thangcd/thread-intelligence-classifier
016
1{2 "best_global_step": 678,3 "best_metric": 0.9908637873754153,4 "best_model_checkpoint": "./classifier_model_refined/checkpoint-678",5 "epoch": 3.0,6 "eval_steps": 500,7 "global_step": 1017,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.14749262536873156,14 "grad_norm": 2.5459585189819336,15 "learning_rate": 1.903638151425762e-06,16 "loss": 0.09093725204467773,17 "step": 5018 },19 {20 "epoch": 0.2949852507374631,21 "grad_norm": 0.011955893598496914,22 "learning_rate": 1.8053097345132743e-06,23 "loss": 0.10499533653259277,24 "step": 10025 },26 {27 "epoch": 0.4424778761061947,28 "grad_norm": 0.4081194996833801,29 "learning_rate": 1.7069813176007865e-06,30 "loss": 0.09049854278564454,31 "step": 15032 },33 {34 "epoch": 0.5899705014749262,35 "grad_norm": 7.041113376617432,36 "learning_rate": 1.608652900688299e-06,37 "loss": 0.07993946075439454,38 "step": 20039 },40 {41 "epoch": 0.7374631268436578,42 "grad_norm": 43.293846130371094,43 "learning_rate": 1.5103244837758112e-06,44 "loss": 0.05462934970855713,45 "step": 25046 },47 {48 "epoch": 0.8849557522123894,49 "grad_norm": 0.4207114577293396,50 "learning_rate": 1.4119960668633234e-06,51 "loss": 0.059678683280944826,52 "step": 30053 },54 {55 "epoch": 1.0,56 "eval_accuracy": 0.9900332225913622,57 "eval_f1_macro": 0.9923847199375366,58 "eval_loss": 0.08367534726858139,59 "eval_runtime": 2.9099,60 "eval_samples_per_second": 413.754,61 "eval_steps_per_second": 13.059,62 "step": 33963 },64 {65 "epoch": 1.0324483775811208,66 "grad_norm": 5.123465538024902,67 "learning_rate": 1.3136676499508356e-06,68 "loss": 0.05710693359375,69 "step": 35070 },71 {72 "epoch": 1.1799410029498525,73 "grad_norm": 7.183027267456055,74 "learning_rate": 1.215339233038348e-06,75 "loss": 0.06182104110717773,76 "step": 40077 },78 {79 "epoch": 1.3274336283185841,80 "grad_norm": 0.10848981887102127,81 "learning_rate": 1.1170108161258603e-06,82 "loss": 0.06060198307037354,83 "step": 45084 },85 {86 "epoch": 1.4749262536873156,87 "grad_norm": 0.09567607939243317,88 "learning_rate": 1.0186823992133726e-06,89 "loss": 0.06725824356079102,90 "step": 50091 },92 {93 "epoch": 1.6224188790560472,94 "grad_norm": 4.66201639175415,95 "learning_rate": 9.203539823008849e-07,96 "loss": 0.07334633827209473,97 "step": 55098 },99 {100 "epoch": 1.7699115044247788,101 "grad_norm": 0.17738792300224304,102 "learning_rate": 8.220255653883972e-07,103 "loss": 0.05149734020233154,104 "step": 600105 },106 {107 "epoch": 1.9174041297935103,108 "grad_norm": 0.7206429839134216,109 "learning_rate": 7.236971484759095e-07,110 "loss": 0.0479090166091919,111 "step": 650112 },113 {114 "epoch": 2.0,115 "eval_accuracy": 0.9908637873754153,116 "eval_f1_macro": 0.9934718048955895,117 "eval_loss": 0.07654637098312378,118 "eval_runtime": 2.9002,119 "eval_samples_per_second": 415.144,120 "eval_steps_per_second": 13.103,121 "step": 678122 },123 {124 "epoch": 2.0648967551622417,125 "grad_norm": 0.30022454261779785,126 "learning_rate": 6.253687315634218e-07,127 "loss": 0.06578935623168945,128 "step": 700129 },130 {131 "epoch": 2.2123893805309733,132 "grad_norm": 0.03679751232266426,133 "learning_rate": 5.270403146509341e-07,134 "loss": 0.033387229442596436,135 "step": 750136 },137 {138 "epoch": 2.359882005899705,139 "grad_norm": 0.12588904798030853,140 "learning_rate": 4.287118977384464e-07,141 "loss": 0.0704490613937378,142 "step": 800143 },144 {145 "epoch": 2.5073746312684366,146 "grad_norm": 0.21027934551239014,147 "learning_rate": 3.3038348082595866e-07,148 "loss": 0.04873675346374512,149 "step": 850150 },151 {152 "epoch": 2.6548672566371683,153 "grad_norm": 41.093448638916016,154 "learning_rate": 2.32055063913471e-07,155 "loss": 0.056772613525390626,156 "step": 900157 },158 {159 "epoch": 2.8023598820058995,160 "grad_norm": 28.944047927856445,161 "learning_rate": 1.3372664700098328e-07,162 "loss": 0.03798616647720337,163 "step": 950164 },165 {166 "epoch": 2.949852507374631,167 "grad_norm": 0.2888835668563843,168 "learning_rate": 3.539823008849557e-08,169 "loss": 0.044747047424316407,170 "step": 1000171 },172 {173 "epoch": 3.0,174 "eval_accuracy": 0.9908637873754153,175 "eval_f1_macro": 0.9934718048955895,176 "eval_loss": 0.0788116306066513,177 "eval_runtime": 2.9073,178 "eval_samples_per_second": 414.135,179 "eval_steps_per_second": 13.071,180 "step": 1017181 }182 ],183 "logging_steps": 50,184 "max_steps": 1017,185 "num_input_tokens_seen": 0,186 "num_train_epochs": 3,187 "save_steps": 500,188 "stateful_callbacks": {189 "TrainerControl": {190 "args": {191 "should_epoch_stop": false,192 "should_evaluate": false,193 "should_log": false,194 "should_save": true,195 "should_training_stop": true196 },197 "attributes": {}198 }199 },200 "total_flos": 1076376257335296.0,201 "train_batch_size": 32,202 "trial_name": null,203 "trial_params": null204}205 