rakhman-llm/Text2Code-Java-CodeT5
014
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 3.0,5 "eval_steps": 1000,6 "global_step": 15462,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.09701202949165696,13 "grad_norm": 0.9203131794929504,14 "learning_rate": 1.9353253136722286e-05,15 "loss": 1.7024,16 "step": 50017 },18 {19 "epoch": 0.19402405898331393,20 "grad_norm": 1.0445950031280518,21 "learning_rate": 1.8706506273444574e-05,22 "loss": 1.3896,23 "step": 100024 },25 {26 "epoch": 0.19402405898331393,27 "eval_loss": 1.1814582347869873,28 "eval_runtime": 49.6691,29 "eval_samples_per_second": 104.351,30 "eval_steps_per_second": 3.262,31 "step": 100032 },33 {34 "epoch": 0.2910360884749709,35 "grad_norm": 0.8253631591796875,36 "learning_rate": 1.8059759410166862e-05,37 "loss": 1.3302,38 "step": 150039 },40 {41 "epoch": 0.38804811796662786,42 "grad_norm": 0.9224209189414978,43 "learning_rate": 1.741301254688915e-05,44 "loss": 1.2914,45 "step": 200046 },47 {48 "epoch": 0.38804811796662786,49 "eval_loss": 1.1394085884094238,50 "eval_runtime": 49.6277,51 "eval_samples_per_second": 104.438,52 "eval_steps_per_second": 3.264,53 "step": 200054 },55 {56 "epoch": 0.48506014745828485,57 "grad_norm": 0.7809765934944153,58 "learning_rate": 1.6766265683611435e-05,59 "loss": 1.2695,60 "step": 250061 },62 {63 "epoch": 0.5820721769499418,64 "grad_norm": 0.6769764423370361,65 "learning_rate": 1.6119518820333723e-05,66 "loss": 1.2523,67 "step": 300068 },69 {70 "epoch": 0.5820721769499418,71 "eval_loss": 1.1180071830749512,72 "eval_runtime": 49.6817,73 "eval_samples_per_second": 104.324,74 "eval_steps_per_second": 3.261,75 "step": 300076 },77 {78 "epoch": 0.6790842064415987,79 "grad_norm": 0.8303701877593994,80 "learning_rate": 1.547277195705601e-05,81 "loss": 1.233,82 "step": 350083 },84 {85 "epoch": 0.7760962359332557,86 "grad_norm": 0.8391227722167969,87 "learning_rate": 1.4826025093778296e-05,88 "loss": 1.2294,89 "step": 400090 },91 {92 "epoch": 0.7760962359332557,93 "eval_loss": 1.1015874147415161,94 "eval_runtime": 49.6378,95 "eval_samples_per_second": 104.416,96 "eval_steps_per_second": 3.264,97 "step": 400098 },99 {100 "epoch": 0.8731082654249127,101 "grad_norm": 0.7971065640449524,102 "learning_rate": 1.4179278230500584e-05,103 "loss": 1.2111,104 "step": 4500105 },106 {107 "epoch": 0.9701202949165697,108 "grad_norm": 2.1875197887420654,109 "learning_rate": 1.353253136722287e-05,110 "loss": 1.2006,111 "step": 5000112 },113 {114 "epoch": 0.9701202949165697,115 "eval_loss": 1.091620922088623,116 "eval_runtime": 49.6213,117 "eval_samples_per_second": 104.451,118 "eval_steps_per_second": 3.265,119 "step": 5000120 },121 {122 "epoch": 1.0671323244082267,123 "grad_norm": 0.729252278804779,124 "learning_rate": 1.2885784503945158e-05,125 "loss": 1.1861,126 "step": 5500127 },128 {129 "epoch": 1.1641443538998835,130 "grad_norm": 0.6466144323348999,131 "learning_rate": 1.2239037640667444e-05,132 "loss": 1.1714,133 "step": 6000134 },135 {136 "epoch": 1.1641443538998835,137 "eval_loss": 1.0831941366195679,138 "eval_runtime": 49.6774,139 "eval_samples_per_second": 104.333,140 "eval_steps_per_second": 3.261,141 "step": 6000142 },143 {144 "epoch": 1.2611563833915405,145 "grad_norm": 0.7486400008201599,146 "learning_rate": 1.1592290777389729e-05,147 "loss": 1.1696,148 "step": 6500149 },150 {151 "epoch": 1.3581684128831975,152 "grad_norm": 0.7059735059738159,153 "learning_rate": 1.0945543914112017e-05,154 "loss": 1.1566,155 "step": 7000156 },157 {158 "epoch": 1.3581684128831975,159 "eval_loss": 1.0764732360839844,160 "eval_runtime": 49.6981,161 "eval_samples_per_second": 104.29,162 "eval_steps_per_second": 3.26,163 "step": 7000164 },165 {166 "epoch": 1.4551804423748544,167 "grad_norm": 0.7840662002563477,168 "learning_rate": 1.0298797050834303e-05,169 "loss": 1.154,170 "step": 7500171 },172 {173 "epoch": 1.5521924718665114,174 "grad_norm": 1.1568676233291626,175 "learning_rate": 9.652050187556591e-06,176 "loss": 1.1596,177 "step": 8000178 },179 {180 "epoch": 1.5521924718665114,181 "eval_loss": 1.07025945186615,182 "eval_runtime": 49.6881,183 "eval_samples_per_second": 104.311,184 "eval_steps_per_second": 3.26,185 "step": 8000186 },187 {188 "epoch": 1.6492045013581684,189 "grad_norm": 0.7761677503585815,190 "learning_rate": 9.005303324278878e-06,191 "loss": 1.1541,192 "step": 8500193 },194 {195 "epoch": 1.7462165308498254,196 "grad_norm": 0.8179004788398743,197 "learning_rate": 8.358556461001164e-06,198 "loss": 1.1483,199 "step": 9000200 },201 {202 "epoch": 1.7462165308498254,203 "eval_loss": 1.0663487911224365,204 "eval_runtime": 49.6731,205 "eval_samples_per_second": 104.342,206 "eval_steps_per_second": 3.261,207 "step": 9000208 },209 {210 "epoch": 1.8432285603414824,211 "grad_norm": 0.7527589201927185,212 "learning_rate": 7.711809597723452e-06,213 "loss": 1.1429,214 "step": 9500215 },216 {217 "epoch": 1.9402405898331394,218 "grad_norm": 0.8005126714706421,219 "learning_rate": 7.065062734445739e-06,220 "loss": 1.1281,221 "step": 10000222 },223 {224 "epoch": 1.9402405898331394,225 "eval_loss": 1.0637460947036743,226 "eval_runtime": 49.672,227 "eval_samples_per_second": 104.344,228 "eval_steps_per_second": 3.261,229 "step": 10000230 },231 {232 "epoch": 2.0372526193247964,233 "grad_norm": 0.6758320927619934,234 "learning_rate": 6.418315871168025e-06,235 "loss": 1.1251,236 "step": 10500237 },238 {239 "epoch": 2.1342646488164534,240 "grad_norm": 0.7758616805076599,241 "learning_rate": 5.771569007890312e-06,242 "loss": 1.1252,243 "step": 11000244 },245 {246 "epoch": 2.1342646488164534,247 "eval_loss": 1.0619136095046997,248 "eval_runtime": 49.6813,249 "eval_samples_per_second": 104.325,250 "eval_steps_per_second": 3.261,251 "step": 11000252 },253 {254 "epoch": 2.2312766783081104,255 "grad_norm": 0.8191428780555725,256 "learning_rate": 5.124822144612599e-06,257 "loss": 1.1191,258 "step": 11500259 },260 {261 "epoch": 2.328288707799767,262 "grad_norm": 0.7311928272247314,263 "learning_rate": 4.478075281334886e-06,264 "loss": 1.1187,265 "step": 12000266 },267 {268 "epoch": 2.328288707799767,269 "eval_loss": 1.0580228567123413,270 "eval_runtime": 49.6253,271 "eval_samples_per_second": 104.443,272 "eval_steps_per_second": 3.264,273 "step": 12000274 },275 {276 "epoch": 2.4253007372914244,277 "grad_norm": 0.6671691536903381,278 "learning_rate": 3.831328418057173e-06,279 "loss": 1.115,280 "step": 12500281 },282 {283 "epoch": 2.522312766783081,284 "grad_norm": 0.7181034088134766,285 "learning_rate": 3.1845815547794594e-06,286 "loss": 1.1025,287 "step": 13000288 },289 {290 "epoch": 2.522312766783081,291 "eval_loss": 1.0561449527740479,292 "eval_runtime": 49.663,293 "eval_samples_per_second": 104.363,294 "eval_steps_per_second": 3.262,295 "step": 13000296 },297 {298 "epoch": 2.619324796274738,299 "grad_norm": 1.3909169435501099,300 "learning_rate": 2.5378346915017466e-06,301 "loss": 1.1197,302 "step": 13500303 },304 {305 "epoch": 2.716336825766395,306 "grad_norm": 0.7657527327537537,307 "learning_rate": 1.8910878282240333e-06,308 "loss": 1.1113,309 "step": 14000310 },311 {312 "epoch": 2.716336825766395,313 "eval_loss": 1.0557653903961182,314 "eval_runtime": 49.6516,315 "eval_samples_per_second": 104.387,316 "eval_steps_per_second": 3.263,317 "step": 14000318 },319 {320 "epoch": 2.813348855258052,321 "grad_norm": 0.6979902386665344,322 "learning_rate": 1.24434096494632e-06,323 "loss": 1.1123,324 "step": 14500325 },326 {327 "epoch": 2.910360884749709,328 "grad_norm": 0.7375929355621338,329 "learning_rate": 5.975941016686069e-07,330 "loss": 1.1155,331 "step": 15000332 },333 {334 "epoch": 2.910360884749709,335 "eval_loss": 1.054747223854065,336 "eval_runtime": 49.5431,337 "eval_samples_per_second": 104.616,338 "eval_steps_per_second": 3.27,339 "step": 15000340 }341 ],342 "logging_steps": 500,343 "max_steps": 15462,344 "num_input_tokens_seen": 0,345 "num_train_epochs": 3,346 "save_steps": 50000,347 "stateful_callbacks": {348 "TrainerControl": {349 "args": {350 "should_epoch_stop": false,351 "should_evaluate": false,352 "should_log": false,353 "should_save": true,354 "should_training_stop": true355 },356 "attributes": {}357 }358 },359 "total_flos": 7.532337163862016e+16,360 "train_batch_size": 32,361 "trial_name": null,362 "trial_params": null363}364 