300153732299805,
|
| 163 |
+
"eval_cosine_accuracy": 0.8914008140563965
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 1.6294227188081938,
|
| 167 |
+
"global_step": 14000,
|
| 168 |
+
"eval_loss": 4.730409145355225,
|
| 169 |
+
"eval_cosine_accuracy": 0.892034649848938
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 1.6876163873370578,
|
| 173 |
+
"global_step": 14500,
|
| 174 |
+
"eval_loss": 4.478672981262207,
|
| 175 |
+
"eval_cosine_accuracy": 0.9077752232551575
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 1.7458100558659218,
|
| 179 |
+
"global_step": 15000,
|
| 180 |
+
"eval_loss": 4.705593109130859,
|
| 181 |
+
"eval_cosine_accuracy": 0.8841115832328796
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 1.8040037243947857,
|
| 185 |
+
"global_step": 15500,
|
| 186 |
+
"eval_loss": 4.561342716217041,
|
| 187 |
+
"eval_cosine_accuracy": 0.8960490226745605
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 1.86219739292365,
|
| 191 |
+
"global_step": 16000,
|
| 192 |
+
"eval_loss": 4.534769535064697,
|
| 193 |
+
"eval_cosine_accuracy": 0.9019649028778076
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.920391061452514,
|
| 197 |
+
"global_step": 16500,
|
| 198 |
+
"eval_loss": 4.725633144378662,
|
| 199 |
+
"eval_cosine_accuracy": 0.913057267665863
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 1.9785847299813781,
|
| 203 |
+
"global_step": 17000,
|
| 204 |
+
"eval_loss": 4.73307991027832,
|
| 205 |
+
"eval_cosine_accuracy": 0.8949925899505615
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 2.036778398510242,
|
| 209 |
+
"global_step": 17500,
|
| 210 |
+
"eval_loss": 4.846031188964844,
|
| 211 |
+
"eval_cosine_accuracy": 0.8875977396965027
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 2.094972067039106,
|
| 215 |
+
"global_step": 18000,
|
| 216 |
+
"eval_loss": 4.649144649505615,
|
| 217 |
+
"eval_cosine_accuracy": 0.8969997763633728
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 2.15316573556797,
|
| 221 |
+
"global_step": 18500,
|
| 222 |
+
"eval_loss": 4.538448810577393,
|
| 223 |
+
"eval_cosine_accuracy": 0.8948869705200195
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 2.211359404096834,
|
| 227 |
+
"global_step": 19000,
|
| 228 |
+
"eval_loss": 4.58160400390625,
|
| 229 |
+
"eval_cosine_accuracy": 0.9011197686195374
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 2.2695530726256985,
|
| 233 |
+
"global_step": 19500,
|
| 234 |
+
"eval_loss": 4.580519199371338,
|
| 235 |
+
"eval_cosine_accuracy": 0.9015423655509949
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 2.3277467411545625,
|
| 239 |
+
"global_step": 20000,
|
| 240 |
+
"eval_loss": 4.5440673828125,
|
| 241 |
+
"eval_cosine_accuracy": 0.8975279927253723
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 2.3859404096834265,
|
| 245 |
+
"global_step": 20500,
|
| 246 |
+
"eval_loss": 4.727934837341309,
|
| 247 |
+
"eval_cosine_accuracy": 0.8922459483146667
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 2.4441340782122905,
|
| 251 |
+
"global_step": 21000,
|
| 252 |
+
"eval_loss": 4.589358806610107,
|
| 253 |
+
"eval_cosine_accuracy": 0.9073526263237
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 2.5023277467411544,
|
| 257 |
+
"global_step": 21500,
|
| 258 |
+
"eval_loss": 4.702400207519531,
|
| 259 |
+
"eval_cosine_accuracy": 0.8977392911911011
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 2.560521415270019,
|
| 263 |
+
"global_step": 22000,
|
| 264 |
+
"eval_loss": 4.601651668548584,
|
| 265 |
+
"eval_cosine_accuracy": 0.9057680368423462
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.618715083798883,
|
| 269 |
+
"global_step": 22500,
|
| 270 |
+
"eval_loss": 4.483080863952637,
|
| 271 |
+
"eval_cosine_accuracy": 0.9074582457542419
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 2.676908752327747,
|
| 275 |
+
"global_step": 23000,
|
| 276 |
+
"eval_loss": 4.641151428222656,
|
| 277 |
+
"eval_cosine_accuracy": 0.8939362168312073
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.735102420856611,
|
| 281 |
+
"global_step": 23500,
|
| 282 |
+
"eval_loss": 4.60220193862915,
|
| 283 |
+
"eval_cosine_accuracy": 0.8949925899505615
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 2.793296089385475,
|
| 287 |
+
"global_step": 24000,
|
| 288 |
+
"eval_loss": 4.529559135437012,
|
| 289 |
+
"eval_cosine_accuracy": 0.8974223732948303
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 2.851489757914339,
|
| 293 |
+
"global_step": 24500,
|
| 294 |
+
"eval_loss": 4.752791881561279,
|
| 295 |
+
"eval_cosine_accuracy": 0.8923515677452087
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 2.9096834264432028,
|
| 299 |
+
"global_step": 25000,
|
| 300 |
+
"eval_loss": 4.698167324066162,
|
| 301 |
+
"eval_cosine_accuracy": 0.8986900448799133
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 2.967877094972067,
|
| 305 |
+
"global_step": 25500,
|
| 306 |
+
"eval_loss": 4.680306434631348,
|
| 307 |
+
"eval_cosine_accuracy": 0.8960490226745605
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 3.026070763500931,
|
| 311 |
+
"global_step": 26000,
|
| 312 |
+
"eval_loss": 4.683339595794678,
|
| 313 |
+
"eval_cosine_accuracy": 0.8993238806724548
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 3.084264432029795,
|
| 317 |
+
"global_step": 26500,
|
| 318 |
+
"eval_loss": 4.631614685058594,
|
| 319 |
+
"eval_cosine_accuracy": 0.896682858467102
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 3.142458100558659,
|
| 323 |
+
"global_step": 27000,
|
| 324 |
+
"eval_loss": 4.696747779846191,
|
| 325 |
+
"eval_cosine_accuracy": 0.8975279927253723
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 3.200651769087523,
|
| 329 |
+
"global_step": 27500,
|
| 330 |
+
"eval_loss": 4.715923309326172,
|
| 331 |
+
"eval_cosine_accuracy": 0.896682858467102
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 3.2588454376163876,
|
| 335 |
+
"global_step": 28000,
|
| 336 |
+
"eval_loss": 4.600804328918457,
|
| 337 |
+
"eval_cosine_accuracy": 0.9014367461204529
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 3.3170391061452515,
|
| 341 |
+
"global_step": 28500,
|
| 342 |
+
"eval_loss": 4.584753036499023,
|
| 343 |
+
"eval_cosine_accuracy": 0.9018592834472656
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 3.3752327746741155,
|
| 347 |
+
"global_step": 29000,
|
| 348 |
+
"eval_loss": 4.553245544433594,
|
| 349 |
+
"eval_cosine_accuracy": 0.9040777683258057
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 3.4334264432029795,
|
| 353 |
+
"global_step": 29500,
|
| 354 |
+
"eval_loss": 4.594734191894531,
|
| 355 |
+
"eval_cosine_accuracy": 0.9052398204803467
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 3.4916201117318435,
|
| 359 |
+
"global_step": 30000,
|
| 360 |
+
"eval_loss": 4.592144012451172,
|
| 361 |
+
"eval_cosine_accuracy": 0.9071413278579712
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 3.5498137802607075,
|
| 365 |
+
"global_step": 30500,
|
| 366 |
+
"eval_loss": 4.61123514175415,
|
| 367 |
+
"eval_cosine_accuracy": 0.9049229025840759
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 3.6080074487895715,
|
| 371 |
+
"global_step": 31000,
|
| 372 |
+
"eval_loss": 4.575016498565674,
|
| 373 |
+
"eval_cosine_accuracy": 0.9083033800125122
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 3.666201117318436,
|
| 377 |
+
"global_step": 31500,
|
| 378 |
+
"eval_loss": 4.594618320465088,
|
| 379 |
+
"eval_cosine_accuracy": 0.9043946862220764
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 3.7243947858473,
|
| 383 |
+
"global_step": 32000,
|
| 384 |
+
"eval_loss": 4.5688347816467285,
|
| 385 |
+
"eval_cosine_accuracy": 0.9060849547386169
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 3.782588454376164,
|
| 389 |
+
"global_step": 32500,
|
| 390 |
+
"eval_loss": 4.608202934265137,
|
| 391 |
+
"eval_cosine_accuracy": 0.9062961935997009
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 3.840782122905028,
|
| 395 |
+
"global_step": 33000,
|
| 396 |
+
"eval_loss": 4.594712734222412,
|
| 397 |
+
"eval_cosine_accuracy": 0.9059792757034302
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 3.898975791433892,
|
| 401 |
+
"global_step": 33500,
|
| 402 |
+
"eval_loss": 4.600465297698975,
|
| 403 |
+
"eval_cosine_accuracy": 0.9069300889968872
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 3.9571694599627563,
|
| 407 |
+
"global_step": 34000,
|
| 408 |
+
"eval_loss": 4.5989670753479,
|
| 409 |
+
"eval_cosine_accuracy": 0.9064018726348877
|
| 410 |
+
}
|
| 411 |
+
],
|
| 412 |
+
"total_epochs": 4.0,
|
| 413 |
+
"total_steps": 34368
|
| 414 |
+
}
|