checkpoint-9376/trainer_state.json
| 1 | { |
| 2 | "best_metric": 1.0019118785858154, |
| 3 | "best_model_checkpoint": "fairface_age_image_detection/checkpoint-9376", |
| 4 | "epoch": 4.0, |
| 5 | "eval_steps": 500, |
| 6 | "global_step": 9376, |
| 7 | "is_hyper_param_search": false, |
| 8 | "is_local_process_zero": true, |
| 9 | "is_world_process_zero": true, |
| 10 | "log_history": [ |
| 11 | { |
| 12 | "epoch": 0.21331058020477817, |
| 13 | "grad_norm": 2.756800413131714, |
| 14 | "learning_rate": 5.710486811065838e-06, |
| 15 | "loss": 1.7206, |
| 16 | "step": 500 |
| 17 | }, |
| 18 | { |
| 19 | "epoch": 0.42662116040955633, |
| 20 | "grad_norm": 3.085395097732544, |
| 21 | "learning_rate": 5.3888054900278794e-06, |
| 22 | "loss": 1.3151, |
| 23 | "step": 1000 |
| 24 | }, |
| 25 | { |
| 26 | "epoch": 0.6399317406143344, |
| 27 | "grad_norm": 3.533961057662964, |
| 28 | "learning_rate": 5.067124168989921e-06, |
| 29 | "loss": 1.212, |
| 30 | "step": 1500 |
| 31 | }, |
| 32 | { |
| 33 | "epoch": 0.8532423208191127, |
| 34 | "grad_norm": 5.493643283843994, |
| 35 | "learning_rate": 4.745442847951962e-06, |
| 36 | "loss": 1.145, |
| 37 | "step": 2000 |
| 38 | }, |
| 39 | { |
| 40 | "epoch": 1.0, |
| 41 | "eval_accuracy": 0.5652, |
| 42 | "eval_loss": 1.115075707435608, |
| 43 | "eval_model_preparation_time": 0.0037, |
| 44 | "eval_runtime": 903.327, |
| 45 | "eval_samples_per_second": 11.07, |
| 46 | "eval_steps_per_second": 1.384, |
| 47 | "step": 2344 |
| 48 | }, |
| 49 | { |
| 50 | "epoch": 1.0665529010238908, |
| 51 | "grad_norm": 5.28899621963501, |
| 52 | "learning_rate": 4.423761526914004e-06, |
| 53 | "loss": 1.1006, |
| 54 | "step": 2500 |
| 55 | }, |
| 56 | { |
| 57 | "epoch": 1.2798634812286689, |
| 58 | "grad_norm": 6.52961540222168, |
| 59 | "learning_rate": 4.102080205876046e-06, |
| 60 | "loss": 1.0519, |
| 61 | "step": 3000 |
| 62 | }, |
| 63 | { |
| 64 | "epoch": 1.493174061433447, |
| 65 | "grad_norm": 10.652497291564941, |
| 66 | "learning_rate": 3.7803988848380875e-06, |
| 67 | "loss": 1.0292, |
| 68 | "step": 3500 |
| 69 | }, |
| 70 | { |
| 71 | "epoch": 1.7064846416382253, |
| 72 | "grad_norm": 5.245483875274658, |
| 73 | "learning_rate": 3.4587175638001284e-06, |
| 74 | "loss": 1.016, |
| 75 | "step": 4000 |
| 76 | }, |
| 77 | { |
| 78 | "epoch": 1.9197952218430034, |
| 79 | "grad_norm": 4.070522308349609, |
| 80 | "learning_rate": 3.1370362427621706e-06, |
| 81 | "loss": 1.0056, |
| 82 | "step": 4500 |
| 83 | }, |
| 84 | { |
| 85 | "epoch": 2.0, |
| 86 | "eval_accuracy": 0.5831, |
| 87 | "eval_loss": 1.030907392501831, |
| 88 | "eval_model_preparation_time": 0.0037, |
| 89 | "eval_runtime": 917.4025, |
| 90 | "eval_samples_per_second": 10.9, |
| 91 | "eval_steps_per_second": 1.363, |
| 92 | "step": 4688 |
| 93 | }, |
| 94 | { |
| 95 | "epoch": 2.1331058020477816, |
| 96 | "grad_norm": 6.568565845489502, |
| 97 | "learning_rate": 2.815354921724212e-06, |
| 98 | "loss": 0.9796, |
| 99 | "step": 5000 |
| 100 | }, |
| 101 | { |
| 102 | "epoch": 2.34641638225256, |
| 103 | "grad_norm": 10.742039680480957, |
| 104 | "learning_rate": 2.4936736006862537e-06, |
| 105 | "loss": 0.9587, |
| 106 | "step": 5500 |
| 107 | }, |
| 108 | { |
| 109 | "epoch": 2.5597269624573378, |
| 110 | "grad_norm": 4.3592424392700195, |
| 111 | "learning_rate": 2.171992279648295e-06, |
| 112 | "loss": 0.9455, |
| 113 | "step": 6000 |
| 114 | }, |
| 115 | { |
| 116 | "epoch": 2.773037542662116, |
| 117 | "grad_norm": 5.696556091308594, |
| 118 | "learning_rate": 1.8503109586103366e-06, |
| 119 | "loss": 0.9363, |
| 120 | "step": 6500 |
| 121 | }, |
| 122 | { |
| 123 | "epoch": 2.986348122866894, |
| 124 | "grad_norm": 5.746116638183594, |
| 125 | "learning_rate": 1.5286296375723786e-06, |
| 126 | "loss": 0.9347, |
| 127 | "step": 7000 |
| 128 | }, |
| 129 | { |
| 130 | "epoch": 3.0, |
| 131 | "eval_accuracy": 0.5883, |
| 132 | "eval_loss": 1.0057746171951294, |
| 133 | "eval_model_preparation_time": 0.0037, |
| 134 | "eval_runtime": 926.1536, |
| 135 | "eval_samples_per_second": 10.797, |
| 136 | "eval_steps_per_second": 1.35, |
| 137 | "step": 7032 |
| 138 | }, |
| 139 | { |
| 140 | "epoch": 3.1996587030716723, |
| 141 | "grad_norm": 4.218583106994629, |
| 142 | "learning_rate": 1.20694831653442e-06, |
| 143 | "loss": 0.9035, |
| 144 | "step": 7500 |
| 145 | }, |
| 146 | { |
| 147 | "epoch": 3.4129692832764507, |
| 148 | "grad_norm": 6.724698543548584, |
| 149 | "learning_rate": 8.852669954964614e-07, |
| 150 | "loss": 0.9081, |
| 151 | "step": 8000 |
| 152 | }, |
| 153 | { |
| 154 | "epoch": 3.6262798634812285, |
| 155 | "grad_norm": 6.658480167388916, |
| 156 | "learning_rate": 5.635856744585031e-07, |
| 157 | "loss": 0.905, |
| 158 | "step": 8500 |
| 159 | }, |
| 160 | { |
| 161 | "epoch": 3.839590443686007, |
| 162 | "grad_norm": 4.382744789123535, |
| 163 | "learning_rate": 2.4190435342054474e-07, |
| 164 | "loss": 0.8969, |
| 165 | "step": 9000 |
| 166 | }, |
| 167 | { |
| 168 | "epoch": 4.0, |
| 169 | "eval_accuracy": 0.5892, |
| 170 | "eval_loss": 1.0019118785858154, |
| 171 | "eval_model_preparation_time": 0.0037, |
| 172 | "eval_runtime": 929.5968, |
| 173 | "eval_samples_per_second": 10.757, |
| 174 | "eval_steps_per_second": 1.345, |
| 175 | "step": 9376 |
| 176 | } |
| 177 | ], |
| 178 | "logging_steps": 500, |
| 179 | "max_steps": 9376, |
| 180 | "num_input_tokens_seen": 0, |
| 181 | "num_train_epochs": 4, |
| 182 | "save_steps": 500, |
| 183 | "stateful_callbacks": { |
| 184 | "TrainerControl": { |
| 185 | "args": { |
| 186 | "should_epoch_stop": false, |
| 187 | "should_evaluate": false, |
| 188 | "should_log": false, |
| 189 | "should_save": true, |
| 190 | "should_training_stop": true |
| 191 | }, |
| 192 | "attributes": {} |
| 193 | } |
| 194 | }, |
| 195 | "total_flos": 2.32490554103808e+19, |
| 196 | "train_batch_size": 32, |
| 197 | "trial_name": null, |
| 198 | "trial_params": null |
| 199 | } |
| 200 | |